You are on page 1of 23

This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2017.Doi Number

A fractional order general type-2 fuzzy PID


controller design algorithm
Shi Jian Zhong
School of Energy and Power Engineering, Nanjing Institute of Technology, Nanjing 211167, P.R. China
Corresponding author: Shi Jian Zhong (e-mail: sjz-ha@163.com).
This work was supported by the scientific research fund project of Nanjing Institute of Technology (YKJ201523).

ABSTRACT Fractional order PID controller was received attentions in control problems for it had 2
freedom adjustable parameters. Practices had proved that better results could be obtained by introduction of
fractional order PID in control problems. And in recently years, fractional order PID combined with fuzzy
logic system was gradually got attention, the typical of which was fractional order type-1 fuzzy PID
controller. For fractional order PID and fractional order type-1 fuzzy PID controller couldn't deal with the
uncertainty of system, so fractional order interval type-2 fuzzy PID controller was applied to solve this
problem. But interval type-2 fuzzy sets were a simplification of type-2 fuzzy sets, and there may be have
the defect of information loss. A fractional order general type-2 fuzzy PID controller was proposed in this
article. The proposed controller based on general type-2 fuzzy logic system, which can make full use of the
advantages of general type-2 fuzzy logic system in describing the uncertainty of the system. The fractional
order general type-2 fuzzy PID controller utilized a simplified type reduction called NT type reduction
algorithm. The NT type reduction algorithm can get the defuzzification result directly and avoided iterative
process as KM type reduction commonly used in interval type-2 fuzzy controller. The simulations of 3
processes and a practical inverted pendulum system show that fractional order general type-2 fuzzy PID
controller can reduce overshoot, improve system response speed and accelerate system stability time in
comparing with other controllers. Especially, when the system has disturbance, parameters uncertainty or
structure uncertainty, the fractional order general type-2 fuzzy PID controller has better control effects than
other compared controllers.

INDEX TERMS Fractional order control; Fractional order fuzzy PID; General type-2 fuzzy sets; General
type-2 fuzzy logic system; Type reduction

I. INTRODUCTION control the system more flexibly and obtain better control
Fractional calculus was an extension of integer calculus, effect. There were some successful FOPID applications,
and the integral or differential order of fractional calculus such as AVR system [2], power electronic [3], robotic systems
was not conventional integer number but real or even [4]
, boiler-turbine systems [5] and so on.
complex one. Fractional calculus provided a powerful With the development of fuzzy logic system theory,
support for the extension of the classic research methods in many researchers focused on combining fractional
control theory and a better expiation of the current results. mathematics with fuzzy logic controller, one of the hot
Many researches paid attentions to fractional order control points was fractional order fuzzy PID controller (FO-FPID).
problems based on fractional order operators and theory of For example, a novel FO-FPID controller was applied in
fractional order differential equations. Among these nonlinear process and an open loop unstable process with
fractional order control problems, fractional order PID time delay control problems [6], an enhanced FO-FPID
(FOPID) controller was widely studied, which was first controller was used for controlling robotic system [7], an
proposed by Podlubny and can be represented as PIλDμ [1]. expert 2DOF FO-FPID controller for nonlinear systems [8].
As the introduction of integral and differential order λ and μ, The FO-FPID was also applied in binary distillation column
FOPID had 2 more adjustable parameters compared with control [9], fractional order processes [10], robot application
tradition PID controller. For the tuning range of controller [11]
, copper removal process [12], grid connected variable
parameters become larger, the FOPID controller could speed wind energy conversion system [13], pumped storage

VOLUME XX, 2017 1

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

hydro unit [14], and so on. The simulation results showed pattern recognition [42-43] and interval type-2 fuzzy logic
that the control performances of FO-FPID was better than applications in intelligent control [44]. There were some
conventional fuzzy PID controller and FOPID controller. others reviews on type-2 fuzzy logic systems, such as
Although FOPID and FO-FPID controller achieved some reviews on industrial applications of type-2 fuzzy sets and
successful applications, but they had the disadvantage that systems [45], type-reduction of type-2 fuzzy sets reviews [46],
they couldn't deal with linguistic and numerical reviews on optimal design of adaptive type-2 neuro-fuzzy
uncertainties associated with dynamic real-world systems [47].
unstructured environments. The uncertainties mainly In recent years, some publications about fractional order
included: definition of fuzzy variables, membership system based on interval type-2 fuzzy logic system
function defined by different experts, disturbances of appeared. The interval type-2 fuzzy logic system applied in
control system, sensor measurement noise, characteristics fractional order system mainly included 2 aspects. The first
changing of controlled object and so on. As the membership was designing of fractional order interval type 2 fuzzy
function of type-1 fuzzy logic system was crisp value, so it controller, such as interval type-2 fractional order fuzzy
couldn't deal with these uncertainties. For describing PID-based power system stabilizer [48], the optimal time
uncertainties and overcoming faults of PID and type-1 domain tuning of interval type-2 fractional order fuzzy PID
fuzzy controller, type-2 fuzzy sets were introduced by controller using ABC-GA [49], interval type-2 fractional
zadeh in 1975[15] as an extension of type-1 fuzzy sets, in order fuzzy controller for a tractor active suspension system
[50]
which uncertainty was represented by the addition of an , interval type-2 fractional order controller for redundant
extra dimension, and provided a better representation of robot [51], interval type-2 fuzzy fractional-order
uncertainty in certain applications. Type-2 fuzzy sets were backstepping sliding mode controller [52] , optimal interval
three dimensional form, which contained 2 fuzzy type-2 fuzzy fractional order super twisting algorithm [53],
membership functions. The membership function of fuzzy interval type-2 fractional order fuzzy logic controllers for
variables was called primary membership function, and the fractional order systems [54] and dynamic stability
membership function of primary membership function was enhancement of power system [55].The other was controlling
called secondary membership function. This definition of fractional order system using interval type 2 fuzzy logic
made the computation complexity of type-2 fuzzy sets very system, such as interval type-2 fuzzy neural network for
high, which limited type-2 fuzzy sets to be deployed into fractional-order chaotic systems [56], interval type-2 fuzzy
practical applications. Mendel and his students made some system for chaotic non-linear fractional order systems [57],
deep researches on type-2 fuzzy sets, and defined a fractional order interval type-2 T-S fuzzy systems [58].
simplified type-2 fuzzy sets, that were interval type-2 fuzzy For interval type-2 fuzzy sets simplified secondary
sets, where the secondary membership function was 1 [16-19]. membership function as 1, so the uncertainty modeling
Based on interval type-2 fuzzy sets, many papers about provided by an interval type-2 fuzzy system was
interval type-2 fuzzy logic system were published. incomplete, because the uncertainty was not uniform as
Kumbasar proposed a simple design method for interval assumed in this model. On the other hand, there existed
type-2 fuzzy PID controller [20], and applied a big bang–big representations that allowed the modeling of non-uniform
crunch optimization algorithm to turn parameters of interval uncertainty. Recently, general type-2 fuzzy sets had
type-2 fuzzy PID controller [21]. Inverse interval type-2 received an increasing attention of interest since they had
fuzzy controller was studied in [22-23] and applied to more design degrees of freedom than interval type-2 fuzzy
neutralization processes. He also analyzed stability of a sets. Thus general type-2 fuzzy logic systems would further
single input interval type-2 fuzzy logic controllers [24]. A extend the uncertainty handling property of fuzzy systems
structure of interval type-2 fuzzy controller was derived in and obtain better performance levels. The main problem of
[25] and [26] revisited Karnik–Mendel algorithms in the general type-2 fuzzy system was its complex computation
framework of linear fractional programming. Hagras done of type reduction. Liu and Mendel introduced an α-plane
some jobs on internal type-2 fuzzy logic, like controlling representation for general type-2 fuzzy sets [59-61]. Based on
for autonomous mobile robots [27], realise ambient this representation, an efficient type reduction method for
intelligence in ubiquitous computing environments [28], general type-2 fuzzy sets was proposed. There were some
mobile field workforce area optimization [29] and mobile successful applications of general type-2 fuzzy controller in
field workforce area optimization [30]. Sepúlveda tested the literatures. The general type-2 fuzzy controller applications
ability of type-1 and interval type-2 handling uncertainty [31] contained mobile robot [62-65], water tank, temperature,
and embedded the interval type-2 fuzzy controller into an mobile robot and beam and ball [66], traffic signal
FPGA [32]. [34-37] discussed some optimization methods scheduling[67], inverted pendulum plant, robotic
for designing type-2 fuzzy logic controllers, Melin studied manipulator with time-varying payloads and 3-PSP parallel
applications of interval type-2 fuzzy neural networks in robot [68-69], 5-agents system, a multi-agent with unknown
face recognition [38] and non-linear identification and time dynamics and unknown time-varying topology [70], two-
series prediction [39]. And he also reviewed design and wheeled self-balancing robot [71], trajectory tracking of
optimization of interval type-2 fuzzy controllers [40-41], type- wheeled mobile robots [72], a class of nonlinear power
2 fuzzy logic applications in clustering, classification and systems [73], water level and DC motor speed [74], aerospace

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

[75]
, airplane flight [76] and so on. Section 5 is the simulation results of 4 processes and
Until now, there were little literatures about applications comparison of FOGT2-FPID with other controllers. And
of general type-2 fuzzy logic systems to fractional order section 6 draws the conclusions and future works.
systems. [77] proposed a general type-2 fuzzy neural II. GENERAL TYPE-2 FUZZY SETS
network controller for fractional-order multi-agent systems A type-1 fuzzy sets A on a universal sets X can be
and [78] proposed a synchronization of uncertain fractional characterized by the membership function as (1).
order chaotic systems based on general type-2 fuzzy system. A  {x,  ( x) | x  X ,  ( x)  [0,1]} (1)
In this paper, a fractional order general type-2 fuzzy PID α cuts of A can be defined as (2).
(FOGT2-FPID) controller is proposed. The proposed A  {x,|  ( x)   ,   [0,1]} (2)
controller based on general type 2 fuzzy logic system, and Aα contains all the elements x in the domain X whose
extended conventional fuzzy PID controller to fractional membership degree is greater than or equal to α, the
order fuzzy PID controller. Furthermore, in the type characteristic function of which is shown as (3).
reduction process of proposed FOGT2-FPID controller, a
1, x  A
simplified and direct type reduction is adapted. The  A   (3)
contributions and originality of this paper can be described 0, x  A
as follows: The definition of fuzzy sets of number multiplication is
1). At present, most researches on fractional order type-2 shown as (4).
fuzzy controller mainly focused on interval type-2 fuzzy  A( x)  
systems. In this paper, a fractional order general type-2 x  X ,  A( x)    A( x)   (4)
fuzzy PID controller based on general type-2 fuzzy logic  A( x) A( x)  
systems is proposed. Due to the secondary membership of Then, type-1 fuzzy sets A can be represented by its α cuts
general type-2 fuzzy sets is a function, the ability of as (5).
FOGT2-FPID controller handing system uncertainty is A=  A (5)
better than fractional order type-1 fuzzy PID (FOT1-FPID)  [0,1]
controller and fractional order interval type-2 fuzzy PID Type-2 fuzzy sets have 2 membership functions, and a
(FOIT2-FPID) controller.
2). The main difference of type-2 and tradition type-1 type-2 fuzzy set A on a universal set X can be characterized
fuzzy logic system was that type-2 fuzzy logic system by the membership function as (6), where x X[80]:
contained a type reduction process, which converted type-2 A  {( x, u ),  A ( x, u ) | x  X , u  J x  [0,1]} (6)
fuzzy sets to type-1 and then got the final defuzzification u is the primary membership function and  A ( x, u ) is the
result. Karnik-Mendel (KM) type reduction [79] was widely
used for interval or general type-2 fuzzy logic systems. The secondary membership function.
KM type reduction was an iterative search process and this Liu extended α cuts of type-1 fuzzy sets to general type-2
was time consuming in real-time controlling processes. In fuzzy sets, and α cuts (α planes)of general type-2 fuzzy sets
this paper, a simplified and direct type reduction called NT A can be defined as (7)[59].
type reduction is adapted in the proposed FOGT2-FPID A  {( x, u ),  A ( x, u )   | x  X , u  J x  [0,1]} (7)
controller. NT type reduction was an approximate of KM
algorithm, and no iterative process was required, this can A general type-2 fuzzy sets A can be represented as the
improve type reduction efficiency of the FOGT2-FPID union of its associated type-2 fuzzy sets A :
controller.
A FOU ( A ) (8)
3). The control performance of FOGT2-FPID controller  [0,1]
is applied to 4 processes to show its effectiveness and
For minimum t-norm operation, centroid type-reduction
practicality. And the control performances of FOGT2-FPID
controller are compared with conventional PID controller, for a general type-2 fuzzy sets A is the union of the
type-1 fuzzy PID (T1-FPID) controller, interval type-2 centroids of its associated type-2 fuzzy sets A , with α∈[0,
fuzzy PID (IT2-FPID) controller, general type-2 fuzzy PID 1]:
(GT2-FPID) controller, fractional order type-1 fuzzy PID
C A( x )   / CA ( x) (9.1)
(FOT1-FPID) controller and fractional order interval type-2  [0,1]

fuzzy PID (FOIT2-FPID) controller using KM type


reduction algorithm. C A ( x )  [l A , rA ] (9.2)
The rests of paper are organized as follows: Section 2 [l A , rA ] represent endpoints of each α-plane calculated
describes the definition of general type-2 fuzzy sets based on
α plane representation. Section 3 describes the definition of by KM type reduction algorithm. So, if number of α planes
mathematic expression of fractional order PID controller and is D, then in order to obtain the defuzzification result of
its numerical solution. General type-2 fractional order fuzzy general type-2 fuzzy sets, D times of KM algorithm will be
PID controller is described in section 4, and the structure of implemented.
general type 2 fuzzy logic systems is depicted detailed.
III. FRATIONAL ORDER PID CONTROLLER

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

A. DEFINITION
The basic operator of fractional calculus is  Dt , t and α is

the high and low limit of operator, λ is the fractional order. λ
can be a real number or a complex number,  Dt can be

 =1  PID
defined as (10). PD
 d
  Re( )  0
 dt
D
 t

 1 Re( )  0 (10)
 t
  (d ) Re( )  0



The most widely used definition of fractional calculus in P  
PI


fractional order control was Grunwald-Letnikov fractional
calculus definition [81], the definition of λ-order calculus for  =1
FIGURE 1. Structure scheme of fractional order PIλDµ controller
continuous differentiable function f(t) can be described as
As shown in Fig.1, abscissa is integral order and
(11).
[( t  )/ h ]
ordinates is differential order. Traditional integer order PD,
 Dt f (t )  lim h 
h 0
j 0
wj f (t  jh) (11) PI, PID controller are the specific points on the plane. If λ
or µ is chosen arbitrarily, then fractional order PID
(1) j (  1) controller will cover the whole plane. So fractional order
where wj  and Г is gamma function. PID controller has more flexible control structure compared
j !(  j  1) with integer order PID controller. As the tuning range of
(11) unified fractional differential and integral, λ>0 controller parameters becomes larger, the FOPID controller
indicated fractional differential and λ<0 indicated fractional can control the controlled object more flexibly and get
integral. [(t - α)/h] indicated rounding, h is step size, t and α better control effect.
represented the upper and lower bounds of integral From the perspective of Bode graph, amplitude frequency
respectively. characteristic’s slope of Bode graph for integer order PID
Suppose initial time t=0, by the classic formula for the controller only permits integer multiple of 20dB/dec.
Laplace transform of the fractional order differential However, amplitude frequency characteristic’s slope of
equations with constant coefficients. fractional order PID controller can be arbitrary shape.
 n 1
B. NUMBERICAL SOLUTION
 e st 0 Dt f (t )dt  s  F (s)   s k 0 Dt  k 1 f (t ) |t 0 (12) The fractional order system has infinite dimension, so the
0
k 0
finite differential equation must be used to approximate the
where n is the integer such that n-1<λ≤n. realization of fractional order controller. In this paper, the
If initial value 0 Dt  k 1 f (t )|t  0 =0 , then Laplace transform fractional order PIλDµ controller is discretized by time
domain numerical method. And the fractional order PIλDµ
of 0 Dt f (t ) can be represented as (13).
controller differential equation is transformed into time
L{ 0 Dt f (t )}  s  F ( s ) (13) domain discrete equation by using the time domain
The transfer function of fractional order PID controller can numerical solution of fractional order differential equation.
be show as (14). The steps of time domain numerical method for
K fractional order PIλDµ controller are as follows [82]:
Gc ( s)  K p  I  K D s  ,  , >0 (14) 1). At the k-th sampling time, calculate error e(k):
s
e( k )  r  y ( k ) (15)
Fig.1 shows the structure scheme of fractional order PIλDµ
controller. where r is the set value and y(k) is the system output.
2). Calculate PIλDµ controller output:
k k
=K P e(k )  K I h  wj  e(k  j )  K D h    wj e(k  j )
u (k)
j 0 j 0

(16)
where u(k) is the controller output, KP, KI, KD are
proportional gain, integral gain and differential gain, h is
sampling period, and e(k-j) is error from previous sampling
time.
The coefficient w can be calculated more simply by the
following recurrence formula:

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

w0   1 (17.1) Output process

(  )  1  
wj   [1 
defuzzifer

]w j 1 (17.2) Output

j Crisp Inputs Fuzzy Rules


Type

w0  1
Fuzzifier
reduction
(17.3) Type-1 fuzzy sets

 1  Fuzzy inference

wj  [1  ]w j 1 (17.4) Input type-2 fuzzy sets Output type-2 fuzzy sets

j FIGURE 4. General type-2 fuzzy controller


If the step h is small enough, the approximate value of The time domain of PIλDµ controller is depicted as (18).
fractional calculus can be directly obtained by (17). d   e(t ) d  e(t )
From (16), fractional differentiation is related to all the u (t )  K P e(t)
+K I  K (18)
dt   dt 
D
past history, or it also has memory function. The closer errors
where KP is proportional gain, KI is integral gain and KD is
have larger response factor, the farer errors have smaller
derivative gain.
response factor, which ensures the influence of the historical
From Fig.3, the FOGT2-FPID controller output u(t) is
information on the present and future. This is different from
defined as (19).
integer order PID, which is only related to some adjacent
d  U (t )
points. In theory, fractional order controller is a filter with u (t )  GPDU (t )+GPI (19)
infinite dimension, so it can improve the control accuracy dt  
and system stability. U(t) is the general type-2 fuzzy controller output and can
be calculated by (24-26) in the section B.
IV. FRACTIONAL ORDER GENERAL TYPE-2 FUZZY The primary membership function of proposed general
PID CONTROLLER type 2 fuzzy sets in this paper is triangular. For simplify,
error and derivative of error shares the same primary
A. GENERAL TYPE-2 FUZZY LOGIC SYSTEM membership function, which can be shown as Fig.5.
An integral order general type-2 fuzzy logic system(GT2FLS) NB NM Z PM PB
almost has the same structure as conventional type-1 fuzzy 1
logic system, the only difference is type reduction of
GT2FLS, a typical structure of integral order general type-2
fuzzy logic system is shown as Fig.2[49]. Here the input
scaling factors GE and GCE normalize the inputs error and
derivative of error to the universe of discourse where the
membership function of the two inputs are defined. Thus, the
inputs error and derivative of error are transformed to E
and E .
r  e(t ) E (t )
GPD
GE
General  u (t ) y (t )
 0
Type2 Fuzzy U (t ) Process
de(t ) GCE
E (t ) Controller 
-Ep -Ep/2 0 Ep/2 Ep
dt GPI 
FIGURE 5. Primary membership function of error and error derivative
FIGURE 2. Integral order general type-2 fuzzy logic system structure As described in Fig.5, the membership function is
de(t ) d  e(t ) d - symmetric and there will be 25 fuzzy rules in this paper, the
In Fig.2, if replace
dt
as
dt 
and  as
dt -
, fuzzy rules are shown as Tab.I.
TABLE I
then the fractional order general type-2 fuzzy logic system GENERAL TYPE-2 FUZZY CONTROLLER RULES
will be obtained as Fig.3. E/ E NB NM Z PM PB
r  e(t ) E (t )
GPD NB NB NB NB NM Z
GE
General  u (t ) y (t ) NM NB NB NM Z PM

Type2 Fuzzy U (t ) Process

Z NB NM Z PM PB
d e(t) GCE
E (t ) Controller 
dt  GPI
d - PM NM Z PM PB PB
dt - PB Z PM PB PB PB
FIGURE 3. Fractional order general type-2 fuzzy logic system structure where the consequent parameters are NB=-1, NM=-0.8, Z=0,
The general type 2 fuzzy controller modular in Fig.3 is a PM=0.8, PB=1 [49].
nonlinear mapping, the inputs of general type-2 fuzzy In fact, at most two membership functions are fired
controller are E and E , the output of general type-2 fuzzy simultaneously, namely Ai and Ai1 in E or B j and B j 1 in
controller is U in Fig.3. And the modular can be described as
Fig.4. E . From Fig.5, sum of adjacent membership functions is 1.
Thus at same time, only 4 fuzzy rules are fired, which show
as follow:

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

Rule 1:If E is Ai and E is B j , then U=y1 k )


f( k )
 f(
k  (k  1,..., M ) (24)
Rule 2:If E is Ai and E is B j 1 , then U = y2 2
M M

Rule 3:If E is Ai1 and E is B j , then U = y3  k  yk  ( f() f())  y


k k k
c( )  k 1
M
 k 1
M
(25)
Rule 4:If E is Ai1 and E is B j 1 , then U = y4
From fuzzy inference of interval type-2 fuzzy logic

k 1

k  ( f() f())
k 1
k k

system, the fired membership degree of fuzzy rule is also an The final output of type reduction is union of all
interval, the 4 rules interval are shown as (20) using centroids of α planes.
‘product’ operator. D

[ f1 , f1 ]  [u A ( E )  u B ( E ), u A ( E )  u B ( E )] (20.1)  c( )   i
U i 1
i j i j
(26)
[ f 2 , f 2 ]  [u Ai ( E )  u B j 1 ( E ), u Ai ( E )  u B j 1 ( E )]
D
(20.2)

i 1
i

[ f3 , f3 ]  [u Ai1 ( E )  u B j ( E ), u Ai1 ( E )  u B j ( E )] (20.3)


Where D is number of α plane, αi is average of 1 to D,
[ f 4 , f 4 ]  [u Ai1 ( E )  u B j1 ( E ), u Ai1 ( E )  u B j1 ( E )] (20.4) that is α1=0, α2=1/D, α3=2/D,…,αD=1.
C. FOGT2-FPID CONTROLLER OUTPUT
[ f k , f k ] is the low and up membership function of general The main different between fractional order general type-2
type-2 fuzzy sets as α=0, when use interval type-2 fuzzy fuzzy PID controller and integral order general type-2 fuzzy
PID controller is the calculation of fractional differential for
logic, [ f k , f k ] will be the low and up membership function
error e(t) and fractional integration for general type-2 fuzzy
of interval type-2 fuzzy sets. controller output U(t).
B. TYPE REDUCTION The procedure of calculating output of FOGT2-FPID u(t)
The secondary membership function in this paper is can be summary as follows:
chosen as triangular, and show as Fig.6. 1). At the k-th sampling time, calculate error e(k).
e( k )  r  y ( k )
u
where r is the set value and y(k) is the system output.
1 d  e(t )
2). Calculate by (17.3) and (17.4).
dt 
d  e(t )   k 
fk =h  w j e(k  j )
dt  j 0
Ape( yk )
3). Calculate the inputs of general type-2 fuzzy controller.
fk E  GE  e(t )
0 d  e(t )
yk y E  GCE 
dt 
4). Calculate the fired membership degree of fuzzy rules
1 interval [ f k , f k ] by (20).
5). Calculate the membership function of low and up limes
 A ( y, u ) for general type-2 fuzzy sets in each α plane by (21-23).
FIGURE 6. Triangular secondary membership function 6). Calculate the general type-2 fuzzy controller final
The primary membership function of apex in each output of type reduction U by (24-26).
triangular is defined as (21). d  U (t )
Ape( yk )  f k  wk  [ f k  f k ] (21) 7). Calculated by (17.1) and (17.2).
dt  
Here, for simplify, wk is the same for each triangular d  U (t ) k
secondary membership function. 
 h  wj U (k  j )
From Fig.6, membership function of low and up limes dt j 0

for general type-2 fuzzy sets in each α plane can be 8). Calculate the FOGT2-FPID controller output u(t) by
described as (22-23). (19).
f k ( )    ( Ape( yk )  f k )  f k (22)
V. SIMULATIONS
f k ( )    ( Ape( yk )  f k )  f k (23)
By NT type reduction, the centroid of each α plane c(α) A. SECOND ORDER STABLE LINEAR PLANT ALONG
is shown as (24-25). WITH DEAD TIME(P1)

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

1
G(s)  e Ls
s  2n s  n2
2

Controller gains are selected form [49] process 1 nominal


process. PID controller gains are 4.1479,4.0079 and 0.1792.
Type-1 fuzzy controller gains are GE=0.5876, GCE=1.0123,
GPD=0.2937, GPI=2.228 and Ep=1 in Fig.5. Interval type-2
fuzzy controller shares the same gains with general type-2
fuzzy controller, which are GE=0.9956, GCE=0.8387,
GPD=0.532, GPI=4.0573 and Ep=1 in Fig.5.
Fractional order type-1 fuzzy controllers are GE=1.8158,
GCE=1.4873, GPD=1.2406, GPI=1.5101, λ=0.6442, μ=1.1357,
and Ep=4 in Fig.5. Fractional order interval type-2 fuzzy
and fractional order general type-2 fuzzy controller shares
the same gains, which are GE=1.5110, GCE=0.8901,
GPD=2.319, GPI=0.2665, λ=1.1465, μ=1.141and Ep=4 in
Fig.5. (c) Error curve
FIGURE 7. Response curves of P1 under 7 controllers in case 1 parameters
Case 1: ε=1.125, ωn=0.45, L=0.4. Case 2: ε=0.5624, ωn=0.225, L=0.4.
Fig.7 depicts system output, controller output and error Fig.8 depicts system output, controller output and error
curve of P1 under unit step disturbance in case 1 parameters curve of P1 under unit step disturbance in case 2 parameters
by 7 controllers. by 5 controllers. In this case, PID and T1-FPID controller
are unstable, and FOT1-FPID controller is constant
amplitude oscillation.

(a) System output curve

(a) System output curve

(b) Controller output curve

(b) Controller output curve

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

(c) Error curve (c) Error output curve


FIGURE 8. Response curves of P1 under 5 controllers in case 2 parameters FIGURE 9. Response curves of P1 under 7 controllers in case 3 parameters
Case 3: ε=1.6875, ωn=0.225, L=0.4. Case 4: ε=0.5624, ωn=0.675, L=0.
Fig.9 depicts system output, controller output and error Fig.10 depicts system output, controller output and error
curve of P1 under unit step disturbance in case 3 parameters curve of P1 under unit step disturbance in case 4 parameters
by 7 controllers. by 7 controllers.

(a) System output curve (a) System output curve

(b) Controller output curve (b) Controller output curve

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

(c) Case 3 system output curve

(c) Error output curve


FIGURE 10. Response curves of P1 under 7 controllers in case 4 parameters
In order to illustrate the advantages of the FOGT2-FPID
controller in dealing with uncertainty, Fig.11 shows the
system output curves of P1 when the set value changing
from 1 to 0 after 30s.

(d) Case 4 system output curve


FIGURE 11. System output curves of P1 under set value change
Fig.12 shows the system output curves of P1 when the
controller adding a disturbance at 30s.

(a) Case 1 system output curve

(a) Case 1 system output curve

(b) Case 2 system output curve

(b) Case 2 system output curve

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

FIGURE 12. System output curves of P1 under controller adding disturbance at


30s
Tab. II shows under unit step disturbance, the P1 control
performance comparisons of FOGT2-FPID with other PID
controllers, that are PID, type 1 fuzzy PID (T1-FPID),
interval type 2 fuzzy PID(IT2-FPID), general type 2 fuzzy
PID(GT2-FPID), fractional order type 1 fuzzy PID (FOT1-
FPID), fractional order interval type 2 fuzzy PID(FOIT2-
FPID). And control performance indexes include steady
state time (ts, that is time when the absolute value of system
error reached the 2% of set value), rising time (tr, that is
time when the system output reach the 2% of set value),
overshoot (OS), three errors integral criterion (that is an
(c) Case 3 system output curve integral function of error between the set value and system
output under the unit step disturbance), mainly contains ISE,
ITSE, ITAE and can be calculated as follows. In Tab II,
simulation time is 30s.
ts
ISE   e(t )2 dt
0
ts
ITSE   t  e(t )2 dt
0
ts
ITAE   t  e(t ) dt
0

(d) Case 4 system output curve


TABLE II
CONTROL PERFORMANCE COMPARISIONS OF GT2FO-PID WITH OTHER CONTROLLERS(P1)
IT2- GT2- IT2FO- GT2FO-
P1 PID T1-FPID T1FO-FPID
FPID FPID FPID FPID
ts(s) 11.53 6.97 4.59 20.2 8.88 3.8
tr(s) 3.23 1.34 1.3 1.99 2.98 2.08
OS(%) 20.99 17.9 21.9 123.8 9.1 15.8
case 1 Oscillation
ISE 1.68 0.89 0.88 3.88 1.60 1.26
ITSE 1.99 0.51 0.49 11.44 1.42 0.89
ITAE 7.12 1.73 1.30 24.71 3.21 1.62
ts(s) >30 14.22 23 8.52
tr(s) 1.21 1.19 2.24 1.77
OS(%) 63.1 71.7 61.3 67.5
case 2 unstable unstable Oscillation
ISE 1.67 1.54 2.36 1.75
ITSE 5.27 2.64 5.88 2.54
ITAE 33.72 9.69 22.97 6.21
ts(s) 24.54 8.36 5.26 20.59 12.75 4.72
tr 2.74 1.28 1.25 1.75 2.68 1.97
OS(%) 59 31.5 36.6 126.7 21.2 14.9
case 3 unstable
ISE 2.51 0.926 0.926 3.82 1.48 1.07
ITSE 7.52 0.619 0.577 11.19 1.49 0.69
ITAE 28.38 2.56 1.82 24.42 5.23 1.65
ts(s) 23.41 13.22 14.83 29.7 17.08 10.65
tr(s) 3.3 1.3 1.27 1.95 3 2.01
OS(%) 3.87 22.8 27.5 118.3 1.6 15.55
case 4 Oscillation
ISE 1.85 0.96 0.95 4.12 1.72 1.26
ITSE 3.98 0.825 0.829 16.29 2.47 0.97
ITAE 21.46 4.33 5.94 48.68 12.38 3.65

Controller gains are selected form [49] process 2 nominal


B. FIRST ORDER UNSTABLE LINEAR PLANT ALONG process. PID controller gains are 4.1479,4.0079 and 0.1792.
WITH DEAD TIME(P2) Type-1 fuzzy controller gains are GE=0.5876, GCE=1.0123,
GPD=0.2937, GPI=2.228 and Ep=1 in Fig.5. Interval type-2
K  Ls fuzzy controller shares the same gains with general type-2
G( s)  e
Ts  1

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

fuzzy controller, which are GE=0.9956, GCE=0.8387,


GPD=0.532, GPI=4.0573 and Ep=1 in Fig.5.
Fractional order type-1 fuzzy controllers are GE=0.7463,
GCE=1.0948, GPD=0.4629, GPI=2.4066, λ=1.0082, μ=1.0069,
and Ep=1in Fig.5. Fractional order interval type-2 fuzzy
and fractional order general type-2 fuzzy controller shares
the same gains, which are GE=0.9847, GCE=0.8478,
GPD=0.522, GPI=4.1333, λ=0.9978, μ=0.9971 and Ep=1 in
Fig.5.
Case 1: K=1, T=10, L=0.2.
Fig.13 depicts system output, controller output and error
curve of P2 under unit step disturbance in case 1 parameters
by 7 controllers.

(c) Error curve


FIGURE 13. Response curves of P2 under 7 controllers in case 1 parameters
Case 2: K=1, T=10, L=0.4.
Fig.14 depicts system output, controller output and error
curve of P2 under unit step disturbance in case 2 parameters
by 7 controllers.

(a) System output curve

(a) System output curve

(b) Controller output curve

(b) Controller output curve

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

(c) Error curve (c) Error curve


FIGURE 14. Response curves of P2 under 7 controllers in case 2 parameters FIGURE 15. Response curves of P2 under 7 controllers in case 3 parameters
Case 3: K=1, T=20, L=0.2. Case 4: K=4, T=20, L=0.35.
Fig.15 depicts system output, controller output and error Fig.16 depicts system output, controller output and error
curve of P2 under unit step disturbance in case 3 parameters curve of P2 under unit step disturbance in case 4 parameters
by 7 controllers. by 7 controllers.

(a) System output curve (a) System output curve

(b) Controller output curve (b) Controller output curve

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

(c) Case 3 system output curve

(c) Error curve


FIGURE 16. Response curves of P2 under 7 controllers in case 4 parameters
Fig.17 shows the system output curves of P2 when the
set value changing from 1 to 0 after 30s.

(d) Case 4 system output curve


FIGURE 17. System output curves of P1 under set value change
Fig.18 shows the system output curves of P2 when the
controller adding a disturbance at 30s.

(a) Case 1 system output curve

(a) Case 1 system output curve

(b) Case 2 system output curve

(b) Case 2 system output curve

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

(c) Case 3 system output curve (d) Case 4 system output curve
FIGURE 18. System output curves of P2 under controller adding disturbance at
30s
Tab. III shows under unit step disturbance, P2 control
performance comparisons of FOGT2-FPID with PID, T1-
FPID, IT2-FPID, GT2-FPID, FOT1-FPID, FOIT2-FPID. In
Tab III, simulation time is 30s.
TABLE III
CONTROL PERFORMANCE COMPARISIONS OF GT2FO-PID WITH OTHER CONTROLLERS(P2)
IT2- IT2FO- GT2FO-
P2 PID T1-FPID GT2-FPID T1FO-FPID
FPID FPID FPID
ts(s) 30 30 28.5 11.64 30 27.48 8.9
tr(s) 1.89 3.76 2.58 2.70 3.15 2.57 2.71
OS(%) 66.6 77.6 58.7 43.9 92.9 49.1 35.8
case 1
ISE 2.30 6.23 2.57 1.77 6.00 1.98 1.57
ITSE 9.91 58.87 10.38 3.04 47.51 5.45 2.13
ITAE 41.37 142.37 44.32 9.4 116.05 26.87 6.38
ts(s) 29.8 30 30 17.5 30 30 13.2
tr(s) 1.94 3.81 2.64 2.74 3.32 2.61 2.73
OS(%) 83.17 90.6 72.7 64.02 93.9 61.6 53.4
case 2
ISE 3.92 10.47 4.65 2.67 7.78 2.87 2.12
ITSE 24.23 134.75 36.39 8.03 81.17 12.65 4.52
ITAE 78.33 223.80 107.70 23.68 162.70 53.44 13.10
ts(s) >30 >30 >30 21.19 >30 >30 16.54
tr 2.9 5.2 3.5 3.7 4.28 3.5 3.7
OS(%) 72.5 84.2 74.3 53.1 134 63.6 45.8
case 3
ISE 4.32 9.43 5.39 2.85 18.33 3.86 2.47
ITSE 33.61 113.25 44.70 9.34 253.11 23.67 6.34
ITAE 101.84 201.53 118.65 30.51 302.69 80.88 20.11
ts(s) 13.32 25.63 13.15 7.05 23.51 10.37 5.39
tr(s) 1.28 2.98 2.03 2.08 2.8 2.03 2.09
OS(%) 60.2 51.9 39.2 38.5 62.7 30.4 28.6
case 4
ISE 1.30 2.36 1.41 1.33 2.31 1.22 1.18
ITSE 2.39 6.99 1.89 1.37 6.55 1.22 0.96
ITAE 8.81 29.33 7.85 3.71 23.18 4.82 2.44

Fractional order type-1 fuzzy controllers are GE=1.6336,


C. SECOND ORDER NONLINEAR PLANT(P3) GCE=1.0516, GPD=1.2995, GPI=2.4564, λ=0.908, μ=1.1848,
and Ep=2 in Fig.5. Fractional order interval type-2 fuzzy
d 2 y (t ) dy (t ) and fractional order general type-2 fuzzy controller shares
2
 2   2 y 2 (t )   2u (t  L) the same gains, which are GE=2.103, GCE=1.1732,
dt dt
GPD=1.3634, GPI=2.5829, λ=0.9686, μ=1.1805 and Ep=2 in
Controller gains are selected form [49] process 4 nominal
Fig.5.
process. PID controller gains are 0.8028,1.8548 and 0.4609.
Case 1: ε=1, σ=1, L=0.
Type-1 fuzzy controller gains are GE=0.3079, GCE=0.1807,
Fig.19 depicts system output, controller output and error
GPD=0.3258, GPI=2.7788 and Ep=1 in Fig.5. Interval type-2
curve of P3 under unit step disturbance in case 1 parameters
fuzzy controller shares the same gains with general type-2
by 7 controllers.
fuzzy controller, which are GE=0.2859, GCE=0.1944,
GPD=0.3501, GPI=3.2681 and Ep=1 in Fig.5.

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

(a) System output curve (a) System output curve

(b) Controller output curve (b) Controller output curve

(c) Error curve (c) Error curve


FIGURE 19. Response curves of P3 under 7 controllers in case 1 parameters FIGURE 20. Response curves of P3 under 7 controllers in case 2 parameters
Case 2: ε=1, σ=1, L=0.1. Case 3: ε=1, σ=1, L=0.2.
Fig.20 depicts system output, controller output and error Fig.21 depicts system output, controller output and error
curve of P3 under unit step disturbance in case 2 parameters curve of P3 under unit step disturbance in case 3 parameters
by 7 controllers. by 7 controllers.

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

(a) System output curve (a) System output curve

(b) Controller output curve (b) Controller output curve

(c) Error curve (c) Error curve


FIGURE 21. Response curves of P3 under 7 controllers in case 3 parameters FIGURE 22. Response curves of P3 under 7 controllers in case 4 parameters
Case 4: ε=1, σ=0.7, L=0. Fig.23 shows the system output curves of P3 when the
Fig.22 depicts system output, controller output and error set value changing from 1 to 2 after 15s.
curve of P3 under unit step disturbance in case 4 parameters
by 7 controllers.

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

(a) case 1 system output curve (a) Case 1 system output curve

(b) case 2 system output curve (b) Case 2 system output curve

(c) case 3 system output curve (c) Case 3 system output curve

(d) case 4 system output curve (d) Case 4 system output curve
FIGURE 23. System output curves of P3 under set value change FIGURE 24. System output curves of P3 under controller adding disturbance at
Fig.24 shows the system output curves of P3 when the 15s

controller adding a disturbance at 15s. Tab. IV shows under unit step disturbance, P3 control
performance comparisons of FOGT2-FPID with PID, T1-

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

FPID, IT2-FPID, GT2-FPID, FOT1-FPID, FOIT2-FPID. In Tab IV, simulation time is 15s.
TABLE IV
CONTROL PERFORMANCE COMPARISIONS OF GT2FO-PID WITH OTHER CONTROLLERS(P3)
IT2- GT2- IT2FO- GT2FO-
P3 PID T1-FPID T1FO-FPID
FPID FPID FPID FPID
ts(s) 6.29 7.05 7.12 7.18 3.96 4.51 2.77
tr(s) 1.4 1.84 1.98 2.44 1.72 1.95 1.74
OS(%) 16.5 25.4 21.0 8.8 19.8 1.6 3.4
case 1
ISE 0.53 1.07 1.11 1.23 0.91 0.85 0.81
ITSE 0.26 0.82 0.82 0.89 0.54 0.42 0.38
ITAE 1.34 2.66 2.45 2.03 1.38 0.82 0.65
ts(s) 6.62 7.41 7.52 5.01 4.06 4.53 2.81
tr(s) 1.41 1.9 2.04 2.47 1.74 1.84 1.66
OS(%) 20.3 29.6 24.7 11.6 23.6 3.0 5.6
case 2
ISE 0.63 1.19 1.23 1.32 0.98 0.90 0.88
ITSE 0.36 1.04 1.01 1.02 0.63 0.46 0.43
ITAE 1.72 3.29 2.99 2.10 1.57 0.87 0.73
ts(s) 6.94 9.94 7.97 5.63 3.67 4.54 2.77
tr 1.44 1.97 2.11 2.52 1.87 1.72 1.63
OS(%) 24.6 34.0 28.5 14.4 13.5 6.15 10.6
case 3
ISE 0.74 1.34 1.35 1.42 1.0 0.98 0.96
ITSE 0.50 1.34 1.25 1.19 0.57 0.52 0.51
ITAE 2.23 4.42 3.67 2.60 1.18 0.99 0.85
ts(s) 7.49 10.37 8.53 7.22 4.74 4.74 2.62
tr(s) 1.29 1.75 1.88 2.25 1.71 1.77 1.61
OS(%) 23.0 35.3 30.5 17.2 22.5 5.25 7.21
case 4
ISE 0.55 1.14 1.58 1.24 0.91 0.86 0.82
ITSE 0.34 1.11 1.02 0.98 0.55 0.43 0.39
ITAE 1.95 4.55 3.71 2.96 1.42 1.05 0.68
 x2   0 
   
D. NONLINEAR INVERTED PENDULUN STSTEM(P4)  (m p  m p )lx22 sin( x1 ) cos( x1 )   cos( x1 ) 
 x1   g sin( x1 )    A  x1   
  (m p  m p  mc )   (m p  m p  mc ) u
 x2    x2   
4l ((m p  m p )l cos( x1 )  4l  ((m p  m p )l cos( x1 )
2
 
2

In this example, a practical inverted pendulum system is    3 
3 ( m   m  m )  ( m   m  m ) 
 p p c  p p c
tested, which is consisted by a cart, a pendulum and a rail
for defining the position of the cart and is shown as Fig.25. where x1   is the angle of the pendulum and x2  x1  
is angular velocity of the pendulum. u is the control force in
 the unit (Newton) applied horizontally to the cart. The
parameters, mc and mp, are, respectively, the mass of the
cart and the mass of the pendulum in the unit (kg), and g=
9.8m/s2 is the gravity acceleration. The parameter l is the
half length of the pendulum in the unit (m). Δmp is the
uncertainty in the mass of the pendulum. ΔA is the
2l structural uncertainty of the inverted pendulum. The
parameters of this example are mc=0.5kg, mc=0.2kg, l=0.5m.
Type-1 fuzzy controller gains are GE=1.6336,
u GCE=1.0516, GPD=1.2995, GPI=2.4564 and Ep=0.5 in Fig.5.
Interval type-2 fuzzy controller shares the same gains with
general type-2 fuzzy controller, which are GE=2.1030,
GCE=1.1732, GPD=1.3634, GPI=8.5829 and Ep=0.5 in Fig.5.
Fractional order type-1 fuzzy controllers are GE=1.6336,
FIGURE 25. Inverted pendulum on a cart GCE=1.0516, GPD=1.2995, GPI=2.4564, λ=0.908, μ=1.1848,
The pendulum is hinged in the center of the top surface and Ep=0.5 in Fig.5. Fractional order interval type-2 fuzzy
of the cart and can rotate around the pivot in the same and fractional order general type-2 fuzzy controller shares
vertical plane with the rail. The cart can move right or left the same gains, which are GE=10.103, GCE=1.1732,
on the rail freely. It is given that no friction exists in the GPD=1.3634, GPI=8.5829, λ=0.9686, μ=1.1805 and Ep=0.5
system between the cart and the rail or between the cart and in Fig.5.
the pendulum. The dynamic equation of the uncertain The initial conditions x1=0.1rad and x2=0rad/s, the set
inverted pendulum system can be expressed as [83]: value is x1=0rad.
Case 1: normal plant

VOLUME XX, 2017 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

In this case, there is no uncertainty in the mass of the


pendulum or structural uncertainty of the inverted
0 0 
pendulum, that is Δmp=0 and A   .
0 0 
Fig.25 depicts pendulum angle output, controller output
and error curve of P4 in case 1 condition by 6 controllers.

(a) Pendulum angle curve

(a) Pendulum angle curve

(b) Controller output curve


FIGURE 27. Response curves of P4 under 5 controllers in case 2 condition
Case 3: uncertainty in pendulum mass(Δmp=2kg)
In this case, there is uncertainty in the mass of the
pendulum that is adding Δmp=2kg at 5s. But the initial
conditions x1=0rad and x2=0rad/s, the set value is x1=0.1rad.
Fig.28 depicts pendulum angle output and controller
(b) Controller output curve output curve of P4 in case 3 condition by 4 controllers.
In this case T1-FPID and FOT1-FPID controller are
unstable.

(c) Error curve


FIGURE 26. Response curves of P4 under 6 controllers in case 1 condition
Case 2: uncertainty in pendulum mass(Δmp=2kg) (a) Pendulum angle curve
In this case, there is uncertainty in the mass of the
pendulum, that is adding Δmp=2kg at 5s.
In this case FOT1-FPID controller is unstable.
Fig.27 depicts pendulum angle output and controller
output curve of P4 in case 2 condition by 5 controllers.

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

Case 5: external disturbance


In this case, an external disturbance value, d=29N is
added to system at 5s.
Fig.30 depicts pendulum angle output and controller
output curve of P4 in case 5 condition by 6 controllers.

(b) Controller output curve


FIGURE 28. Response curves of P4 under 4 controllers in case 3 condition (Set
value is 0.1)
Case 4: structure uncertainty
In this case, there is structural uncertainty of the inverted
0.075 0.075
pendulum, that is adding A    at 5s.
(a) Pendulum angle curve
0.075 0.075
Fig.29 depicts pendulum angle output and controller
output curve of P4 in case 4 condition by 6 controllers.

(b) Controller output curve


FIGURE 30. Response curves of P4 under 6 controllers in case 5 condition
Tab. V shows P4 control performance comparisons of
FOGT2-FPID with T1-FPID, IT2-FPID, GT2-FPID, FOT1-
(a) Pendulum angle curve
FPID, FOIT2-FPID. In Tab V, simulation time is 10s.
In order to compared with results in [84], the root mean
square error (RMSE) and the integral of absolute error (IAE)
criteria are added.
1 N
RMSE  
N i 1
e(i )2
ts
IAE   e(t ) dt
0

(b) Controller output curve


FIGURE 29. Response curves of P4 under 6 controllers in case 4 condition
TABLE V
CONTROL PERFORMANCE COMPARISIONS OF GT2FO-PID WITH OTHER CONTROLLERS(P4)
IT2- GT2- IT2FO- GT2FO-
P4 IT2F-PID[84] T1-FPID T1FO-FPID
FPID FPID FPID FPID
ISE 0.0360 0.0078 0.0062 0.0063 0.0425 0.0034 0.0031
ITSE - 0.0041 0.0034 0.0036 0.1127 0.00062 0.00054
case 1 ITAE - 0.1049 0.118 0.1313 1.8633 0.0130 0.0114
RMSE 0.0085 0.0279 0.0250 0.0252 0.0652 0.0184 0.0177
IAE 1.8001 0.1177 0.1140 0.1184 0.4954 0.0435 0.0403
ISE 0.0581 0.0078 0.0062 0.0063 612 0.0032 0.0030
case 2
ITSE - 0.0041 0.0034 0.0036 >1000 0.00056 0.00049

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

ITAE - 0.1492 0.1178 0.1312 351 0.0115 0.0103


RMSE 0.0053 0.0279 0.0250 0.0252 7.8278 0.0179 0.0173
IAE 3.5 0.1228 0.1141 0.1184 39 0.041 0.0383
ISE - 0.0080 0.0073 0.0051 0.0031
ITSE - 0.0180 0.0149 0.0168 0.005
case 3 ITAE - unstable 0.4625 0.5009 unstable 0.5337 0.1804
RMSE - 0.0283 0.0271 0.0225 0.0175
IAE - 0.1655 0.1690 0.1150 0.0617
ISE 0.053 0.0196 0.0074 0.0065 0.0421 0.0032 0.0030
ITSE - 0.0229 0.0036 0.0035 0.1095 0.00056 0.00049
case 4 ITAE - 0.6245 0.1022 0.1244 1.8489 0.0115 0.0103
RMSE 0.0072 0.0443 0.0272 0.0256 0.0649 0.0179 0.0173
IAE 1.996 0.2799 0.1157 0.1176 0.4936 0.041 0.0383
ISE 81.85 0.0158 0.0096 0.0094 0.0467 0.0059 0.0055
ITSE - 0.0484 0.0218 0.0205 0.1368 0.0144 0.0132
case 5 ITAE - 0.7756 0.5953 0.5968 2.1307 0.2637 0.2422
RMSE 0.2861 0.0398 0.0309 0.0307 0.0684 0.0244 0.0234
IAE 95.1 0.2313 0.1937 0.1955 0.5419 0.0894 0.0829
[6] Das S, Pan I, Das S, et al, “A novel fractional order fuzzy PID
controller and its optimal time domain tuning based on integral
performance indices”, Engineering Applications of Artificial
VI. CONCLUSIONS Intelligence, vol.25, no.2, pp. 430-442, Mar.2012.
The proposed FOGT2-FPID controller in this paper based [7] Sharma R, Rana K P S, Kumar V, “Performance analysis of
on general type-2 fuzzy systems. The simulations show that fractional order fuzzy PID controllers applied to a robotic
the FOGT2-FPID controller can achieve better control manipulator”, Expert Systems with Applications, vol.41, no.9, pp.
4274-4289, Jul.2014.
efforts compared with other controls, even if the system [8] Mohan, V, Chhabra, H, Rani A, et al, “An expert 2DOF fractional
contains parameters uncertainty, structure uncertainty and order fuzzy PID controller for nonlinear systems”, Neural
disturbances. The FOGT2-FPID controller utilize NT Computing and Applications, vol.31, no.8, pp. 4253-4270,
directly type reduction algorithm and time domain Aug.2109.
[9] Mishra P, Kumar V, Rana KPS, “A fractional order fuzzy PID
numerical method to approximate fractional order operators, controller for binary distillation column control”, Expert Systems
thus reduce the type reduction consuming time and make with Applications, vol.42, no.22, pp. 8533-8549, Dec.2014.
the implementation of the controller easier. [10] Das S, Pan I, Das S, “Performance comparison of optimal fractional
The next research mainly concerns 3 aspects: order hybrid fuzzy PID controllers for handling oscillatory
fractional order processes with dead time”, ISA Transactions, vol.52,
1). Although there were some literatures about stability no.4, pp. 550-566, Jul.2013.
of interval type-2 fuzzy controller, however, the stability of [11] Sharma R, Gaur P, Mittal A P, “Design of two-layered fractional
general type-2 fuzzy controller has not yet proved. And the order fuzzy logic controllers applied to robotic manipulator with
stability of proposed FOGT2-FPID controller will be variable payload”, Applied Soft Computing, vol.47, pp. 565-576,
Oct.2016.
studied to improve its theory. [12] Zhang F X, Yuan C H, Zhou X J, et al, “Fractional order fuzzy PID
2). Optimize the parameters of FOGT2-FPID controller optimal control in copper removal process of zinc hydrometallurgy”,
by some swarm intelligence optimization algorithms, such Hydrometallurgy, vol.178, pp. 60-76, Jun.2018.
as PSO, ACO, GA and so on. Through global search [13] Beddar A, Bouzekri H, Babes B, et al, “Experimental enhancement
of fuzzy fractional order PI+I controller of grid connected variable
algorithm of these intelligent optimization algorithms, more speed wind energy conversion system”, Energy Conversion and
proper parameters of FOGT2-FPID controller will be Management, vol.123, pp. 569-580, Sep.2016.
obtained and the performance of the control system will be [14] Xu Y, Zhou J, Xue X, et al, “An adaptively fast fuzzy fractional
better. order PID control for pumped storage hydro unit using improved
gravitational search algorithm”, Energy Conversion and
3). Apply the proposed FOGT2-FPID controller in real Management, vol.111, pp.67-78, Mar.2016.
time system to verify its practicality and robustness. [15] Zadeh LA, “The concept of a linguistic variable and its application
to approximate reasoning-I”, Information Sciences, vol.8, no.3, pp.
REFERENCES 199-249, 1975.
[1] Podlubny I, “Fractional-order systems and PIλDμ –controllers[J]”, [16] Mendel J M, John R I B, “Type-2 fuzzy sets made simple”, IEEE
IEEE transactions on automatic control, vol.44, no.1, pp. 208-214, Transactions on Fuzzy Systems, vol.10, no.2, pp. 117-127,
Jan.1999. Aug.2002.
[2] Pan I, Das S, “Chaotic multi-objective optimization based design of [17] Karnik N N, Mendel J M, “Type-2 fuzzy logic systems: type-
fractional order PIλDμ controller in AVR system”, International reduction”, IEEE International Conference on Systems, Man, and
Journal of Electrical Power & Energy Systems, vol.43, no.1, Cybernetics, San Diego, CA, USA,1998, pp.2046-2051.
pp.393-407, Dec. 2012. [18] Liang Q, Mendel J M, “Interval type 2 fuzzy logic systems: Theory
[3] Calderón AJ, Vinagre BM, Feliu V, “Fractional order control and Design”, IEEE Transactions on Fuzzy Systems, vol.8, no.5, pp.
strategies for power electronic buck converters”, Signal Process, 535-550, Oct.2000.
vol.86, no.10, pp. 2803-2819, Oct.2006. [19] Mendel J M, John R I, Liu F, “Interval Type-2 Fuzzy Logic
[4] Sharma R, Gaur P, Mittal A P, “Performance analysis of two-degree Systems Made Simple”, IEEE Transactions on Fuzzy Systems,
of freedom fractional order PID controllers for robotic manipulator vol.14, no.6, pp. 808-821, Nov.2007.
with payload”, ISA Transactions, vol.58, pp. 279-291, Sep.2015. [20] Kumbasar T, “A simple design method for interval type-2 fuzzy pid
[5] Wang D F, Wang X Y, Han P, “Design of fractional order controllers”, Soft computing, vol.18, no.7, pp. 1293-1304, Oct .2014.
controllers for a boiler-turbine system”, China proceedings of the
CSEE, vol.30, no.5, pp.113-119, Feb.2010.

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

[21] Kumbasar T, Hagras H, “Big bang–big crunch optimization based [41] Castillo O, Melin P, “Optimization of type-2 fuzzy systems based
interval type-2 fuzzy PID cascade controller design strategy”, on bio-inspired methods: A concise review”, Information Sciences,
Information Sciences, vol.282, pp. 277-295, Oct.2014. vol.205, pp.1-19, Dec.2012.
[22] Kumbasar T, Eksin I, Guzelkaya M, et al, “Type-2 fuzzy model [42] Melin P, Castillo O.” A review on the applications of type-2 fuzzy
based controller design for neutralization processes”, ISA logic in classification and pattern recognition”, Expert Systems with
Transactions, vol.51, no.2, pp. 277-287, Mar.2012. Applications, vol.40, no.13, pp. 5413-5423, Oct.2013.
[23] Kumbasar T, Eksin I, Guzelkaya M, et al, “Exact inversion of [43] Melin P, Castillo O, “A review on type-2 fuzzy logic applications in
decomposable interval type-2 fuzzy logic systems”, International clustering, classification and pattern recognition”, Applied Soft
Journal of Approximate Reasoning, vol.54, no.2, pp. 253-272, Computing, vol.21, pp. 568-577, Aug.2014.
Feb.2013. [44] Castillo O, Melin P, “A review on interval type-2 fuzzy logic
[24] Kumbasar T, “Robust stability analysis and systematic design of applications in intelligent control”, Information Sciences, vol.279,
single input interval type-2 fuzzy logic controllers”, IEEE pp. 615-631, Sep.2014.
Transactions on Fuzzy Systems, vol.24, no.3, pp. 675-694, [45] Türkay D, Baykasoglu A, Altun K, et al, “Industrial applications of
Aug.2015. type-2 fuzzy sets and systems: a concise review”, Computers in
[25] Aliasghary M, Eksin I, Guzelkaya M, et al, “General derivation and Industry, vol.62, no.2, pp. 125-137, Feb.2011.
analysis for input–output relations in interval type-2 fuzzy logic [46] Torshizi A D, Zarandi M H F, Zakeri H, “On type-reduction of
systems”, Soft Computing, vol.19, no.5, pp. 1283-1293, May.2015. type-2 fuzzy sets: a review”, Applied Soft Computing, vol.27,
[26] Kumbasar T, “Revisiting Karnik–Mendel algorithms in the pp.614-627, Feb.2015.
framework of linear fractional programming”, International Journal [47] Hassan S, Khanesar M A, Kayacan E, et al, “Optimal design of
of Approximate Reasoning, vol.82, pp.1-21, Mar.2017. adaptive type-2 neuro-fuzzy systems: a review”, Applied Soft
[27] Hagras H, “A hierarchical type-2 fuzzy logic control architecture for Computing, vol.44, pp. 134-143, Jul.2016.
autonomous mobile robots”, IEEE Transactions on Fuzzy Systems, [48] Ray P K, Paital S R, Foo Y S E, et al, “A hybrid firefly-swarm
vol.24, no.4, pp. 524 -539, Aug.2004. optimized fractional order interval type-2 fuzzy PID-PSS for
[28] Doctor F, Hagras H, Callaghan V, “A type-2 fuzzy embedded agent transient stability improvement”, IEEE transactions on industry
to realise ambient intelligence in ubiquitous computing applications, vol.55, no.6, pp. 6486-6498, Aug.2019.
environments”, Information Sciences, vol.171 no.4, pp. 309-334 [49] Kumar A, Kumar V, “A novel interval type-2 fractional order fuzzy
May.2005. PID controller: Design, performance evaluation, and its optimal
[29] Starkey A, Hagras H, Shakya S, et al, “A multi-objective genetic time domain tuning”, ISA Transactions, vol.68, pp.251-275,
type-2 fuzzy logic based system for mobile field workforce area May.2107.
optimization”, Information Sciences, 2015, 329:390-411.Feb [50] Mohammadikia R, Aliasghary M, “Design of an interval type-2
[30] Acampora G, Alghazzawi D, Hagras H, et al, “An interval type-2 fractional order fuzzy controller for a tractor active suspension
fuzzy logic based framework for reputation management in peer to system”, Computers and Electronics in Agriculture, vol.167,
peer e-commerce”, Information Sciences, vol.333, pp. 88-107, Mar. 105049, Dec.2019.
2015. [51] Kumar A, Kumar V, “Design of interval type-2 fractional-order
[31] Sepúlveda R, Montiel O, Castillo O, et al, “Embedding a high speed fuzzy logic controller for redundant robot with artificial bee colony”,
interval type-2 fuzzy controller for a real plant into an FPGA”, Arabian Journal for Science and Engineering, vol.44, no.3, pp.
Applied Soft Computing, vol.12, no.3, pp. 988–998, Mar.2012. 1883–1902, Mar.2019.
[32] Sepúlveda R, Castillo O, Melin P, et al, “Experimental study of [52] Moezi S A, Zakeri E, Eghtesad M, “Optimal adaptive interval type-
intelligent controllers under uncertainty using type-1 and type-2 2 fuzzy fractional-order backstepping sliding mode control method
fuzzy logic”, Information Sciences, Vol.177, no.10, pp. 2023-2048, for some classes of nonlinear systems”, ISA Transactions, vol.93,
May.2007. pp.23-39, Oct.2019.
[33] Castillo O, Melin P, Alanis A, et al, “Optimization of interval type- [53] Zakeri E, Moezi S A, Eghtesad M, “Optimal interval type-2 fuzzy
2 fuzzy logic controllers using evolutionary algorithms”, Soft fractional order super twisting algorithm: A second order sliding
Computing, vol.15, no.6, pp. 1145-1160, Mar.2011. mode controller for fully-actuated and under-actuated nonlinear
[34] Castillo O, Melin P, Pedrycz W, “Design of interval type-2 fuzzy systems”, ISA Transactions, vol.85, pp.13-32, Feb.2019.
models through optimal granularity allocation”, Applied Soft [54] Kumar A, Vijay K, “Performance analysis of optimal hybrid novel
Computing, vol.11, no.8, pp. 5590-5601, Dec.2011. interval type-2 fractional order fuzzy logic controllers for fractional
[35] Hidalgo D, Melin P, Castillo O, “An optimization method for order systems”, Expert Systems with Applications, vol.93, pp.435-
designing type-2 fuzzy inference systems based on the footprint of 455, Mar.2018.
uncertainty using genetic algorithms”, Expert Systems with [55] Abdulkhader H K, Jacob J, Mathew A T, “Robust type-2 fuzzy
Application, vol.39, no.4, pp. 4590-4598, Mar.2012. fractional order PID controller for dynamic stability enhancement of
[36] Martí nez-Soto R, Castillo O, Aguilar L T, et al, “A hybrid power system having RES based microgrid penetration”, Electrical
optimization method with PSO and GA to automatically design Power and Energy Systems, vol10, pp. 357-371, Sep.2019.
Type-1 and Type-2 fuzzy logic controllers”, International Journal [56] Mohammadzadeh A, Ghaemi S, “A modified sliding mode
of Machine Learning and Cybernetics, vol.6, no.2, pp. 175-196, approach for synchronization of fractional-order chaotic/
Apr.2015. hyperchaotic systems by using new self-structuring hierarchical
[37] Amador-Angulo L, Castillo O, “A new fuzzy bee colony type-2 fuzzy neural network”, Neurocomputing, vol.191, pp.200-
optimization with dynamic adaptation of parameters using interval 213, May.2016.
type-2 fuzzy logic for tuning fuzzy controllers”, Soft Computing, [57] Pourya J, Mohammad T, Mahsan T K, “Adaptive type-2 fuzzy
vol.22, no.2, pp. 571-594, Jan.2018. system for synchronization and stabilization of chaotic nonlinear
[38] Melin P, Mendoza O, Castillo O, “Face recognition with an fractional order systems”, IET Control Theory & Applications,
improved interval type-2 fuzzy logic sugeno integral and modular vol.12, no.2, pp. 183-193, Jan.2017.
neural networks”, IEEE Transactions on Systems, Man and [58] Sakthivel R, Kavikumar R, Ma Y K, et al, “Observer-Based H∞
Cybernetics, Part A: Systems and Humans, vol.41, no.5, pp. 1001- Repetitive Control for Fractional-Order Interval Type-2 TS Fuzzy
1012, Sep. 2011. Systems”, IEEE Access, vol.6, pp. 49828-49837, Sep.2018.
[39] Castillo O, Castro J R, Melin P, et al, “Application of interval type- [59] Liu F, “An efficient centroid type-reduction strategy for general
2 fuzzy neural networks in non-linear identification and time series type-2 fuzzy logic system”, Information Sciences, vol.178, no.9, pp.
prediction”, Soft Computing, vol.18, no.6, pp. 1213-1224, Jun.2014. 2224-2236, May.2008.
[40] Castillo O, Melin P, “A review on the design and optimization of [60] Mendel J M, Liu F, Zhai D, “α-Plane representation for type-2 fuzzy
interval type-2 fuzzy controllers”, Applied Soft Computing, vol.12, sets: theory and applications”, IEEE Transactions on Fuzzy Systems,
no.4, pp. 1267-1278, Apr.2012. vol.175, no.5, pp. 1189-1207, Oct.2009.

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2980686, IEEE Access

[61] Mendel J M, “Comments on "α-Plane representation for type-2 [82] Xue D Y, Chen Y Q, “MATLAB solutions to mathematical
fuzzy sets: theory and applications"”, IEEE Transactions on Fuzzy problems in control”, Beijing, CHINA: Tsinghua University, 2007.
Systems, vol.18, no.1, pp.229-230, Feb.2010. [83] Becerikli Y, Celik B K, “Fuzzy control of inverted pendulum and
[62] Wagner C, Hagras H, “Towards general type-2 fuzzy logic systems concept of stability using Java application”, Mathematical and
based on zSlices[J]”, IEEE Transactions on Fuzzy Systems, vol.18, Computer Modelling, vol.46, no.1-2, pp.24-37, Jul.2007.
no.4, pp.637-660, Mar.2010. [84] El-Nagar A M, El-Bardini M”, Interval type-2 fuzzy PID controller
[63] Kumbasar T, Hagras H, “A self-tuning zSlices-based general type-2 for uncertain nonlinear inverted pendulum system”, ISA
fuzzy PI controller”, IEEE Transactions on Fuzzy Systems, vol.23, Transactions, vol.53, no.3, pp. 732-743, May. 2014.
no.4, pp. 991-1013, Jul.2015.
[64] Sanchez M A, Castillo O, Castro J R, “Generalized type-2 fuzzy
systems for controlling a mobile robot and a performance
comparison with Interval type-2 and type-1 fuzzy systems”, Expert
Systems with Applications, vol.42, no.14, pp.5904-5914, Aug.2015.
[65] Amador-Angulo L, Castillo O, Castro J R, “A generalized type-2
fuzzy logic system for the dynamic adaptation the parameters in a
bee colony optimization algorithm applied in an autonomous mobile
robot control”, IEEE International Conference on Fuzzy Systems,
Vancouver, BC, Canada ,2016, pp.537-544.
[66] Oscar C, Leticia A A, Castro J R, et al, “A comparative study of
type-1 fuzzy logic systems, interval type-2 fuzzy logic systems and
generalized type-2 fuzzy logic systems in control problems”,
Information Sciences, vol.354, pp.257-274, Aug.2016.
[67] Khooban M H, Vafamand N, Liaghat A, et al, “An optimal general
type-2 fuzzy controller for Urban Traffic Network”, ISA
Transactions, vol.66, pp.335-343, Jan.2016.
[68] Ontiveros E, Melin P, Castillo O, “High order α-planes integration:
a new approach to computational cost reduction of general type-2 SHI JIAN ZHONG was born in Jiangsu province,
fuzzy systems”, Engineering Applications of Artificial Intelligence, China in 1984. He received the B.E. degrees in automatic department from
vol.74, pp. 186-197, Sep.2018. North China Electric Power University in 2006 and the Ph.D. degree in
[69] Baghbani F, M.-R. A T, Alireza A, “Indirect adaptive robust mixed automatic department from North China Electric Power University in 2012.
H2/H∞ general type-2 fuzzy control of uncertain nonlinear systems”, From 2014 to now, he is a lecturer in School of Energy and Power
Applied Soft Computing, vol.72, pp. 392-418, Nov.2018. Engineering of Nanjing Institute of Technology. His research interests
[70] Mohammadzadeh A, Kaynak O, “A novel general type-2 fuzzy include fuzzy identification and fuzzy control, adaptive inverse control and
controller for fractional-order multi-agent systems under unknown neural networks identification.
time-varying topology”, Journal of the Franklin Institute, vol.36,
no.10, pp.5151-5171, Jul.2019.
[71] Zhao T, Yu Q, Dian S, et al, “Non-singleton general type-2 fuzzy
control for a two-wheeled self-balancing robot”, International
Journal of Fuzzy Systems, vol.21, no.6, pp. 1724-1737, Sep.2019.
[72] Dian S, Han J, Guo R, et al, “Double closed-loop general type-2
fuzzy sliding model control for trajectory tracking of wheeled
mobile robots”, International Journal of Fuzzy Systems, vol.21, no.7,
pp.2032-2042, Oct.2019.
[73] Khooban M H, Niknam T, Sha-Sadeghi M, “A time-varying general
type-II fuzzy sliding mode controller for a class of nonlinear power
systems”, Journal of Intelligent and Fuzzy Systems, vol.30, no.5, pp.
2927-2937, Apr.2016.
[74] Ontiveros-Robles E, Melin P, Castillo O, “Comparative analysis of
noise robustness of type 2 fuzzy logic controllers”, Kybernetika,
vol.54, no.1, pp. 175-201, Jan.2018.
[75] Castillo O, Cervantes L, Soria J, et al, “A generalized type-2 fuzzy
granular approach with applications to aerospace”, Information
Sciences, vol.354, pp.165-177, Aug.2016.
[76] Cervantes L, Castillo O, “Type-2 fuzzy logic aggregation of
multiple fuzzy controllers for airplane flight control”, Information
Sciences, vol.324, pp.247-256, Dec.2015.
[77] Mohammadzadeh A, Kaynak O, “A novel general type-2 fuzzy
controller for fractional-order multi-agent systems under unknown
time-varying topology”, Journal of the Franklin Institute, vol.356,
pp.5151-5171, Jul.2019.
[78] Mohammadzadeh A, Ghaemi S, Kaynak O, et al, “Observer-based
method for synchronization of uncertain fractional order chaotic
systems by the use of a general type-2 fuzzy system”, Applied Soft
Computing, vol.49, pp.544-560, Dec.2106.
[79] Karnik N N, Mendel J M, “Centroid of a type-2 fuzzy set”,
Information Sciences, vol.132, pp.195-220, Feb.2001.
[80] Mendel J M, “Uncertain rule based fuzzy logic systems:
introduction and new directions”, 2nd ed., New York, NY, USA:
Springer:2017.
[81] Podlubny I, “Fractional differential equations”, San Diego, CA,
USA: Academic,1999.

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.

You might also like