mathematics

Article
Machine-Learning-Based Improved Smith Predictive Control
for MIMO Processes
Xinlan Guo 1, *, Mohammadamin Shirkhani 2, * and Emad M. Ahmed 3,4

1 College of Rail Transportation, Nanjing Vocational Institute of Transport Technology, Nanjing 211188, China
2 Department of Electrical Engineering, Ilam University, Ilam 69315-516, Iran
3 Department of Electrical Engineering, College of Engineering, Jouf University, Sakaka 72388, Saudi Arabia
4 Department of Electrical Engineering, College of Engineering, Aswan University, Aswan 81542, Egypt
* Correspondence: [email protected] (X.G.); [email protected] (M.S.)

Abstract: Controlling time-delayed processes is one of the challenges in today’s process industries.
If the multi-input/multi-output system is dynamically coupled, the delay problem becomes more
critical. In this paper, a new method based on Smith’s predictive method, with the help of a type-2
fuzzy system to control the system with the mentioned features, is presented. The variability in the
time delay, the existence of disturbances and the existence of structural and parametric uncertainty
lead to the poor performance of the traditional Smith predictor. Even if the control system is set
up correctly at the beginning of the setup, it will eventually wear out, and the above problems will
appear. Therefore, computational intelligence is used here, and by updating the parameters of the
control system at the same time as the system changes, the control system adapts itself to achieve the
best performance. To evaluate the proposed control system, a complex process system is simulated,
the results of which show the good performance of Smith’s prediction method based on a type-2
fuzzy system.

Keywords: machine learning; parameter uncertainty; Smith predictive; MIMO control

Citation: Guo, X.; Shirkhani, M.;
Ahmed, E.M. Machine-Learning-Based MSC: 93-08
Improved Smith Predictive Control for
MIMO Processes. Mathematics 2022,
10, 3696. https://doi.org/10.3390/
math10193696 1. Introduction
Academic Editors: António Lopes Many processes in the chemical industry have time delays. Time delays in processes
and Luigi Fortuna cause the behavior of the open-loop function to be non-minimum system controlled, and
therefore, it is difficult to control these systems [1–5]. The presence of time delay in processes
Received: 14 August 2022
is one of the major challenges in all control systems. In these processes, the control signal is
Accepted: 6 October 2022
generated and applied to the process, but the output does not change for a certain period
Published: 9 October 2022
of time. Therefore, if the output needs to change suddenly, the control system will not
Publisher’s Note: MDPI stays neutral work. In order to solve this problem, different methods have been proposed, the most
with regard to jurisdictional claims in common of which is the use of model predictive control (MPC). This control system has
published maps and institutional affil- excellent performance for small time delays because it regularly predicts the output state.
iations.
However, if the time delay of the process is big, unfortunately, the efficiency of MPC will
decrease. Another solution is that the control system has a time delay just like the process.
However, this idea is not effective if the amount of time delay is variable. So far, the most
practical and comprehensive method for controlling processes with large time delays is
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
Smith’s method. The two basic challenges of Smith’s method are the non-minimum phase
This article is an open access article
of the process and the large variation in the time delay. In this article, we overcome both
distributed under the terms and problems with a new method. In single-input/output systems, using Smith’s predictor for
conditions of the Creative Commons processes with long time delays can be effective in improving the system’s response. In this
Attribution (CC BY) license (https:// method with process model prediction, the time delay parameter will be removed from
creativecommons.org/licenses/by/ the equation [6–8]. The problems in this method can be attributed to its high sensitivity to
4.0/). model error, a decrease in the control in response to turbulence control of integral processes

Mathematics 2022, 10, 3696. https://doi.org/10.3390/math10193696 https://www.mdpi.com/journal/mathematics

Mathematics 2022, 10, 3696 2 of 19

and the inability to control unstable processes [9,10]. Many researchers have provided
various methods to solve these problems, known as “time delay compensators” [11–13].
There are also two general new methods that are based on output prediction with no delay
in the form of discrete-time control systems provided [14,15]. The authors of a number of
references tried to expand Smith’s method multi-input/output systems [16,17]. However,
the fact is that the usage of this method for such systems has more restrictions than single-
input/output systems. The existence of interference effects between control loops is one of
the biggest limitations to these methods. To avoid these effects, even if we want to use a
decoupler due to the high sensitivity of Smith’s method on model error, the selected model
should not have many mistakes because it will make the control system more unstable.
Compared to a conventional feedback method, to avoid interactions between loops, the
product of the process transfer function in the controller G p (s)G c (s) (which is a condition
of the absence of interactions between the loops in the conventional feedback method)
and multiplication to the process transfer function without time delay in the controller
G0 p (s)G c (s) must also be a diagonal matrix. In addition to this, in Smith’s method for
processes in which time delay parameters are not separable from the process transfer
method, the process is not applicable. Therefore, by focusing on the mentioned limitations,
the application of Smith’s method for controlling multi-input/output systems with time
latency has not received much attention from researchers, and in most cases, researchers
use the same common system feedback to control these systems. In [18,19], methods based
on Smith’s predictor are proposed to control multi-input/multi-output systems with time
delay. To adjust the parameters of the controllers in MIMO systems, many design methods
are mentioned in the references, which generally can be divided into five categories:
• Detuning methods [20,21];
• Methods of closing loops consecutively [22];
• Iterative methods or guess and error [23];
• Methods of the simultaneous solution of equations or optimization [24];
• Independent methods [25].
In detuning methods, each controller in the system is accordingly designed based on
the element of the diameter, and the interactions of the loops with each other are ignored.
Then, the controller is readjusted, taking into account the internal effects of the loops. The
simplicity of these methods is their main advantage, but their disadvantage is due to the
fact that the performance of the loop and its stability is not clearly and explicitly considered
in these methods. In the second method, the loops are sequential; they are closed and
adjusted. Usually, this starts with the fastest loop, and therefore, the dynamic interactions in
this loop are considered in setting the controller of the next loop, and the same is performed
for the others [26]. Some of the disadvantages of this method are reported in [27]. Some
disadvantages include the dependence of the last controller’s design on the order in which
the controllers were designed, and also, the selection of repetitive methods in response to
the designed loops.
In the iterative method (the third method), first, the controller parameters are set
as sequentially similar to the loop closure method, and after all the loops are closed, the
controllers are set again one after the other. This method continues until the answers
converge [28]. In the guessing and error method, the PID controller parameters are de-
termined step by step in such a way as to ensure the stability of the system. This type of
design is usually accompanied by “Relay Feedback” experiments known as “self-adjusting
variables” [29]. One of its main disadvantages is that this method not only requires the
successful testing of feedback relays, but also, in this method, there is no strong rela-
tionship between parameter regulation and system performance. Controller design in
multi-input/output loops by the simultaneous solution of equations (the fourth method) is
numerically difficult and complex. In [30], In this regard, one design method for PI/PID
controllers is provided in multi-input/output loops. In this reference, the modified Ziegler–
Nichols method is used, and the effects of interference between the loops are considered,
but there is no guarantee. For the existence of the answer, because the calculations are
Mathematics 2022, 10, 3696 3 of 19

nonlinear and complex, this method is only used for two-input/output systems, and its
expansion to higher-dimensional systems seems to be difficult and has not been reported
so far. In [31], a multi-loop design is proposed as a nonlinear optimization problem. How-
ever, this formulation does not include systems with different latencies. Simultaneously
optimizing the equations of multi-input/output systems numerically is very difficult, and
the results are highly dependent on the conditions defined in the objective function, and
if the loops are closed in a distinctive order, the system may eventually become unstable.
In [32,33], independent design methods (fifth method) are used. In these methods, the
controllers are designed separately and with specific limits to ensure stability and proper
performance. However, detailed information about the dynamics of the controllers is not
used in other loops, so the final performance may be poor. In [34], a new method is pro-
posed for controlling time-delayed processes in single-input/output systems. The function
Gmb (S), which is often the first-order function with a predominant or equal interest to the
open-loop function of the system control, is added to the open-loop function of the control
system, and thus all the zeros on the right due to the time delay of the function are moved
to the left. The interval loops are moved to the left, thus improving the performance of the
control system.
Most of the used PID controllers consider time delays as a part of the system. These
dead times cause disturbances in the system. One of the ways to compensate for these
dead times is to use Smith’s technique [35–38]. In [39], a fuzzy PID controller is designed
based on Smith’s predictor technique for a heating system. By using this method, we can
compensate for the dead time. In these methods, it is possible to increase the speed of the
system by tuning the PID controller parameters. This has been completed in [40].
It should be noted that in any control system, its asymptotic stability is a basic re-
quirement. For this, a controller based on the nominal model is used, but the presence of
uncertainties in the system can cause it to operate incorrectly. In order to guarantee the
proper functioning of the system, robust controllers are used. Every robust control system
must have two conditions: closed-loop stability in nominal conditions and closed-loop
stability despite different uncertainties. In robust controllers, if the range of acceptable
values is greater, it indicates that the controller is more robust [41].
Fuzzy Smith has been used in articles [34–39], but it has some fundamental differences
with our work, which can be listed as follows:
1. In the mentioned articles, the main controller was PID, which has poor performance in
dealing with systems with high delay and a non-minimum phase, and unfortunately,
there was no mention of this issue in the mentioned articles. However, in our proposed
method, by designing and setting a first-order transformation function (with dominant
gain), the system was released from the non-minimum phase state and the system
was tamed (Tractable).
2. In all the mentioned articles, a type-1 fuzzy system was used, while we used type-2
fuzzy system. It is clear that a type-2 fuzzy system has more parameters than a type-1
system and therefore has more degrees of freedom and shows higher accuracy.
3. The last thing is that in our article, we applied some types of parametric uncertainties
and disturbances, but in the aforementioned articles, uncertainty and disturbance
were not applied in such a comprehensive and complete way.
The discussion of system robustness is fully presented in this paper. The rest of
the paper is organized as follows: Section 2 introduces Smith’s predictive control. The
proposed control system is described in Section 3. Section 4 shows the simulation results,
and Section 5 provides the conclusion.

2. Smith’s Predictive Control of Time-Delayed MIMO Systems

Smith’s predictive control method is a common and suitable solution for controlling
delayed systems. In this method, knowing the system model and the amount of delay is
necessary. Figure 1 shows Smith’s predictive control system for the time-delayed MIMO
system, and the proposed control system is described in Section 3.
 2. Smith’s Predictive Control of Time-Delayed MIMO Systems
Smith’s
2. Smith’s predictive
Predictive control
Control method is a common
of Time-Delayed and suitable solution for controlling
MIMO Systems
delayed systems. In this method, knowing the system model and the amount of delay is
Smith’s predictive control method is a common and suitable solution for controlling
necessary. Figure 1 shows Smith’s predictive control system for the time-delayed MIMO
Mathematics 2022, 10, 3696
delayed systems. In this method, knowing the system model and the amount of delay4 is of 19
system, and the proposed control system is described in Section 3.
necessary. Figure 1 shows Smith’s predictive control system for the time-delayed MIMO
As shown in Figure 1, Smith’s method must specify the exact system model (𝐺(𝑠))
system, and the proposed control system is described in Section 3.
and the amount of time delay (𝜏𝑑 ); otherwise, the correct answer will not be received.
As shown in Figure 1, Smith’s method must specify the exact system model (𝐺(𝑠))
As shown in Figure 1, Smith’s method must specify the exact system model (G (s))
and the amount of time delay (𝜏𝑑 ); otherwise, the correct answer will not be received.
and the amount of time delay (τd ); otherwise, the correct answer will not be received.

Figure 1. Smith’s predictive control method.

Figure 1. Smith’s predictive control method.
Figure 1. Smith’s predictive control method.
3. The Proposed Control System
3. The Proposed Control System
The proposed
3. The Proposed Controlmethod
Systemin this paper has two phases: one is stabilizing and changes
The proposed method in this paper has two phases: one is stabilizing and changes
the system behavior from a non-minimum phase to minimum phase, and the other is the
theThe proposed
system method
behavior frominathis paper has two
non-minimum phases:
phase one is stabilizing
to minimum andthe
phase, and changes
other is
precise control of the system (process) by considering system dynamic changes and para-
thethe
system behavior
precise controlfrom a non-minimum
of the system (process) phase to minimumsystem
by considering phase,dynamic
and the other is the
changes and
metric uncertainty. For the first phase, it is proposed to add the dominant gain transfer
parametric
precise controluncertainty. For(process)
of the system the first phase, it is proposed
by considering to add
system the dominant
dynamic changes gain transfer
and para-
function to the open-loop equations of the system (Figure 2).
function
metric to the open-loop
uncertainty. equations
For the first phase,ofitthe system (Figure
is proposed to add2).the dominant gain transfer
function to the open-loop equations of the system (Figure 2).

Figure2.2.Open-loop
Figure Open-loopequations
equationsof of
thethe system
system with
with thethe addition
addition of dominant
of the the dominant
gaingain transfer
transfer func-
function.
tion.
Figure 2. Open-loop equations of the system with the addition of the dominant gain transfer func-
It is assumed that the measurement transfer function is equal to 1 (Gm (s) = 1). The
tion.
closed-loop transfer functions of this system are in the form of relations (1) and (2).

−1
−1
= G p,n (s) Gc,n (s) 1 + Gmb,n (s) + Gc,n (s) Gn (s) 1 − e−τd s + G p,n (s) Gc,n (s)



(Yn (s)) Ysp,n (s) (1)

(Yn (s))(dn (s))−1 = ( Gd,n (s) + Gmb,n (s) Gd,n (s) + Gd,n (s) Gn (s)(1 − e−τd s ) Gc,n (s))

−1 (2)
1 + Gmb,n (s) + Gc,n (s) Gn (s)(1 − e−τd s ) + Gc,n (s) G p,n (s)
In the above relationships, Yn (S), Ysp,n (S), dn (S), Gmb,n (S), G p,n (S), Gc,n S and Gn (s)
from left to right, respectively, represent the response matrix, the reference input matrix,
Mathematics 2022, 10, 3696 5 of 19

the input perturbation matrix, the matrix of predictive functions, the matrix of the process
include time delay, matrix functions of the controllers, and finally, the matrix of the process
without time delay. In this system, as in the proposed control system in single systems, the
input/output of the Gmb,n (S) function, which predicts the nth loop in the control system,
must be a function with a minimum phase behavior and a higher gain than other functions
in the open-loop function of that loop. The ideal mode for controlling multi-input/output
systems is the absence of loop interaction with each other. This happens when the matrix
G p,2 (s) Ge,2 (s) is a diagonal matrix, in which case, the matrix of the open-loop transfer
function of the proposed control system will be as follows:
 
Gmb1 (s) + G p11 (s) Gc11 (s)
X (s) = (3)
Gmb2 (s) + G p22 (s) Gc22 (s)

The conditions of the dominant interest concept must be established in a way that the
conditions of Equations (4) and (5) in the loops at all frequencies in phase diagrams do not
exceed the range from 0 to −180◦ .

| Gmb1 (s)| ≥ G p11 (s) Gc11 (s) (4)

| Gmb2 (s)| ≥ G p22 (s) Gc22 (s) (5)

This method is effectively expandable for higher-dimensional systems, and it is only
necessary to have the necessary conditions to create a dominant interest constraint in the
open-loop function of each control loop [29]. If there are interaction effects between the
loops, then, for example, in a 2 × 2 open-loop control system, the control system is obtained
as follows:



Q(s) = 1 + Gmb,11 (s) + G p11 (s) Gc11 (s) 1 + Gmb,22 (s) + G p22 (s) Gc22 (s) −
(6)
G p12 (s) G p21 (s) Gc11 (s) Gc22 (s)

which, in order to establish the dominant interest constraint, must have the following
equation in all or most of the curves so that in the phase diagram of the open-loop function,
the control system does not oscillate around the non-minimum phase function diagram
and does not exceed 180◦ :
G p22 (s) Gc22 (s) + G p11 (s) Gc11 (s)
+ G p11 (s) G p22 (s) Gc11 (s) Gc22 (s)
Gmb,11 (s) + Gmb,22 (s) + Gmb,11 (s) Gmb,22 (s) ≥ + G p11 (s) Gc11 (s) Gmb,22 (s) (7)
+ G p22 (s) Gc22 (s) Gmb,11 (s)
− G p11 (s) G p21 (s) Gc11 (s) Gc22 (s)

Of course, a simpler method can be used to apply the proposed method to control
systems in which the loops interact. In this method, first, the interactions of the loops are
eliminated from various methods, such as controller adjustment methods, and then, the
proposed method is used for the assumed system. This method is used in the example
given in this article. In the case of selecting the degree of the Gmb,j (s) function given that in
multi-input/output systems, the Gmb,n (s) function exists for each loop, we select the same
loop separately for the open-loop function; so, selecting the degree of this function is the
same as for single-input/output systems, which will be discussed below. In the case of
these systems, the general open-loop function can be represented as Equation (8):

overall open − loop trans f er f unction = G∞,j (s) = Gmb,j (s) + Gsol,j (s)
(mo )
Kmb,j Ksol,j Nsol,j (s) −st (8)
= ( pb) + ( pb) e d,j , j = 1, 2, . . . , n
Dmb,j (s) Dsol,j (s)

In the above relation, Gsol,j (s) is a simple open-loop function in the j-th loop. Addi-
tionally, Gool,j (s) is the general open-loop function for the j-th loop. Kmb,j and Ksol,j are
the compensatory gains and the simple open-loop function, respectively, and mo, pb, and
Mathematics 2022, 10, 3696 6 of 19

po are the degrees of the polynomials. Thus, the equation characteristic of the proposed
method will be in the form of relation (9).
( po ) (mo ) ( pb)
Kmb,j Dsol,j (s) + Ksol,j Nsol,j (s) Dmo,j e−std j = 0 (9)

In order for the dominant interest condition to be true in the general open-loop function
of each loop, Equation (10) must be true.

po ≥ mo + pb → pb ≤ po − mo (10)

Now, if Gsol,j (s) is a suitable function, that is, po = mo, for the function to have a
( po )
minimum phase behavior in relation (10), (i.e., Kmb,j Dsol,j (s)) must be pb = 0, whereas if
Gsol,j (s) is a strictly suitable function, in order for the function to have a minimum phase
behavior in relation (10), it must be po > mo; in which case, we have pb ≥ 1. In this case,
pb = 1 is the most desirable degree because the first-degree functions have a finite phase.
Additionally, their phase curves will never reach −180◦ , so they are always stable. In
general, the condition for the disappearance of all zeros to the right due to the time delay
parameter of the general open-loop function is to establish relation (11):
( po ) (mo ) ( pb)
Kmb,j Dsol,j ( jw) ≥ Ksol,j Nsol,j ( jw) Dmb ( jw)e− jwtd,j (11)

which is equivalent to relation (12):

Gmb,j ( jw) ≥ Gsol,j ( jw) (12)

If there are interference effects between the control loops, then by using Equation (8)
and applying a method similar to the one mentioned above, the appropriate degree and
gain of the compensating function can be found.
The function Gmb is a first-order transfer function, and it is enough to apply in the con-
ditions of relations (5), (6) and (10). A first-order transfer function only has two parameters:
one is the gain and the other is the position of its pole. To set these two parameters, it is
enough to meet the mentioned conditions. Of course, various functions can be found that
satisfy the mentioned conditions, but definitely only one function can give us the optimal
answer. Currently, this optimal transfer function has been obtained by trial and error, but
we intend to provide a mechanism to obtain the optimal Gmb in future research.
This function gives us the task of stabilizing the control system and guaranteeing its
convergence; the next step is to accurately estimate the amount of delay using a type-2
fuzzy neural network and use it for Smith’s predictive control. Therefore, introducing
the existence of such a function is one of the innovations of this article, and the details
regarding this function and how to calculate the optimal Gmb will definitely be discussed in
future papers.

3.1. Sensitivity Analysis of the Proposed Method

In this section, the robustness of the proposed method to model error is investigated.
For simplicity, we discuss the inter-loop control system when there are no interactions.
Obviously, this method can be easily generalized to systems with interactions. The char-
acteristic equation of the proposed system is in the form of relation (13) if there are no
interactions in the control system.

1 + Gmb,n (s) + G p,n (s) Gc,n (s) = 0 (13)

The error in the model can be represented as δG p,n ( jω ) = G p,n ( jω ) − Gn,n ( jω ).

δG p,n ( jω ) is the difference between the actual model G p,n ( jω ) and the nominal model
Gn,n ( jω ). By placing δG p,n ( jω ) in Equation (14), we will obtain Equation (14).

1 + Gmb,n ( jω ) + Gn,n ( jω ) Gc,n ( jω ) + δGn ( jω ) Gc,n ( jω ) = 0 (14)

 The error in the model can be represented as 𝛿𝐺𝑝,𝑛 (𝑗𝜔) = 𝐺𝑝,𝑛 (𝑗𝜔) − 𝐺𝑛,𝑛 (𝑗𝜔) .
𝛿𝐺𝑝,𝑛 (𝑗𝜔) is the difference between the actual model 𝐺𝑝,𝑛 (𝑗𝜔) and the nominal model
𝐺𝑛,𝑛 (𝑗𝜔). By placing 𝛿𝐺𝑝,𝑛 (𝑗𝜔) in Equation (14), we will obtain Equation (14).
Mathematics 2022, 10, 3696 7 of 19
1 + 𝐺𝑚𝑏,𝑛 (𝑗𝜔) + 𝐺𝑛,𝑛 (𝑗𝜔)𝐺𝑐,𝑛 (𝑗𝜔) + 𝛿𝐺𝑛 (𝑗𝜔)𝐺𝑐,𝑛 (𝑗𝜔) = 0 (14)

Through Equation (15), we can obtain the norm of the highest error range of the
model, whichEquation
Through will be Equation
(15), we can(15)obtain the norm of the highest error range of the model,
which will be Equation (15)
|1 + 𝐺𝑚𝑏,𝑛 (𝑗𝜔) + 𝐺𝑛,𝑛 (𝑗𝜔)𝐺𝑐,𝑛 (𝑗𝜔)|
|𝛿𝐺𝑝,𝑛 (𝑗𝜔)| =1 + G (15)
mb,n ( jω )|𝐺+𝑐,𝑛G(𝑗𝜔)|
n,n ( jω ) Gc,n ( jω )
δG p,n ( jω ) = (15)
| Gc,n ( jω )|
In this way, we can obtain the norm obtained for a typical feedback loop:
In this way, we can obtain the norm obtained for a typical feedback loop:
|1 + 𝐺𝑛,𝑛 (𝑗𝜔)𝐺𝑐,𝑛 (𝑗𝜔)|
|𝛿𝐺𝑝,𝑛 (𝑗𝜔)| =|1 + Gn,n ( jω ) Gc,n ( jω )| (16)
δG p,n ( jω ) = |𝐺 (𝑗𝜔)| (16)
| Gc,n𝑐,𝑛( jω )|
According to Equations (15) and (16), under equal conditions, due to the 𝐺 (𝑠)
According to Equations (15) and (16), under equal conditions, due to the G𝑚𝑏,𝑛 mb,n ( s )
function, the norm of the highest model error range in the proposed method is larger than
function, the norm of the highest model error range in the proposed method is larger than a
a normal feedback loop. Therefore, it can be said that the proposed method is more re-
normal feedback loop. Therefore, it can be said that the proposed method is more resistant
sistant
to model toerror.
modelAserror. As aitresult,
a result, can beitsaid
can that
be said
the that the proposed
proposed method
method is is moreinflexible
more flexible terms
in terms of removing
of removing model errors.model errors.
In the
In the second
second phase,
phase, itit is
is time
time to
to accurately
accurately estimate
estimate the
the amount
amount of
of system
system latency.
latency.
This is effectively completed using a type-2 fuzzy neural network (Figure
This is effectively completed using a type-2 fuzzy neural network (Figure 3). 3).

Figure 3.
Figure 3. The proposed
proposed control
control system.
system.

Since the analysis is based on the transfer functions and is unlike the state space
equations, the input cannot be separated, so the sensitivity analysis is presented only based
on the uncertainty of the model. A control system, if it is designed correctly, can consider
disturbance as an uncertainty of the model, reduce its effect, and provide the appropriate
response. Two important challenges of a control system are uncertainty in the model and
the presence of disturbances. Therefore, in order to evaluate the control system, we had to
apply both challenges, and fortunately, the proposed method came out victorious.
Figure 3 shows the general structure of the process control system (including both
phases). The T2FNN inputs are: the process inputs, process outputs, as well as the process
outputs of the previous moments. The output of the type-2 fuzzy neural network is the
instantaneous estimation of the time delay of the process. Any change in process dynamics
and any factor that changes the system time delay (such as time-lapse and system wear) is
immediately observed by the type-2 fuzzy system, and a new Td is generated.
Mathematics 2022, 10, 3696 8 of 19

3.2. Type-2 Fuzzy System

In this section, the structure of type-2 fuzzy systems is introduced. Figure 3 shows the
proposed type 2 fuzzy system.
The calculations of the first layer are as follows:
  2
  x − c 1 x j < c1ji


  j

 ji
c1ji ≤ x j ≤ c2ji


 ∅ ji x j =


 1

 
   2
 x j − c2

x j > c2ji
 ji (17)

   2 c 1 + c2

 x j − c2ji x j ≤ ji 2 ji

 

x =

∅
 ji j

2 c1 + c2

  
 x j − c1 x j > ji 2 ji
 
ji

The upper and lower of the ith neuron and the jth input are denoted by ∅ ji and ∅ ji ,
respectively. Therefore, the outputs of the first layer are as follows:
n +1 e
  
∅i ( x ) = exp − ∑ j=1 2ji ( j )
 ∅ x
 σi
 n +1
 (18)
∑ ∅ ji ( x j )
∅i ( x ) = exp − j=1 σ2


i

where ∅i and ∅i are the upper and lower of the ith neuron h i (i = 1, 2, . . . , m), respectively.
1 2


x ∈ x j , j = 1, . . . , n is the input vector, and c ji ∈ c ji , c ji is the center of all the type-2
fuzzy neurons. Next, the left and right endpoints of the second layer are as follows:
 q 2 m 1
 ŷl = ∑i=1 q∅i ( x)cwi σwi +∑im=q+1 ∅i ( x)cwi σwi

∑i=1 ∅i ( x )σwi +∑i=q+1 ∅i ( x )σwi

p (19)
∑i=1 ∅i ( x )c1wi σwi +∑im= p+1 ∅i ( x )c2wi σwi
 ŷr =

p
 m
∑i=1 ∅i ( x )σwi +∑i= p+1 ∅i ( x )σwi

where p and q are the left and right switching points the type-2 fuzzy system, which can
be calculated using the trial-and-error method or the Karnik–Mendel (KM) algorithm.
Additionally, m, wi , cwi , and σwi are the mean value of the first-layer neurons, the weights,
the center of the weights, and the spread of the weights, respectively. Lastly, the general
output of the network can be derived as follows:
q r m l
1 ∑i=1 ŷr cwi σwi + ∑i=q+1 ŷl cwi σwi
ŷ = q (20)
2 ∑i=1 ŷr σwi + ∑im=q+1 ŷl σwi

In Equation (20), crwi and clwi are the centers of W

e r and W
e l , respectively. The gradient
descent method is used to teach the network. See [42] for more details.

4. Simulation
In the following, there are two examples of using the proposed system to control
multi-input/output processes and how it is used for these systems.
Example 1: The process function of the distillation tower Wood-Berry, which is a well-
known function in the field of multi-input/output systems, is considered as follows [30]:
−s −3s
" #
12.8e −18.9e
G p (s) = 16.7s+1 21s+1
6.6e−7s −19.4e−3s
10.9s+1 14.4s+1

The first point to consider in solving these problems is the interaction of control loops
Is on each other. In order for the loops not to interact with each other, the product of
G p × Gc must be the product of a diagonal matrix. Various methods are used to diameter
 12.8𝑒 −𝑠 −18.9𝑒 −3𝑠

𝐺𝑝 (𝑠) = [16.7𝑠+1
6.6𝑒 −7𝑠
21𝑠+1
−19.4𝑒 −3𝑠
]
10.9𝑠+1 14.4𝑠+1

The first point to consider in solving these problems is the interaction of control loops
Mathematics 2022, 10, 3696 9 of 19
Is on each other. In order for the loops not to interact with each other, the product of
𝐺𝑝 × 𝐺𝑐 must be the product of a diagonal matrix. Various methods are used to diameter
the open-loop function matrix of the control system. In some references, this is achieved
the open-loop
using decouplers function
[31–35],matrix
and of
in the control
others, system. Inthe
by adjusting some references,
controller this is achieved
parameters [36–40].
using decouplers [31–35], and in others, by adjusting the controller
This operation is performed. In this example, the second method is used. A method parameters [36–40].
ac-
This operation is performed. In this example, the second method
cording to which the parameters of the controllers are adjusted so that the loops do not is used. A method
according
interact to which
with the parameters
each other is presented ofin
the[37].
controllers
Based onarethis
adjusted
method, so for
thatthis
theexample,
loops do thenot
interact with each other is presented in [37]. Based on this method, for this example, the
0.0915
parameters of the controllers for the two control loops are obtained as 𝐺𝑐1 = 0.547 + 0.0915
parameters of the controllers for the two control loops are obtained as G = 0.547 + 𝑠
0.021 c1 s
and 𝐺 𝑐2 = −0.107 − 0.021 for the PI controllers.
and Gc2 = −0.107 − 𝑠s for the PI controllers.
In
In thethe proposed
proposed method,
method, to to design
design the the control
control structure,
structure, we we must
must first
first select
select thethe
𝐺
G𝑚𝑏,𝑛
mb,n
(𝑠)
( s ) function.
function. Given
Given that
thatin this
in example,
this example, we want
we wantto show
to the
show resistance
the of
resistance theof pro-
the
posed
proposed method
method to the model
to the modelerror
errorforfor
allall
time delay
time delay parametersinin𝐺G𝑝𝑖𝑗pij(𝑠),
parameters (s),the
theprobability
probability
of error is set as up to to +50%.
+50%. We We consider
consider the percentage
percentage in the proposed method, the
𝐺 mb,n(𝑠)
G𝑚𝑏,𝑛 function must
(s) function must bebe selected
selected so that it has the ability to cover the highest error limit
so that at at most
mostfrequencies,
frequencies,the thedominant
dominant interest
interest condition
condition is observed,
is observed, andand the open-
the open-loop
1 1
function
loop has ahas
function minimum
a minimum phase behavior.
phase behavior.By selecting Gmb1 𝐺
By selecting ) =(𝑠)
(s𝑚𝑏1 s+=1 for the
forfirst looploop
the first and
𝑠+1
0.5 0.5
G ( s )
mb2𝐺𝑚𝑏2 (𝑠)
and = s+=1 for the second loop, as shown in Figures 3 and 4, the dominant
for the second loop, as shown in Figures 3 and 4, the dominant interest interest
𝑠+1
condition for the highest model error range norm, with the conditions mentioned, is met.
condition for the highest model error range norm, with the conditions mentioned, is met.
Figures 5 and 6 show the frequency response diagram for the open-loop function of the
Figures 5 and 6 show the frequency response diagram for the open-loop function of the
first and second loops in the proposed control structure.
first and second loops in the proposed control structure.

Mathematics 2022, 10, x FOR PEER REVIEW 10 of 1

Figure
Figure 4.
4. The
The proposed
proposed type
type 22 fuzzy
fuzzy system.
system.

(a) (b)
Figure 5. Frequency
Figure 5. Frequency response ofresponse of function
open-loop open-loopforfunction
the first for the(a)
loop. first loop. (a) Ratio
Amplitude Amplitude Ratio (b)
(b) Phase.
Phase
 (a) (b)
(a) (b)
Figure 5. Frequency response of open-loop function for the first loop. (a) Amplitude Ratio (b)
Mathematics 2022, 10, 3696 10 of 19
Phase Figure 5. Frequency response of open-loop function for the first loop. (a) Amplitude Ratio (b)
Phase

(a) (b)
(a) (b)
Figure 6. Frequency response of open-loop function for the second loop. (a) Amplitude Ratio (b)
Phase Figureresponse
Figure 6. Frequency 6. Frequency response
of open-loop of open-loop
function function
for the second for (a)
loop. theAmplitude
second loop. (a)(b)
Ratio Amplitude
Phase. Ratio (
Phase
As can
As can bebe seen
seen in in these
thesefigures,
figures,the thecondition
conditionfor forthe
thesuperiority
superiorityofofthe theGGmb,n (s) gain
mb,n(s) gain
in most frequencies
in most frequencies is such that
As can be seen in these the phase
phase diagram
figures,
diagram is limited
the condition
is limited for before reaching
the superiority
before the
reaching theof−180° − the ◦Gmb,n (s)
180
point in
point in these
these two
intwo
most loops. Therefore,
frequencies
loops. Therefore, ititcan
is such can
that bethe
be concluded
phase diagram
concluded thatthe
that theis selection
limitedof
selection of GGmb,ii
before (s) is
is the −
reaching
mb,ii(s)
appropriate in
appropriate in both
both
point loops.
inloops. Additionally,
these Additionally,
two the minimum
loops. Therefore,
the minimum
it can be phase behavior
concluded
phase behavior thatof the
of the selection
open-loop
open-loop of Gmb,ii (s
function in
function in the
the proposed method
appropriate
proposed method
in both is is one
one of
loops. of the advantages
advantages
Additionally,
the the ofof this method
minimum
this methodphase compared
behaviorto
compared toofthe
the open-l
the
methods in
methods in which
which
function thein
the open-loop
the proposed
open-loop function
function method of the
of the control
is one of the
control system has non-minimum
advantages
system has non-minimum
of this method phase
compared to
phase
behavior. Figures
behavior. Figures methods7–10
7–10 show
in show the
which the performance
the open-loop
performance of Smith’s
function predictive control
of the predictive
of Smith’s control system systems,
control Smith’s
has systems,
non-minimum ph
predictor
Smith’s with
predictor dominant
behavior.
with dominant gain, and
Figures 7–10
gain,the proposed
show
and control
theproposed
the performance system
control (Smith’s
ofsystem
Smith’s predictor
predictive
(Smith’s with
control syste
predictor
dominant
with dominant gain and
Smith’s type-2
gain and fuzzy
predictor
type-2with estimator
fuzzy dominant
estimatorfor time
gain, delay).
and the
for time In this
proposed
delay). example,
In thiscontrol
example, y is
system distillate
1 y1 is (Smith’s
dis- predi
composition,
tillate y2 isdominant
with
composition, bottom composition,
gain and
y2 is bottom type-2
composition, u isand
andfuzzy refluxu flow
estimator rate.
for
is reflux timeIt should
flow rate. be
delay). In noted
It this
should that
example,
be y1 is
the reference for y
tillate is the step
composition,
noted that the reference for y1 is the
1 signal
y with
is the
bottom step time
composition,t = 50 s,
andand u for
is y ,
reflux
2 step signal with the step time t = 50 s, and for y2 , this2 this is
flow the step
rate. It should
with
the the
stepstep time
noted t = 200
that the s. tThe
= 200control
reference signals
Theycontrol
s. for ustep
1 and uu2 are shown inshown
Figures t =8 50and 9, for y ,
is with the step time 1 is thesignals signal with
1 and uthe
2 arestep time in s, and
Figures 2
respectively. is the step with the step time t = 200 s. The control signals u1 and u2 are shown in Fig
8 and 9, respectively.
8 and 9, respectively.

Mathematics 2022, 10, x FOR PEER REVIEW 11 of 19

Figure 7.
Figure 7. Performance
Performance of
of three control methods
three control for y𝑦1for
methods for forexample
example1.1.
1

Figure 8.
Figure 8. Performance
Performance of
of three control methods
three control for y𝑦22for
methods for forexample
example1.1.
Mathematics 2022, 10, 3696 11 of 19
Figure
Figure 8.
8. Performance
Performance of
of three
three control
control methods for 𝑦𝑦22 for
methods for for example
example 1.
1.

Figure
Figure9.
Figure 9.The
9. Thecontrol
The control signal 𝑢u
controlsignal
signal for
𝑢111 for all
forall three
allthree methods
threemethods for
methodsfor example1.
forexample
example1.
1.

Figure
Figure10.
Figure 10.The
10. Thecontrol
The control signal 𝑢u
controlsignal
signal 𝑢22 for
forall three
allthree methods
threemethods for
methodsfor example
forexample 1.
example1.
1.

Itis
It
It isclear
is clearfrom
clear fromFigures
from Figures7–10
Figures 7–10that
7–10 thatthe
that theproposed
the proposedmethod
proposed method(Smith’s
method (Smith’spredictor
(Smith’s predictorwith
predictor withdom-
with dom-
dom-
inantgain
inant
inant gainand
gain andtype-2
and type-2fuzzy
type-2 fuzzyestimator)
fuzzy estimator)
estimator) performs
performs
performs better
better
better both
both
both in in terms
in terms
terms of of accuracy
of accuracy
accuracy andand
and re-
re-
response
sponse speed.
speed. In In addition,
addition, thethe proposed
proposed method
method has
has a a minimum
minimum
sponse speed. In addition, the proposed method has a minimum control cost. In order control
control cost.
cost. In
In order
order to
to
Mathematics 2022, 10, x FOR PEER REVIEW 12 of to
19
challenge
challenge the
the control
control systems,
systems, aa disturbance
disturbance signal
signal (a
(a pulse
pulse with
with a
a height
height
challenge the control systems, a disturbance signal (a pulse with a height of 1 and a width of
of 11 and
and aa width
width
of222s)
of
of s)is
s) isapplied
is appliedat
applied atthe
at the momentttt===150
themoment
moment 150sss(Figures
150 (Figures11–14).
(Figures 11–14).
11–14).

Figure
Figure11.
11.Performance
Performanceof
ofthree
threecontrol
controlmethods for 𝑦
methodsfor by applying
y11 by applying disturbance.
disturbance.
Mathematics 2022, 10, 3696 12 of 19
Figure 11.
Figure 11. Performance
Performance of
of three
three control
control methods for 𝑦𝑦1 by
methods for by applying
applying disturbance.
disturbance.
1

Figure
Figure 12.
Figure12. Performance
12.Performance of
Performanceof three
ofthree control
threecontrol methods
controlmethods for 𝑦y𝑦222 by
for
methodsfor byapplying
by applyingdisturbance.
applying disturbance.
disturbance.

Mathematics 2022, 10, x FOR PEER REVIEW 13 of 19

Figure 13.
Figure13.
Figure The control
Thecontrol
13.The signal 𝑢11 for
signal 𝑢u
controlsignal for all
allthree
threemethods
methodsby
byapplying
applyingdisturbance.
disturbance.
disturbance.
1

Figure 14.
Figure 14. The control signal u𝑢22for
control signal forall
allthree
threemethods
methodsby
byapplying
applyingdisturbance.
disturbance.

It
It can
can bebe seen
seen from
fromFigures
Figures11–14
11–14that
thatthe
theperformance
performanceofofthe theproposed
proposedmethod
method with
witha
type-2 fuzzy estimator is very good, and the system reaches its equilibrium
a type-2 fuzzy estimator is very good, and the system reaches its equilibrium point in less point in less
than 2020 s.
s. In the meantime,
meantime, the worst performance is related to Smith’s Smith’s pure
pure predictive
predictive
method,
method, which
which fluctuates
fluctuates moremore than
than 200
200 ss after
after applying
applyingdisturbance.
disturbance.
Another
Another challenge
challenge for for the
the control
controlsystem
systemisisthetheuncertainty
uncertaintyor orchanges
changesin
inthe
theparame-
param-
ters of the controlled system (process). This challenge inevitably arises
eters of the controlled system (process). This challenge inevitably arises because systems because systems
wear
wear out
out over
over time
time and
and their
their behavior
behavior changes. Next, it
changes. Next, it is
is assumed
assumed that
that the
the system
system pa-
pa-
rameters (numerator and denominators coefficients of the transfer function
rameters (numerator and denominators coefficients of the transfer function as well as time as well as time
delay)
delay) are
are doubled
doubled at at tt == 150
150 ss (Figures
(Figures 15–18).
15–18).
 than 20 s. In the meantime, the worst performance is related to Smith’s pure predictive
method, which
method, which fluctuates
fluctuates more
more than
than 200
200 ss after
after applying
applying disturbance.
disturbance.
Another challenge for the control system is the uncertainty
Another challenge for the control system is the uncertainty or or changes
changes inin the
the param-
param-
eters of
eters of the
the controlled
controlled system
system (process).
(process). This
This challenge
challenge inevitably
inevitably arises
arises because
because systems
systems
wear out over time and their behavior changes. Next, it is assumed that
wear out over time and their behavior changes. Next, it is assumed that the system pa- the system pa-
Mathematics 2022, 10, 3696 rameters (numerator and denominators coefficients of the transfer function as
rameters (numerator and denominators coefficients of the transfer function as well as time well as
13time
of 19
delay) are doubled at t = 150 s (Figures 15–18).
delay) are doubled at t = 150 s (Figures 15–18).

Figure 15. Performance

Figure Performance of three methods for 𝑦𝑦1 by
three control methods by applying
applying uncertainty.
uncertainty.
Figure 15.
15. Performance of
of three control
control methods for
for y11 by applying uncertainty.

Mathematics 2022, 10, x FOR PEER REVIEW 14 of 19

Figure 16.
Figure Performance of
16. Performance of three
three control
control methods for y𝑦𝑦22by
methods for byapplying
applyinguncertainty.
uncertainty.
2 by applying uncertainty.

Figure 17.
Figure 17. The control signal u𝑢11for
control signal forall
allthree
threemethods
methodsby
byapplying
applyinguncertainty.
uncertainty.

Figure 18. The control signal 𝑢2 for all three methods by applying uncertainty.
Mathematics 2022, 10, 3696 14 of 19

Figure 17. The control signal 𝑢1 for all three methods by applying uncertainty.

Figure 18. The control signal u𝑢22for

forall
allthree
threemethods
methodsby
byapplying
applyinguncertainty.
uncertainty.

Interesting results
results can
can be
beseen
seenininFigures
Figures11–18.
11–18.Smith’s
Smith’s pure
pure prediction
prediction method
method di-
diverges
verges against large changes in parameters. Since after the initial adjustment Smith’s
against large changes in parameters. Since after the initial adjustment of
prediction method, it is assumed that the system does not change behavior (or at least
changes
changes very
very little),
little), this
this method
method has has nothing
nothing to to say
say in
in the
the face
face of
of major
major changes.
changes. On On the
the
other
other hand,
hand, itit is observed that
is observed that the convergence of
the convergence of the
the control
control system
system isis adjusted
adjusted by adding
by adding
the dominant gain
the dominant gaintransfer
transferfunction
function(𝐺(G 𝑚𝑏mb
) to
) to Smith’s
Smith’s predictive
predictive method.
method. Finally,
Finally, withwith
the
the addition of a type-2 fuzzy estimator to the control system, the accuracy
addition of a type-2 fuzzy estimator to the control system, the accuracy and speed of con- and speed of
convergence is dramatically
vergence is dramatically improved. It should be noted that for example
improved. It should be noted that for example 1, the phase mar- 1, the phase
◦
margin is° 43
gin is 43 andand thethe gain
gain margin
margin is 18.6
is 18.6 dB.dB.
Example
Example 2: A 33×
2: A ×33subsystem
subsystemofofthe theshell
shellheavy-oil
heavy-oilfractionator
fractionatorisisasasfollows
follows[31]:
[31]:
4.05e− 27s 1.77e−28s
−28𝑠 5.88e−27s −27𝑠
 
−27𝑠
4.05𝑒50s+1 1.77𝑒
60s+1 5.88𝑒
50s +1
 5.39e−18s 5.27e−8s −15s 
G (s) = 50𝑠 +1
50s+1
60𝑠 + 1 6.9e
60s+1
50𝑠 + 1
60s+1 
−18𝑠 −8𝑠
5.39𝑒
4.38e−20s 5.27𝑒
4.42e −22s 6.9𝑒 −15𝑠
7.2
𝐺(𝑠) = 33s+1 44s+1 19s+1
50𝑠 + 1 60𝑠 + 1 60𝑠 + 1
−20𝑠 −22𝑠
In order to avoid cluttering the 4.38𝑒 article 4.42𝑒
and confusing 7.2readers, we do not intend
the
to carry out all the steps in Example [ 33𝑠 +1 1for this 44𝑠example,
+1 19𝑠 but 1 ] to show the ability
+ only
of the control system for the system with any number of inputs and outputs and any
amount In order to avoid cluttering
of interaction. Based on thethis
article and confusing
method, the readers,
for this example, thewe do not intend
parameters to
of the
carry out all the steps in Example 1 for this example, but
controllers for the three control loops are obtained as Gc1 = 0.662 + 0.0365 only to show the ability of the
0.019
s , Gc2 = 1 + s
control system for 0.077the system with any number of inputs and outputs and any amount of
and Gc3 = 4 + s . Figures 19–21 show the performance of Smith’s predictive control
interaction. Based on this method, for this example, the parameters of the controllers for
systems, Smith’s predictor with dominant gain, and the proposed control
0.0365 0.019system (Smith’s
the three control
predictor loops aregain
with dominant obtained Gc1 =fuzzy
and aastype-2 0.662estimator
+ , Gfor
c2 = 1 +delay).
time and Gc3 = 4be
It should +
s s
0.077
noted that the19–21
. Figures reference
show y1 is
forthe the step signal
performance step time tcontrol
with thepredictive
of Smith’s for y2 , this
= 300 s;systems, is the
Smith’s
s
step with the step time t = 600 s; and for y3 , this is the step with the step time t = 900. The
control signals u1 , u2 , and u3 are shown in Figures 22–24, respectively. In this example, y1
is the top endpoint composition, y2 is the side end point composition, and y3 is the bottom
reflux temperature. The inputs are the top drawn flow rate (u1 ), the side drawn flow rate
(u2 ), and the bottom reflux heat duty (u3 ). It should be noted that the reference for y1 is the
step signal with the step time t = 300 s; for y2 , this is the step with the step time t = 600 s;
and for y3 , this is the step with the step time t = 900.
 the
the step
thestep time
timettt===600
steptime 600
600s;s;s;and
and foryyy333,,,this
for
andfor this isisthe
thisis the step
thestepstepwith with
withthethe step
thestep
steptime time
timettt===900.
900. The
900.The control
Thecontrol
control
signals
signals u
u , u
u , and u
u are shown in Figures 22–24, respectively.
signals u111 , u222 , and u333 are shown in Figures 22–24, respectively. In this example, y111 is
, , and are shown in Figures 22–24, respectively. In
In this
this example,
example, yy isisthe
the
the
top
top endpoint
endpoint composition,
composition, yy is
is the
the side
side end
end point
point
top endpoint composition, y222 is the side end point composition, and y333 is the bottom composition,
composition, and
and y
y is
is the
the bottom
bottom
reflux
reflux temperature.
refluxtemperature.
temperature.The The inputs
Theinputs are
inputsarearethe the top
thetop drawn
topdrawndrawnflow flow rate
flowraterate(u(u111),),),the
(u the side
theside drawn
sidedrawn
drawnflowflow
flowraterate
rate
(u
(u ),
), and the bottom reflux heat duty (u ).
). It should be
(u2 ), and the bottom reflux heat duty (u3 ). It should be noted that the reference for y11 is
22 and the bottom reflux heat duty (u 33 It should be noted
noted that
that the
the reference
reference for
for yy 1 isis
Mathematics 2022, 10, 3696 the
the step
step signal
signal with
with the
the step
step time
time tt == 300
300 s;
s; for
for yy ,, this
this is
is the
the step
step with
with the
the step
step time
time tt15
=
= of
600
600 19
the step signal with the step time t = 300 s; for y2 , this is the step with the step time t = 600
2
2
s;s;and
s; and
andforforyyy33,,,this
for
3
this isisthe
thisis the step
thestep with
stepwith
withthe the step
thestep time
steptime timettt===900.900.
900.

Figure
Figure
Figure
19.
Figure19.
19.
Performance
19.Performance of
Performanceof three
ofthree control
threecontrol
three
methods
controlmethods
control for𝑦y𝑦𝑦
for
methodsfor
methods for
for
11 for
11for
example
example
forexample 2.2.
2.
example2.

Figure
Figure
Figure 20.
Figure20.
20. Performance
20.Performance of
Performanceof three
ofthree control
threecontrol
three methods
controlmethods
control for𝑦y𝑦𝑦
for
methodsfor
methods for forexample
for
222 for example2.
example 2.
2.
2 for example 2.

Figure
Figure
Figure 21.
Figure21.
21. Performance
21.Performance of
Performanceof
Performance three
ofthree
of control
threecontrol
three methods
controlmethods
control for𝑦y𝑦𝑦
for
methodsfor
methods for 333... .
3
Mathematics
Mathematics 2022, 10, x FOR
FORPEER
PEERREVIEW 16 of
ofof19
Mathematics 2022,
2022, 10,
10, x3696 REVIEW 1616 19
19

Figure 22. The control signal 𝑢1 for

forall
allthree
threemethods
methodsfor
forexample
example2.
Figure
Figure 22.
22. The
Thecontrol signal 𝑢
controlsignal u11 for all three methods for example 2.
2.

Figure 23. The control signal 𝑢2 forforall

allthree
threemethods
methodsfor
forexample 2.
Figure
Figure 23.
23. The
Thecontrol signal 𝑢
controlsignal u22 for all three methods for example2.
example 2.

Figure
Figure24.
Figure 24.The
24. Thecontrol
The control signal𝑢𝑢
controlsignal
signal u33 for all three methods.
3 for all three methods.

Example222isis
Example
Example isaaasystem
systemwith
system withhigh
with highlatency
high latencyand
latency andrelatively
and relativelystrong
relatively stronginteraction.
strong interaction.ItItItcan
interaction. canbe
can be
be
seen fromFigures
seen Figures 18–20 that that the proposed
proposed has been able to successfully provide a
seen from
from Figures 18–2018–20 that the the proposed method
method has been able able to
to successfully
successfully provide
provide
aasuitable
suitable answer.
answer. ItItisiscarefully
carefully observed
observed ininFigure
Figure 1818
that byby
that applying
applying thethe
suitable answer. It is carefully observed in Figure 18 that by applying the second step,
second
secondstep, the
step,
Smith’s pure
the predictive method has moremorethan 150% 150%overshoot, Smith’sSmith’s
predictive method
the Smith’s
Smith’s pure
pure predictive
predictive method
method hashas more thanthan 150% overshoot,
overshoot, Smith’s predictive
predictive
with
method dominant gain has 60%, and finally, this method with a type-2 fuzzy estimatorestima-
has less
methodwithwithdominant
dominantgain gainhashas60%,
60%,and
andfinally,
finally,this
thismethod
methodwithwithaatype-2
type-2fuzzy
fuzzy estima-
40%.
tor The minimum cost of the control signal of the proposed method is clearly shown in
torhas
hasless
less40%.
40%.The
Theminimum
minimumcost costof
ofthe
thecontrol
controlsignal
signalofofthe
theproposed
proposedmethodmethodisisclearly
clearly
Figure
shown 22. Table 1 shows a comparison between some methods based on root mean square
shown in in Figure
Figure 22.
22. Table
Table 11 shows
shows aa comparison
comparison between
between some
some methods
methods based based onon root
root
Mathematics 2022, 10, 3696 17 of 19

◦
error (RMSE) criterion. It should be noted that for example 2, the phase margin is 29 and
the gain margin is 12.2 dB.

Table 1. A comparison between some methods for example 2.

y1 y2 y3
SP 0.135 0.089 0.051
SP + Gmb 0.073 0.059 0.038
SP + Gmb + T2Fuzzy 0.061 0.045 0.032
Method of [43] 0.063 0.055 0.040
Method of [44] 0.061 0.049 0.034

It can be seen from Table 1 that the proposed method has the best answer in terms of
accuracy (RMSE criterion). Although in control y1 , the method of [41] has an RMSE equal
to our method, for the other two outputs, our proposed method works better.

5. Conclusions
In this paper, a new method based on Smith’s predictor for controlling time-delayed
MIMO systems is presented. The proposed control system had two phases: one is the
addition of a transfer function with a dominant gain to stabilize the system, and the second
phase is the use of a type-2 fuzzy estimator to estimate the system time delay online. In the
simulation with two examples, it was shown that the proposed method is very robust and
can control the system (process) well with proper accuracy and speed of response. Control
challenges, disturbance and parametric uncertainty were also applied, which showed that
the proposed system is highly efficient in facing these two challenges. The simulation
results clearly showed how much the addition of the transfer function with the dominant
gain can improve the performance of the control system. In addition, a type-2 fuzzy system
with the ability to accurately estimate the amount of time delay was able to provide a
very suitable answer. As a suggestion to continue the work, in addition to estimating the
amount of time delay, the process parameters can also be estimated using the fuzzy system.
In addition, other more accurate estimators (type-3 fuzzy, etc.) can be used.

Author Contributions: Conceptualization, X.G. and M.S.; data curation, X.G., M.S. and E.M.A.; for-
mal analysis, X.G. and M.S.; funding acquisition, E.M.A.; investigation, X.G. and M.S.; methodology,
X.G.; project administration, X.G.; resources, X.G., M.S. and E.M.A.; software, X.G. and E.M.A.;
supervision, X.G. and E.M.A.; validation, X.G., M.S. and E.M.A.; visualization, X.G. and E.M.A.;
writing—original draft, X.G. and M.S.; writing—review and editing, X.G. and M.S. All authors have
read and agreed to the published version of the manuscript.
Funding: This paper is the research result of the 5th Jiangsu Province vocational education teaching
reform research topic “Research on the Quality Evaluation of Vocational College Talent Training
under the 1+X Certificate System” (ZYB564) project, which is also supported by the 2022 Jiangsu
Universities “Qinglan Project” Excellent Teaching Team Project.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: The data are available upon request due to restrictions, e.g., privacy or
ethical. The data presented in this study are available on request from the corresponding author. The
data are not publicly available due to personal reasons.
Conflicts of Interest: The authors declare no conflict of interest.
Mathematics 2022, 10, 3696 18 of 19

Nomenclature
G(s) Exact system model
τd Time delay
Gm (s) Measurement transfer function
Yn (S) Response matrix
Ysp,n (S) Reference input matrix
dn (S) Input perturbation matrix
Gmb,n (S) Matrix of predictive functions
Gc,n S Matrix functions of the controllers
Gp,n (S) Matrix of process including time delay
Gn (s) Matrix of process without time delay

References
1. An, Q.; Guo, H.; Zheng, Y. On Robust Stability and Stabilization of Networked Evolutionary Games with Time Delays. Mathematics
2022, 10, 2695. [CrossRef]
2. Liu, X.; Li, W.; Yao, C.; Li, Y. Finite-Time Guaranteed Cost Control for Markovian Jump Systems with Time-Varying Delays.
Mathematics 2022, 10, 2028. [CrossRef]
3. Huang, H.; Shirkhani, M.; Tavoosi, J.; Mahmoud, O. A New Intelligent Dynamic Control Method for a Class of Stochastic
Nonlinear Systems. Mathematics 2022, 10, 1406. [CrossRef]
4. Tavoosi, J.; Shirkhani, M.; Azizi, A.; Din, S.U.; Mohammadzadeh, A.; Mobayen, S. A hybrid approach for fault location in power
distributed networks: Impedance-based and machine learning technique. Electr. Power Syst. Res. 2022, 210, 108073. [CrossRef]
5. Danyali, S.; Aghaei, O.; Shirkhani, M.; Aazami, R.; Tavoosi, J.; Mohammadzadeh, A.; Mosavi, A. A New Model Predictive Control
Method for Buck-Boost Inverter-Based Photovoltaic Systems. Sustainability 2022, 14, 11731. [CrossRef]
6. Chuong, V.L.; Vu, T.N.L.; Truong, N.T.N.; Jung, J.H. An Analytical Design of Simplified Decoupling Smith Predictors for
Multivariable Processes. Appl. Sci. 2019, 9, 2487. [CrossRef]
7. Tavoosi, J.; Shirkhani, M.; Abdali, A.; Mohammadzadeh, A.; Nazari, M.; Mobayen, S.; Asad, J.H.; Bartoszewicz, A. A New General
Type-2 Fuzzy Predictive Scheme for PID Tuning. Appl. Sci. 2021, 11, 10392. [CrossRef]
8. Feliu-Batlle, V.; Rivas-Perez, R. Control of the temperature in a petroleum refinery heating furnace based on a robust modified
Smith predictor. ISA Trans. 2020, 112, 251–270. [CrossRef]
9. Sbaaei, E.S.; Fouad, M.M.K.; Ahmed, T.S. Dynamic Modeling and Analysis of Process Constraints for Improvement for an
Industrial Unipol®Polypropylene Fluidized Bed Reactor. Arab. J. Sci. Eng. 2021, 47, 6135–6154. [CrossRef]
10. Mohammadi, F.; Mohammadi-Ivatloo, B.; Gharehpetian, G.B.; Ali, M.H.; Wei, W.; Erdinç, O.; Shirkhani, M. Robust control
strategies for microgrids: A review. IEEE Syst. J. 2021, 16, 2401–2412. [CrossRef]
11. Yang, Z.; Yang, F.; Hu, W.; Zhang, Z.; Zhou, X. Delay Compensation Control Strategy for Electric Vehicle Participating in
Frequency Regulation Based on MPC Algorithm. Electronics 2022, 11, 2341. [CrossRef]
12. Karrakchou, J.; Lahmidi, F.; Namir, A.; Rachik, M. Stability, observers and compensators for discrete-time delay systems 1. In
Systems Modelling and Optimization; Routledge: London, UK, 2022; pp. 71–79.
13. Aazami, R.; Heydari, O.; Tavoosi, J.; Shirkhani, M.; Mohammadzadeh, A.; Mosavi, A. Optimal Control of an Energy-Storage
System in a Microgrid for Reducing Wind-Power Fluctuations. Sustainability 2022, 14, 6183. [CrossRef]
14. Shah, D.H.; Mehta, A. Discrete-Time Sliding Mode Control for Networked Control System; Springer: Singapore, 2018.
15. Li, Y.; Liu, G.-P.; Sun, S.; Tan, C. Prediction-based approach to finite-time stabilization of networked control systems with time
delays and data packet dropouts. Neurocomputing 2019, 329, 320–328. [CrossRef]
16. Sharkawy, A.N.; Koustoumpardis, P.N.; Aspragathos, N. Human–robot collisions detection for safe human–robot interaction
using one multi-input–output neural network. Soft Comput. 2020, 24, 6687–6719. [CrossRef]
17. Alavudeen Basha, A.; Vivekanandan, S. A fuzzy-based adaptive multi-input–output scheme in lieu of diabetic and hypertension
management for post-operative patients: An human–machine interface approach with its continuum. Neural Comput. Appl. 2020,
13, 1–7. [CrossRef]
18. Kim, D.-E.; Li, A.; Dau, M.-N.; Kim, H.-H.; Chung, W.-Y. Deep learning-based smith predictor design for a remote grasping
control system. J. Mech. Sci. Technol. 2022, 36, 2533–2545. [CrossRef]
19. Giraldo, S.A.C.; Flesch, R.C.C.; Normey-Rico, J.E.; Sejas, M.Z.P. A method for designing decoupled filtered Smith predictor for
square MIMO systems with multiple time delays. IEEE Trans. Ind. Appl. 2018, 54, 6439–6449. [CrossRef]
20. Elliott, A.J.; Cammarano, A.; Neild, S.A.; Hill, T.L.; Wagg, D.J. Comparing the direct normal form and multiple scales methods
through frequency detuning. Nonlinear Dyn. 2018, 94, 2919–2935. [CrossRef]
21. Shah, N.A.; Agarwal, P.; Chung, J.D.; El-Zahar, E.R.; Hamed, Y.S. Analysis of Optical Solitons for Nonlinear Schrödinger Equation
with Detuning Term by Iterative Transform Method. Symmetry 2020, 12, 1850. [CrossRef]
22. Ruan, S.; Xiong, S.; You, J.; Yang, Y. Generation of streamwise helical vortex loops via successive reconnections in early pipe
transition. Phys. Fluids 2022, 34, 054112. [CrossRef]
Mathematics 2022, 10, 3696 19 of 19

23. Huang, J.; Wang, H.; Yang, H. Int-deep: A deep learning initialized iterative method for nonlinear problems. J. Comput. Phys.
2020, 419, 109675. [CrossRef]
24. Emamjomeh, M.M.; Mousazadeh, M.; Mokhtari, N.; Jamali, H.A.; Makkiabadi, M.; Naghdali, Z.; Hashim, K.; Ghanbari, R.
Simultaneous removal of phenol and linear alkylbenzene sulfonate from automotive service station wastewater: Optimization of
coupled electrochemical and physical processes. Sep. Sci. Technol. 2020, 55, 3184–3194. [CrossRef]
25. Guerrier, S.; Dorize, C.; Awwad, E.; Renaudier, J. Introducing coherent MIMO sensing, a fading-resilient, polarization-independent
approach to φ-OTDR. Opt. Express 2020, 28, 21081–21094. [CrossRef] [PubMed]
26. Huang, L.; Xin, H.; Li, Z.; Ju, P.; Yuan, H.; Lan, Z.; Wang, Z. Grid-synchronization stability analysis and loop shaping for
PLL-based power converters with different reactive power control. IEEE Trans. Smart Grid 2019, 11, 501–516. [CrossRef]
27. Euzébio, T.A.; Yamashita, A.S.; Pinto, T.V.; Barros, P.R. SISO approaches for linear programming based methods for tuning
decentralized PID controllers. J. Process Control. 2020, 94, 75–96. [CrossRef]
28. Garrido, J.; Ruz, M.L.; Morilla, F.; Vázquez, F. Iterative Method for Tuning Multiloop PID Controllers Based on Single Loop
Robustness Specifications in the Frequency Domain. Processes 2021, 9, 140. [CrossRef]
29. Prieto, P.J.; Cazarez-Castro, N.R.; Aguilar, L.T.; Cardenas-Maciel, S.L. Self-tuning for a SISO-type fuzzy control based on the relay
feedback approach. In Numerical and Evolutionary Optimization; Springer: Cham, Switzerland, 2017; pp. 171–186.
30. Carlucho, I.; De Paula, M.; Acosta, G.G. An adaptive deep reinforcement learning approach for MIMO PID control of mobile
robots. ISA Trans. 2020, 102, 280–294. [CrossRef]
31. Veerasamy, V.; Wahab, N.I.A.; Ramachandran, R.; Othman, M.L.; Hizam, H.; Kumar, J.S.; Irudayaraj, A.X.R. Design of single-and
multi-loop self-adaptive PID controller using heuristic based recurrent neural network for ALFC of hybrid power system. Expert
Syst. Appl. 2022, 192, 116402. [CrossRef]
32. Çelik, E.; Öztürk, N. A hybrid symbiotic organisms search and simulated annealing technique applied to efficient design of PID
controller for automatic voltage regulator. Soft Comput. 2018, 22, 8011–8024. [CrossRef]
33. Li, M.; Cheung, S. Isolation enhancement for MIMO dielectric resonator antennas using dielectric superstrate. IEEE Trans.
Antennas Propag. 2020, 69, 4154–4159. [CrossRef]
34. Luan, X.; Chen, Q.; Albertos, P.; Liu, F. Conversion of SISO processes with multiple time-delays to single time-delay processes. J.
Process Control. 2018, 65, 84–90. [CrossRef]
35. Pang, H.; Fu, W.-Q.; Liu, K. Stability analysis and fuzzy smith compensation control for semi-active suspension systems with
time delay. J. Intell. Fuzzy Syst. 2015, 29, 2513–2525. [CrossRef]
36. Huang, H.; Zhang, S.; Yang, Z.; Tian, Y.; Zhao, X.; Yuan, Z.; Hao, S.; Leng, J.; Wei, Y. Modified Smith fuzzy PID temperature
control in an oil-replenishing device for deep-sea hydraulic system. Ocean. Eng. 2018, 149, 14–22. [CrossRef]
37. Tang, Y.; Du, F.; Cui, Y.; Zhang, Y. New Smith predictive fuzzy immune PID control algorithm for MIMO networked control
systems. EURASIP J. Wirel. Commun. Netw. 2018, 2018, 1–5. [CrossRef]
38. Iranmehr, H.; Aazami, R.; Tavoosi, J.; Shirkhani, M.; Azizi, A.-R.; Mohammadzadeh, A.; Mosavi, A.H.; Guo, W. Modeling the
Price of Emergency Power Transmission Lines in the Reserve Market Due to the Influence of Renewable Energies. Front. Energy
Res. 2022, 13, 9. [CrossRef]
39. Dehghani, A.; Khodadadi, H. Designing a neuro-fuzzy PID controller based on smith predictor for heating system. In Proceedings
of the 17th International Conference on Control, Automation and Systems (ICCAS), Jeju, Korea, 18 October 2017; IEEE: Piscataway
Township, NJ, USA; pp. 15–20.
40. Rong, H.-G.; Zheng, H.; Li, Z.-Q.; Xia, Y.-X. Tuning of fuzzy PID controller for Smith predictor. J. Cent. South Univ. Technol. 2010,
17, 566–5671. [CrossRef]
41. Fortuna, L.; Frasca, M.; Buscarino, A. Optimal and Robust Control: Advanced Topics with Matlab®; CRC Press: Boca Raton, FL, USA, 2021.
42. Tavoosi, J.; Suratgar, A.A.; Menhaj, M.B. Nonlinear system identification based on a self-organizing type-2 fuzzy RBFN. Eng.
Appl. Artif. Intell. 2016, 54, 26–38. [CrossRef]
43. Nagarsheth, S.H.; Sharma, S.N. The combined effect of fractional filter and Smith Predictor for enhanced closed-loop performance
of integer order time-delay systems: Some investigations. Arch. Control. Sci. 2020, 30, 47–76.
44. Li, Z.; Bai, J.; Zou, H. Modified two-degree-of-freedom Smith predictive control for processes with time-delay. Meas. Control.
2020, 53, 691–697. [CrossRef]