processes

Review
A Review of Convex Approaches for Control,
Observation and Safety of Linear Parameter Varying
and Takagi-Sugeno Systems
Francisco-Ronay López-Estrada 1, *,† , Damiano Rotondo 2,3, *,† and
Guillermo Valencia-Palomo 4, *,†
1 TURIX-Dynamics Diagnosis and Control Group, Tecnológico Nacional de México/Instituto Tecnológico de
Tuxtla Gutiérrez, Carretera Panamericana km 1080, C.P. 29050 Tuxtla Gutierrez, Mexico
2 Institut de Robòtica i Informàtica Industrial, CSIC-UPC, Llorens i Artigas 4-6, 08028 Barcelona, Spain
3 Research Center for Supervision, Safety and Automatic Control, Universitat Politècnica de Catalunya (UPC),
Rambla Sant Nebridi, 22, 08022 Terrassa, Spain
4 Tecnológico Nacional de México/Instituto Tecnológico de Hermosillo, Av. Tecnológico y Periférico Poniente,
S/N, 83170 Hermosillo, Sonora, Mexico
* Correspondence: [email protected] (F.-R.L.-E.); [email protected] (D.R.);
[email protected] (G.V.-P.)
† These authors contributed equally to this work.




Received: 25 September 2019; Accepted: 30 October 2019; Published: 4 November 2019


Abstract: This paper provides a review about the concept of convex systems based on Takagi-Sugeno,
linear parameter varying (LPV) and quasi-LPV modeling. These paradigms are capable of hiding the
nonlinearities by means of an equivalent description which uses a set of linear models interpolated
by appropriately defined weighing functions. Convex systems have become very popular since they
allow applying extended linear techniques based on linear matrix inequalities (LMIs) to complex
nonlinear systems. This survey aims at providing the reader with a significant overview of the
existing LMI-based techniques for convex systems in the fields of control, observation and safety.
Firstly, a detailed review of stability, feedback, tracking and model predictive control (MPC) convex
controllers is considered. Secondly, the problem of state estimation is addressed through the design
of proportional, proportional-integral, unknown input and descriptor observers. Finally, safety of
convex systems is discussed by describing popular techniques for fault diagnosis and fault tolerant
control (FTC).

Keywords: linear parameter varying (LPV) systems; Takagi-Sugeno systems; convex systems; linear
matrix inequalities (LMIs); fault diagnosis; fault tolerant control (FTC)

1. Introduction
Confucius once said ”the beginning of wisdom is to call things by their proper name”. In this regard,
it can be noticed that within the control community there is a big disagreement to call a certain class
of multiple model systems by its proper name, in other words, to call a spade a spade. Multiple
models were proposed in order to reduce the complexity of controller design for nonlinear systems
by describing the latter as a combination of local linear models. To this end, several approaches have
been proposed in the literature to deal with this problem, such as the linear parameter varying (LPV),
the quasi-LPV (qLPV) and the Takagi-Sugeno (TS).
LPV systems were introduced by Refs. [1,2] as models used to design controllers that guarantee
a suitable closed-loop performance for nonlinear plants working under time-varying operating
conditions. This was achieved by embedding the plant’s nonlinearities inside the so-called scheduling

Processes 2019, 7, 814; doi:10.3390/pr7110814 www.mdpi.com/journal/processes

Processes 2019, 7, 814 2 of 39

parameters. The term LPV was coined to differentiate the resulting class of systems from both linear
time invariant (LTI) and linear time varying (LTV) systems. The difference with respect to LTI systems
is clear because LPV systems are non-stationary. On the other hand, LPV systems are distinguished
from LTV systems in the perspective taken on both analysis and synthesis. LPV systems can be seen as
a generalization of a group of LTV systems, each one obtained by means of a predetermined trajectory
of the weighing functions. Therefore, properties such as stability, disturbance rejection and tracking,
among others, hold for a family of LTV systems, rather than for a single LTV system [3]. A typical LPV
system is described by:

ẋ (t) = A(θ (t)) x (t) + B(θ (t))u(t) (1)

y(t) = C (θ (t)) x (t) (2)

where x (t) ∈ Rnx denotes the state vector, u(t) ∈ Rnu is the input vector, y(t) ∈ Rny is the output
vector and θ (t) ∈ Rnθ is the vector of varying parameters, which can be a function of exogenous or
endogenous variables (in the latter case, the system is referred to as quasi-LPV) and that takes values
within a region Θ, that is, θ (t) ∈ Θ ⊂ Rnθ .
TS models are similar to LPV systems, since they are obtained by considering some collection
of linear models, although their overall blending is obtained by means of a set of fuzzy IF ... THEN
rules [4]. At first, they were obtained by performing linearization of the nonlinear plant about different
operating points [5]. Nevertheless, this conception was changed in the work by Ohtake et al. [6], who
proposed a convex modeling technique via the so-called sector nonlinearity approach. In this case,
the main idea is to obtain a convex system such that the global model matches the nonlinear system
exactly in a compact subset of the state space. The number of sub-models is directly related to the
number of nonlinear terms. For each nonlinear term, two sub-models are obtained such that for k
nonlinear terms, the global model is composed of h = 2k sub-models. Therefore, the bigger is the
number of nonlinear terms, the bigger becomes the conservatism of the global convex system and the
computational burden of both analysis and synthesis.
The TS approach was adopted rapidly by the control community and was applied to
state estimation [7], control [8], fault detection [7], descriptor systems [9], state observers [10],
waste-water treatment plants [11], bioreactors [12], process industry [13,14], mechatronics [15,16],
aeronautics [17,18] and automotive [19,20], among others. Comprehensive material about the topic
can be found in Refs. [8,21–23] and the references therein. On the other hand, another school of
thought named these approaches as quasi-LPV (qLPV) in order to differentiate fuzzy approaches
from model-based approaches. Nonetheless, models obtained by means of the sector nonlinearity
approach are not fuzzy, since the weighting functions are completely deterministic, as detailed in
Ref. [24]. Literature on qLPV systems can be found in Refs. [25–33], just to mention a few.
It is clear that LPV and TS systems have been developed independently but recently some works
have started discussing about the analogies between these paradigms [23,34,35]. For this reason, we
find it appropriate to consider a terminology that includes both schools of thought and in this review
we propose to denote both LPV and TS systems as convex systems. The idea of unifying these two
paradigms under a single name is not new, as it was originally proposed in Ref. [36] and retaken in
Refs. [37–40]. Nonetheless, in spite of the success of these paradigms, there is no literature review that
allows tasting all the flavors offered by the vastness of convex approaches. Therefore, in this paper,
three main aspects of convex systems are reviewed: control, observation and safety. The objective is to
help the reader to locate themselves in the area of convex systems by learning about the main used
techniques. It is worth highlighting that, although real-life applications of the reviewed methods are
discussed whenever appropriate, the level of detail is kept low, since the main focus of this review
is theoretical. The reader interested in a more extensive survey of experimental applications and
validations based on high-fidelity simulations is referred to the excellent work in Ref. [41] and the
references therein.
Processes 2019, 7, 814 3 of 39

The overall structure of this review is provided in Figure 1.

1. INTRODUCTION

2. CONTROL OF CONVEX SYSTEMS: it is dedicated to discuss

advances and control techniques for convex systems
2.1 Convex state-feedback control
2.2 Convex output-feedback control
2.3 Convex tracking controller
2.4 Model predictive control fo convex systems
2.5 Final comments on control of convex systems

3. OBSERVATION OF CONVEX SYSTEMS: it is dedicated to

observers
3.1 Convex observers
3.2 Robust observers
3.3 Proportional-integral observers
3.4 Descriptor observers

4. SAFETY IN CONVEX SYSTEMS: it is dedicated to fault diagnosis

and fault tolerant control of convex systems
4.1 Residual generation for fault detection
4.2 Unknown input observers-based fault isolation
4.3 Observer-based fault estimation
4.4 Multiple model adaptive estimators
4.5 Sliding mode fault tolerant control
4.6 Fault tolerant control based on controller reconfiguration
4.7 Fault-hiding via virtual actuators and sensors

5. CONCLUSIONS

Figure 1. Structure of the review.

Processes 2019, 7, 814 4 of 39

Notation: The notation used in this article is quite standard. Rm×n denotes the set of all matrices
with m rows and n columns. If a square matrix A ∈ Rn×n is symmetric, this fact will be denoted
by A ∈ Sn . Given a matrix A ∈ Sn , A  0 (A ≺ 0) denotes positive (negative) definiteness, that
is, that all its eigenvalues are positive (negative). Similarly, A  0 (A  0) denotes positive (negative)
semi-definiteness. For a matrix A ∈ Rm×n , A T and A† denote its transpose and pseudo-inverse,
respectively. If A ∈ Rn×n is non-singular, A−1 will denote its inverse. The symbol ∗ denotes the
transposed element in a symmetric position of a matrix. Finally, He{ A} is used as a shorthand notation
for A + A T .

2. Control of Convex Systems

Convex systems can arise from three possible interpretations [42]: (i) they can be seen as linear
systems subject to uncertainties for which the synthesis of a controller must be approached from a
robust control perspective; (ii) they can be seen as a family of parameter varying systems, for which the
instantaneous value of the varying parameters can be injected directly in the control structure, leading
to a gain scheduled control [43,44]; and (iii) the two previous situations can be combined, as suggested
by Ref. [45], where a double-layer polytopic framework was considered to this end.
In the last years, significant progress has been made in the control of convex systems. For
example, in the presence of uncertainties or disturbances, LPV robust control techniques have shown to
provide better performance than robust LTI controllers [46,47]. Indeed, many linear matrix inequality
(LMI)-based solutions for LTI systems have been extended to LPV systems, for example, an LPV
stabilizing controller was proposed for an arm-driven inverted pendulum in Ref. [48] and was shown
to outperform classical robust control techniques, such as H∞ and µ-synthesis. However, the method in
Ref. [48] does not guarantee that the closed-loop system exhibits a robust performance. To handle this
problem, a parametrized LPV H∞ control was presented in Ref. [49], which showed good performance
when applied to a turbofan jet engine. Other H∞ controllers for systems affected by time-varying
parametric uncertainties can be consulted in Refs. [50,51]. In order to improve the performance of
H∞ controllers, a switching controller designed with multiple Lyapunov functions was proposed by
Ref. [52]. Similarly, an LPV control for switched systems with slow-varying parameters was proposed
for an F-16 aircraft model in Ref. [53] by adopting the blending method developed by Ref. [54], which
achieves the separation of the entire parameter set into overlapped subsets, such that the overall
LPV controller can be blended over the entire region by means of regional controllers. In spite of the
good achieved performance, the method is applicable only under the assumption that the scheduling
parameters can be measured on-line, which is often difficult to satisfy in practice. To solve this problem,
a robust compensator which considers prior and non-real-time knowledge of the varying parameters
was proposed by Ref. [55] for stable polytopic LPV plants. Robust convex controllers have been also
proposed in the context of networked nonlinear systems [56,57] where the communication channel is
affected by package dropouts intermittently.

2.1. Convex State-Feedback Control

The first convex developments were proposed in Ref. [58] and subsequent papers [2,44,59].
The main difference with respect to the robust control theory is that the varying parameters are
assumed to be known and they can be used to schedule the time-varying controller gain. The most
widely applied control approach is the state-feedback, for which a conceptual scheme is given in
Figure 2. This approach computes the control law as follows:

u(t) = K (θ (t)) x (t) (3)

where K (θ (t)) ∈ Rnu ×nx denotes the controller gain. It is the simplest control law that can be
considered but its implementation requires knowing the full state of the system. Combining (1)
and (3), the closed-loop system is described by the following autonomous convex system:
Processes 2019, 7, 814 5 of 39

h i
ẋ (t) = A(θ (t)) + B(θ (t))K (θ (t)) x (t) (4)

Hence, by using the Lyapunov candidate function V ( x (t)) = x T (t) Px (t) > 0, with P ∈ Snx
and requiring V̇ ( x (t)) < 0, the so-called quadratic stability condition is obtained, as follows:

He{ PA(θ ) + PB(θ )K (θ )} ≺ 0 ∀θ ∈ Θ (5)

𝜃(𝑡)

Convex 𝑢(𝑡) 𝑥(𝑡)

Convex plant
controller
𝐴 𝜃 , 𝐵(𝜃)
𝐾(𝜃)

Figure 2. Conceptual scheme of convex control.

Equation (5) is a bilinear matrix inequality (BMI) as the unknown variables K (θ ) and P appear in
the same product of matrices PB(θ )K (θ ). However, it is possible to transform (5) into an LMI by pre-
and post-multipliying (5) by Q = P−1 , thus obtaining [60] (similarity transformations do not change
the eigenvalues of a matrix, hence its positive/negative definiteness):

He{ QPA(θ ) Q + QPB(θ )K (θ ) Q} ≺ 0 ∀θ ∈ Θ (6)

Note that in this case PQ = ( PQ) T = I and therefore the following is obtained:

He{ A(θ ) Q + B(θ )K (θ ) Q} ≺ 0 ∀θ ∈ Θ (7)

Finally, the quadratic term is eliminated by using the change of variables Γ(θ ) = K (θ ) Q, so
that (7) becomes:

He{ A(θ ) Q + B(θ )Γ(θ )} ≺ 0 ∀θ ∈ Θ (8)

which is in an LMI form. It is important to mention that, in the case that multiple specifications are
desired, the above change of variables introduces some conservatism, since it forces to use the same
Lyapunov matrix Q for all specifications, whereas using different matrices for different specifications
would lead to better performance. However, using LMIs instead of BMIs is convenient due to the
computational efficiency of available LMI solvers, whereas BMIs are non-convex, so that there is no
guarantee of obtaining a global minimum. Equation (8) represents an infinite number of constraints,
therefore it presents a computational problem. Unfortunately, the direct application of a polytopic
approach is not straightforward. One could rewrite (8) as:

Mij := He{ Ai Q + Bj Γi } ≺ 0 ∀i, j = 1, . . . , h (9)

and achieve stabilization by using u(t) = K (θ (t)) x (t), with the feedback controller gain obtained as
K (θ (t)) = Γ(θ (t)) Q−1 , where Γ(θ (t)) = ∑ih=1 ρi (θ (t))Γi and ρi (θ (t)) denotes the coefficients of the
following polytopic decomposition:
Processes 2019, 7, 814 6 of 39

" # " #
h h
A(θ (t)) Ai
B(θ (t))
= ∑ ρi (θ (t)) Bi
, ∑ ρi (θ (t)) = 1, ρi (θ (t)) ≥ 0 ∀θ ∈ Θ (10)
i =1 i =1

However, this solution has the drawback that a vertex gain Ki must be robust with respect to all
possible values of B(θ (t)), which corresponds to a high degree of conservatism. For this reason,
alternative solutions can be found in the literature, such as that proposed by Ref. [61] which consists
in pre-filtering the control input u(t). The combination of the filter and the system (1) leads to a
convex system with constant input matrix since B(θ (t)) appears embedded into the state matrix of
the augmented system. However, it must be mentioned that some recent work has questioned the
advantages of the pre-filter against using directly the LMIs (9) for the controller design [62,63].
Other alternative solutions aim at relaxing (9), although usually the requirement of low
conservatism is associated with an increase in the computational load. Among these solutions, it is
worth mentioning the conditions proposed by Ref. [64], who presented a fuzzy control application
of the Polya’s theorems on positive forms in the standard simplex. The result is a set of sufficient
conditions to prove the positiveness of double sums, which are progressively less conservative as a
complexity parameter n increases. These conditions are asymptotically exact, that is, necessary and
sufficient when n tends to infinity. Other conditions are those obtained by generating partitions of the
polytope through the triangulation method [65], which allows to obtain a family of sufficient conditions
for positivity/negativity of double sums and, in parallel, another family of necessary conditions, which
become asymptotically exact by decreasing the size of the partitions. In addition, one can recall the
conditions proposed by Ref. [66], that allow to relax the conditions of double polytopic sum to take
into account, for example, the existence of gaps in the set Θ. Nonetheless, the most popular relaxation
is the one proposed by Tuan et al. [67], which considers that an LMI in the form of (9) is equivalent to:

Mij ≺ 0 i ∈ [1, 2, ..., h] (11)

2
M + Mij + M ji  0 1 ≤ i 6= j ≤ h (12)
h − 1 ii

which reduces the conservatism and increases the applicability of the controller.
For convex qLPV and TS systems, Equations (9) and (11) are also known as parallel distributed
compensation (PDC) [68]. In this case, the feedback controller and the convex system share the same
weighting functions and the LMI conditions are obtained with the direct Lyapunov method. However,
the more local models the convex representation has, the greater is the conservatism of the LMI solution.
This fact follows from the necessity of finding a feasible solution that employs a common matrix P for all
the local models. A possible strategy to reduce the conservatism is to consider nonquadratic Lyapunov
functions (NQLFs) as done, for example, in Refs. [69–72]. The solution obtained through NQLFs,
which is also known as non-PDC [73,74], reduces considerably the conservatism and maintains the
same weighting functions for both the convex model and the controller. However, non-PDC controllers
are harder to design than PDC controllers, since the weighting functions involve time derivatives of
the NLQF, leading to local results [75]. This problem does not arise in convex systems dependent on
exogenous time-varying parameters, because the NLQLF would not involve time derivatives of the
states, hence global solutions can be obtained for this case [72,76–78].

2.2. Convex Output-Feedback Control

A variant of the state-feedback control strategy previously described consists in using directly
the output y(t) for feedback, which is easier to implement in cases where the state is unavailable for
measurement. Nevertheless, some conditions have to be ensured to make it possible to synthesize
these controllers, by means of approaches initially developed in the robust context and later extended
to the LPV framework [79–82].
Processes 2019, 7, 814 7 of 39

The next result allows the quadratic stabilization of a convex system of the form (1)–(2) using a
convex dynamic output-feedback controller defined as:

ẋK (t) = AK (θ (t)) xK (t) + BK (θ (t))y(t) (13)

u(t) = CK (θ (t)) xK (t) + DK (θ (t))y(t) (14)

where xK (t) ∈ Rnx is the internal state of the controller and AK (θ (t)), BK (θ (t)), CK (θ (t)), DK (θ (t)) are
matrix-valued functions, such that the closed-loop system obtained by the connection of (1)–(2) and
(13)–(14) is stable. In particular, the closed-loop system is described by the following autonomous
convex system:
" # " #" #
ẋ (t) A(θ (t)) + B(θ (t)) DK (θ (t))C (θ (t)) B(θ (t))CK (θ (t)) x (t)
= (15)
ẋK (t) BK (θ (t))C (θ (t)) AK (θ (t)) xK (t)

Due to the presence of C (θ (t)) post-multiplying BK (θ (t)) and DK (θ (t)) in (15), the procedure to
obtain LMIs for design of the controller’s matrices is somehow more complex. The system (1)–(2) is
quadratically stabilizable using the convex controller (13)–(14) if there exist a positive definite matrix
P ∈ S2nx such that Acl (θ ) P + PAcl (θ ) ≺ 0, ∀θ ∈ Θ, where Acl (θ ) is the state matrix of the autonomous
system (15). Following Refs. [83,84], this condition is achieved if and only if there exist matrices Q  0
and S  0 and matrix-valued functions ĈK (θ ) and B̂K (θ ) such that the following holds ∀θ ∈ Θ:

He{ A(θ ) Q + B(θ )ĈK (θ )} ≺ 0 (16)

He{SA(θ ) + B̂K (θ )C (θ )} ≺ 0 (17)

A possible methodology to obtain the controller’s matrices after solving (16)–(17) is described
hereafter [85]. If (16)–(17) and the following condition hold:
!
Q I
0 (18)
I S

then, by letting M, N be non-singular matrices such that MN T = I − SQ and choosing DK (θ (t)) = 0,

the controller’s matrices AK (θ (t)), BK (θ (t)) and CK (θ (t)) can be computed as follows:
h i
BK (θ (t)) = M−1 B̂K (θ (t)) − SB(θ (t)) DK (θ (t)) (19)
h i
CK (θ (t)) = ĈK (θ (t)) − DK (θ (t))C (θ (t)) Q N −T (20)

−1
AK (θ (t)) = M ÂK (θ (t)) − SB(θ (t))CK (θ (t)) N T − MBK (θ (t))C (θ (t)) Q (21)
  
− S A(θ (t)) + B(θ (t)) DK (θ (t))C (θ (t)) Q N −T

On the other hand, if (18) does not hold, then the matrices have to be adjusted using Qλ = λQ,
Sλ = λS, B̂Kλ (θ (t)) = λ B̂K (θ (t)) and ĈKλ (θ (t)) = λĈK (θ (t)), where λ > 1, until (18) holds with these
new variables and the controller’s matrices can be computed.
Equation (18) guarantees the existence of the invertible matrices M and N used for controller
computation. In the same way, the conditions to perform H∞ control, control with guaranteed cost
or to achieve other specifications can be obtained. Note that double polytopic sums appear due to
the terms B(θ )ĈK (θ ) and B̂K (θ )C (θ ) in (16)–(17). If the controller is restricted to the case where B̂K (θ )
and ĈK (θ ) are constant, then the LMIs (16)–(17) can be reduced to a finite number of conditions easily,
otherwise the discussion provided in the previous section about possible relaxations would apply
with slight modifications. It is worth remarking that in convex systems in which θ (t) depends on
Processes 2019, 7, 814 8 of 39

unmeasured states, the analysis and design of the output feedback control become more complicated,
see for example, Ref. [86].

2.3. Convex Tracking Controller

Consider a convex system subject to unknown inputs ans sensor noise, described by (for the sake
of simplicity, the output matrix is assumed to be constant):

ẋ (t) = A(θ (t)) x (t) + B(θ (t))u(t) + R(θ (t))d(t) (22)

y(t) = Cx (t) + Gd(t) (23)

where d(t) ∈ Rnd is the disturbance vector comprising both unknown inputs and noise and R(θ (t))
and G are matrices of appropriate dimensions. As illustrated in Figure 3, a convex tracking controller
can be considered for this system, with control law:
h iT
u(t) = K1 (θ (t)) x (t) + K2 (θ (t))e(t) = K(θ (t)) x (t) e(t) (24)

where K1 (θ (t)) and K2 (θ (t)) are the gains to be designed and e(t) is the integration error, added to
compensate steady-state errors and reach the desired output w(t):

ė(t) = w(t) − y(t) = w(t) − Cx (t) − Gd(t) (25)

Proportional
action
𝐾1 (𝜃)
𝜃(𝑡)
− +
𝑤(𝑡) + ⋵ (𝑡) Integral
+ Convex plant
න action 𝐶
𝐾2 (𝜃) 𝑢(𝑡) 𝐴 𝜃 , 𝐵(𝜃) 𝑥(𝑡) 𝑦(𝑡)
𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑜𝑟
𝑐𝑜𝑚𝑝𝑎𝑟𝑎𝑡𝑜𝑟 𝜃(𝑡)
Figure 3. Convex tracking controller scheme.

The system augmented with the integrator can be rewritten in a compact form by introducing the
augmented state vector xc (t) = [ x T (t) e T (t)] T :

ẋc (t) = Āc (θ (t)) xc (t) + B̄c (θ (t))u(t) + R̄c (θ (t))d(t) + B̄w w(t) (26)

with:
" # " # " # " #
A(θ (t)) 0 B(θ (t)) 0 R(θ (t))
Āc (θ (t)) = B̄c (θ (t)) = B̄w = R̄c (θ (t)) = (27)
−C 0 0 I −G

The closed loop system has the form:

h h
ẋc (t) = ∑ ρi (θ (t)) ∑ ρ j (θ (t)) Āci − B̄ci K j xc (t) + B̄Rwi d¯w (t)


(28)
i =1 j =1
Processes 2019, 7, 814 9 of 39

with:
h i h iT
B̄Rwi = R̄i B̄w , d¯w (t) = d(t) w(t) (29)

Sufficient conditions for the existence of the controller are given in Ref. [35] and presented
hereafter. Consider the system (22)–(23), the feedback control law defined by (24), the integrator, and
let the attenuation level be given by γc > 0. The closed loop system error (25) is globally stable with H∞
performance if k xc (t) k22 < γc2 k d¯ω (t) k22 and if there exists a matrix X  0 such that ∀i, j ∈ [1, 2, ..., h],
the following holds: "   #
T + Ξ T B̄ T + B̄
He X Āci T
j ci Rwi B̄Rwi X
≺0 (30)
∗ −γc2 I
h i
Then, the controller gain matrices are computed by K j = K1j K2j = X −1 Ξ j .
This is possible because if we consider the L2 -gain from d¯ω (t) to xc (t) such that:

Jxc d = V̇ (t) + xcT (t) xc (t) − γc2 d¯ω

T
(t)d¯ω (t) < 0 (31)

where V (t) is a quadratic Lyapunov function, the LMI (30) is obtained after solving the performance
criteria (31). Complete procedures are described in detail in Ref. [35].
An alternative approach is to use a reference model as originally proposed by Ref. [87] and later
applied by Refs. [26,88], which has the advantage that the tracking error is described by an autonomous
system, so its convergence to zero can be guaranteed even without the use of an integrator.

2.4. Model Predictive Control for Convex Systems

Model predictive control (MPC) is a control strategy that is based on the use of a mathematical
model to predict the system’s behavior in a future time window and then finds the optimal input
sequence by minimizing a cost function [89,90]. Only the first calculated input is applied to the system
and the remaining are discarded, repeating this prediction-optimization process at every sample. This
control technique is popular because it can take into account systematically complex dynamics, as well
as physical and process quality constraints [91].
Consider the discrete-time convex system:

x (k + 1) = A(θ (k)) x (k) + B(θ (k))u(k ) (32)

where A(θ (k)) = ∑lj=1 θ j (k) A j and B(θ (k )) = ∑lj=1 θ j (k) Bj . Therefore, θ (k) belongs to a convex
polytope Θ defined by the values θ j (k) such that ∑lj=1 θ j (k ) = 1, with, 0 ≤ θ j (k) ≤ 1. On the
other hand, when θ (k) varies in the polytope Θ, the system matrices vary in the polytope Ω defined
as follows:

[ A(θ (k)), B(θ (k))] ∈ Ω = Co {[ A1 , B1 ], [ A2 , B2 ], . . . , [ Al , Bl ]} (33)

where [ Ai , Bi ] are the vertex matrices obtained when θi = 1 and θ j = 0 for j 6= i. Hereafter,
for illustrative purposes, it is assumed that there is no model uncertainty and that both the scheduling
variable θ (k) and the state x (k ) are known at time k. However, the future evolution of the model is
uncertain since future values of θ (k) are unknown.
Let us define the following quadratic cost function:
∞
J (k) = x (k |k) T Qx (k|k ) + u(k|k ) T Ru(k|k ) + ∑ x (k + i |k) T Qx (k + i |k) + u(k + i |k ) T Ru(k + i |k) (34)
| {z } i =1
J0 (k ) | {z }
J1 (k )
Processes 2019, 7, 814 10 of 39

where Q, R are weighting matrices with appropriate dimensions and the notation x (k + i |k) represents
the predicted value for the state variable x at the future sample k + i calculated at sample k. Hence,
J0 (k) correspond to the first prediction step and J1 (k) correspond to the remaining of the prediction.
Let U (k) be the sequence of inputs computed at sample k, that is, U (k) = [u(k|k), U1 (k )] =
[u(k|k), u(k + 1|k), . . .]. Then, the optimal control sequence is obtained by minimizing the maximum
value that the cost function (34) can take for all the possible future trajectories of the parameter
θ (k), that is,

U ∗ (k) = min max J (k) (35)

U (k) [ A(θ (k+i )),B(θ (k+i ))]∈Ω i ≥0

where ∗ denotes optimality. The first element of U ∗ (k), that is, u∗ (k |k), is applied to the system, while
the remaining of the sequence U1∗ (k) can be proven to be equivalent to a state feedback control law
whose gain does not depend on the instantaneous value of θ (k) (see Ref. [92] for further details), that is,

U1∗ (k) = {u(k + i |k ) = K (k ) x (k + i |k), i ≥ 1} (36)

Following Ref. [92], instead of solving (35), an upper bound for the term J1 (k) can be defined,
as follows:

max J1 (k) ≤ V ( x (k + 1|k )) = x (k + 1|k) T Px (k + 1|k ) P(k)  0. (37)

[ A(θ (k+i )),B(θ (k+i ))]∈Ω, i ≥0

Then, an upper bound of the worst case of J (k) is minimized instead of (35), as follows:

U ∗ (k) = min x (k|k) T Qx (k|k ) T + u(k|k) T Ru(k|k ) T + x (k + 1|k ) T Px (k + 1|k) (38)

u(k |k ),P(k )

The optimization problem (38) can be reformulated as the following minimization problem:

min γ (39)
γ,u(k |k ),Q̃(k ),Y (k )

subject to LMIs:
1 1 
x̂ (k + 1|k) T x̂ (k|k ) T Q 2 u(k|k)T R 2

1
 x̂ (k + 1|k) Q̃(k) 0 0 
0 (40)
 
1
 Q 2 x (k|k)

0 γI 0 
1
R 2 u(k|k) 0 0 γI
1 1
Γ̃ j (k) T Y (k)T R 2
 
Q̃(k) Q̃(k ) Q 2
 Γ̃ j (k) Q̃(k) 0 0 
0 ∀ j = 1, . . . , l (41)
 
 1
 Q 2 Q̃(k) 0 γI 0 
1
R 2 Y (k) 0 0 γI

with x̂ (k + 1|k) = [ A(θ (k)) x (k|k) + B(θ (k ))u(k|k )], Γ̃ j (k ) = A j Q̃(k) + Bj Y (k) and Q̃(k)  0. The gain
in (36) is computed as K (k) = Y (k ) Q̃−1 (k ), which guarantees that the state evolves in an ellipsoidal
invariant set.
Considering the system output as y(k) = Cx (k), the cost function (34) may be subject to
constraints [93]:

u(k|k ) ≤ umax (42)

 
C A(θ (k )) x (k|k ) + B(θ (k ))u(k|k ) ≤ ymax (43)
2
Processes 2019, 7, 814 11 of 39

Constraints on the inputs are satisfied if there exists a matrix X  0 such that:
" #
X Y
0 with Xii ≤ u2max (44)
YT Q̃

In a similar way, the constraints on the outputs are equivalent to the LMI:
" #
Q̃ [ A j Q̃ + Bj Y ] T C T
0 j = 1, . . . , l (45)
C [ A j Q̃ + Bj Y ] y2max

The constrained MPC algorithm with control law (36) can be obtained by solving the optimization
problem (39) subject to the LMIs (40), (41), (44), (45) and constraints (42) and (43). However,
although this algorithm does not impose u(k |k) and y(k + 1|k) to invariant ellipsoid constraints, still
includes constraints on all future inputs and outputs. A method to improve the conservatism is to relax
the future constraints (44) and (45) and bound only u(k|k ) and y(k + 1|k ) [93]. To guarantee stability
an additional constraint that ensures that the cost function decreases monotonously (φ(k) < φ(k − 1)
with φ(k) = x (k|k) T Qx (k |k) T + u(k|k ) T Ru(k|k ) T + x (k + 1|k) T Px (k + 1|k)) has to be included in
the optimization:

min γ (46)
γ,u(k |k ),Q̂(k ),Y (k )

subject to LMIs:
1 1 
x̂ (k + 1|k ) T x (k|k)T Q 2 u(k|k)T R 2

γ
 x̂ (k + 1|k) Q̂(k) 0 0 
0 (47)
 
1
 Q 2 x (k|k)

0 I 0 
1
R 2 u(k|k) 0 0 I
1 1
Γ̂ j (k) T Y (k)T R
 
φ ( k − 1) Q̂(k) Q 2 2
 Γ̂ (k) Q̂(k) 0 0 
j
0 ∀ j = 1, . . . , l (48)
 
 1
 Q 2 Q̂(k) 0 I 0 
1
R 2 Y (k) 0 0 I
" #
u(k|k) − umax
≤0 (49)
−umax − u(k|k)
"   #
C A(θ (k)) x (k|k) + B(θ (k))u(k|k) − ymax
  ≤0 (50)
−ymax − C A(θ (k)) x (k|k) + B(θ (k))u(k|k)

with Γ̂ j (k) = A j Q̂(k) + Bj Y (k ) and Q̂(k )  0. The gain of the control law (36) is computed as
K (k) = Y (k ) Q̂−1 (k). To initialize the algorithm, in k = 0, the Lyapunov constraint φ(k ) < φ(k − 1)
is not taken into account. The resulting control strategy, which is depicted in Figure 4, provides
guaranteed closed-loop stability provided that a feasible solution has been found. A parameter
dependant feedback law instead of (36) can also be considered as in Ref. [93].
Processes 2019, 7, 814 12 of 39

MODEL PREDICTIVE CONTROLLER

Convex model
𝐴 𝜃 , 𝐵(𝜃)

𝑥(𝑘
ො + 1|𝑘)

Online
Convex plant
optimization
𝑢 ∗ (𝑘|𝑘) 𝐴 𝜃 , 𝐵(𝜃) 𝑥(𝑘|𝑘)
(𝐿𝑀𝐼𝑠)

Figure 4. Convex model predictive control scheme.

Most of the MPC strategies for convex systems are based on the algorithm proposed by Ref. [92]
since it stabilises robustly an LPV system for all possible parameter variations. However, such
algorithm was not thought for application to LPV systems and therefore it suffers from being
conservative and computationally demanding. The strategies proposed by Refs. [93,94], which consider
bounds in the parameter variations, show less conservatism and a decreased computational load when
compared to Ref. [92]. A modification of Ref. [93] that involves updating the polytope Ω while keeping
it defined by the least possible number of vertices has been presented in Ref. [95]. This innovation is
motivated by the fact that the fewer vertices are used to describe Ω, the less likely it is that infeasibility
problems could occur. In Ref. [96], an extension to nonlinear systems has been presented, where a
linearized model is obtained from the nonlinear model at each sample and then an LPV model that
varies in a politope Ω is used for obtaining the state prediction. Other existing approaches are focused
on the use of Lyapunov functions that depend on θ (k) that enlarge the feasible region [97,98]. In line
with this work, an algorithm that uses closed-loop predictions with good achieved performance and
low computational requirements was presented in Ref. [99]. More recently, Ref. [100] has presented
a class of nonlinearly parameterized Lyapunov functions to achieve more efficient relaxed stability
conditions. A robust MPC scheme for LPV systems where the varying parameters are assumed to
be measured online and exploited for feedback has been derived in Ref. [101]. Explicit MPC for
convex systems has been also proposed in order to avoid the need of online optimization [102,103].
In general, MPC for convex systems has been a topic that has received an intense interest by the
research community in the last few years, for which the interested reader is referred to Refs. [104–107]
and the references therein.

2.5. Final Comments on Control of Convex Systems

It is worth noting that what has been discussed in this section about the control of convex systems
does not apply only to the problem of controller design for quadratic stabilization but also to the
case of other specifications, such as D -stabilization [108], H∞ control [109], control with guaranteed
cost [110] and many more. The described methods can be adapted to deal with convex systems with
piecewise constant parameters, which provide a unifying concept lying in between the robust and the
gain-scheduled perspectives, including both as extremal cases [111].
On the other hand, there are some cases in which the direct application of the convex techniques
described so far would not work, for example, due to the loss of controllability for some points of the
design polytope, so that alternative strategies must be employed. For instance, consider the following
simplified model of a unicycle mobile robot
Processes 2019, 7, 814 13 of 39

   
ẋ (t) cos(α(t)) 0 " #
 v
 ẏ(t)  =  sin(α(t)) 0  (51)
  
w
α̇(t) 0 1

where x, y are the spacial 2D coordinates, v and w are the mobile robot translational and angular
velocities, respectively and α denotes the orientation with respect to the fixed frame. If α ∈ [−π, π ]
were considered, one would get a non-controllable convex models in the vertices although the
underlying nonlinear system is actually controllable. For example, the solution proposed in Ref. [88]
was based on dividing the parameter space in regions and use a switched approach in order to avoid
the above-mentioned singularities.
Another important theoretical point to be remarked is that the LMI-based assessment of stability
(or some other goal) in convex systems arising from an underlying nonlinear system could mislead to
believe that global stability (or performance) would hold for the original system. This fact is in general
not true, as shown remarkably by Ref. [112] with a simple second order autonomous nonlinear system,
that is, the well known Van der Pol equation. Fortunately, Ref. [112] also shows that it is possible to
estimate the region of attraction for the nonlinear system, based on the Lyapunov function obtained
for the convex system. This fact was further studied by Refs. [113–115] and was used by Ref. [34] to
create a metric to compose different convex models obtained for the same nonlinear system.
Finally, it is worth mentioning that analysis and control problems for convex systems with
delays have also attracted some recent interest [116,117]. These systems belong to the intersection
of convex systems and time-delay systems, so they inherit the difficulties of each one. In particular,
the stability analysis of these systems must be performed using tools such as Lyapunov-Razumikhin
functions and Lyapunov-Krasovskii functionals, which increase the number of decision variables [118].
The interested reader is referred to the monograph [119] and the references therein.

3. Observation of Convex Systems

In many real-world applications there are some internal state variables that cannot be measured
with the available sensors. Nevertheless, many control techniques are based on the assumption that
the whole state is available, which is not always true. In practice, the available information concerns
the input u(t) and the output y(t), rather than the state x (t). In such case, the observability properties
state that when the system is observable, the initial state can be determined and, therefore, the state
trajectory can be reconstructed from input and output measurements, by means of the so-called state
observer. State observers are dynamical systems that are designed to estimate asymptotically the
state vector x (t). Applications of convex observers can be found in UAVs [120], electric vehicles [121],
networked systems [122], DC motors [123], wind turbines [124], riderless bicycles [125], to mention
a few.

3.1. Convex Observers

The most common state observer is the one named after Luenberger, which for a convex systems
has the following form [126] (see Figure 5):

x̂˙ (t) = A(θ (t)) x̂ (t) + B(θ (t))u(t) + L(θ (t)) (y(t) − ŷ(t)) (52)
ŷ(t) = C x̂ (t) (53)

where x̂ (t) ∈ Rnx denotes the state estimate, ŷ(t) ∈ Rny denotes the output estimate and the meaning
of the remaining variables can be inferred from the previous section. The scheduled observer gains
L(θ (t)) are designed to guarantee closed-loop stability of the estimation error dynamics for all values
Processes 2019, 7, 814 14 of 39

of θ (t), such that the estimation error between the observer (52)–(53) and the system (1)–(2) converges
towards zero. Let us define the estimation error as follows:

e(t) := x (t) − x̂ (t). (54)

𝑢(𝑡) Convex plant 𝑦(𝑡)

𝐴 𝜃 ,𝐵 𝜃 ,𝐶

𝜃(𝑡)

Convex observer 𝑦(𝑡)

ො
𝐴 𝜃 , 𝐵 𝜃 , 𝐶, 𝐿(𝜃)

Figure 5. Convex state observer scheme.

Then, the estimation error dynamics is defined as:

ė(t) = ( A(θ (t)) − L(θ (t)C ))e(t) (55)

The stability condition of the above differential equation can be obtained by means of LMI-based
techniques, for example by considering the quadratic stability concept. In this case, one seeks the
existence of a quadratic Lyapunov function V (e(t)) = e(t) T Pe(t) ≥ 0, P  0, whose derivative over
the error dynamics is given by:

V (e(t)) =ė(t) T Pe(t) + e(t) T Pė(t) = e(t) T (He{ PA (θ (t)) − PL (θ (t)) C }) e(t) < 0

In order to eliminate the quadratic term, the change of variable W (θ (t) = PL(θ (t) is considered,
such that the following LMI is obtained:

He{ PA(θ ) − W (θ )C } ≺ 0 ∀θ ∈ Θ (56)

where the observer gain matrix can be computed later from its solution as L(θ (t)) = P−1 W (θ (t)).
Furthermore, to improve the speed convergence of the state observer, a decay rate α < 0 can be
added as requirement, by asking that:

V̇ (e(t)) + 2αP < 0 (57)

which is also known in the literature as α-stabilization. As a result, the LMI (56) is replaced by
the following:
He{ PA(θ ) − W (θ )C } + 2αP ≺ 0 ∀θ ∈ Θ (58)

It should be noticed that the stability of a state observer is guaranteed if the LMI (58) has a solution.
Nevertheless, the approach described so far does not consider disturbances or measurement noise,
which affect all physical systems.
Processes 2019, 7, 814 15 of 39

3.2. Robust Observers

Dynamical systems can be affected by external disturbances and measurement noise and,
moreover, there exists always a mismatch between the real plant and its model used for control. These
effects can lead to the loss of stability or performance if not taken into account appropriately. A control
system that remains stable and with none (or little) performance loss despite the disturbances/noise is
said to be robust. In other words, robustness means that the system remains stable and with almost
the same performance even in the presence of disturbance, model mismatches or noise. For example,
a wind turbine system must keep its efficiency even in the presence of air velocity changes. In contrast,
an example of unwanted disturbance amplification is the Tacoma Narrows suspension bridge, where
strong winds caused resonant oscillations of increasing magnitude in the bridge structure, which
ultimately led to its destruction.
In particular, the robust state observer design problem is related to finding the observer gains
such that it is always possible to estimate the real states within prescribed tolerances, despite the effects
of uncertainties.
A possible technique to deal with uncertainties is by means of the robust H∞ approach, which
has been developed since the beginning of the eighties and has been applied intensely, with successful
results, to convex systems [11,120,127,128].
The H∞ approach assumes that the disturbance d(t) ∈ Rnd belongs to a set of norm bounded
functions. The idea is to minimize the worst error that can arise from any disturbance in the
following set:
Z ∞ 1/2
T
k d ( t ) k2 = d (τ )d(τ )dτ (59)
0

For instance, let us consider a convex model affected by the above sources of uncertainty:

ẋ (t) = A(θ (t)) x (t) + B(θ (t))u(t) + R(θ (t))d(t) (60)

y(t) =Cx (t) + Gd(t) (61)

where the meaning of each variable and matrix is kept as previously and the matrices A (θ (t)), B (θ (t)),
R (θ (t)) satisfy the polytopic property:
   
A(θ (t)) h Ai h
 B(θ (t))  = ∑ ρi (θ (t))  Bi  , ∑ ρi (θ (t)) = 1, ρi (θ (t)) ≥ 0 ∀θ ∈ Θ (62)
   
R(θ (t)) i =1 Ri i =1

Let us consider the Luenberger observer given by (52)–(53), for which the dynamics of the
estimation error defined as in (55), can be described after some algebraic manipulations as follows:

ė(t) = ( A(θ (t)) − L(θ (t))C ) e(t) + ( R(θ (t)) + L(θ (t)) G )d(t) (63)

Then, the design problem can be formulated as the one of guaranteeing asymptotic stability of
the estimation error (63) while at the same time minimizing, by means of the H∞ technique, the ratio
between the `2 norm of the output vector and the `2 norm of the disturbance vector against the
disturbance vector d(t), that is,

||y||`2
min γ : < γ, ||d||`2 6= 0 (64)
γ >0 ||d||`2

where γ > 0 is the prescribed attenuation level (upper bound on the above mentioned ratio).
Then, by considering a bound on the L2 gain from d(t) to e(t) given by the Lyapunov function
V (e(t)) = e(t) T Pe(t), P  0, the above performance criterion is satisfied if the following holds:
Processes 2019, 7, 814 16 of 39

V̇ (e(t)) + e(t) T e(t) − γ2 d(t) T d(t) < 0 (65)

Similarly to the procedure described above in Section 3.1, the following LMI is obtained from
manipulations on (65): " #
He{ PAi − Ξi C } + I PEi − Ξi G
≺0 (66)
∗ −γ̄I

Once solved the above LMI, the observer gain matrix can be computed as Li = P−1 Ξi , which achieves
√
an attenuation level γ = γ̄.

3.3. Proportional-Integral Observers

Proportional-integral observers (PIOs) have become popular in recent years due to their
robustness against constant or slowly varying disturbances. In a PIO an additional term, which is
proportional to the integral of the output estimation error, is added in order to increase the robustness
performance [129], as depicted in Figure 6. This term gives an additional degree of freedom that can
be used for the estimation of unknown inputs such as disturbances [130,131], battery charge [132] and
faults [133,134], among others.

𝑑(𝑡)
Convex plant 𝑦(𝑡)
𝑢(𝑡)
𝐴 𝜃 , 𝐵 𝜃 , 𝑅 𝜃 , 𝐶, 𝐺

𝜃(𝑡)

Proportional observer 𝑦(𝑡) ො

𝐴 𝜃 , 𝐵 𝜃 , 𝐶, 𝐿 𝜃 , 𝐾𝑝 (𝜃)

−
Φ න
+ መ
𝑑(𝑡)

Figure 6. Convex proportional integral observer (PIO) scheme.

A PIO for a system in the form (60)–(61) is described by the following equations:

x̂˙ (t) = A (θ (t)) x (t) + B (θ (t)) u(t) + L(θ (t))(C x̂ (t) − y(t)) + K p (θ (t))dˆ(t) (67)
dˆ˙(t) = Φ(y(t) − C x̂ (t)) (68)

where x̂ and dˆ denote the estimated state and unknown input vectors, respectively and L(θ (t)),
K p (θ (t)), and Φ are the observer gain matrices to be computed. The addition of an integrator provides
more robustness to the observer so that it can deal with measurement noise or modeling uncertainties.
Let us define the estimation errors as (55) and ed (t) = d(t) − dˆ(t). In order to get a suitable design
procedure, it can be considered that the unknown input d(t) is varying slowly, which means that
d˙(t) ≈ 0. Then, the dynamics of the estimation errors can be computed as:
Processes 2019, 7, 814 17 of 39

ė(t) = ( A(θ (t)) − L(θ (t))C )e(t) + ( R(θ (t)) − L(θ (t)) G )d(t) − K p (θ (t))dˆ(t) (69)
ėd (t) = −ΦCe(t) − ΦGd(t) (70)

By considering:

R(θ (t)) − L(θ (t)) G = K p (θ (t)) (71)

ΦG = 0 (72)
T
and an extended error vector defined as ē T (t) := e T (t)edT (t) , the overall error dynamics is


rewritten as:
ē˙(t) = ( Ae (θ (t)) − Le (θ (t))Ce )ē(t) (73)

with:
" # " #
A(θ (t) K p (θ (t) L(θ (t))
Ae (θ (t)) = Le (θ (t)) = Ce = [C, 0] (74)
0 0 Φ

Then, by considering a Lyapunov equation V (ē(t)) = ē T (t)P ē(t), the solution is obtained in the LMI
formulation, similarly to (56), as:

He{P Ae (θ ) − We (θ )Ce } ≺ 0 ∀θ ∈ Θ (75)

where P = diag( P, Q), P, Q  0 and We (θ (t)) = P Le (θ (t)). The observer gain matrices are obtained
from the solution of (75) as Le (θ (t)) = P −1 We (θ (t)).
It is worth remarking that the slow variation assumption is very common in the literature. From a
practical point of view, it can be relaxed, as done for example in Ref. [135]. It is also worth noticing that
later, in Section 4.2, unknown input observers, which allow to obtain asymptotic convergence of the
estimation error despite the presence of d(t) without making the assumption that d(t) is approximately
constant, will be reviewed.

3.4. Descriptor Observers

The observers discussed above are designed for regular systems whose dynamics are represented
only by ordinary differential equations. Nevertheless, some mathematical models are composed of both
differential and static (algebraic) equations. Designing state observers for descriptor systems is harder
than for regular systems because descriptor systems usually have three types of modes (finite dynamic,
impulsive and non-dynamic [136]) and the observer must deal with all of them. Nonetheless, there are
plenty of fields in which descriptor systems are applied, for example, aircraft modeling [137], complex
systems [11], microgrids [138], electrical [9], mechanical and hydraulic systems [136] and biological
processes [12], among others. Furthermore, there are some mechanical systems for which possible
variations in time of masses and/or inertias rely on a natural descriptor model as in Refs. [139–141].
In these cases, although convex systems can be transformed into a regular system, it has been proved
that it is possible to reduce the number of linear models by considering a descriptor representation,
which in general makes the LMI-based constraints less conservative [142,143].
A convex descriptor system affected by disturbances is described by

E ẋ (t) = A(θ (t)) x (t) + B(θ (t))u(t) + R(θ (t))d(t) (76)

y(t) = Cx (t) + Gd(t) (77)

where E is a constant matrix with rank( E) = r ≤ n x . Note that in the particular case of E = I,
the descriptor system becomes a regular system and the observer can be computed as described
Processes 2019, 7, 814 18 of 39

previously. On the other hand, unlike regular systems, a descriptor system has different modes that
are given by differential and algebraic equations. As a result, different types of observability condition
should be verified such as R-observability [144], which means that:
" #
sE − Ai
rank = nx ∀i ∈ [1, 2, ...N ] (78)
C

and I-observability [144], that is,

 
E Ai
rank  0 E  = nx + r ∀i ∈ [1, 2, ..., N ] (79)
 
0 C

R-observability characterizes the ability to reconstruct only the reachable state from the output
data. However, due to the algebraic equations, impulsive terms can appear, which are not desirable
since they can saturate the state response or, in general, have negative effects on the system. On the
other hand, I-observability guarantees the ability to estimate impulse terms given by the algebraic
equations [12].
Then if the convex descriptor system is both R- and I-observable, the following observer can
be proposed:

ż(t) = N (θ (t))z(t) + J (θ (t))u(t) + L(θ (t))y(t) (80)

x̂ (t) = z(t) + T2 y(t) (81)
ŷ(t) = C x̂ (t) (82)

where z(t) represents the observer state and x̂ (t) stands for the estimated states. N (θ (t)), J (θ (t)),
L(θ (t)) and T2 are unknown gain matrices of appropriate dimensions to be computed. Based
on (76)–(77) and (80)–(82), the estimation error e(t) is:

e(t) = x (t) − x̂ (t) = ( I − T2 C ) x (t) − z(t) − T2 Gd(t) (83)

Assuming that there exists a matrix T1 ∈ Rnx ×nx such that:

I − T2 C = T1 E (84)

the estimation error becomes:

e(t) = T1 Ex (t) − z(t) − T2 Gd(t). (85)

Assuming that the disturbances is slowly varying, d˙(t) ≈ 0, the dynamics of e(t) is given by:

ė(t) = T1 E ẋ (t) − ż(t) (86)

= T1 ( A(θ (t)) x (t) + B(θ (t))u(t) + R(θ (t))d(t)) − ( N (θ (t))z(t) + J (θ (t))u(t) + L(θ (t))y(t))
=( T1 A(θ (t)) − L(θ (t))C − N (θ (t)) T1 E) x (t) + ( T1 B(θ (t)) − J (θ (t)))u(t)
+ ( T1 R(θ (t)) − L(θ (t)) G + N (θ (t)) T2 G )d(t) + N (θ (t))e(t) (87)

In order to guarantee convergence to zero of the estimation error, the following conditions
are considered:
Processes 2019, 7, 814 19 of 39

T1 A(θ (t)) − L(θ (t))C − N (θ (t)) T1 E = 0 (88)

T1 B(θ (t)) − J (θ (t)) = 0 (89)

After some algebraic manipulations, the following equations equivalences are obtained:

N (θ (t)) = T1 A(θ (t)) + K (θ (t))C (90)

Γ(θ (t)) = L(θ (t)) − N (θ (t)) T2 (91)
R̄(θ (t)) = T1 R(θ (t)) − Γ(θ (t)) G (92)

A particular solution of both matrices T1 and T2 is computed as:

i E †
" #
h i h
T1 T2 = Inx 0 (93)
C

The estimation error becomes:

ė(t) = N (θ (t))e(t) + R̄(θ (t))d(t)

and, in order to guarantee robustness, the following H∞ performance criterion is considered:

V̇ (e(t)) + e(t) T e(t) − γ2 d(t) T d(t) < 0 (94)

with attenuation level γ > 0 and quadratic Lyapunov function V (e(t)) := e(t) T Pe(t), P  0, such that
the following BMI is obtained:
" #
He{ PT1 A(θ (t)) + PK (θ (t))C } + I PT1 R(θ (t)) − PΓ(θ (t)) G
≺0 (95)
∗ −γI

Then, by considering the change of variable Ξ(θ (t)) = PK (θ (t)) and Ω = PΓ(θ (t)), the above
LMI becomes:
" #
He{ PT1 A(θ (t)) + Ξ(θ (t))C } + I PT1 R(θ (t)) − Ω(θ (t)) G
≺0 (96)
∗ −γI

4. Safety in Convex Systems

Due to the increased demand of safety and reliability in complex systems, fault diagnosis
techniques have attracted a great amount of attention in the past few decades. Concerning the recent
developments of fault diagnosis for convex systems, hereafter we will review: (i) residual generation
for fault detection; (ii) unknown input observers (UIO)-based fault isolation; (iii) observer-based
fault estimation; and (iv) multiple model adaptive estimators (MMAEs). Fault tolerant control (FTC)
systems aim at maintaining closed-loop stability and desired performances in the face of faults in some
components, for example, actuators or sensors. The majority of the literature has focused the attention
on LTI systems, although one can find ad-hoc approaches developed for nonlinear systems, see for
example, Ref. [145]. It should not be surprising that some recent research has attempted to extend FTC
strategies developed originally for LTI systems to the convex case, in order to enlarge their applicability
to a wider class of nonlinear systems. This research has focused mainly on the following techniques,
which will be reviewed in the following: (i) sliding mode control (SMC); (ii) control reconfiguration;
and (iii) virtual actuators/sensors.
Processes 2019, 7, 814 20 of 39

4.1. Residual Generation for Fault Detection

Fault detection aims at detecting accurately the appearance of a fault and is usually performed
generating residual signals which act as fault indicators [146]. For the following convex system with
unknown disturbances d(t) and faults f (t):

ẋ (t) = A (θ (t)) x (t) + B (θ (t)) u(t) + R (θ (t)) d(t) + F (θ (t)) f (t) (97)
y(t) = Cx (t) + Gd(t) + H f (t) (98)

let us consider a detection filter with the following structure:

ż(t) = N (θ (t)) z(t) + J (θ (t)) u(t) + L (θ (t)) y(t) (99)

x̂ (t) = z(t) − Ey(t) (100)
ŷ(t) = C x̂ (t) (101)

where N (θ (t)), J (θ (t)), L (θ (t)) are filter gains to be determined through design. By defining the
estimation error signal e(t) = x (t) − x̂ (t), the residual signal r (t) = y(t) − ŷ(t) and the matrix
T = I + EC, if the following constraints hold:

TA (θ (t)) − N (θ (t)) T − L (θ (t)) C = 0 (102)

TB (θ (t)) − J (θ (t)) = 0 (103)
h i
E G H =0 (104)

then one obtains that the residual has a dynamics described by:

ė(t) = N (θ (t)) e(t) + Bd (θ (t)) d(t) + B f (θ (t)) f (t) (105)

r (t) = Ce(t) + Dd(t) + H f (t) (106)

with:
Bd (θ (t)) = [ TR (θ (t) − L (θ (t)) G − N (θ (t)) EG )]

B f (θ (t)) = [ TF (θ (t) − L (θ (t)) H − N (θ (t)) EH )]

Then, in order to achieve the fault detection goal, one must ensure the asymptotic stability of the error
system while making the signal r(t) as sensitive as possible to faults and as insensitive as possible to
disturbances, which is usually achieved by means of a mix of H∞ and H− index optimization. This
approach was initially proposed for filter design in the full-frequency domain, see for example, Refs. [147,
148]. However, for some practical systems, fault and disturbance frequencies ranges are known
beforehand, which motivated recent research on filter design in a finite-frequency domain [149–151],
using the so-called generalized Kalman-Yakubovich-Popov (GKYP) lemma [152]. Another recent line of
research worth of mentioning is the one that investigates the behavior of the fault detection observer
when unmeasurable scheduling parameters are considered, see for example, Ref. [153].

4.2. Unknown Input Observers-Based Fault Isolation

In the last years, UIOs have shown to be a promising technique for fault detection purposes, due
to their ability to provide the system state estimate even in the presence of unknown inputs, such as
faults and disturbances. The approaches proposed for UIO design can be basically split into two classes:
in the first one, the state estimation is decoupled from the unknown inputs, for example, by means of
some structural conditions [154]. In the second case, joint estimation of the state and unknown inputs
is achieved [155].
Processes 2019, 7, 814 21 of 39

Let us consider the following convex system (for the sake of simplicity, the whole state is assumed
to be measured):

ẋ (t) = A (θ (t)) x (t) + B (θ (t)) u(t) + F (θ (t)) f (t) (107)

y(t) = x (t) (108)

and let R (θ (t)) and H (θ (t)) be some given matrix functions. Let us choose:

T (θ (t)) = I − R (θ (t)) (109)

S1 (θ (t)) = R (θ (t)) A (θ (t)) − H (θ (t)) (110)
S2 (θ (t)) = H (θ (t)) T (θ (t)) (111)

then:
 
ż(t) = H (θ (t)) z(t) + R (θ (t)) B (θ (t)) u(t) + S (θ (t)) − Ṫ (θ (t)) y(t) (112)
x̂ (t) = z(t) + T (θ (t)) y(t) (113)

where Ṫ (θ (t)) is the time derivative of T (θ (t)) and:

S (θ (t)) = S1 (θ (t)) + S2 (θ (t)) (114)

is an unknown input observer for (107)–(108) [156], for which the dynamics of the estimation error
e(t) = x (t) − x̂ (t) is given by:

ė(t) = H (θ (t)) e(t) + R (θ (t)) F (θ (t)) f (t) (115)

In fact, taking into account (107)–(108) and (112)–(113), one finds:

ė(t) = [ A (θ (t)) − S (θ (t)) − T (θ (t)) A (θ (t))] x (t) − H (θ (t)) z(t) (116)

+ [ I − R (θ (t)) − T (θ (t))] B (θ (t)) u(t) + [ I − T (θ (t))] F (θ (t)) f (t)

which, using (109)–(111), can be rewritten as follows:

ė(t) = [ H (θ (t)) − H (θ (t) T (θ (t)))] x (t) − H (θ (t)) z(t) + R (θ (t)) F (θ (t)) f (t) (117)

Then, it is easy to check that (115) follows from (117) taking into account (113).
The main feature of the estimation error dynamics in (115) is that convergence of e(t) to zero
when f (t) = 0 can be ensured by a proper choice of the matrix H (θ (t)) (for example, as a diagonal
matrix with strictly negative parameter-varying elements on the main diagonal). Moreover, the matrix
R (θ (t)) can be used to constrain the range of the matrix R (θ (t)) F (θ (t)), in such a way that different
directions of e(t) can be assigned to different faults, such that not only fault detection but also fault
isolation can be achieved.
In the last years, one can recognize a trend in research that goes towards robustification of this
technique, which was started by Ref. [157]. For instance, a few recent works have merged UIOs with
interval observers [158–160], in such a way that instead of a single trajectory for the estimation error,
lower and upper bounds which are compatible with the uncertainty are computed. On the other hand,
other works have considered the case in which the scheduling variables are measured inexactly, see for
example, Refs. [128,161,162]. Further improvements have been provided by Ref. [163], who have not
restricted the parameter dependency of the UIO to mimic the one of the system, so that the decoupling
conditions can be relaxed and have also considered the case in which the output equation of the convex
system is not restricted to be parameter-independent.
Processes 2019, 7, 814 22 of 39

4.3. Observer-Based Fault Estimation

As the name itself suggests, in observer-based fault estimation techniques, an observer is used
to estimate the fault as though as if it were another (unmeasurable) state of the system. In order to
exemplify the main idea behind these techniques, let us consider the following convex system:

ẋ (t) = A (θ (t)) x (t) + B (θ (t)) u(t) + F (θ (t)) f (t) (118)

y(t) = Cx (t) (119)

and let us assume that the dynamics of f (t) is described by:

f˙(t) = A f f (t) (120)

with known matrix A f (this assumption can be relaxed using, for example, an interval formulation).
T
Then, it is possible to consider an augmented state x̄ (t) = x (t) T , f (t) T such that the resulting


augmented system is described by:

x̄˙ (t) = Ā (θ (t)) x̄ (t) + B̄ (θ (t)) u(t) (121)

y(t) = C̄x (t) (122)

with: " # " #

A (θ (t)) F (θ (t)) B (θ (t)) h i
Ā (θ (t)) = B̄ (θ (t)) = C̄ = C 0
0 Af 0

Hence, a state observer designed to provide an estimate x̄ˆ (t) of x̄ (t) would provide an estimate f¯(t)
of f (t).
Among recent works developing further this concept, an adaptive polytopic observer which
could estimated time-varying actuator faults was presented in Ref. [164] for convex descriptor systems,
differing from most of other papers which assume generally that the actuator faults are constant.
Sliding mode observers have been investigated by Refs. [165,166], which have considered the case of
erroneous scheduling parameters. The case in which the fault’s frequency content is not distributed
within the whole frequency domain but in a finite interval of frequencies was addressed by Ref. [167]
based on the GKYP lemma. An improvement of the design conditions has been brought by Ref. [168],
which have developed a robust fault estimator via homogeneous polynomially parameter-dependent
Lyapunov functions. It is worth highlighting that, although the majority of the results found in the
literature consider the case of additive faults, some recent work has proposed a switched observer
formulation to estimate actuator multiplicative faults in discrete-time convex systems [135]. Successful
applications of observer-based fault estimation, either using high-fidelity simulations or through
experimental validation, can be found in the areas of aviation [165,166], bioreactors [12], distillation
columns [169], automotive suspension systems [170] and renewable microgrids [171,172].

4.4. Multiple Model Adaptive Estimators

The main idea behind MMAEs is to choose a set of models that represent the possible system
behavior patterns and to obtain the state estimate as a combination of the estimates obtained from
local state observers which run in parallel, each one based on the individual models that match these
patterns [173,174]. The above mentioned combination is achieved as a weighted sum, where the
weights represent the likelihood that the corresponding model is indeed true. Under certain conditions,
the weight associated with the correct model converges to 1, while the other weights converge to 0,
which allows an adaptive identification of the correct model [175]. This approach has been developed
for discrete-time systems and it is exemplified hereafter.
Processes 2019, 7, 814 23 of 39

Following Ref. [176], let us consider the discrete-time convex system:

x (k + 1) = A (ζ (k), θ (k)) x (k) + B (ζ (k), θ (k)) u(k) (123)

y(k) = C (ζ (k), θ (k)) x (k ) (124)

where using standard notation, k ∈ Z denotes a sample. Moreover, ζ (k) denotes an uncertain
parameter, for which a finite set of candidate parameter values {ζ 1 , ζ 2 , . . . , ζ N } is considered.
For the system (123)–(124), state estimation is achieved by means of the following convex
MMAE [176]:

h
x̂ (k) = ∑ pi (k)x̂(k|ζ i ) (125)
i =1
h
ŷ(k) = ∑ pi (k)ŷ(k|ζ i ) (126)
i =1
ζ̂ (k) = ζ i∗ (k) , i∗ (k) = arg max pi (k) (127)
i ∈{1,...,h}

where x̂ (k), ŷ(k) and ζ̂ (k) denote the estimates of the state x (k ), the output y(k ) and the unknown
parameter vector ζ, respectively and pi (k ) are dynamic weights, which can be interpreted as a
time-varying indicator of how likely it is that ζ = ζ i . In (125)–(126), each x̂ (k|ζ i ), ŷ(k|ζ i ) correspond to
local estimates, obtained under the assumption that ζ = ζ i .
The dynamic weights pi (k ) appearing in (125)–(127) can be generated as follows:

p i ( k ) β i ( k ) e − ωi ( k )
p i ( k + 1) = (128)
N
−ω j (k)
∑ p j (k) β j (k)e
j =1

where β i (k) is a positive weighting matrix function and ωi (k ) is the error measuring function, which
describes how different is each local output estimate ŷ(k |ζ i ) from the observed output y(k).
The above described convex MMAE, for which a conceptual scheme is provided in Figure 7,
has some relevant properties. First of all, if pi (0) > 0 ∀i ∈ {1, . . . , h}, it can be proven that all
the weights pi (k) generated by (128) are non-negative, uniformly bounded and contained in [0, 1],
with ∑ih=1 pi (k) = 1, ∀k > 0. Moreover, it can be demonstrated that under some conditions the
parameter estimate ζ̂ (k ) will converge to a value ζ i∗ with pi∗ (k) → 1 as k → ∞, which corresponds to
the local estimate that exhibits the smallest error measuring function.
These properties have been exploited for fault identification purposes by Ref. [177], where it was
shown that a convex MMAE could be used to achieve icing diagnosis in unmanned aerial vehicles
(UAVs) with the relevant feature that information about the icing location could be obtained. In this
case, the idea is to assign different faulty models to different parameters ζ i , in such a way that the
dynamic weights would suggest which model is the one that fits data coming from the sensors the
best. A similar idea was employed in Ref. [178], where a bank of observers, each one corresponding to
a system description taking into account the presence of a particular fault, was used to address the
problem of fault detection and isolation in near-space vehicles (NSVs) with actuator faults.
Processes 2019, 7, 814 24 of 39

𝑢(𝑘) Plant 𝑦(𝑘) 𝑦(𝑘|𝜁1 )

ζ(t) Local state
observer ζ1 𝑝1(𝑘)
𝑦(𝑘|𝜁2 ) 𝑝2(𝑘)
Dynamic
Local state ·
· weights ·
observer ζ2 ·
·

·
· generator 𝑝𝑁 (𝑘)
·
·
𝑦(𝑘|𝜁𝑁 )
Local state
observer ζN

Figure 7. Multiple model adaptive estimator (MMAE) conceptual scheme.

4.5. Sliding Mode Fault Tolerant Control

The robustness of sliding mode control (SMC) against matched uncertainties makes it an
interesting choice for FTC [179], although its direct application is impeded by the loss of regularity of
the sliding mode when complete losses of actuators are considered. In fact, the first applications of
SMC to convex systems were in a fault-free context [180]. The above mentioned problem was solved
for overactuated systems by Ref. [181] by means of control allocation (CA), which provides an effective
mechanism for distributing a virtual control signal among the available actuators.
More specifically, a convex plant subject to actuator faults represented by a diagonal semipositive
definite matrix W (t) with diagonal entries that represent the effectiveness level of actuators was
considered in Ref. [181], as follows:

ẋ (t) = A (θ (t)) x (t) + B f E (θ (t)) W (t)u(t) (129)

where E(θ ) is invertible for all θ ∈ Θ and B f is factored as B f = [ B1 , B2 ] T with B2 B2T = Il , l < nu
and k B2 k > k B1 k, so that B2 represents the dominant contribution of the distribution of the control
action within the channels of the system.
For the system (129), the control law is chosen as (see Figure 8 for an illustrative scheme):
  −1
u(t) = − ( E (θ (t)))−1 B2T B2 E (θ (t)) Ŵ (t) ( E (θ (t)))−1 B2T (νl (t) + νn (t)) (130)

where Ŵ (t) is an estimate of W (t), νl (t) is the linear component of the virtual control, chosen to be a
standard state-feedback νl (t) = − Fx (t) and νn (t) is the nonlinear discontinuous part, which induces
sliding and provides robustness:

σ(t)
νn (t) = −κ (t) for σ (t) 6= 0 (131)
kσ(t)k

where κ (t) is an adaptive modulation function given by:

κ (t) = k F k k x (t)k κ̄ (t) + η (132)

κ̄˙ (t) = − βκ̄ (t) + γe0 k F k k x (t)k kσ (t)k (133)
Processes 2019, 7, 814 25 of 39

with β, γ, e0 , η positive scalars and σ defines the sliding surface σ (t) = 0:

" # ! !
B1 B2T
  −1 Z t
σ(t) = B2 B Tf B f B Tf x ( t ) − x (0) − A (θ (τ )) − F x (τ )dτ (134)
0 Il

Finally, Ref. [145] shows that if F is designed such that the fault-free closed-loop system is quadratically
stable, then it is possible to prove that for any faults/failures inside the set:

We = {(w1 , . . . , wnu ) ∈ [0, 1] × . . . × [0, 1] : H (θ )T H (θ ) > eI } (135)

with e small scalar which satisfies 0 < e  1 and H (θ ) = B2 E(θ )W (t) ( E(θ ))−1 B2T , the sliding motion
will be stable if:  
c
γ0 γ1 1 + √ <1 (136)
e

where γ0 is the L2 gain of G̃ (s) = F (sI − A(θ ) + Bν F )−1 [ In−l , 0] T , γ1 = | B1 | (which, by assumption
is small) and c represents the worst-case condition number (over Θ) of E(θ ).
The design of the state-feedback controller in Ref. [181] was based on the assumption that all the
plant states are available. This assumption was later relaxed by Ref. [182], where an unknown input
observer (UIO) was used to estimate the unavailable plant states. Further research has led to develop
some conditions based on the Lyapunov-Krasovskii functional approach that do not only guarantee
the passivity and asymptotical stability of the closed-loop system but also cover the issue of actuator
saturation and the existence of time-varying delays [183].

𝑊(𝑡)
𝑢(𝑡) 𝑥(𝑡)
Actuators Plant

FDI

෡
𝑊(𝑡) 𝑣𝑙 (𝑡) Baseline
Controller F
Control
Allocation
Sliding mode
𝑣𝑛 (𝑡) control

Figure 8. Sliding mode fault tolerant control (FTC) scheme.

4.6. Fault Tolerant Control Based on Controller Reconfiguration

On the other hand, as the name itself suggests, in controller reconfiguration, some modification of
the control law is performed in order to compensate for the fault effect and make the faulty system
behave as close as possible to the nominal system. The reconfiguration can be performed either by
considering the fault estimation fˆ(t) as an additional scheduling variable θ f (t), as in Refs. [26,184,185]
or by introducing a component in the control law, which is responsible to achieve fault tolerance,
Processes 2019, 7, 814 26 of 39

as proposed by Refs. [186–189]. In the first case, if we restrict our attention to a state-feedback control
law (for the sake of simplicity), then it would have the following structure:
 
u ( t ) = K θ ( t ), θ f ( t ) x ( t ) (137)

with the advantage that fault tolerance would be achieved employing exactly the same LMI-based
techniques employed for standard control design. On the other hand, in the second case, the control
law is obtained as follows (see Figure 9):

u(t) = un (t) + u f (t) (138)

where un (t) is the nominal state-feedback controller in fault-free condition and u f (t) is used to
accommodate the faults. The advantage of this approach lies in that it eases the integrated design of
fault estimator and fault tolerant controller, as discussed deeply in Ref. [188]. In order to illustrate
this fact, let us consider the following convex system, which is a simplification of the class of systems
considered in Ref. [188] by neglecting disturbances and parametric uncertainties in the state matrix:

ẋ (t) = A (θ (t)) x (t) + B (θ (t)) u(t) + F (θ (t)) f a (t) (139)

y(t) = Cx (t) + Fs f s (t) (140)

where f a (t) and f s (t) denote actuator and sensor faults, respectively. By augmenting the state as
x̄ (t) = [ x (t), f a (t), f s (t)] T , the system (139)–(140) becomes:

x̄˙ (t) = Ā (θ (t)) x̄ (t) + B (θ (t)) u(t) (141)

y(t) = C̄ x̄ (t) (142)

for which an observer can be proposed, as follows:

ż(t) = M (θ (t)) z(t) + J (θ (t)) u(t) + L (θ (t)) y(t) (143)

x̂ (t) = z(t) + Hy(t) (144)

where z(t) and x̂ (t) are the observer internal state and the estimate of x̄ (t), respectively. Under the
assumption that:

L (θ (t)) = L1 (θ (t)) + L2 (θ (t)) (145)

L2 (θ (t)) = (Ξ Ā (θ (t)) − L1 (θ (t)) C̄ ) H (146)
M (θ (t)) = Ξ Ā (θ (t)) − L1 (θ (t)) C̄ (147)
J (θ (t)) = Ξ B̄ (θ (t)) (148)

with Ξ = I − H C̄, the dynamics of the estimation error e(t) = x (t) − x̂ (t) is described by:

ė(t) = M (θ (t)) e(t) = [Ξ Ā (θ (t) − L1 (θ (t)) C̄ )] e(t) (149)

If the control law is chosen as:

h i
u(t) = K (θ (t)) x̂ (t) = Kx (θ (t)) K f (θ (t)) x̂ (t) (150)

where Kx (θ (t)) and K f (θ (t)) are the state-feedback and actuator fault compensation gains respectively,
then if K f (θ (t)) is chosen as K f (θ (t)) = − B (θ (t))† F (θ (t)) (under the assumption that the actuator
fault f a (t) is in the range space of the control input), one obtains:
Processes 2019, 7, 814 27 of 39

ẋ (t) = [ A (θ (t)) + B (θ (t)) Kx (θ (t))] x (t) + E (θ (t)) e(t) (151)

with: h i
E (θ (t)) = − B (θ (t)) Kx (θ (t)) F (θ (t)) 0 (152)

On the other hand, the sensor fault can be compensated as:

yc (t) = y(t) − Fs fˆs (t) (153)

where yc (t) is the compensated system output and fˆs (t) is the sensor fault estimate. Since (149) and
(151) describe an autonomous convex system, the integrated FE/FTC design can be formulated as an
LMI-based stabilization problem (H∞ optimization if there are uncertainties and/or disturbances).

𝑓𝑎 (𝑡) 𝑓𝑠 (𝑡)
𝑢(𝑡) 𝑥(𝑡) 𝑦(𝑡) 𝑦𝑐 (𝑡)
Actuators Plant Sensors

𝑓መ𝑠 (𝑡)
Observer
𝑢𝑓 (𝑡) Fault
compensator 𝑓መ𝑎 (𝑡)

Nominal
𝑢𝑛 (𝑡) controller 𝑥(𝑡)
ො
Figure 9. Controller reconfiguration FTC scheme.

4.7. Fault-Hiding via Virtual Actuators and Virtual Sensors

Contrarily to the previously described approach, the fault-hiding paradigm aims at reconfiguring
the faulty plant instead of the controller/observer when a fault occurs [88]. The reconfiguration block
hides the fault from the controller/observer point of view, such that it will see the same plant as
before the fault and thus can be kept without modifying or retuning it (see Figure 10). The advantage
of doing so is that fault tolerance can be added to an already existing control scheme by means of
a plug-and-play philosophy. In case of actuator faults, the reconfiguration block is named virtual
actuator [190], whereas it is named virtual sensor in the case of sensor faults [191]. Although virtual
sensors and virtual actuators were initially considered separately, an overall scheme that employs both
of them in order to tolerate simultaneous actuator and sensor faults was later developed [192,193].
Some recent work has also studied issues related to the existence of input constraints (saturations) and
fault isolation delays [194].
Let us consider the following convex system:

ẋ (t) = A (θ (t)) x (t) + B (θ (t)) W (t) (u(t) + f a (t)) (154)

y(t) = V (t)Cx (t) + f s (t) (155)

where, consistently with the previously described approaches, W (t) and V (t) denote losses of
effectiveness in the actuators and sensors, respectively, whereas f a (t) and f s (t) denote additive
actuator/sensor faults.
Processes 2019, 7, 814 28 of 39

𝑊(𝑡), 𝑓𝑎 (𝑡) 𝑉(𝑡), 𝑓𝑠 (𝑡)

𝑥(𝑡)
𝑢(𝑡) Actuators Plant Sensors

𝑦(𝑡)
෡
𝑊(𝑡), 𝑓መ𝑎 (𝑡) FDI ෠
𝑉(𝑡), 𝑓መ𝑠 (𝑡)
Virtual Virtual
actuator sensor
𝑥𝑣𝑎 (𝑡)

𝑢𝑐 (𝑡) 𝑥(𝑡)
ො State 𝑦𝑐 (𝑡)
Controller
observer

Figure 10. FTC using virtual sensors and virtual actuators.

The structure of the convex virtual actuator depends on the following rank condition (Ŵ denotes
an estimation of W) [192]:

∀θ ∈ Θ



rank B θ (t)Ŵ (t) = rank ( B (θ (t))) (156)

which describes whether fault tolerance can be achieved through a simple redistribution of the control
inputs. In the first case, the reconfiguration structure is as follows:

u(t) = Nva (θ (t)) uc (t) − fˆa (t) (157)

where fˆa (t) is an estimation of f a (t) and Nva (θ (t)) is given by:
 †
Nva (θ (t)) = B (θ (t)) Ŵ (t) B (θ (t)) (158)

In case (156) does not hold, the virtual actuator becomes a dynamical system with state equation:

ẋva (t) = [ A (θ (t)) + B∗ (θ (t)) Mva (θ (t))] xva (t) + [ B (θ (t) − B∗ (θ (t)))] uc (t) (159)

and output equation (reconfiguration structure):

h i
u(t) = Nva (θ (t)) uc (t) − Mva (θ (t)) xva (t) − fˆa (t) (160)

where xva (t) is the virtual actuator state, Mva (θ (t)) denotes the virtual actuator gain and the matrix
B∗ (θ (t)) is obtained as:
B∗ (θ (t)) = B (θ (t)) Ŵ (t) Nva (θ (t)) (161)

Similarly, the structure of the virtual sensor depends on the following rank condition (V̂ is the
estimation of V) [192]:


rank V̂ (t)C = rank(C ) (162)

so that if it holds, then the virtual sensor is a static block, whereas if the above condition does not hold,
then it is a dynamical system with internal state xvs (t) and dynamics described by the equations:
Processes 2019, 7, 814 29 of 39

ẋvs (t) = [ A (θ (t)) + Mvs (θ (t)) C ∗ ] xvs (t) + B (θ (t)) uc (t) (163)
 
− Mvs (θ (t)) Nvs (t) y(t) + V̂ (t)Cxva (t) − fˆs (t)
 
yc (t) = Nvs (t) y(t) + V̂ (t)Cxva (t) − fˆs (t) + (C − C ∗ ) xvs (t) (164)

where fˆs (t) denotes an estimation of f s (t), Mvs (θ (t)) is the virtual sensor gain and:
 †
Nvs (t) = C V̂ (t)C (165)
∗
C = Nvs (t)V̂ (t)C (166)

Then, it is possible to show that if the control loop consists of a state-feedback law with controller
gain K (θ (t)) and a Luenberger observer with gain L (θ (t)), then thanks to the introduction of the
virtual actuator/sensor in the loop, one can find an appropriate similarity transformation of the overall
augmented state such that the dynamics in the new state coordinates x̆ is described by:

A(θ ) + Mvs (θ )C ∗
 
0 0 0
 ? A(θ ) + L(θ )C 0 0 
x̆˙ (t) =   x̆ (t) (167)
 
 ? ? A(θ ) + B(θ )K (θ ) 0 
? ? ? A(θ ) + B∗ (θ ) Mva (θ )

where ? denotes some generic non-zero terms, which is in a block-triangular structure. Then, since
A(θ ) + L(θ )C and A(θ ) + B(θ )K (θ ) are already stable due to the stability of the faultless system,
one can ensure overall stability under fault occurrence by designing the gains Mvs (θ ) and Mva (θ )
so that A(θ ) + Mvs (θ )C ∗ and A(θ ) + B∗ (θ ) Mva (θ ) are stable (the reader is referred to Ref. [192] for
further details on the method, along with a discussion about the quadratic stability of block-triangular
convex systems).
Finally, it is worth mentioning that some recent works have combined model predictive control
(MPC) with the convex formulation in order to take into account possible input and state constraints
associated to actuator saturation and other physical limitations [195,196]. The convex MPC framework
has been used to go one step further than FTC, that is, to perform health-aware control on the basis
of the information about the system reliability provided by a prognosis and health management
(PHM) module [197,198]. This type of control strategy increases the overall reliability, anticipates the
apparition of faults and reduces the operational costs.

5. Conclusions
In this paper, we have performed a review of the most applied techniques in control, observation
and safety of convex systems. With this terminology we have wished to unify the concepts of linear
parameter varying and Takagi-Sugeno systems, with the purpose of allowing the reader to taste all the
flavors of techniques offered by the humongous existing literature about these classes of systems. Due
to the huge amount of papers, the review is in no way meant to be exhaustive but it is meant to be a
helpful document to look for any reader who wishes to locate himself/herself in this field and learn
about the main used techniques. We feel that we have done our best to provide a discussion about
the state-of-the-art of the topic. However, in spite of our best efforts, many publications could not be
included and for this reason, we would like to apologize in advance for any omission.
In addition, it is important to mention that this paper is mainly focused on discussing the
advances of polytopic convex systems. However, it is acknowledged that other approaches that lead to
an LPV representation exist, such as grid-based LPV [199], linear fractional transformation (LFT)-based
LPV [200,201], polynomial LPV approaches [202] and tensor model-based transformation [203,204],
among others. Also, the reader should note that, throughout the review, only methods based on
Processes 2019, 7, 814 30 of 39

quadratic Lyapunov functions have been discussed. Nevertheless, less conservative solutions can be
obtained based on non-quadratic Lyapunov functions, for example, the polyquadratic, as proposed
in Refs. [69–71]. In general, all these topics are currently investigated and, therefore, the above
references are recommended to the interested reader.

Author Contributions: Conceptualization, F.-R.L.-E., D.R., and G.V.-P.; investigation, F.-R.L.-E., D.R., and G.V.-P.;
writing-original draft preparation, F.-R.L.-E., D.R., and G.V.-P.; writing-review and editing F.-R.L.-E., D.R.,
and G.V.-P.
Funding: This work has been supported by Tecnológico Nacional de México grants 6723.18-P and 5400.19-P,
the Instituto de Ciencia Tecnológia e Innovación, Chiapas (ICTIECH) 1123/2019. This work has been partially
funded by the Spanish State Research Agency (AEI) and the European Regional Development Fund (ERFD)
through the projects SCAV (ref. MINECO DPI2017-88403-R) and DEOCS (ref. MINECO DPI2016-76493), and also
by AGAUR ACCIO RIS3CAT UTILITIES 4.0 – P7 SECUTIL. This work has been also supported by the AEI through
the Maria de Maeztu Seal of Excellence to IRI (MDM-2016-0656) and the grant Juan de la Cierva-Formacion
(FJCI-2016-29019).
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Shamma, J.; Athans, M. Guaranteed properties for nonlinear gain scheduled control systems. In Proceedings
of the 27th IEEE Conference on Decision and Control, Austin, TX, USA, 7–9 December 1988; Volume 3, pp.
2202–2208.
2. Shamma, J.; Athans, M. Analysis of gain scheduled control for nonlinear plants. IEEE Trans. Autom. Control
1990, 35, 898–907. [CrossRef]
3. Shamma, J. An Overview of LPV Systems. In Control of Linear Parameter Varying Systems with Applications;
Mohammadpour, J., Scherer, C.W., Eds.; Springer: Boston, MA, USA, 2012; pp. 3–26.
4. Tanaka, K.; Wang, H.O. Fuzzy Control Systems Design and Analysis: A Linear Matrix Inequality Approach; John
Wiley & Sons: Hoboken, NJ, USA, 2004. [CrossRef]
5. Jadbabaie, A.; Jamshidi, M.; Titli, A. Guaranteed-cost design of continuous-time Takagi-Sugeno fuzzy
controllers via linear matrix inequalities. In Proceedings of the IEEE International Conference on Fuzzy
Systems Proceedings, IEEE World Congress on Computational Intelligence, Anchorage, AK, USA, 4–9 May
1998; Volume 1, pp. 268–273.
6. Ohtake, H.; Tanaka, K.; Wang, H.O. Fuzzy modeling via sector nonlinearity concept.
Integr. Comput.-Aided Eng. 2003, 10, 333–341. [CrossRef]
7. Ichalal, D.; Marx, B.; Ragot, J.; Maquin, D. State estimation of Takagi–Sugeno systems with unmeasurable
premise variables. IET Control Theory Appl. 2010, 4, 897–908. [CrossRef]
8. Lendek, Z.; Guerra, T.M.; Babuska, R.; De Schutter, B. Stability Analysis and Nonlinear Observer Design Using
Takagi-Sugeno Fuzzy Models; Springer: Berlin/Heidelberg, Germany, 2011. [CrossRef]
9. López-Estrada, F.R.; Astorga-Zaragoza, C.M.; Theilliol, D.; Ponsart, J.C.; Valencia-Palomo, G.; Torres, L.
Observer synthesis for a class of Takagi–Sugeno descriptor system with unmeasurable premise variable.
Application to fault diagnosis. Int. J. Syst. Sci. 2017, 48, 3419–3430. [CrossRef]
10. Wang, Z.; Li, F.; Qin, Y.; Li, D.; Ma, G.; Ma, J. A Novel Dual Nonlinear Observer for Vehicle System Roll Behavior
With Lateral and Vertical Coupling; SAE Technical Paper; SAE International: Warrendale, PA, USA, 2019.
[CrossRef]
11. Nagy Kiss, A.M.; Marx, B.; Mourot, G.; Schutz, G.; Ragot, J. State estimation of two-time scale multiple
models. Application to wastewater treatment plant. Control Eng. Pract. 2011, 19, 1354–1362. [CrossRef]
12. López-Estrada, F.R.; Ponsart, J.C.; Astorga-Zaragoza, C.M.; Camas-Anzueto, J.L.; Theilliol, D. Robust sensor
fault estimation for descriptor-LPV systems with unmeasurable gain scheduling functions: Application to
an anaerobic bioreactor. Int. J. Appl. Math. Comput. Sci. 2015, 25, 233–244. [CrossRef]
13. Zhao, Z.; Wang, Y.; Liu, F. A multi-way LPV modeling method for batch processes. J. Process Control 2018,
65, 56–67. [CrossRef]
Processes 2019, 7, 814 31 of 39

14. Gómez-Peñate, S.; López-Estrada, F.R.; Valencia-Palomo, G.; Rotondo, D.; Enríquez-Zárate, J. Actuator
and sensor fault estimation based on a proportional-integral quasi-LPV observer with inexact scheduling
parameters. In Proceedings of the 3rd IFAC Workshop on Linear Parameter-Varying Systems, Eindhoven,
The Netherlands, 4–6 November 2019.
15. Lopez-Estrada, F.R.; Astorga-Zaragoza, C.M.; Valencia-Palomo, G.; Rios-Rojas, C.; Galicia-Gonzalez, C.;
Escobar-Gomez, E. Observer-based LPV stabilization system for a riderless bicycle. IEEE Latin Am. Trans.
2018, 16, 1076–1083. [CrossRef]
16. López-Estrada, F.R.; Theilliol, D.; Astorga-Zaragoza, C.M.; Ponsart, J.C.; Valencia-Palomo, G.;
Camas Anzueto, J. Fault diagnosis observer for descriptor Takagi-Sugeno systems. Neurocomputing
2019, 331, 10–17. [CrossRef]
17. Pfifer, H.; Moreno, C.P.; Theis, J.; Kotikapuldi, A.; Gupta, A.; Takarics, B.; Seiler, P. Linear parameter
varying techniques applied to aeroservoelastic aircraft: In memory of Gary Balas. IFAC-PapersOnLine 2015,
48, 103–108. [CrossRef]
18. Huang, B.; Lu, B.; Li, Q. A proportional–integral-based robust state-feedback control method for linear
parameter-varying systems and its application to aircraft. Proc. Inst. Mech. Eng. Part G 2019, 233, 4663–4675.
[CrossRef]
19. Gutjahr, B.; Gröll, L.; Werling, M. Lateral vehicle trajectory optimization using constrained linear
time-varying MPC. IEEE Trans. Intell. Transp. Syst. 2016, 18, 1586–1595. [CrossRef]
20. Gómez-Peñate, S.; López-Estrada, F.R.; Valencia-Palomo, G.; Osornio-Ríos, R.; Zepeda-Hernández, J.;
Rios-Rojas, C.; Camas-Anzueto, J. Sensor fault diagnosis observer for an electric vehicle modeled as a
Takagi-Sugeno system. J. Sens. 2018, 2018, 3291639. [CrossRef]
21. Chadli, M.; Borne, P. Multiple Models Approach in Automation: Takagi-Sugeno Fuzzy Systems; Wiley Online
Library: London, UK, 2013. [CrossRef]
22. Rotondo, D. Advances in Gain-Scheduling and Fault Tolerant Control Techniques; Springer: Cham, Switzerland,
2017. [CrossRef]
23. Bernal, M.; Estrada, V.; Márquez, R. Diseño e Implementación de Sistemas de Control Basados en Estructuras
Convexas Y Desigualdades Matriciales Lineales; Pearson: México city, Mexico, 2019.
24. Rodrigues, M.; Hamdi, H.; BenHadj-Braiek, N.; Theilliol, D. Observer-based fault tolerant control design for
a class of LPV descriptor systems. J. Frankl. Inst. 2014, 351, 3104–3125. [CrossRef]
25. Rotondo, D.; Nejjari, F.; Puig, V. Quasi-LPV modeling, identification and control of a twin rotor MIMO
system. Control Eng. Pract. 2013, 21, 829–846. [CrossRef]
26. Rotondo, D.; Nejjari, F.; Puig, V. Robust quasi–LPV model reference FTC of a quadrotor UAV subject to
actuator faults. Int. J. Appl. Math. Comput. Sci. 2015, 25, 7–22. [CrossRef]
27. He, D.F.; Huang, H.; Chen, Q.X. Quasi-min–max MPC for constrained nonlinear systems with guaranteed
input-to-state stability. J. Frankl. Inst. 2014, 351, 3405–3423. [CrossRef]
28. He, Z.; Zhao, L. Quadrotor trajectory tracking based on quasi-LPV system and internal model control.
Math. Probl. Eng. 2015, 2015, 857291. [CrossRef]
29. Rizzello, G.; Naso, D.; Turchiano, B.; Seelecke, S. Robust position control of dielectric elastomer actuators
based on LMI optimization. IEEE Trans. Control Syst. Technol. 2016, 24, 1909–1921. [CrossRef]
30. Rizzello, G.; Ferrante, F.; Naso, D.; Seelecke, S. Robust interaction control of a dielectric elastomer actuator
with variable stiffness. IEEE/ASME Trans. Mech. 2017, 22, 1705–1716. [CrossRef]
31. Pérez-Estrada, A.J.; Osorio-Gordillo, G.L.; Darouach, M.; Alma, M.; Olivares-Peregrino, V.H. Generalized
dynamic observers for quasi-LPV systems with unmeasurable scheduling functions. Int. J. Robust
Nonlinear Control 2018, 28, 5262–5278. [CrossRef]
32. Robles, R.; Sala, A.; Bernal, M. Performance-oriented quasi-LPV modeling of nonlinear systems. Int. J.
Robust Nonlinear Control 2019, 29, 1230–1248. [CrossRef]
33. Baranyi, P. Extracting LPV and qLPV structures from state-space functions: A TP model transformation
based framework. IEEE Trans. Fuzzy Syst. 2019. [CrossRef]
34. Rotondo, D.; Puig, V.; Nejjari, F.; Witczak, M. Automated generation and comparison of Takagi–Sugeno and
polytopic quasi-LPV models. Fuzzy Sets Syst. 2015, 277, 44–64. [CrossRef]
35. López-Estrada, F.R.; Ponsart, J.C.; Theilliol, D.; Zhang, Y.; Astorga-Zaragoza, C.M. LPV model-based
tracking control and robust sensor fault diagnosis for a quadrotor UAV. J. Intell. Robot. Syst. 2016, 84, 163–177.
[CrossRef]
Processes 2019, 7, 814 32 of 39

36. Ramírez, M.; Villafuerte, R.; González, T.; Bernal, M. Exponential estimates of a class of time–delay nonlinear
systems with convex representations. Int. J. Appl. Math. Comput. Sci. 2015, 25, 815–826. [CrossRef]
37. Arceo, J.C.; Vázquez, D.; Estrada-Manzo, V.; Márquez, R.; Bernal, M. Nonlinear convex control of the Furuta
pendulum based on its descriptor model. In Proceedings of the 13th International Conference on Electrical
Engineering, Computing Science and Automatic Control (CCE), Mexico City, Mexico, 26–30 September 2016;
pp. 1–6.
38. Quintana, D.; Estrada-Manzo, V.; Bernal, M. A methodology for real-time implementation of nonlinear
observers via convex optimization. In Proceedings of the 15th International Conference on Electrical
Engineering, Computing Science and Automatic Control (CCE), Mexico City, Mexico, 5–7 September 2018;
pp. 1–6.
39. Arceo, J.C.; Sánchez, M.; Estrada-Manzo, V.; Bernal, M. Convex stability analysis of nonlinear singular
systems via linear matrix inequalities. IEEE Trans. Autom. Control 2018, 64, 1740–1745. [CrossRef]
40. López-Estrada, F.R.; Hernández-de León, H.R.; Estrada-Manzo, V.; Bernal, M. LMI-based fault detection and
isolation of nonlinear descriptor systems. In Proceedings of the IEEE International Conference on Fuzzy
Systems (FUZZ-IEEE), Naples, Italy, 9–12 July 2017; pp. 1–5.
41. Hoffmann, C.; Werner, H. A survey of linear parameter-varying control applications validated by
experiments or high-fidelity simulations. IEEE Trans. Control Syst. Technol. 2014, 23, 416–433. [CrossRef]
42. Apkarian, P.; Biannic, J.M.; Gahinet, P. Self-scheduled H∞ control of missile via linear matrix inequalities.
J. Guid. Control Dyn. 1995, 18, 532–538. [CrossRef]
43. Rugh, W.J.; Shamma, J.S. Research on gain scheduling. Automatica 2000, 36, 1401–1425. [CrossRef]
44. Shamma, J.S.; Athans, M. Gain scheduling: Potential hazards and possible remedies. IEEE Control Syst. Mag.
1992, 12, 101–107.
45. Rotondo, D.; Nejjari, F.; Puig, V. Robust state-feedback control of uncertain LPV systems: An LMI-based
approach. J. Frankl. Inst. 2014, 351, 2781–2803. [CrossRef]
46. Sato, M. Gain-scheduled output-feedback controllers depending solely on scheduling parameters via
parameter-dependent Lyapunov functions. Automatica 2011, 47, 2786–2790. [CrossRef]
47. Sato, M.; Peaucelle, D. Gain-scheduled output-feedback controllers using inexact scheduling parameters for
continuous-time LPV systems. Automatica 2013, 49, 1019–1025. [CrossRef]
48. Kajiwara, H.; Apkarian, P.; Gahinet, P. LPV Techniques for Control of an Inverted Pendulum. IEEE Control
Syst. 1999, 19, 44–54.
49. Bruzelius, F.; Breitholtz, C.; Pettersson, S. LPV-based gain scheduling technique applied to a turbo fan
engine model. In Proceedings of the International Conference on Control Applications, Glasgow, UK, 18–20
September 2002; pp. 713–718.
50. Scherer, C.W. Robust Mixed Control and LPV Control with Full Block Scaling; Technical Report; Delft University
of Technology, Mechanical Engineering Systems and Control Group: Delft, The Netherlands, 2004.
51. Mohammadpour, J.; Scherer, C. Control of Linear Parameter Varying Systems with Applications; Springer: New
York, NY, USA, 2012. [CrossRef]
52. Xu, H.E.; Jun, Z.; M, D.G.; Chao, C. Switching control for LPV polytopic systems using multiple Lyapunov
functions. In Proceedings of the 30th Chinese Control Conference, Yantai, China, 22–24 July 2011;
pp. 1771–1776.
53. Xu, H.E.; Jun, Z.; Dimirovski, G.M. A blending method control of switched LPV systems with slow-varying
parameters and its application to an F-16 aircraft model. In Proceedings of the 30th Chinese Control
Conference, Yantai, China, 22–24 July 2011; pp. 1765–1770.
54. Shin, J.; Balas, G.; Kaya, A.M. Blending methodology of linear parameter varying control synthesis of F-16
aircraft system. J. Guid. Control Dyn. 2002, 25, 1040–1048. [CrossRef]
55. Xie, W.; Kamiya, Y.; Eisaka, T. Robust control system design for polytopic stable LPV systems. IMA J. Math.
Control Inf. 2003, 20, 201–216. [CrossRef]
56. Qiu, J.; Feng, G.; Gao, H. Fuzzy-Model-Based Piecewise H∞ Static-Output-Feedback Controller Design for
Networked Nonlinear Systems. IEEE Trans. Fuzzy Syst. 2010, 18, 919–934. [CrossRef]
57. Yin, X.; Zhang, L.; Zhu, Y.; Wang, C.; Li, Z. Robust control of networked systems with variable communication
capabilities and application to a semi-active suspension system. IEEE/ASME Trans. Mech. 2016, 21, 2097–2107.
[CrossRef]
Processes 2019, 7, 814 33 of 39

58. Shamma, J.S. Analysis and Design of Gain Scheduled Control Systems. Ph.D. Thesis, Massachusetts Institute
of Technology, Cambridge, MA, USA, 1988.
59. Shamma, J.S.; Athans, M. Guaranteed properties of gain scheduled control for linear parameter-varying
plants. Automatica 1991, 27, 559–564. [CrossRef]
60. Goebel, R.; Hu, T.; Teel, A.R. Dual matrix inequalities in stability and performance analysis of linear
differential/difference inclusions. In Current Trends in Nonlinear Systems and Control; Springer: Boston, MA,
USA, 2006; pp. 103–122. [CrossRef]
61. Apkarian, P.; Gahinet, P.; Becker, G. Self-scheduled H∞ control of linear parameter-varying systems: A design
example. Automatica 1995, 31, 1251–1261. [CrossRef]
62. Pandey, A.; Sehr, M.; de Oliveira, M. Pre-filtering in gain-scheduled and robust control. In Proceedings of
the American Control Conference (ACC), Boston, MA, USA, 6–8 July 2016; pp. 3698–3703.
63. Sehr, M.A.; de Oliveira, M.C. Pre-filtering and post-filtering in gain-scheduled output-feedback control.
Int. J. Robust Nonlinear Control 2017, 27, 3259–3279. [CrossRef]
64. Sala, A.; Arino, C. Asymptotically necessary and sufficient conditions for stability and performance in fuzzy
control: Applications of Polya’s theorem. Fuzzy Sets Syst. 2007, 158, 2671–2686. [CrossRef]
65. Kruszewski, A.; Sala, A.; Guerra, T.M.; Ariño, C. A triangulation approach to asymptotically exact conditions
for fuzzy summations. IEEE Trans. Fuzzy Syst. 2009, 17, 985–994. [CrossRef]
66. Sala, A.; Arino, C. Relaxed stability and performance LMI conditions for Takagi–Sugeno fuzzy systems
with polynomial constraints on membership function shapes. IEEE Trans. Fuzzy Syst. 2008, 16, 1328–1336.
[CrossRef]
67. Tuan, H.D.; Apkarian, P.; Narikiyo, T.; Yamamoto, Y. Parameterized linear matrix inequality techniques in
fuzzy control system design. IEEE Trans. Fuzzy Syst. 2001, 9, 324–332. [CrossRef]
68. Wang, H.O.; Tanaka, K.; Griffin, M. Parallel distributed compensation of nonlinear systems by Takagi-Sugeno
fuzzy model. In Proceedings of the 1995 IEEE International Conference on Fuzzy Systems, Yokohama, Japan,
20–24 March 1995; Volume 2, pp. 531–538.
69. Wang, L.K.; Zhang, H.G.; Liu, X.D. H∞ Observer Design for Continuous-Time Takagi-Sugeno Fuzzy
Model with Unknown Premise Variables via Nonquadratic Lyapunov Function. IEEE Trans. Cybern. 2016,
46, 1986–1996. [CrossRef]
70. Márquez, R.; Guerra, T.M.; Bernal, M.; Kruszewski, A. A non-quadratic Lyapunov functional for H∞ control
of nonlinear systems via Takagi-Sugeno models. J. Frankl. Inst. 2016, 353, 781–796. [CrossRef]
71. Márquez, R.; Guerra, T.M.; Bernal, M.; Kruszewski, A. Asymptotically necessary and sufficient conditions for
Takagi–Sugeno models using generalized non-quadratic parameter-dependent controller design. Fuzzy Sets
Syst. 2017, 306, 48–62. [CrossRef]
72. Pandey, A.; de Oliveira, M.C. Quadratic and poly-quadratic discrete-time stabilizability of linear
parameter-varying systems. IFAC-PapersOnLine 2017, 50, 8624–8629. [CrossRef]
73. Lam, H.K.; Wu, L.; Zhao, Y. Linear matrix inequalities-based membership-function-dependent stability
analysis for non-parallel distributed compensation fuzzy-model-based control systems. IET Control
Theory Appl. 2014, 8, 614–625. [CrossRef]
74. Cherifi, A.; Guelton, K.; Arcese, L. Quadratic design of d-stabilizing non-pdc controllers for quasi-lpv/ts
models. IFAC-PapersOnLine 2015, 48, 164–169. [CrossRef]
75. Guerra, T.M.; Bernal, M.; Guelton, K.; Labiod, S. Non-quadratic local stabilization for continuous-time
Takagi–Sugeno models. Fuzzy Sets Syst. 2012, 201, 40–54. [CrossRef]
76. Daafouz, J.; Bernussou, J. Parameter dependent Lyapunov functions for discrete time systems with time
varying parametric uncertainties. Syst. Control Lett. 2001, 43, 355–359. [CrossRef]
77. Chadli, M.; Daafouz, J.; Darouach, M. Stabilisation of singular LPV systems. IFAC Proc. Vol. 2008,
41, 9999–10002. [CrossRef]
78. Pandey, A.P.; de Oliveira, M.C. On the necessity of LMI-based design conditions for discrete time LPV filters.
IEEE Trans. Autom. Control 2018, 63, 3187–3188. [CrossRef]
79. El Ghaoui, L.; Oustry, F.; AitRami, M. A cone complementarity linearization algorithm for static
output-feedback and related problems. IEEE Trans. Autom. Control 1997, 42, 1171–1176. [CrossRef]
80. Prempain, E.; Postlethwaite, I. Static H∞ loop shaping control of a fly-by-wire helicopter. Automatica 2005,
41, 1517–1528. [CrossRef]
Processes 2019, 7, 814 34 of 39

81. Henrion, D.; Lasserre, J.B. Convergent relaxations of polynomial matrix inequalities and static output
feedback. IEEE Trans. Autom. Control 2006, 51, 192–202. [CrossRef]
82. Apkarian, P.; Noll, D. Nonsmooth H∞ synthesis. IEEE Trans. Autom. Control 2006, 51, 71–86. [CrossRef]
83. Scherer, C.; Gahinet, P.; Chilali, M. H∞ design with pole placement constraints: An LMI approach. IEEE Trans.
Autom. Control 1996, 41, 358–367.
84. Gahinet, P. Explicit controller formulas for LMI-based H∞ synthesis. Automatica 1996, 32, 1007–1014.
[CrossRef]
85. Amato, F. Robust Control of Linear Systems Subject to Uncertain Time-Varying Parameters; Springer:
Berlin/Heidelberg, Germany, 2006; Volume 325. [CrossRef]
86. Kose, I.E.; Jabbari, F. Control of LPV systems with partly measured parameters. IEEE Trans. Autom. Control
1999, 44, 658–663. [CrossRef]
87. Abdullah, A.; Zribi, M. Model reference control of LPV systems. J. Frankl. Inst. 2009, 346, 854–871. [CrossRef]
88. Rotondo, D.; Puig, V.; Nejjari, F.; Romera, J. A fault-hiding approach for the switching quasi-LPV
fault-tolerant control of a four-wheeled omnidirectional mobile robot. IEEE Trans. Ind. Electron. 2014,
62, 3932–3944. [CrossRef]
89. Valencia-Palomo, G.; Rossiter, J.A. Auto-tuned predictive control based on minimal plant information.
IFAC Proc. Vol. 2009, 42, 554–559. [CrossRef]
90. Valencia-Palomo, G.; Rossiter, J.; López-Estrada, F. Improving the feed-forward compensator in predictive
control for setpoint tracking. ISA Trans. 2014, 53, 755–766. [CrossRef] [PubMed]
91. Valencia-Palomo, G.; Hilton, K.; Rossiter, J.A. Predictive control implementation in a PLC using the IEC
1131.3 programming standard. In Proceedings of the 2009 European Control Conference (ECC), Budapest,
Hungary, 23–26 August 2009; pp. 1317–1322.
92. Kothare, M.V.; Balakrishnan, V.; Morari, M. Robust constrained model predictive control using linear matrix
inequalities. Automatica 1996, 32, 1361–1379. [CrossRef]
93. Lu, Y.; Arkun, Y. Quasi-min-max MPC algorithms for LPV systems. Automatica 2000, 36, 527–540. [CrossRef]
94. Park, P.; Jeong, S.C. Constrained RHC for LPV systems with bounded rates of parameter variations.
Automatica 2004, 40, 865–872. [CrossRef]
95. Lu, Y.; Arkun, Y. Polytope updating in quasi-min-max MPC algorithms. IFAC Proc. Vol. 2000, 33, 407–412.
[CrossRef]
96. Lu, Y.; Arkun, Y. A scheduling quasi–min-max model predictive control algorithm for nonlinear systems.
J. Process Control 2002, 12, 589–604. [CrossRef]
97. Lee, S.; Won, S. Model predictive control for linear parameter varying systems using a new parameter
dependent terminal weighting matrix. IEICE Trans. Fund. Electron. Commun. Comput. Sci. 2006, 89, 2166–2172.
[CrossRef]
98. Wada, N.; Saito, K.; Saeki, M. Model predictive control for linear parameter varying systems using parameter
dependent Lyapunov function. IEEE Trans. Circuits Syst. II 2006, 12, 1446–1450. [CrossRef]
99. Pluymers, B.; Rossiter, J.; Suykens, J.; De Moor, B. The efficient computation of polyhedral invariant sets for
linear systems with polytopic uncertainty. In Proceedings of the American Control Conference, Portland,
OR, USA, 8–10 June 2005; pp. 804–809. [CrossRef]
100. Garone, E.; Casavola, A. Receding horizon control strategies for constrained LPV systems based on a class of
nonlinearly parameterized Lyapunov functions. IEEE Trans. Autom. Control 2012, 57, 2354–2360. [CrossRef]
101. Yu, S.; Böhm, C.; Chen, H.; Allgöwer, F. Model predictive control of constrained LPV systems. Int. J. Control
2012, 85, 671–683. [CrossRef]
102. Besselmann, T.; Lofberg, J.; Morari, M. Explicit MPC for LPV systems: Stability and optimality. IEEE Trans.
Autom. Control 2012, 57, 2322–2332. [CrossRef]
103. Zhang, J.; Xiu, X. Kd tree based approach for point location problem in explicit model predictive control.
J. Frankl. Inst. 2018, 355, 5431–5451. [CrossRef]
104. Ariño, C.; Querol, A.; Sala, A. Shape-independent model predictive control for Takagi–Sugeno fuzzy systems.
Eng. Appl. Artif. Intell. 2017, 65, 493–505. [CrossRef]
105. Hanema, J.; Lazar, M.; Tóth, R. Stabilizing tube-based model predictive control: Terminal set and cost
construction for LPV systems. Automatica 2017, 85, 137–144. [CrossRef]
106. Ding, B.; Wang, P.; Hu, J. Dynamic output feedback robust MPC with one free control move for LPV model
with bounded disturbance. Asian J. Control 2018, 20, 755–767. [CrossRef]
Processes 2019, 7, 814 35 of 39

107. Morato, M.M.; Nguyen, M.Q.; Sename, O.; Dugard, L. Design of a fast real-time LPV model predictive
control system for semi-active suspension control of a full vehicle. J. Frankl. Inst. 2019, 356, 1196–1224.
[CrossRef]
108. Ghersin, A.S.; Pena, R.S.S. Applied LPV control with full block multipliers and regional pole placement.
J. Control Sci. Eng. 2010, 2010, 3. [CrossRef]
109. Ostertag, E. Mono-and Multivariable Control and Estimation: Linear, Quadratic and LMI Methods; Springer:
Berlin/Heidelberg, Germany, 2011. [CrossRef]
110. Rotondo, D.; Puig, V.; Nejjari, F. Linear quadratic control of LPV systems using static and shifting
specifications. In Proceedings of the European Control Conference (ECC), Linz, Austria, 15–17 July 2015;
pp. 3085–3090.
111. Briat, C. Stability analysis and control of a class of LPV systems with piecewise constant parameters.
Syst. Control Lett. 2015, 82, 10–17. [CrossRef]
112. Bruzelius, F.; Pettersson, S.; Breitholtz, C. Region of attraction estimates for LPV-gain scheduled control
systems. In Proceedings of the European Control Conference (ECC), Cambridge, UK, 1–4 September 2003;
pp. 892–897. [CrossRef]
113. Pitarch, J.L.; Sala, A.; Arino, C.V. Closed-form estimates of the domain of attraction for nonlinear systems
via fuzzy-polynomial models. IEEE Trans. Cybern. 2013, 44, 526–538. [CrossRef]
114. Lendek, Z.; Lauber, J. Local stability of discrete-time TS fuzzy systems. IFAC-PapersOnLine 2016, 49, 7–12.
[CrossRef]
115. Lendek, Z.; Nagy, Z.; Lauber, J. Local stabilization of discrete-time TS descriptor systems. Eng. Appl.
Artif. Intell. 2018, 67, 409–418. [CrossRef]
116. Zhang, X.; Tsiotras, P.; Knospe, C. Stability analysis of LPV time-delayed systems. Int. J. Control 2002,
75, 538–558. [CrossRef]
117. Wu, F.; Grigoriadis, K.M. LPV systems with parameter-varying time delays: Analysis and control. Automatica
2001, 37, 221–229. [CrossRef]
118. Briat, C.; Sename, O.; Lafay, J.F. Memory-resilient gain-scheduled state-feedback control of uncertain
LTI/LPV systems with time-varying delays. Syst. Control Lett. 2010, 59, 451–459. [CrossRef]
119. Briat, C. Linear Parameter-Varying and Time-Delay Systems: Analysis, Observation, Filtering Control; Springer:
Berlin/Heidelberg, Germany, 2014; Volume 3. [CrossRef]
120. Guzmán-Rabasa, J.A.; López-Estrada, F.R.; González-Contreras, B.M.; Valencia-Palomo, G.; Chadli, M.;
Pérez-Patricio, M. Actuator fault detection and isolation on a quadrotor unmanned aerial vehicle modeled
as a linear parameter-varying system. Meas. Control 2019. [CrossRef]
121. Zhang, H.; Zhang, G.; Wang, J. H∞ Observer Design for LPV Systems With Uncertain Measurements
on Scheduling Variables: Application to an Electric Ground Vehicle. IEEE/ASME Trans. Mech. 2016,
21, 1659–1670. [CrossRef]
122. Li, H.; Wu, C.; Yin, S.; Lam, H.K. Observer-based fuzzy control for nonlinear networked systems under
unmeasurable premise variables. IEEE Trans. Fuzzy Syst. 2015, 24, 1233–1245. [CrossRef]
123. Ibrir, S.; Sabir, A. Robust observer-based stabilization and tracking of uncertain linear systems with L2-gain
performance: Application to DC motors. In Proceedings of the 2016 IEEE International Energy Conference
(ENERGYCON), Leuven, Belgium, 4–8 April 2016; pp. 1–6. [CrossRef]
124. Gauterin, E.; Kammerer, P.; Kühn, M.; Schulte, H. Effective wind speed estimation: Comparison between
Kalman Filter and Takagi–Sugeno observer techniques. ISA Trans. 2016, 62, 60–72. [CrossRef]
125. Brizuela-Mendoza, J.A.; Astorga-Zaragoza, C.M.; Zavala-Río, A.; Pattalochi, L.; Canales-Abarca, F. State and
actuator fault estimation observer design integrated in a riderless bicycle stabilization system. ISA Trans.
2016, 61, 199–210. [CrossRef]
126. Bergsten, P.; Palm, R.; Driankov, D. Observers for Takagi-Sugeno fuzzy systems. IEEE Trans. Syst. Man
Cybern. Part B 2002, 32, 114–121. [CrossRef]
127. Zhang, L.; Yin, X.; Ning, Z.; Ye, D. Robust filtering for a class of networked nonlinear systems with switching
communication channels. IEEE Trans. Cybern. 2016, 47, 671–682. [CrossRef]
128. Gómez-Peñate, S.; Valencia-Palomo, G.; López-Estrada, F.R.; Astorga-Zaragoza, C.M.; Osornio-Rios, R.A.;
Santos-Ruiz, I. Sensor fault diagnosis based on a sliding mode and unknown input observer for
Takagi-Sugeno systems with uncertain premise variables. Asian J. Control 2019, 21, 339–353. [CrossRef]
Processes 2019, 7, 814 36 of 39

129. Busawon, K.K.; Kabore, P. Disturbance attenuation using proportional integral observers. Int. J. Control
2001, 74, 618–627. [CrossRef]
130. Youssef, T.; Chadli, M.; Karimi, H.R.; Zelmat, M. Design of unknown inputs proportional integral observers
for TS fuzzy models. Neurocomputing 2014, 123, 156–165. [CrossRef]
131. Kang, D. Design of a disturbance observer for discrete-time linear systems. In Proceedings of the 14th
International Conference on Control, Automation and Systems (ICCAS), Seoul, South Korea, 22–25 October
2014; pp. 1381–1383.
132. Xu, J.; Mi, C.C.; Cao, B.; Deng, J.; Chen, Z.; Li, S. The state of charge estimation of lithium-ion batteries based
on a proportional-integral observer. IEEE Trans. Veh. Technol. 2013, 63, 1614–1621.
133. Youssef, T.; Chadli, M.; Karimi, H.R.; Wang, R. Actuator and sensor faults estimation based on proportional
integral observer for TS fuzzy model. J. Frankl. Inst. 2017, 354, 2524–2542. [CrossRef]
134. Rotondo, D.; Cristofaro, A.; Johansen, T.A.; Nejjari, F.; Puig, V. Detection of icing and actuators faults in
the longitudinal dynamics of small UAVs using an LPV proportional integral unknown input observer.
In Proceedings of the 3rd Conference on Control and Fault-Tolerant Systems (SysTol), Barcelona, Spain, 7–9
Deptember 2016; pp. 690–697. [CrossRef]
135. Rotondo, D.; López-Estrada, F.R.; Nejjari, F.; Ponsart, J.C.; Theilliol, D.; Puig, V. Actuator multiplicative
fault estimation in discrete-time LPV systems using switched observers. J. Frankl. Inst. 2016, 353, 3176–3191.
[CrossRef]
136. Duan, G.R. Analysis and Design of Descriptor Linear Systems; Springer: New York, NY, USA, 2010. [CrossRef]
137. Masubuchi, I.; Kato, J.; Saeki, M.; Ohara, A. Gain-scheduled controller design based on descriptor
representation of LPV systems: Application to flight vehicle control. In Proceedings of the 43rd IEEE
Conference on Decision and Control (CDC), Nassau, Bahamas, 14–17 December 2004; Volume 1, pp. 815–820.
138. Baghaee, H.R.; Mirsalim, M.; Gharehpetian, G.B.; Talebi, H.A. A generalized descriptor-system robust H∞
control of autonomous microgrids to improve small and large signal stability considering communication
delays and load nonlinearities. Int. J. Electr. Power Energy Syst. 2017, 92, 63–82. [CrossRef]
139. González, A.; Estrada-Manzo, V.; Guerra, T.M. Gain-scheduled H∞ admissibilisation of LPV discrete-time
systems with LPV singular descriptor. Int. J. Syst. Sci. 2017, 48, 3215–3224. [CrossRef]
140. Arceo, J.C.; Villafuerte, R.; Estrada-Manzo, V.; Bernal, M. LMI-Based Controller Design for Time-Delay
Nonlinear Descriptor Systems with Guaranteed Exponential Estimates. IFAC-PapersOnLine 2018, 51, 585–590.
[CrossRef]
141. González, A.; Guerra, T.M. Enhanced Predictor-Based Control Synthesis for Discrete-Time TS Fuzzy
Descriptor Systems With Time-Varying Input Delays. IEEE Trans. Fuzzy Syst. 2018, 27, 402–410. [CrossRef]
142. Guerra, T.M.; Estrada-Manzo, V.; Lendek, Z. Observer design for Takagi–Sugeno descriptor models: An LMI
approach. Automatica 2015, 52, 154–159. [CrossRef]
143. Estrada-Manzo, V.; Lendek, Z.; Guerra, T.M. Generalized LMI observer design for discrete-time nonlinear
descriptor models. Neurocomputing 2016, 182, 210–220. [CrossRef]
144. Hamdi, H.; Rodrigues, M.; Mechmeche, C.; Theilliol, D. Fault detection and isolation for linear parameter
varying descriptor systems via proportional integral observer. Int. J. Adapt. Control Signal Proc. 2012,
26, 224–240. [CrossRef]
145. Alwi, H.; Edwards, C. Fault tolerant longitudinal aircraft control using non-linear integral sliding mode.
IET Control Theory Appl. 2014, 8, 1803–1814. [CrossRef]
146. Gertler, J. Fault Detection and Diagnosis; John Wiley & Sons, Ltd.: Hoboken, NJ, USA, 2008. [CrossRef]
147. Wei, X.; Verhaegen, M. LMI solutions to the mixed H∞ /H− fault detection observer design for linear
parameter-varying systems. Int. J. Adapt. Control Signal Proc. 2011, 25, 114–136. [CrossRef]
148. Chadli, M.; Abdo, A.; Ding, S.X. H− /H∞ fault detection filter design for discrete-time Takagi-Sugeno fuzzy
system. Automatica 2013, 49, 1996–2005. [CrossRef]
149. Chibani, A.; Chadli, M.; Shi, P.; Braiek, N.B. Fuzzy fault detection filter design for T–S fuzzy systems in the
finite-frequency domain. IEEE Trans. Fuzzy Syst. 2016, 25, 1051–1061. [CrossRef]
150. Wang, Z.; Shi, P.; Lim, C.C. H− /H∞ fault detection observer in finite frequency domain for linear
parameter-varying descriptor systems. Automatica 2017, 86, 38–45. [CrossRef]
151. Chibani, A.; Chadli, M.; Ding, S.X.; Braiek, N.B. Design of robust fuzzy fault detection filter for polynomial
fuzzy systems with new finite frequency specifications. Automatica 2018, 93, 42–54. [CrossRef]
Processes 2019, 7, 814 37 of 39

152. Iwasaki, T.; Hara, S. Generalized KYP lemma: Unified frequency domain inequalities with design
applications. IEEE Trans. Autom. Control 2005, 50, 41–59. [CrossRef]
153. Estrada, F.L.; Ponsart, J.C.; Theilliol, D.; Astorga-Zaragoza, C.M. Robust H− /H∞ fault detection observer
design for descriptor-LPV systems with unmeasurable gain scheduling functions. Int. J. Control 2015,
88, 2380–2391. [CrossRef]
154. Darouach, M.; Zasadzinski, M.; Xu, S.J. Full-order observers for linear systems with unknown inputs.
IEEE Trans. Autom. Control 1994, 39, 606–609. [CrossRef]
155. Rotondo, D.; Witczak, M.; Puig, V.; Nejjari, F.; Pazera, M. Robust unknown input observer for state and fault
estimation in discrete-time Takagi–Sugeno systems. Int. J. Syst. Sci. 2016, 47, 3409–3424. [CrossRef]
156. Rotondo, D.; Cristofaro, A.; Johansen, T.A.; Nejjari, F.; Puig, V. Icing detection in unmanned aerial vehicles
with longitudinal motion using an LPV unknown input observer. In Proceedings of the Conference on
Control Applications (CCA), Sydney, Australia, 21–23 September 2015; pp. 984–989. [CrossRef]
157. Chadli, M.; Karimi, H.R. Robust observer design for unknown inputs Takagi–Sugeno models. IEEE Trans.
Fuzzy Syst. 2012, 21, 158–164. [CrossRef]
158. Meyer, L.; Ichalal, D.; Vigneron, V. Interval observer for LPV systems with unknown inputs. IET Control
Theory Appl. 2017, 12, 649–660. [CrossRef]
159. Rotondo, D.; Cristofaro, A.; Johansen, T.A.; Nejjari, F.; Puig, V. State estimation and decoupling of unknown
inputs in uncertain LPV systems using interval observers. Int. J. Control 2018, 91, 1944–1961. [CrossRef]
160. Rotondo, D.; Cristofaro, A.; Johansen, T.A.; Nejjari, F.; Puig, V. Robust fault and icing diagnosis in unmanned
aerial vehicles using LPV interval observers. Int. J. Robust Nonlinear Control 2018. [CrossRef]
161. Hassanabadi, A.H.; Shafiee, M.; Puig, V. Actuator fault diagnosis of singular delayed LPV systems with
inexact measured parameters via PI unknown input observer. IET Control Theory Appl. 2017, 11, 1894–1903.
[CrossRef]
162. Xu, F.; Tan, J.; Wang, Y.; Wang, X.; Liang, B.; Yuan, B. Robust fault detection and set-theoretic UIO for
discrete-time LPV systems with state and output equations scheduled by inexact scheduling variables.
IEEE Trans. Autom. Control 2019. [CrossRef]
163. Marx, B.; Ichalal, D.; Ragot, J.; Maquin, D.; Mammar, S. Unknown input observer for LPV systems. Automatica
2019, 100, 67–74. [CrossRef]
164. Rodrigues, M.; Hamdi, H.; Theilliol, D.; Mechmeche, C.; BenHadj Braiek, N. Actuator fault estimation based
adaptive polytopic observer for a class of LPV descriptor systems. Int. J. Robust Nonlinear Control 2015,
25, 673–688. [CrossRef]
165. Chandra, K.P.B.; Alwi, H.; Edwards, C. Fault detection in uncertain LPV systems with imperfect scheduling
parameter using sliding mode observers. Eur. J. Control 2017, 34, 1–15. [CrossRef]
166. Chen, L.; Edwards, C.; Alwi, H. Sensor fault estimation using LPV sliding mode observers with erroneous
scheduling parameters. Automatica 2019, 101, 66–77. [CrossRef]
167. Zhang, K.; Jiang, B.; Shi, P.; Xu, J. Analysis and design of robust H− /H∞ fault estimation observer with
finite-frequency specifications for discrete-time fuzzy systems. IEEE Trans. Cybern. 2014, 45, 1225–1235.
[CrossRef] [PubMed]
168. Xie, X.; Yue, D.; Zhang, H.; Xue, Y. Fault estimation observer design for discrete-time Takagi–Sugeno fuzzy
systems based on homogenous polynomially parameter-dependent Lyapunov functions. IEEE Trans. Cybern.
2017, 47, 2504–2513. [CrossRef] [PubMed]
169. Lopez-Estrada, F.R.; Ponsart, J.C.; Theilliol, D.; Astorga-Zaragoza, C.; Flores-Montiel, M. Robust state and
fault estimation observer for discrete-time D-LPV systems with unmeasurable gain scheduling functions.
Application to a binary distillation column. IFAC-PapersOnLine 2015, 48, 1012–1017. [CrossRef]
170. Morato, M.M.; Sename, O.; Dugard, L.; Nguyen, M.Q. Fault estimation for automotive Electro-Rheological
dampers: LPV-based observer approach. Control Eng. Pract. 2019, 85, 11–22. [CrossRef]
171. Morato, M.M.; Mendes, P.R.; Normey-Rico, J.E.; Bordons, C. Robustness conditions of LPV fault estimation
systems for renewable microgrids. Int. J. Electr. Power Energy Syst. 2019, 111, 325–350. [CrossRef]
172. Morato, M.M.; Regner, D.J.; Mendes, P.R.; Normey-Rico, J.E.; Bordons, C. Fault analysis, detection and
estimation for a microgrid via H2 /H∞ LPV observers. Int. J. Electr. Power Energy Syst. 2019, 105, 823–845.
[CrossRef]
173. Li, X.R.; Bar-Shalom, Y. Multiple-model estimation with variable structure. IEEE Trans. Autom. Control 1996,
41, 478–493.
Processes 2019, 7, 814 38 of 39

174. Hassani, V.; Aguiar, A.P.; Athans, M.; Pascoal, A.M. Multiple model adaptive estimation and model
identification usign a minimum energy criterion. In Proceedings of the American Control Conference (ACC),
St. Louis, MO, USA, 10–12 June 2009; pp. 518–523. [CrossRef]
175. Xiong, K.; Wei, C.; Liu, L. Robust multiple model adaptive estimation for spacecraft autonomous navigation.
Aerosp. Sci. Technol. 2015, 42, 249–258. [CrossRef]
176. Rotondo, D.; Hassani, V.; Cristofaro, A. A multiple model adaptive architecture for the state estimation in
discrete-time uncertain LPV systems. In Proceedings of the American Control Conference (ACC), Seattle,
WA, USA, 24–26 May 2017; pp. 2393–2398.
177. Rotondo, D.; Cristofaro, A.; Hassani, V.; Johansen, T.A. Icing diagnosis in unmanned aerial vehicles using an
LPV multiple model estimator. IFAC-PapersOnLine 2017, 50, 5238–5243. [CrossRef]
178. Yang, G.H.; Wang, H. Fault detection and isolation for a class of uncertain state-feedback fuzzy control
systems. IEEE Trans. Fuzzy Syst. 2014, 23, 139–151. [CrossRef]
179. Alwi, H.; Edwards, C.; Tan, C.P. Fault Detection and Fault-Tolerant Control Using Sliding Modes; Springer:
London, UK, 2011. [CrossRef]
180. Sivrioglu, S.; Nonami, K. Sliding mode control with time-varying hyperplane for AMB systems. IEEE/ASME
Trans. Mech. 1998, 3, 51–59. [CrossRef]
181. Alwi, H.; Edwards, C.; Stroosma, O.; Mulder, J.; Hamayun, M.T. Real-time implementation of an ISM
fault-tolerant control scheme for LPV plants. IEEE Trans. Ind. Electron. 2014, 62, 3896–3905.
182. Hamayun, M.T.; Ijaz, S.; Bajodah, A.H. Output integral sliding mode fault tolerant control scheme for LPV
plants by incorporating control allocation. IET Control Theory Appl. 2017, 11, 1959–1967. [CrossRef]
183. Selvaraj, P.; Kaviarasan, B.; Sakthivel, R.; Karimi, H.R. Fault-tolerant SMC for Takagi–Sugeno fuzzy systems
with time-varying delay and actuator saturation. IET Control Theory Appl. 2017, 11, 1112–1123. [CrossRef]
184. Shin, J.Y.; Wu, N.E.; Belcastro, C. Adaptive linear parameter varying control synthesis for actuator failure.
J. Guid. Control Dyn. 2004, 27, 787–794. [CrossRef]
185. Sloth, C.; Esbensen, T.; Stoustrup, J. Robust and fault-tolerant linear parameter-varying control of wind
turbines. Mechatronics 2011, 21, 645–659. [CrossRef]
186. Jia, Q.; Chen, W.; Zhang, Y.; Li, H. Fault reconstruction and fault-tolerant control via learning observers in
Takagi–Sugeno fuzzy descriptor systems with time delays. IEEE Trans. Ind. Electron. 2015, 62, 3885–3895.
[CrossRef]
187. Li, X.; Zhu, F.; Chakrabarty, A.; Żak, S.H. Nonfragile fault-tolerant fuzzy observer-based controller design
for nonlinear systems. IEEE Trans. Fuzzy Syst. 2016, 24, 1679–1689. [CrossRef]
188. Lan, J.; Patton, R.J. Integrated design of fault-tolerant control for nonlinear systems based on fault estimation
and T–S fuzzy modeling. IEEE Trans. Fuzzy Syst. 2016, 25, 1141–1154. [CrossRef]
189. Li, X.; Lu, D.; Zeng, G.; Liu, J.; Zhang, W. Integrated fault estimation and non-fragile fault-tolerant control
design for uncertain Takagi–Sugeno fuzzy systems with actuator fault and sensor fault. IET Control
Theory Appl. 2017, 11, 1542–1553. [CrossRef]
190. Rotondo, D.; Nejjari, F.; Puig, V.; Blesa, J. Model reference FTC for LPV systems using virtual actuators and
set-membership fault estimation. Int. J. Robust Nonlinear Control 2015, 25, 735–760. [CrossRef]
191. Nazari, R.; Seron, M.M.; De Doná, J.A. Fault-tolerant control of systems with convex polytopic linear
parameter varying model uncertainty using virtual-sensor-based controller reconfiguration. Annu. Rev.
Control 2013, 37, 146–153. [CrossRef]
192. Rotondo, D.; Nejjari, F.; Puig, V. A virtual actuator and sensor approach for fault tolerant control of LPV
systems. J. Process Control 2014, 24, 203–222. [CrossRef]
193. Tabatabaeipour, S.M.; Stoustrup, J.; Bak, T. Fault-tolerant control of discrete-time LPV systems using virtual
actuators and sensors. Int. J. Robust Nonlinear Control 2015, 25, 707–734. [CrossRef]
194. Rotondo, D.; Ponsart, J.C.; Nejjari, F.; Theilliol, D.; Puig, V. Virtual actuator-based FTC for LPV systems with
saturating actuators and FDI delays. In Proceedings of the 3rd Conference on Control and Fault-Tolerant
Systems (SysTol), Barcelona, Spain, 7–9 September 2016; pp. 831–837.
195. Witczak, P.; Luzar, M.; Witczak, M.; Korbicz, J. A robust fault-tolerant model predictive control for linear
parameter-varying systems. In Proceedings of the 19th International Conference on Methods and Models in
Automation and Robotics (MMAR), Miedzyzdroje, Poland, 2–5 September 2014; pp. 462–467.
196. Acevedo-Valle, J.M.; Puig, V.; Tornil-Sin, S.; Witczak, M.; Rotondo, D. Predictive Fault Tolerant Control for
LPV systems using model reference. IFAC-PapersOnLine 2015, 48, 30–35. [CrossRef]
Processes 2019, 7, 814 39 of 39

197. Pour, F.K.; Puig, V.; Cembrano, G. Health-aware LPV-MPC based on system reliability assessment for
drinking water networks. In Proceedings of the IEEE Conference on Control Technology and Applications
(CCTA), Copenhagen, Denmark, 21–24 August 2018; pp. 187–192.
198. Pour, F.K.; Puig, V.; Cembrano, G. Health-aware LPV-MPC based on a reliability-based remaining useful life
assessment. IFAC-PapersOnLine 2018, 51, 1285–1291. [CrossRef]
199. Wu, F.; Yang, X.H.; Packard, A.; Becker, G. Induced L2-norm control for LPV systems with bounded
parameter variation rates. Int. J. Robust Nonlinear Control 1996, 6, 983–998. [CrossRef]
200. Packard, A. Gain scheduling via linear fractional transformations. Syst. Control Lett. 1994, 22, 79–92.
[CrossRef]
201. Veenman, J.; Scherer, C.W. Stability analysis with integral quadratic constraints: A dissipativity based proof.
In Proceedings of the 52nd IEEE Conference on Decision and Control, Florence, Italy, 10–13 December 2013;
pp. 3770–3775.
202. Brizuela-Mendoza, J.; Sorcia-Vázquez, F.J.; Guzmán-Valdivia, C.H.; Osorio-Sánchez, R.; Martínez-García, M.
Observer design for sensor and actuator fault estimation applied to polynomial LPV systems: A riderless
bicycle study case. Int. J. Syst. Sci. 2018, 49, 2996–3006. [CrossRef]
203. Baranyi, P.; Yam, Y.; Várlaki, P. Tensor Product Model Transformation in Polytopic Model-Based Control; CRC
Press: Boca Raton, FL, USA, 2013.
204. Takarics, B.; Baranyi, P. Tensor-product-model-based control of a three degrees-of-freedom aeroelastic model.
J. Guid. Control Dyn. 2013, 36, 1527–1533. [CrossRef]

c 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).