World Academy of Science, Engineering and Technology

International Journal of Mechanical and Mechatronics Engineering

Vol:10, No:7, 2016

Fractional-Order PI Controller Tuning Rules for

Cascade Control System
Truong Nguyen Luan Vu, Le Hieu Giang, Le Linh

 retuning the primary controller, which is often unwieldy in the

Abstract—The fractional–order proportional integral (FOPI) industry.
controller tuning rules based on the fractional calculus for the cascade Recently, fractional calculus [6], [7] that has been an
control system are systematically proposed in this paper. increasing attention paid to fractional-order processes, which
Accordingly, the ideal controller is obtained by using internal model
are really useful to represent the different stable physical
control (IMC) approach for both the inner and outer loops, which
gives the desired closed-loop responses. On the basis of the fractional phenomena with anomalous decay, both from the academic
Open Science Index, Mechanical and Mechatronics Engineering Vol:10, No:7, 2016 publications.waset.org/10004885/pdf

calculus, the analytical tuning rules of FOPI controller for the inner and control engineers for the modeling and control issues due
loop can be established in the frequency domain. Besides, the outer to its flexibility and advancement in terms of computation
loop is tuned by using any integer PI/PID controller tuning rules in power. Besides, the fractional-order differential equations
the literature. The simulation study is considered for the stable (FODE) can be obtained by using fractional calculus and is
process model and the results demonstrate the simplicity, flexibility,
also a generalization of the ordinary differential equations
and effectiveness of the proposed method for the cascade control
system in compared with the other methods. (ODE). The generalization of the PID controller, which is so-
called the PIλDμ [8], is involved two extra parameters as the
Keywords—Fractional calculus, fractional–order proportional fractional-order integrator (λ) and fractional-order
integral controller, cascade control system, internal model control differentiator (µ). The fractional-order PID controller affords
approach. more flexibility in PID controller design due to the selection
of five controller parameters that include the proportional
I. INTRODUCTION gain, the integral gain, the derivative gain, the integral order,

T HE performance of cascade control system largely

depends on tuning of both inner and outer loops. For the
design method based on the frequency response, the tuning
and the derivative order. However, the tuning rules of
FOPI/PID controller are much more complex in compared
with standard (integer) PID controller that has only three
rules given by [1]-[3] are usually recommended to design the parameters [8], [9]. In order to pose the same ease of use of
controllers in terms of higher order dynamics and/or time standard PID controller, there are many different ways to
delay in the open loop transfer function of outer loop. design the FOPI/PID controller. The first mention involving
However, the frequency response methods also have a major the use of fractional structure in a feedback-loop was early
weakness due to many trial and error graphical calculations. made by [10] and then it was extended by [11], where a
Krishnaswamy [4] introduced the tuning charts, which can be feedback amplifier was obtained by considering a feedback-
predicted the primary controller parameters by using the loop in terms of the performance of the closed-loop that was
integral time absolute error (ITAE) criterion for the load invariant to changes in amplifier gain. However, this idea was
disturbance on the secondary loop. However, this method is not concretized and remained for decades as a simple
also limited to use for the first-order plus dead time (FOPDT) proposition. Oustaloup [12] introduced the fractional-order
model and just focused on the proportional integral/ algorithms for the control of dynamic system based on non-
proportional (PI/P) configuration. Therefore, the overall integer derivative and demonstrated the significant
performance of control system can be poor for higher order improvement of the CRONE (French abbreviation of
process models. To overcome this problem, [5] introduced the Commande Robuste d’Ordre Non Entier) controller in
tuning rules to obtain desired closed loop responses for the compared with the integer PID controller. In general, due to its
cascade control system and showed the enhanced overall two extra parameters (λ and µ), the fractional order PID
performance. In general, there are two steps for the tuning of controller can be achieved better performance in compared
cascade control system: the secondary controller is firstly with the classical PID controller and it has been become a new
tuned based on the inner process model and the primary trend to solve many industrial control problems [13], [14]. In
controller is then tuned based on the outer process model. accordance with the literature, the tuning method of PIλDμ can
Accordingly, if the secondary controller is retuned for some be generally classified as analytic [9] and heuristic methods
uncertainties, an additional identification step is essential for [15]. In fact, most of the analytic methods are often tuned by
considering the nonlinear objective function, which is
depended on the specification imposed by the users [15].
In this paper, our aim is to design an analytic method of
L. Le, N.L.V. Truong and H.G. Le are with the Faculty of Mechanical
Engineering, Ho Chi Minh City University of Technology and Education, generalized FOPI controller for enhanced performances of
Vietnam (e-mails: vuluantn@hcmute.edu.vn, linhle@hcmute.edu.vn, both integer and fractional processes with time delay for
gianglh@hcmute.edu.vn).

International Scholarly and Scientific Research & Innovation 10(7) 2016 889 ISNI:0000000091950263
 World Academy of Science, Engineering and Technology
International Journal of Mechanical and Mechatronics Engineering
Vol:10, No:7, 2016

cascade control systems. It is mainly based on the concepts of III. ANALYTICAL TUNING RULES OF FOPI CONTROLLERS FOR
fractional calculus [6] and IMC approach [16]. By using the CASCADE CONTROL SYSTEM
frequency domain, the proposed PI tuning rules can be directly A. Design of FOPI Controller in Frequency Domain
derived for many typical process models without introducing
any nonlinear objective function. The fractional integro-differential equation of the FOPI
controller is described by
II. PRELIMINARIES
u  t   K C e  t   K I Dt λ e  t  ,  λ  0 (4)
Some basic fundamentals of fractional calculus, together
with the problem statement that need to understand the
fractional system, as well as the controller are briefly where K C , K I , and  denote the proportional term, integral
introduced in this section. term, and fractional order in the FOPI controller, respectively.
The continuous transfer function of the FOPI controller can
A. Fractional Calculus be obtained through Laplace transformation as:
Fractional calculus [6] is generalization of the ordinary
calculus, which is developed a functional operator D, KI
GC  s   KC  (5)
Open Science Index, Mechanical and Mechatronics Engineering Vol:10, No:7, 2016 publications.waset.org/10004885/pdf

associated to an order v that is not restricted to integer s

numbers. It generalizes usual notions of derivative for a
positive v and integrals for a negative v. From (5), it is clear that the FOPI controller involves three
It is clear that there are various kinds of definitions for parameters (KC, KI, and ) to tune, since the fractional order λ
fractional derivative. However, the most commonly use is the is not necessarily integer.
Riemann-Liouville definition [6], which is generalized two The FOPI controller is represented in the frequency domain
equalities easily proved for integer orders: by substituting s  j into (5):

KI
x  t (6)
n 1
G C  jω   K C 
0 Dx f  x    f  t  dt,
x
n
nN  jω 
λ
(1)
c
 n  1!
Hence, the convenient form is given as:
It is important to note that the generalized definition of D
becomes c Dxv f  x  . The Laplace transform of D pursues the  j 

    cos  I  j sin  I  ,  I 
 (7)
2
well-known rule for zero initial condition as
L  0 Dx f  x   s F  s  . It is implied that under initial
v v
The FOPI controller in terms of the complex equation is
condition, the system with a dynamic behavior described by established by substituting (7) into (6):
differential equations involving fractional derivative give rise
to transfer functions with fractional power of s. More details  K cos    K sin   (8)
GC  j    K C  I  I   j I  I
are given in [6].    

B. Fractional Linear Model B. FOPI Controller Design Procedure for General Process
According to a single-input, single-output (SISO) linear Models
time invariant (LTI) system, the FODE, provided both input
and output signals u(t) and y(t) that is relaxed at t = 0, can be
expressed by differential equation:

n m

 a D y  t    b D u  t 

i 0
i
j 0
j
(2)
i 0 j 0

As a result, (2) can be described in the Laplace domain by

the following transfer function: Fig. 1 Block diagram of cascade control system

Y  s  bm s m  bm 1s m1    b0 s 0 The cascade control system is shown in Fig. 1; the closed
G s   (3)
U  s  an sn  an 1sn1    a0 s0 loop transfer functions for inner and outer loops are obtained
by:
where i and i are arbitrary real positive.
Y2 GC 2GP 2
= (9)
R2 1  GC 2GP 2

International Scholarly and Scientific Research & Innovation 10(7) 2016 890 ISNI:0000000091950263
 World Academy of Science, Engineering and Technology
International Journal of Mechanical and Mechatronics Engineering
Vol:10, No:7, 2016

Y1 GC1GC 2GP1GP 2 realizable. Therefore, it should be transformed into the

= (10)
R1 1  GC 2GP 2  GC1GC 2GP1GP 2 complex form, and then compared with (8). Finally, the
analytical tuning rules can be simply derived for a number of
Here, the controllers GC1 and GC2 have to be designed to process models.
satisfy set-point tracking (R1) and disturbance (D1, D2) Consider the FOPDT process model as following:
regulating requirements.
K
G s  e  s (16)
C. Design of Secondary Controller  s 1
A secondary controller has to be designed to such that set-
point tracking (Y2/R2) gives a stable overdamping response. In In accordance with the above-mention procedure for the
accordance with the IMC parameterization introduced by [16], design of IMC-based controller, the ideal feedback controller
the process model G P  s  is factored into two parts: equivalent to the IMC controller can be found by
  s   p  s  p  s  , where p  s  is the portion of the model
G P m A m  s 1
Gc 2  s   (17)
inverted by the controller (minimum phase), pA  s  is the K  c 2 s  1  e  s 
Open Science Index, Mechanical and Mechatronics Engineering Vol:10, No:7, 2016 publications.waset.org/10004885/pdf

 
portion of the model not inverted by the controller (it is the
non-minimum phase that may be included the dead time By substituting s  jω into (17), it yields
and/or right half plane zeros and chosen to be all-pass), and
the requirement that p A  0   1 is necessary for the controlled
1  cos     c 2  sin   
variable to track its set-point. Then, the IMC controller q  s  Gc 2  j   
   
K  4sin 2     c 2 sin     c22 2 
can be designed as q  s   pm1  s  f  s  . For the 1DOF control   2   (18)
structure, the IMC filter f  s  is chosen for enhanced  1  cos      c 2  sin  
performance as: j  
 2  2 2
K  4sin     c 2 sin     c 2 
 2 
1
f s  (11)
( c s  1) r
where,

where  c is an adjustable parameter, which can be utilized for

e  j  cos    j sin   ,    (19)
the tradeoffs between the performance and robustness. The
integer r is selected to be large enough for the IMC controller
By comparing (18) with (6), the analytical tuning rules can
proper. Then, the IMC controller is obtained by
be found as:
1
q  s   p m1  s  (12)   sin     c 2   1  cos   
( c s  1) r KI 2  (20)
   
K sin  I   4sin 2     c 2 sin     c22 2 
Substituting (11) into (9), the closed-loop transfer functions  2
  
for the desired set-point is simplified as:
1  cos     c 2  sin    K I 2 cos  I  (21)
Y2 1 Kc 2  
 p A2  s  (13)     
R2 ( c 2 s  1) r 2 K  4sin 2     c 2 sin     c22 2 
  
2 
The ideal feedback controller G C 2 that yields the desired D. Design of Primary Controller
loop responses can be constituted by The closed loop transfer function for the outer loop can be
approximately represented by
q2 (14)
GC 2 
1  G P 2 q2 PA2
GC1GP1
 c 2 s  1
r2
Y1
= (22)
Therefore, the ideal feedback controller can be found by R1 1  G G PA2
 c 2 s  1
C 1 P1 r2

pm12
GC 2  (15)
 c 2 s  1
r2
 p A2 Therefore, the process model of the outer loop is considered as

The resulting controller given by (15) does not have the PA2 (23)
G1  GP1
 c 2 s  1
r2
FOPI-type controller form despite that it is physically

International Scholarly and Scientific Research & Innovation 10(7) 2016 891 ISNI:0000000091950263
 World Academy of Science, Engineering and Technology
International Journal of Mechanical and Mechatronics Engineering
Vol:10, No:7, 2016

Now, consider a stable process model of the outer loop of

the form:

G1  s   pm1  s  p A1  s  (24)

Here, our purpose is also to design the controller, GC1, so that

the closed loop transfer function of the outer loop, Y1/R1, has
the form given by

Y1 1
 p A1  s  (25)
R1 ( c1s  1) r1

Fig. 4 Closed loop response due to set point change (Y1/D1) for the
Then, the controller transfer function of the outer loop is
illustrative example
represented by:
Open Science Index, Mechanical and Mechatronics Engineering Vol:10, No:7, 2016 publications.waset.org/10004885/pdf

IV. SIMULATION STUDY

Pm11  s  c 2 s  1
r2
q1
GC1  s    (26) In order to have a fair comparison, the IAE criterion is
1  G1q1 P  s   s  1r1  P  s   considered here for the set-point tracking [17]:
A2
 c1 A1


The primary controller GC1 can be approximated to the
PI/PID controller form as shown in [5].
IAE   e  t  dt
0
(27)

In this section, the following process model introduced by

[17] and [5] was studied as the illustrated example to
demonstrate the performance of the proposed method in
comparison with those of other well-known methods.
TABLE I
TUNING VALUES BY ALL OF COMPARATIVE DESIGN METHODS FOR THE
ILLUSTRATIVE EXAMPLE
Controller
Proposed Seborg et al. [17] Lee et al. [5]
parameters
Outer loop controller
K C1 6.2 3.5 6.2
τ I1 6.2 5.3 6.2
Fig. 2 Closed loop response due to set point change (Y1/R1) for the
- -
illustrative example τ D1 1.48
Inner loop controller
K C2 1.68 4 5
τ I2 0.82 - -
λ2 0.9 - -
IAE 2.42 2.75 2.77

4
GP1 
 2s  1 4s  1
5
GP 2 
 s  1
(28)
1
GD1 
 3s  1
Fig. 3 Closed loop response due to set point change (Y1/D2) for the
illustrative example GD 2  1
Gm1  0.05
Gm 2  0.2

The PID controllers for inner and outer loops for the above
process were tuned by the proposed tuning rules. The results
were compared with those by the frequency method [17] and

International Scholarly and Scientific Research & Innovation 10(7) 2016 892 ISNI:0000000091950263
 World Academy of Science, Engineering and Technology
International Journal of Mechanical and Mechatronics Engineering
Vol:10, No:7, 2016

[5]. The resulting PID parameters are listed in Table I. Since [15] C. A Monje, B. M. Vinagre, V. Feliu, and Y.Q. Chen, “Tuning and auto-
tuning of fractional order controllers for industry applications,” Control
the PID controllers in cascade control should be tuned Engineering Practice, vol. 16, pp. 798–812, 2008.
considering all the closed loop performances both for set-point [16] M. Morari, E. Zafiriou, Robust Process Control. NJ, USA: Prentice Hall,
tracking (Y1/R1) and disturbance rejection (Y1/D1 and Y1/D2), 1989.
[17] D. E. Seborg, T. F. Edgar, and D. A. Mellichamp, Process Dynamics
the tuning methods were tested in terms of all these and Control. New York, USA: Wiley, 1989.
performances. Figs. 2-4 show the closed loop responses tuned
by the proposed method and the frequency response method
for the unit step change in R1, L2, and L1, respectively. The
results shown in the figures illustrate the superior performance
of the proposed method.

IV. CONCLUSION
An analytical design method of FOPI controller for the
cascade control systems was proposed based on fractional
Open Science Index, Mechanical and Mechatronics Engineering Vol:10, No:7, 2016 publications.waset.org/10004885/pdf

calculus and IMC approach to provide improved performance

for both disturbance rejection and set-point tracking. The
simulation studies demonstrate that can be applied to a large
number of dynamic models, consistently afforded the superior
performance with fast and well-balanced closed-loop time
responses.

ACKNOWLEDGMENT
This work was supported by the HCM City University of
Technology and Education (HCMUTE), Vietnam.

REFERENCES
[1] F. D. Jury, Fundamentals of Three-Mode Controllers. NY, USA: Fisher
Controls Company-Technical Monogragh, 1973.
[2] J. O. Hougen, Measurement and Control Applications. Pittsburgh, USA:
Instrument Society of America, 1979.
[3] T. F. Edgar, R. C. Heeb, and J. O. Hougen, “Computer-aided process
control system design using interactive graphics,” Comput. Chem. Eng,
vol. 5, no. 4, pp. 225-231, 1982.
[4] P. R. Krishnaswamy and G. P. Rangaiah, “When to use cascaded
control,” Ind. Eng. Chem. Res., vol. 29, no. 1, pp. 2163-2166, 1990.
[5] Y. Lee, S. Park, and M. Lee, “PID controller tuning to obtain desired
closed loop responses for cascade control systems,” Ind. Eng. Chem.
Res., vol. 37, no. 4, pp. 1859-1865, 1998.
[6] K. S. Miller and B. Ross, An Introduction to the Fractional calculus and
Fractional Differential Equations. NY, USA: A Wiley-Interscience
Publication, 1993.
[7] I. Podlubny, Fractional differential equations: an introduction to
fractional derivatives, fractional differential equations to methods of
their solution and some of their applications. San Diego, USA:
Academic Press, 1999.
[8] I. Podlubny, “Fractional-order systems and PIλDμ-controllers,” IEEE
Trans Automatic Control, vol. 44, no. 1, pp. 208–14, 1999.
[9] N. L. V. Truong and M. Lee, “Analytical design of fractional-order
proportional-integral controllers for time-delay processes,” ISA Trans.,
vol. 52, pp. 583-591, 2013.
[10] H. Bode, “Relations between attenuation and phase in feedback
amplifier design,” Bell System Technical Journal, vol.19, no. 1, pp. 421–
454, 1940.
[11] R. S. Barbosa, T. Machado, and I. M. Ferreira, “Tuning of PID
controllers based on Bode’s ideal transfer function,” Nonlinear
Dynamics, vol. 38, no. 2, pp. 305–21, 2004.
[12] A. Oustaloup and M. Benoit, La Commande CRONE: Du Calaire Au
Multivariable. Paris, France: Hermès, 1999.
[13] B. M. Vinagre, C.A Monje, A. J. Calderon, and J. I. Suarez, “ Fractional
PID controllers for industry application: a brief introduction,” Journal of
Vibration Control, vol. 13, pp. 1419–1429, 2007.
[14] D. Valério, J. S. Da Costa, “Tuning of fractional PID controllers with
Ziegler–Nichols-type rules,” Signal Process, vol. 86, no. 1, pp. 2771–
2784, 2006.

International Scholarly and Scientific Research & Innovation 10(7) 2016 893 ISNI:0000000091950263