mathematics

Article
Approximation-Based Quantized State Feedback Tracking of
Uncertain Input-Saturated MIMO Nonlinear Systems with
Application to 2-DOF Helicopter
Byung Mo Kim and Sung Jin Yoo *

School of Electrical and Electronics Engineering, Chung-Ang University, 84 Heukseok-Ro, Dongjak-Gu,

Seoul 06974, Korea; [email protected]
* Correspondence: [email protected]

Abstract: This paper addresses an approximation-based quantized state feedback tracking problem
of multiple-input multiple-output (MIMO) nonlinear systems with quantized input saturation. A
uniform quantizer is adopted to quantize state variables and control inputs of MIMO nonlinear
systems. The primary features in the current development are that (i) an adaptive neural network
tracker using quantized states is developed for MIMO nonlinear systems and (ii) a compensation
mechanism of quantized input saturation is designed by constructing an auxiliary system. An
adaptive neural tracker design with the compensation of quantized input saturation is developed by
deriving an augmented error surface using quantized states. It is shown that closed-loop stability
analysis and tracking error convergence are conducted based on Lyapunov theory. Finally, we





 give simulation and experimental results of the 2-degrees-of-freedom (2-DOF) helicopter system for
verifying to the validity of the proposed methodology where the tracking performance of pitch and
Citation: Kim, B.M.; Yoo, S.J.
Approximation-Based Quantized
yaw angles is measured with the mean squared errors of 0.1044 and 0.0435 for simulation results,
State Feedback Tracking of Uncertain and those of 0.0656 and 0.0523 for experimental results.
Input-Saturated MIMO Nonlinear
Systems with Application to 2-DOF Keywords: quantized feedback control; state and input quantization; input saturation; MIMO
Helicopter. Mathematics 2021, 9, 1062. nonlinear systems; 2-DOF helicopter
https://doi.org/10.3390/math
9091062

Academic Editor: Rami Ahmad 1. Introduction

El-Nabulsi
In industrial systems, the operating ranges of actuators are restricted because of the
physical limitation and specification [1]. The control input saturation closely influences
Received: 24 March 2021
the performance of the control system and the stability of closed-loop systems [2]. There-
Accepted: 6 May 2021
Published: 9 May 2021
fore, considerable attention has been devoted to control uncertain nonlinear system in
the presence of input saturation. In [3], an adaptive control design approach using the
hyperbolic tangent function and the Nussbaum function was presented to compensate for
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
saturation nonlinearities of uncertain nonlinear systems. First-order-filter-based auxiliary
published maps and institutional affil-
systems were introduced to analyze the effect of input saturation in uncertain nonlinear
iations. systems such as nonlinear strict-feedback systems [4] and nonlinear stochastic systems [5].
Auxiliary systems using high-order filters were constructed to design adaptive controllers
for input-saturated nonlinear systems with model uncertainties such as nonlinear stochastic
systems [6] and nonlinear strict-feedback systems [7]. By combining these approaches us-
ing auxiliary systems with the function approximation technique, some study results were
Copyright: © 2021 by the authors.
recently developed for various uncertain nonlinear systems in strict-feedback and pure-
Licensee MDPI, Basel, Switzerland.
This article is an open access article
feedback forms. In [8], an observer-based adaptive fuzzy tracking controller was designed
distributed under the terms and
for nonlinear systems with time delay and input saturation. In [9], a disturbance-observer-
conditions of the Creative Commons based adaptive fuzzy control problem was investigated for nonlinear state constrained
Attribution (CC BY) license (https:// systems with input saturation. A robust adaptive control approach was proposed for state-
creativecommons.org/licenses/by/ constrained nonlinear systems with input saturation and unknown control direction [10].
4.0/). Neural-network-based adaptive control problems of pure-feedback nonlinear systems

Mathematics 2021, 9, 1062. https://doi.org/10.3390/math9091062 https://www.mdpi.com/journal/mathematics

Mathematics 2021, 9, 1062 2 of 16

were addressed using a prescribed performance control technique [11] and a function
transformation technique [12]. Furthermore, these adaptive control results were applied to
multiple-input multiple-output (MIMO) nonlinear systems with various applications to
deal with more practical systems. In [13], an adaptive tracking control was proposed for a
class of uncertain MIMO nonlinear systems with non-symmetric input constraints. In [14],
the problem of finite-time adaptive fuzzy tracking control was investigated for MIMO
nonlinear systems with input saturation. An adaptive neural tracking control problem
was studied for MIMO stochastic nonlinear systems with input saturation [15]. In [16],
an adaptive backstepping output feedback tracking problem was considered for MIMO
stochastic pure-feedback nonlinear systems with input saturation. An adaptive neural
tracking control approach was proposed for MIMO pure-feedback time-delay nonlinear
systems with input saturation [17]. In [18], an upper limb robotic exoskeleton with input
saturation was considered for an adaptive control design. However, the aforementioned
study efforts cannot be applied to an adaptive control problem of MIMO nonlinear sys-
tems under capacity-limited network environment with state quantization. The primary
challenge of this problem is how to deal with the multi-input saturation problem using
quantized states in the adaptive control structure and the effects of quantization errors on
the system performance.
Network-based control allows reducing reconfiguration and maintenance cost of the
controller and improving the control efficiency [19]. Since the digital network resources
are limited in the practical communication environment, signal quantization that aims
to map a continuous signal into a discrete set has been widely studied for the control
problems of nonlinear systems. For a variety of nonlinear systems with input quantization,
many interesting study results have been presented. In [20], an adaptive backstepping
stabilization problem was considered for nonlinear uncertain systems with input quantiza-
tion. Adaptive asymptotic tracking control problems were studied for nonlinear uncertain
systems with input quantization [21] and actuator faults [22]. In [23], an output feedback
control approach was presented for uncertain nonlinear systems with input quantization.
In [24], an adaptive backstepping quantized control problem was addressed for a class of
nonlinear systems. Furthermore, the input quantization problem has been studied with
the input saturation problem to consider both the physical limitation of actuators and
lower communication rates. In [25], the problem of adaptive output feedback quantised
tracking control was considered for stochastic nonstrict-feedback nonlinear systems with
asymmetric input saturation. An adaptive tracking method was presented for a class
of uncertain nonlinear systems with input quantization and unknown parameters [26].
However, all the developed control schemes [20–26] only focused on the input quantization
problem, namely the state feedback information should be continuously measured for
the controller design. To consider state quantization, quantized-states-based adaptive
control methods have been recently studied for uncertain nonlinear systems. In [27], an
adaptive control problem of matched nonlinear systems was addressed via the backstep-
ping technique. In [28], a command-filtered-based recursive design was introduced for
uncertain unmatched nonlinear systems to overcome the discontinuity problem of the
derivatives of virtual control laws using quantized states. However, nonlinear functions of
systems concerned in [27,28] were assumed to be known and linearly parameterized, i.e.,
adaptive techniques were only used to deal with parametric uncertainties. To relax this
restriction on uncertainties, an approximation-based adaptive tracker in the presence of
state quantization and time delays was designed in [29]. In [30], an event-triggered adap-
tive neural network control approach was studied for nonlinear systems with unknown
nonlinearities. However, the resultant nonlinear adaptive control approaches [25–30]
involve the following restrictions for more improvement.
(I) The quantized-states-based adaptive control approaches [25–30] focused on the
control problems of single-input single-output systems. Since a large number of practical
systems possess multivariable characteristics and the stability analysis of MIMO systems is
much complicated owing to the interconnected nonlinear dynamics between states and
Mathematics 2021, 9, 1062 3 of 16

inputs, the quantized-states-based tracker design methodology for the networked-based

control of MIMO nonlinear systems with state quantization needs to be further investigated.
(II) The quantized input saturation problems of uncertain nonlinear systems were
considered without state quantization in [25,26] and the input saturation effects were not
considered in [27–30]. To the authors’ knowledge, there is still no reported work on the
input saturation compensation problem in the presence of state quantization. An input
saturation compensation strategy using quantized states should be derived for uncertain
nonlinear systems with state quantization.
In this work, we propose a quantized-states-based adaptive neural control design
for uncertain MIMO nonlinear systems subject to input saturation that overcomes the
above restrictions (I) and (II). It is assumed that all system nonlinearities are completely
unknown, and the full state variables and control inputs quantized by an uniform quantizer
are transmitted to the controller and the MIMO nonlinear systems, respectively. An
augmented error surface using auxiliary variables is defined to design a neural-network-
based adaptive tracking controller using quantized full state information. The unknown
nonlinear functions and quantization errors are compensated by constructing the adaptive
tuning laws using quantized states. The compensation signal is introduced with the
auxiliary system to attenuate the quantized multi-input saturation influence. Based on
the Lyapunov stability theory and some bounding lemmas, the stability of the proposed
quantized feedback system is successfully analyzed with the convergence of the tracking
error. For a practical application of the proposed theoretical result, we simulate and
experiment a 2-degrees-of-freedom (2-DOF) helicopter system.
The rest of the paper is structured as follows. The approximation-based quantized state
feedback tracking problem of MIMO nonlinear systems with quantized input saturation is
formulated in Section 2. The proposed adaptive quantized control design and its stability
analysis are discussed in Section 3. Section 4 introduces a mathematical model of the 2-DOF
helicopter system and its simulation and experimental results are presented. Finally, the
conclusion is given in Section 5.

2. Problem Statement
Consider a class of uncertain MIMO nonlinear systems with quantized input satura-
tion represented by

ẋi = xi+1 , i = 1, . . . , n − 1
ẋn = f ( x̄n ) + g ( x̄n )u(q(v)) (1)
y = x1

where xi = [ xi,1 , . . . , xi,m ]> ∈ Rm is the ith state vector, x̄n = [ x1> , . . . , xn> ]> ∈ Rnm , y =
[y1 , . . . , ym ]> ∈ Rm denotes the system output vector, f (·) ∈ Rm and g (·) ∈ Rm×m are the
unknown smooth function vector and matrix, respectively, v = [v1 , . . . , vm ]> ∈ Rm is the
actual control input vector, q(v) = [q(v1 ), . . . , q(vm )]> ∈ Rm is the quantized signal of v,
and u(q(v)) = [u(q(v1 )), . . . , u(q(vm ))]> ∈ Rm is the quantized input saturation.

Assumption 1 ([31]). The matrix g (·) satisfies 0 < g ≤ |λ( g (·))| ≤ ḡ, ∀ x̄n ∈ Ω x̄n where g > 0
and ḡ > 0 are unknown constants, λ is the eigenvalue operator, and Ω x̄n is a compact set.

The Assumption 1 implies that g (·) is strictly either positive or negative definite.
Without losing generality, it is assumed that g (·) > 0.

Assumption 2. The desired signal xd (t) = [ x1,d , . . . xm,d ] ∈ Rm is continuously differentiable

up to the nth order, bounded, and available for the controller design.
Mathematics 2021, 9, 1062 4 of 16

In this paper, the network-based control environment with state and input quanti-
zation is considered for the system (1). As the input and state quantizer, the uniform
quantizer is selected by the following function

l
≤ h < Lι + 2l

 Lι ,
 Lι − 2
q
h , q(h) = 0, − 2l
≤ h < 2l (2)

− Lι , − Lι − 2l ≤ h < − Lι + l

2

where h = xi,j , v j with i = 1, . . . , n and j = 1, . . . , m, ι ∈ Z+ , l means the length of the

quantization interval, L1 = l, and Lι+1 = Lι + l. Hence, the state and input quantization
q q
errors δxi,j , xi,j − xi,j and δv j , v j − v j satisfy |δxi,j | ≤ l and |δv j | ≤ l, respectively [32].

q q q
Assumption 3 ([27]). The quantized state vector xi = [ xi,1 , . . . , xi,m ]> , i = 1, . . . , n, is available
for feedback instead of the state vector xi .

Remark 1. According to quantization levels, there exist various quantizers such as uniform
quantizer, hysteresis-uniform quantizer, logarithmic-uniform quantizer, etc. In this paper, we use
the uniform quantizer (2) for the simple analysis and implementation, as states and input quantizers.
However, the uniform quantizers can be easily replaced with other quantizers in our control design.
q
The saturation of the jth quantized input u(v j ) is described by
( q q
q q sign(v j )u j,M , |v j | ≥ u j,M
u(v j ) = sat(v j ) = q q (3)
vj , |v j | < u j,M

q
where j = 1, . . . , m, u j,M is the saturation bound of v j , and sign(·) denotes the sign function.

Lemma 1 ([33]). For any e > 0 and s ∈ R, it is ensured that 0 ≤ |s| − s tanh(s/e) ≤ 0.2785e.

Problem 1. The aim of this study is to design a quantized-states-based adaptive control vector v
for uncertain MIMO nonlinear systems (1) ensuring that the output y follows the desired signal xd
in the presence of quantized input saturation.

3. Adaptive Neural Tracking Control in the Presence of State Quantization and

Quantized Input Saturation
3.1. Quantized-States-Based Adaptive Tracker Design Using Neural Networks
To achieve the presented adaptive tracking control objective, we define an augmented
error vector s given by
 n −1 n −1 
n−1
 
d
s=
dt
+Λ z1 = ∑ k
Λk zn−k (4)
k =0

( i −1) ( n −1)
where s = [s1 , . . . , sm ]> ∈ Rm , zi = xi − xd , i = 1, . . . , n − 1, zn = xn − xd −
n −1
ΨTanhφ, Λ = diag[λ1 , . . . , λm ] with constants λ j > 0, j = 1, . . . , m, and ( k ) denote
binomial coefficients. Here, Ψ = diag[ψ1 , . . . , ψm ]; ψj > 0, j = 1, . . . , m, are constants,
Tanhφ = [tanh φ1 , . . . , tanh φm ], and φj , j = 1, . . . , m, are the compensation variables to
deal with the influence of input saturation to be designed later.

Remark 2. Compared with the existing works related to the input saturation [8–18], the compen-
( n −1)
sation variable vector φ employed in the error surface zn = xn − xd − ΨTanhφ is provided
by an auxiliary system using the quantized state feedback information in order to overcome the
quantized input saturation problem. Furthermore, the tanh function form is used to ensure the
boundedness of the compensation signal φ.
Mathematics 2021, 9, 1062 5 of 16

From (4), the time derivative of s is

n −1 
n−1

∑
(n)
ṡ = Λk zn−k+1 + f ( x̄) + g ( x̄)u(vq ) − ΨCosh I (φ)φ̇ − xd (5)
k =1
k

q q
where Cosh I (φ) = diag[1/cosh2 (φ1 ), . . . , 1/cosh2 (φm )] and u(vq ) = [u(v1 ), . . . , u(vm )]> .
A Lyapunov function is defined as V = (1/2)s> s. Then, the time derivative of V
along (2) and (5) becomes
 n −1 
n−1
 
∑
(n)
V̇ = s> Λk zn−k+1 + f ( x̄) + g ( x̄)u(vq ) − ΨCosh I (φ)φ̇ − xd . (6)
k =1
k

Here, unknown nonlinear function vector f ( x̄) can be approximated via the universal
approximation property of radial basis function neural networks (RBFNNs) [34] in the
compact set f ⊂ Rnm as follows:

f ( x̄) = W > Q( x̄) + ε( x̄), x̄ ∈ f (7)

where W = diag[W 1 , . . . , W m ] ∈ RmM×m is the ideal bounded weighting matrix, W j =

[Wj,1 , . . . , Wj,M ]> , j = 1, . . . , m, satisfying kW j k ≤ W̄j with constants W̄j > 0, M is the
number of neural nodes, ε ∈ Rm represents an approximation reconstruction error such
that kεk ≤ ε̄ with an unknown constant ε̄ > 0, Q = [ Q1> , . . . , Q> m]
> ∈ RmM denotes
M
the Gaussian function vector with Q j ∈ R , j = 1, . . . , m. The vector Q j is bounded
as k Q j k ≤ Q̄ j with a constant Q̄ j > 0 from the inherent property of Gaussian basis
functions [35,36].
By employing (7) to estimate f ( x̄) and defining an un-quantized signal v̆, (6) becomes
 n −1 
n−1

V̇ = s> ∑ k Λk zn−k+1 + W > Q(x̄) + ε(x̄) + g (x̄)u(vq ) + v̆ − v
k =1

(n)
+ (v − v̆) − ΨCosh I (φ)φ̇ − xd . (8)

Then, v̆ is chosen as
n −1 
n−1

>
∑
(n)
v̆ = −ζs − Λk zn−k+1 + xd − Ŵ Q( x̄) − B̂Tanh(s/e) (9)
k =1
k

where Tanh(s/e) = [tanh(s1 /e1 ), . . . , tanh(sm /em )]> ∈ Rm , ζ = diag[ζ 1 , . . . , ζ m ]; ζ j > 0,

e j > 0, j = 1, . . . , m, are design constants, Ŵ = diag[Ŵ 1 , . . . , Ŵ m ]; Ŵ j = [Ŵj,1 , . . . , Ŵj,M ]> ,
j = 1, . . . , m, and B̂ are estimates of W j and B, respectively. Here, the positive unknown
constant B is derived later.
Applying (9) into (8) gives
>
V̇ = −s> ζs − s> W̃ Q( x̄) + s> ε( x̄) + s> ( g ( x̄)u(vq ) − v − ΨCosh I (φ)φ̇)
+ s> (v − v̆) − B̃s> Tanh(s/e) − Bs> Tanh(s/e) (10)

where W̃ = Ŵ − W and B̃ = B̂ − B are the estimation errors.

To construct a quantized-states-based actual control law v, a quantized-states-based
augmented error s∗ is defined as

n −1 
n−1

s∗ = ∑ k
Λk z∗n−k (11)
k =0
Mathematics 2021, 9, 1062 6 of 16

where the error surfaces zi∗ , i = 1, . . . , n, with quantized state variables are given by

q ( i −1)
zi∗ = xi − xd , i = 1, . . . , n − 1,
(12)
q ( n −1)
z∗n = xn − xd − ΨTanhφ
q q q
where x j = [ x j,1 , . . . , x j,m ]> , j = 1, . . . , m, and the saturation compensation vector
φ = [φ1 , . . . , φm ]> is provided by the following auxiliary system using quantized states
q q
φ̇j = (cosh2 φj )(−κ j tanh φj + u(v j ) − v j )/ψj , φj (0) = 0, j = 1, . . . , m, (13)

with a design constant κ j > 0.

Based on the quantized-states-based augmented error s∗ , we propose an actual control
law v with adaptation laws for Ŵ j and B̂ as follows:

n −1 
n−1

>
∑
(n)
v = −ζs∗ − Λk z∗n−k+1 + xd − Ŵ Q( x̄q ) − B̂Tanh(s∗ /e) (14)
k =1
k
˙ = Γ (s∗ Q ( x̄q ) − σ |s∗ |Ŵ ),
Ŵ j = 1, . . . , m (15)
j W,j j j W j j

B̂˙ = Γ B (s∗> Tanh(s∗ /e) − σB ks∗ k B̂) (16)

q q
where x̄q = [( x1 )> , . . . , ( xn )> ]> , ΓW,j = diag[γw,j,1 , . . . , γw,j,M ]; γw,j,i > 0, i = 1, . . . , M,
and Γ B > 0 are tuning gains, σW > 0 and σB > 0 are small constants for σ-modification,
and s∗j is the jth element of s∗ .
Substituting (13) into (10) yields
>
V̇ = −s> ζs − s> W̃ Q( x̄) + s> ε( x̄) + s> ( g ( x̄)u(vq ) − u(vq ) + κTanhφ − δv )
+ s> (v − v̆) − B̃s> Tanh(s/e) − Bs> Tanh(s/e) (17)

where κ = diag[κ1 , . . . , κm ] and δv = v − vq .

Using Assumption 1, (3), and Young’s inequality, we have

s> ( g ( x̄)u(vq ) − u(vq )) ≤ ksk( ḡ + 1)u M

s> κTanhφ ≤ kskκ̄
√
−s> δv ≤ ksk ml
q q
where κ̄ = ∑m 2
j=1 κ j and u M = ∑m 2
j=1 u j,M . Then, (17) becomes

> √
V̇ ≤ −s> ζs − s> W̃ Q( x̄) + s> ε( x̄) + ksk(( ḡ + 1)u M + κ̄ + ml )
> > >
+ s (v − v̆) − B̃s Tanh(s/e) − Bs Tanh(s/e). (18)

Remark 3. In the previous adaptive control design to deal with quantized input saturation prob-
lem [25,26], the state quantization problem was not considered. Different from [25,26], the quantized
state variables x̄q are used in the proposed control law (14) instead of original states variables x̄.
Thus, the stability analysis for quantization signal errors between original states and quantized
states are necessary. In the proposed quantized-states-based adaptive tracking control structure
shown in Figure 1, the compensation term ΨTanhφ provided by the auxiliary system (13) is adopted
in the augmented error (11) and control law (14) to compensate for multi-input saturation effects.
To compensate for system uncertainties and unknown parameters, the quantized-states-based neu-
>
ral network Ŵ Q( x̄q ) and parameter adaptation laws (15) and (16) are derived for uncertain
MIMO nonlinear systems. The closed-loop stability is analyzed by establishing the boundedenss of
quantization errors in the next subsection.
Mathematics 2021, 9, 1062 7 of 16

Figure 1. Block diagram of the proposed quantized-states-based adaptive tracking system in the
presence of quantized input saturation.

3.2. Quantization Errors and Closed-Loop Stability Analysis

In this section, the stability analysis of the closed-loop system is carried out by Lya-
punov theory.

Lemma 2. For the adaptation law (15), there exists a compact set ΩWj = {W̃ j |kW̃ j k ≤ ΞWj }
with an unknown constant ΞWj such that W̃ j (t) ∈ ΩWj for all t ≥ 0 provided that W̃ j (0) ∈ ΩWj
where j = 1, . . . , m.

−1 >
Proof. Consider a Lyapunov function candidate VWj = (1/2)W̃ j ΓW,j W̃ j . Then, differenti-
ating VWj with respect to time gives

> >
V̇Wj = W̃ j (s∗j Q j ( x̄q ) − σW |s∗j |Ŵ j ) = W̃ j (s∗j Q j ( x̄q ) − σW |s∗j |(W̃ j + W j )). (19)

Since there exist constants W̄j and Q̄ such that kW j k ≤ W̄j and k Q j k ≤ Q̄ j , respectively,
V̇Wj becomes

V̇Wj ≤ kW̃ j k|s∗j |( Q̄ j + σW W̄j − σW kW̃ j k). (20)

From (20), V̇Wj < 0 when kW̃ j k > ΞWj with ΞWj = ( Q̄ j + σW W̄j )/σW . Thus, W̃ j are
bounded within ΩWj . Thus, if W̃ j (0) ∈ ΩWj , it holds that W̃ j (t) ∈ ΩWj for all t ≥ 0.

Lemma 3. For the adaptation law (16). there exists a compact set Ω B = { B̃|| B̃| ≤ Ξ B } with an
unknown constant Ξ B such that B̃(t) ∈ Ω B for all t ≥ 0 provided that B̃(0) ∈ Ω B .
√
Proof. Consider VB = (1/2) B̃2 /Γ B . From kTanh(s∗ /e)k ≤ m and B̂ = B + B̃, V̇B is
represented by
√
V̇B ≤ | B̃|ks∗ k( m + σB B − σB | B̃|). (21)
√
Then, by defining Ξ B , ( m + σB B)/σB and reasoning the proof of Lemma 2, B̃(t) ∈ Ω B ,
∀t ≥ 0 is ensured if B̃(0) ∈ Ω B .

Lemma 4. Define the quantization errors of the augmented error surface and the control input
vector as

δs = s − s∗ , δv̆ = v̆ − v. (22)
Mathematics 2021, 9, 1062 8 of 16

Then, there exist positive constants ∆s and ∆v̆ such that kδs k ≤ ∆s and kδv̆ k ≤ ∆v̆ , respectively.

Proof. Define δQ = Q( x̄) − Q( x̄q ) and δt = Tanh(s/e) − Tanh(s∗ /e). From k Q j k ≤ Q̄ j ,

√
there exists a constant Q̄ such that k√Qk ≤ Q̄. Using kTanh(·)k ≤ m, we obtain the
inequalities kδQ k ≤ 2Q̄ and kδt k ≤ 2 m. Using the property |δxi,j | ≤ l, δs becomes

n −1  n −1 
n−1 n−1
 
kδs k = ∑ k Λ zn−k − ∑ k Λk z∗n−k
k
k =0 k =0
n −1 
n−1

∑ k Λk ( xn−k − xn−k )
q
=
k =0
n −1  √
n−1

≤ ∑ k Λk l m , ∆s . (23)
k =0

Thus, using (9) and (14), δv̆ is obtained as

n −1 
n−1

>
δv̆ = −ζδs − ∑ k
Λk (zn−k+1 − z∗n−k+1 ) − Ŵ δQ − B̂δt . (24)
k =1

Using Lemmas 2 and 3, δv̆ is bounded as

n −1  √ m √
n−1

kδv̆ k ≤ kζ k∆s + ∑ k
kΛkk l m + 2Q̄ ∑ (ΞWj + W̄j ) + 2 m(Ξ B + B) , ∆v̆ . (25)
k =1 j =1

Thus, it is ensured that kδs k ≤ ∆s and kδv̆ k ≤ ∆v̆ .

Based on Lemmas 2–4, the main result of our study is presented as follows.

Theorem 1. Consider the uncertain MIMO nonlinear system (1) with state quantization and
quantized input saturation controlled by the proposed adaptive quantized state feedback tracker
consisting of (13)–(16). Then, all closed-loop signals are uniformly ultimately bounded and the
tracking error z1 converges to an adjustable neighborhood of the origin.

Proof. From (18) and kδv̆ k ≤ ∆v̆ , the time derivative of V is

> √
V̇ ≤ −s> ζs − s> W̃ Q( x̄) + s> ε( x̄) + ksk(( ḡ + 1)u M + κ̄ + ml + ∆v̆ )
> >
− B̃s Tanh(s/e) − Bs Tanh(s/e). (26)

Owing to kεk < ε̄ with a constant ε̄ and the boundedness of W̃ from Lemma 2, (26) becomes

√ m 
V̇ ≤ −s> ζs + ksk ( ḡ + 1)u M + κ̄ + ml + ∆v̆ + Q̄ ∑ ΞWj + ε̄
j =1
> >
− B̃s Tanh(s/e) − Bs Tanh(s/e). (27)
√
By defining B = ( ḡ + 1)u M + κ̄ + ml + ∆v̆ + Q̄ ∑m
j=1 ΞWj + ε̄ and using Lemma 1, we
obtain the following inequality
m m
− Bs> Tanh(s/e) + Bksk ≤ ∑ B(|s j | − s j tanh(s j /ej )) ≤ ∑ 0.2785Bej . (28)
j =1 j =1
Mathematics 2021, 9, 1062 9 of 16

Additionally, using the inequality | B̃s> Tanh(s/e)| ≤ ksk2 /2 + (Ξ B )2 /2 and selecting the
design parameter as ζ = (1/2) I + ζ̄ I with an identity matrix I ∈ Rm×m and a positive
constant ζ̄, (27) can be represented by

V̇ ≤ −ζ̄V + C (29)

where C = (Ξ B )2 /2 + ∑m
j=1 0.2785Be j . Integrating both sides of (29), the following inequal-
ity holds

C
V (t) ≤ V (0)e−ζ̄t + (1 − e−ζ̄t ), ∀t ≥ 0. (30)
ζ̄

This inequality demonstrates that V (t) is eventually bounded within the value C/ζ̄ which
can be reduced arbitrarily small. Thus, all the closed-loop signals are uniformly ultimately
bounded. Since s is bounded, the tracking error vector z1 is also bounded. Furthermore,
the control law v is bounded and thus ku(vq ) − vq k ≤ ς is ensured with a constant ς. From
this fact, we can check the boundedness of the auxiliary system variable φ by defining the
Lyapunov function candidate Vφ = (1/2)φ> Ψφ. Then, the time derivative of Vφ along
(13) is represented by

V̇φ = cosh> φ cosh φ(−φ> κTanhφ + φ> (u(vq ) − vq ))

≤ cosh> φ cosh φ[κ (kφk − φ> tanh φ) − κ kφk + ςkφk] (31)

where cosh φ = [cosh φ1 , . . . , cosh φm ]> and κ = min{κ1 , . . . , κm }.

By choosing κ = ς + κ ∗ with a constant κ ∗ > 0, (31) becomes

V̇φ ≤ cosh> φ cosh φ(0.2785mκ − κ ∗ kφk) (32)

From (32), V̇φ < 0 when kφk > 0.2785mκ/κ ∗ . Thus, owing to φ(0) = 0, φ is bounded
in a compact set Ξφ = {φ(t)|kφ(t)k ≤ 0.2785mκ/κ ∗ }, ∀t ≥ 0.

4. Application to 2-DOF Helicopter

A 2-DOF helicopter system shown in Figure 2 is a twin rotor experiment equipment
for advanced aerospace applications. The helicopter system driven by two DC motor
consists of a main rotor that controls the pitch and a tail rotor that controls the yaw where
each angle is measured by its encoder. The 2-DOF helicopter system is simulated and
experimented to illustrate the effectiveness of the proposed control approach.

Figure 2. Quanser Aero 2-DOF Helicopter [37].

Mathematics 2021, 9, 1062 10 of 16

4.1. Mathematical Model

The dynamics of the 2-DOF helicopter system is represented by the following nonlinear
equations [38]:

( J p + mb lb2 )ϑ̈ = −mb glb cos ϑ − D p ϑ̇ − mb lb2 ϕ̇2 sin ϑ cos ϑ + K pp Vp + K py Vy

(33)
( Jy + mb lb2 cos2 ϑ) ϕ̈ = − Dy ϕ̇ + 2mb lb2 ϑ̇ ϕ̇ sin ϑ cos ϑ + Kyp Vp + Kyy Vy

where ϑ and ϕ denote pitch and yaw angles denoting the outputs of the system, respectively,
ϑ̇ and ϕ̇ denote angular velocities of pitch and yaw angles, respectively, J p and Jy are the
moments of inertia about the pitch and yaw, respectively, mb is the total mass of the body,
lb is the distance between the center of mass and the origin of the body-fixed frame, g is
the gravity acceleration, D p and Dy are the viscous friction coefficients, K pp , K py , Kyp , and
Kyy are the thrust torque constants, and Vp and Vy are the input voltages injected to the DC
motors for controlling two propellers. The system parameters are given in Table 1.

Table 1. Parameters of the 2-DOF Helicopter system [38].

Symbol Value SI Unit

mb 1.0750 kg· m2
g 9.8065 m/s2
lb 0.002 m
Jp 0.0215 kg· m2
Jy 0.0237 kg· m2
Dp 0.0071 N/V
Dy 0.0220 N/V
K pp 0.0220 N· m/V
K py 0.0221 N· m/V
Kyp −0.0227 N· m/V
Kyy 0.0022 N· m/V

By considering the quantized input saturation and defining the variables x1,1 = ϑ,
x1,2 = ϕ, x2,1 = ϑ̇, x2,2 = ϕ̇, v1 = Vp , and v2 = Vy . Then, the system (33) can be rewritten
in the MIMO nonlinear form (1):

ẋ1 = x2
ẋ2 = f ( x̄2 ) + g ( x̄2 )u(q(v)) (34)
y = x1

where x1 = [ x1,1 , x1,2 ]> , x2 = [ x2,1 , x2,2 ]> , v = [v1 , v2 ]> , and
2 2
 
−mb glb cos x1,1 − D p x2,1 −mlb x2,2 sin x1,1 cos x1,1  K pp K py

J p +mb lb2 J p +mb lb2 J p +mb lb2
f = , g= .
 
Kyp Kyy

− Dy x2,2 +2mb lb2 x2,1 x2,2 sin x1,1 cos x1,1
2 2
Jy +mb lb cos x1,1 Jy +mb lb2 cos2 x1,1 2
Jy +mb lb cos2 x1,1

Notice that the unknown function matrix g in system (34) satisfies Assumption 1.

4.2. Design of Quantized-States-Based Adaptive Tracker

Define the augmented error using quantized states as

s∗ = z2∗ + Λz1∗ (35)

where s∗ = [s1∗ , s2∗ ]> , z1∗ = x1∗ − xd , z2∗ = x2∗ − ẋd − ΨTanhφ, xd = [ x1,d , x2,d ]> , φ =
   
> ψ1 0 λ1 0
[φ1 , φ2 ] , Ψ = , and Λ = . Here, φ is given by
0 ψ2 0 λ2
q q
φ̇j = (cosh2 φj )(−κ j tanh φj + u(v j ) − v j )/ψj , φj (0) = 0, j = 1, 2. (36)
Mathematics 2021, 9, 1062 11 of 16

Using the similar reasoning in Section 3.1, the adaptive quantized state feedback
tracker for system (34) can be derived as follows:

(n) >
v = −ζs∗ − Λz2∗ + xd − Ŵ Q( x̄q ) − B̂Tanh(s∗ /e) (37)
˙ = Γ (s∗ Q ( x̄q ) − σ |s∗ |Ŵ ),
Ŵ j = 1, 2 (38)
j W,j j j W j j

B̂˙ = Γ B (s∗> Tanh(s∗ /e) − σB ks∗ k B̂) (39)

   
ζ1 0 γ 0
where ζ = and ΓW,j = w,1 .
0 ζ2 0 γw,2

Remark 4. In the existing nonlinear control results dealing with the 2-DOF helicopter sys-
tem [38–41], the input saturation problem was not considered even though input saturation
practically occurs in the 2-DOF helicopter system (34). Besides, the previous results [38–41] did
not consider both the state and input quantization problems and thus cannot be applied to the
network-based state-quantized control problem. However, this paper considers the state quantization
and the quantized input saturation effects.

4.3. Simulation Results

Prior to the implementation, a numerical simulation for system (34) is proceeded.
The initial conditions for the simulations are set to x1 (0) = [−0.05, 0.05]> (rad/s) and
x2 (0) = [0, 0]> . Hence, the input saturation of Vp and Vy are set to u1,M = u2,M = 24
from the DC motor specification. The length of the quantization intervals are chosen as
l = 0.01 for state variables and l = 1 for control inputs. The design parameters are chosen
as ζ 1 = ζ 2 = 1, λ1 = λ2 = 3, ψ1 = ψ2 = 1, κ1 = κ2 = 200, e = 0.1, γw,1 = γw,2 = 10, Γ B = 1,
and σW = σB = 0.001. The mean squared errors for the pitch and yaw angles are 0.1044
and 0.0435, respectively. The tracking results for simulation are shown in Figure 3. The DC
motor voltages denoting the control inputs are shown in Figure 4. The output of RBFNNs
and the estimation results of unknown parameter are displayed in Figure 5. From these
figures, we show that the proposed adaptive quantized state feedback tracker achieves the
robust tracking in the presence of state quantization and the quantized input saturation of
the MIMO nonlinear system (34).

40 40

35 35

30 30

25 25

20 20

15 15

10 10

5 5

0 0

-5 -5
0 5 10 15 20 25 30 0 5 10 15 20 25 30

(a) (b)
Figure 3. Tracking results for simulation (a) x1,1 and x1,d (b) x1,2 and x2,d .
Mathematics 2021, 9, 1062 12 of 16

20

-20
0 5 10 15 20 25 30

20

-20
0 5 10 15 20 25 30

Figure 4. Control input voltages Vp and Vy for simulation.

2 24.1

24.09
0
24.08
-2
24.07
-4
24.06

-6 24.05

24.04
-8
24.03
-10
24.02
-12
24.01

-14 24
0 5 10 15 20 25 30 0 5 10 15 20 25 30

(a) (b)
> >
Figure 5. RBFNNs outputs and adaptive parameters for simulation (a) Ŵ 1 Q1 and Ŵ 2 Q2 (b) B̂.

4.4. Experimental Results

The experiment setup is shown in Figure 6. The experiment results are displayed
in Figures 7–9. The initial conditions for experiment are set to x1 (0) = [0, 0]> (rad/s)
and x2 (0) = [0, 0]> . The input saturation is set as u1,M = u2,M = 24. The length of the
quantization intervals are chosen as l = 0.01 for state variables and l = 1 for control inputs.
The design parameters are chosen as ζ 1 = ζ 2 = 30, λ1 = λ2 = 3, ψ1 = ψ2 = 1, κ1 =
κ2 = 200, e = 1, γw,1 = γw,2 = 7, Γ B = 1, and σW = σB = 10−5 . The tracking results for
the experiment are shown in Figure 7. The mean squared errors for the pitch and yaw
angles are 0.0656 and 0.0523, respectively. Figure 8 shows the input voltages. The RBFNNs
outputs and estimation results are displayed in Figure 9. From these figures, we can see
that the proposed approach is successfully validated via the 2-DOF helicopter system (34).
Mathematics 2021, 9, 1062 13 of 16

Figure 6. Experiment setup.

40 40

35 35

30 30

25 25

20 20

15 15

10 10

5 5

0 0

-5 -5
0 5 10 15 20 25 30 0 5 10 15 20 25 30

(a) (b)
Figure 7. Tracking results for experiment (a) x1,1 and x1,d (b) x1,2 and x2,d .

20

10

-10
0 5 10 15 20 25 30

25

20

15

10

0
0 5 10 15 20 25 30

Figure 8. Control input voltage Vp and Vy for experiment.

Mathematics 2021, 9, 1062 14 of 16

0 24.04

-2 24.035

-4 24.03

-6 24.025

-8 24.02

-10 24.015

-12 24.01

-14 24.005

-16 24

-18 23.995

-20 23.99
0 5 10 15 20 25 30 0 5 10 15 20 25 30

(a) (b)
> >
Figure 9. RBFNNs outputs and adaptive parameters for experiment (a) Ŵ 1 Q1 and Ŵ 2 Q2 (b) B̂.

5. Conclusions
A neural-network-based adaptive quantized state feedback control design has been
developed for uncertain MIMO nonlinear systems with state quantization and quantized
input saturation. In the design of the proposed tracker scheme, unknown system nonlin-
earities and quantization errors are compensated by RBFNNs and adaptive techniques,
respectively. The auxiliary system using quantized states and compensation signals has
been introduced to analyze the effects of the quantized multi-input saturation. The stabil-
ity of the closed-loop error signals with quantization errors has been analyzed by some
theoretical lemmas. Finally, we have validated the effectiveness of the proposed quan-
tized state feedback tracker by presenting the simulation and experimental results for the
2-DOF helicopter system. Comparison studies of the adaptive neural tracker design to the
deterministic artificial intelligence approach reported in [42] will be explored in the future.

Author Contributions: Conceptualization, methodology, software, writing—original draft prepara-

tion, B.M.K.; Conceptualization, methodology, writing—review and editing, supervision, S.J.Y. All
authors have read and agreed to the published version of the manuscript.
Funding: This research was supported by the National Research Foundation of Korea (NRF) grant
funded by the Korea government (NRF-2019R1A2C1004898).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Karason, S.P.; Annaswamy, A.M. Adaptive control in the presence of input constraints. IEEE Trans. Autom. Cont. 1994, 39,
2325–2330. [CrossRef]
2. Zhong, Y.S. Globally stable adaptive system design for minimum phase SISO plants with input saturation. Automatica 2005, 41,
1539–1547. [CrossRef]
3. Wen, C.; Zhou, J.; Liu, Z.; Su, H. Robust adaptive control of uncertain nonlinear systems in the presence of input saturation and
external disturbance. IEEE Trans. Autom. Cont. 2011, 56, 1672–1678. [CrossRef]
4. Li, Y.; Tong, S.; Li, T. Composite adaptive fuzzy output feedback control design for uncertain nonlinear stric-feedback systems
with input saturation. IEEE Trans. Cyber. 2015, 45, 2299–2308. [CrossRef] [PubMed]
5. Shi, S.; Tong, S.; Li, Y. Observer-based fuzzy adaptive prescribed performance trackingcontrol for nonlinear stochastic systems
with input saturation. Neurocomputing 2015, 158, 100–108.
6. Gao, Y.F.; Sun, X.; Wen, C.; Wang, W. Adaptive tracking control for a class of stochastic uncertain nonlinear systems with input
saturation. IEEE Trans. Autom. Cont. 2017, 62, 2498–2504. [CrossRef]
Mathematics 2021, 9, 1062 15 of 16

7. Gao, Y.F.; Sun, X.; Wen, C.; Wang, W. Oberver-based adaptive NN control for a class of uncertain nonlinear systems with
nonsymmetric input saturation. IEEE Trans. Nerual Netw. Learn. Syst. 2017, 28, 1520–1530. [CrossRef]
8. Zhou, Q.; Wu, C.; Shi, P. Observer-based adaptive fuzzy tracking control of nonlinear systems with time delay and input
saturation. Fuzzy Sets Syst. 2017, 316, 49–68. [CrossRef]
9. Zhang, Q.; Dong, J. Disturbance-observer-based adaptive fuzzy control for nonlinear state constrained systems with input
saturation and input delay. Fuzzy Sets Syst. 2020, 392, 77–92. [CrossRef]
10. Wu, Y.; Xie, X. Robust adaptive control for state-constrained nonlinear systems with input saturation and unknown control
direction. IEEE Trans. Syst. Man Cyber. Syst. 2021, 51, 1192–1202. [CrossRef]
11. Wang, Y.; Hu, J.; Wang, J.; Xing, X. Adaptive neural novel prescribed performance control for non-affine pure-feedback systems
with input saturation. Nonlinear Dyn. 2018, 93, 1241–1259. [CrossRef]
12. Zerari, N.; Chemachema, M.; Essounbouli, N. Neural network based adaptive tracking control for a class of pure feedback
nonlinear systems with input saturation. IEEE/CAA J. Autom. Sin. 2019, 6, 278–290. [CrossRef]
13. Chen, M.; Ge, S.S.; Ren, B. Adaptive tracking control of uncertain MIMO nonlinear systems with input constraints. Automatica
2011, 47, 452–465. [CrossRef]
14. Cui, G.; Yu, J.; Wang, Q. Finite-time adaptive fuzzy control for MIMO nonlinear systems with input saturation via inproved
command-filtered backstepping. IEEE Trans. Syst. Man Cyber. Syst. 2020. [CrossRef]
15. Si, W.; Dong, X. Adaptive neural control for MIMO stochastic nonlinear pure-feedback systems with input saturation and
full-state constraints. Neurocomputing 2018, 275, 298–307. [CrossRef]
16. Sui, S.; Tong, S.; Li, Y. Adaptive fuzzy backstepping output feedback tracking control of MIMO stochastic pure-feedback nonlinear
systems with input saturation. Fuzzy Sets Syst. 2014, 254, 26–46. [CrossRef]
17. Yang, Y.; Yue, D.; Yuan, D. Adaptive neural tracking control of a class of MIMO pure-feedback time-delay nonlinear systems with
input saturation. Int. J. Syst. Sci. 2015, 47, 3730–3740. [CrossRef]
18. He, W.; Li, Z.; Dong, Y.; Zhao, T. Design and adaptive control for an upper limb robotic exoskeleton in presence of input saturation.
IEEE Trans. Nerual Netw. Learn. Syst. 2019, 30, 97–108. [CrossRef]
19. Lian, F.L.; Moyne, J.R.; Tilbury, D.M. Performance evaluation of control networks: Ethernet, controlnet, and devicenet. Control
Syst. IEEE 2001, 21, 66–83.
20. Zhou, J.; Wen, C.Y.; Yang, G.H. Adaptive backstepping stabilization of nonlinear uncertain systems with quantized input signal.
IEEE Trans. Autom. Control. 2014, 59, 460–464. [CrossRef]
21. Lai, G.Y.; Liu, Z.; Philip Chen, C.L.; Zhang, Y. Adaptive asymptotic tracking control of uncertain nonlinear system with input
quantization. Syst. Control. Lett. 2016, 96, 23–29. [CrossRef]
22. Li, Y.X.; Yang, G.H. Adaptive asymptotic tracking control of uncertain nonlinear systems with input quantization and actuator
faults. Automatica 2016, 72, 177–185. [CrossRef]
23. Xing, L.T.; Wen, C.Y.; Su, H.Y. Robust control for a class of uncertain nonlinear systems with input quantization. Int. J. Robust
Nonlinear 2016, 26, 1585–1596. [CrossRef]
24. Yu, X.W.; Lin, Y. Adaptive backstepping quantized control for a class of nonlinear systems. IEEE Trans. Autom. Control. 2017, 62,
981–985. [CrossRef]
25. Yang, Y.; Yu, Z.; Li, S. Adaptive output feedback quantised tracking control for stochastic nonstrict-feedback nonlinear systems
with input saturation. Int. J. Syst. Sci. 2018, 49, 3130–3145. [CrossRef]
26. Xing, L.; Wen, C.; Liu, Z.; Cai, J.; Zhang, M. Adaptive control for a class of uncertain nonlinear systems subject to saturated input
quantization. In Proceedings of the 2020 16th International Conference Control, Automation, Robotics and Vision (ICARCV),
Shenzhen, China, 13–15 December 2020; pp. 472–477.
27. Zhou, J.; Wen, C.; Wang, W.; Yang, F. Adaptive backstepping control of nonlinear uncertain systems with quantized states. IEEE
Trans. Autom. Cont. 2019, 64, 4756–4763. [CrossRef]
28. Choi, Y.H.; Yoo, S.J. Quantized feedback adaptive command filtered backstepping control for a class of uncertain nonlinear
strict-feedback systems. Nonlinear Dyn. 2020, 99, 2907–2918. [CrossRef]
29. Choi, Y.H.; Yoo, S.J. Neural-networks-based adaptive quantized feedback tracking of uncertain nonlinear strict-feedback systems
with unknown time delays. J. Frankl. Inst. 2020, 357, 10691–10715. [CrossRef]
30. Choi, Y.H.; Yoo, S.J. Quantized-feedback-based adaptive event-triggered control of a class of uncertain nonlinear systems.
Mathematic 2020, 8, 1603. [CrossRef]
31. Meng, W.; Yang, Q.; Sun, Y. Adaptive neural control of nonlinear MIMO systems with time-varying output constraints. IEEE
Trans. Neural Netw. Learn. Syst. 2015, 26, 1074–1085. [CrossRef]
32. Brockett, R.W.; Liberzon, D. Quantized feedback stabilization of linear systems. IEEE Trans. Autom. Cont. 2000, 45, 1279–1289.
[CrossRef]
33. Polycarpou, M.M. Stable adaptive neural control scheme for nonlinear systems. IEEE Trans. Autom. Cont. 1996, 41, 447–451.
[CrossRef]
34. Park, J.; Sandberg, I.W. Universal approximation using radial-basis-function networks. Neural Comput. 1991, 3, 246–257.
[CrossRef]
35. Wang, C.; Hill, D.J.; Ge, G.G.; Chen, G. An ISS-modular approach for adaptive neural control of pure-feedback systems. Automatica
2006, 42, 625–635. [CrossRef]
Mathematics 2021, 9, 1062 16 of 16

36. Kurdila, A.J.; Narcowich, F.J.; Ward, J.D. Persistency of excitation in identification using radial basis function approximants.
SIAM J. Cont. Optim. 1995, 33, 625–642. [CrossRef]
37. Quanser Inc. Quanser AERO Laboratory Guide; Technical Report; Quanser: Arkham, ON, Canada, 2016.
38. Labdai, S.; Chrifi-Alaoui, L.; Drid, S.; Delahoche, L.; Bussy, P. Real-time implementation of an optimized fractional sliding mode
controller on the quanser-aero helicopter. In Proceedings of the 2020 International Conference Control, Automation and Diagnosis
(ICCAD), Paris, France, 7–9 October 2020.
39. Lambert, P.; Reyhanoglu, M. Observer-based sliding mode control of a 2-DoF Helicopter system. In Proceedings of the IECON
2018 44th Annual Conference of the IEEE Industrial Electronics Society, Washington, DC, USA, 21–23 October 2018.
40. Schlanbusch, S.; Zhou, J. Adaptive backstepping control of a 2-DOF helicopter. In Proceedings of the 2019 IEEE 7th International
Conference on Control, Mechatronics and Automation, Delft, The Netherlands, 6–8 November 2019.
41. Schlanbusch, S.; Zhou, J. Adaptive backstepping control of a 2-DOF helicopter system with uniform quantized inputs. In
Proceedings of the IECON 2020 46th Annual Conference of the IEEE Industrial Electronics Society, Singapore, 18–21 October 2020.
42. Sands, T. Control of DC motors to guide unmanned underwater vehicles. Appl. Sci. 2021, 11, 2144. [CrossRef]