Optimal Design of Fractional Order PID Controller using

Lightning Search Algorithm

Wei Gao† Youliang Zhao Yimin Zhang

School of Physics and Electronics School of International Education School of Physics and Electronics
Henan University Henan University Henan University
Kaifeng Henan, China Kaifeng Henan, China Kaifeng Henan, China
[email protected] [email protected] [email protected]

ABSTRACT forms of fractional controllers have also been proposed. A

comparison of four fractional-order controllers can be found in [7].
This paper investigates the parameter tuning problem of Due to the addition of order parameters λ and µ, the FOPID
fractional-order proportional-integrator-derivative (FOPID) controllers have potential performance advantages over the IOPID
controller. A new meta-heuristic algorithm, namely Lightning ones. However, compared to the IOPID controller, the tuning of
Search Algorithm (LSA), which was inspired by the lightning the FOPID controller is more complicated.
phenomena in nature, is employed to optimize the design of the Currently, the tuning strategy of the IOPID controller mainly
FOPID controller. The simulation results show the effectiveness includes the frequency-domain method [8], [9], and numerical
of the proposed method. Compared to the integer-order PID optimization method [10], [11]. The frequency-domain method is
controller, the controlled benchmark system with the LSA tuned usually based on the given gain and phase margin, and the
FOPID controller provides better responses with lower overshoot controller parameters could be analytically calculated. However,
and faster settling time. this method cannot be directly applied to the tune of the FOPID
parameters [11].
CCS CONCEPTS Hence, numerical optimization methods got more attention
• Applied computing~Physical sciences and [11]. In this study, a lightning search algorithm-based
engineering~Engineering optimization method is employed to design the FOPID controller
for a benchmark DC motor position servomechanism system.
KEYWORDS Motor position control simulations are established, and the
Fractional calculus, FOPID controller, Lightning search advantage of the proposed method is demonstrated.
algorithm, Numerical optimization The significant contributions of this paper include: 1) A newly
nature-inspired meta-heuristic algorithm for the optimal tuning of
the FOPID controller is proposed; 2) A detailed simulation study
1 Introduction using a benchmark system to make our results reproducible; 3)
The PID controller is the most widely used control strategy in the Show that, under the same optimization constriction, the best
industry area [1]. In the last ten years, fractional-order calculus FOPID controller outperforms the best IOPID controller and the
and its application have caused great attention of researchers in exist literature results.
different fields [2-6]. Centered on the fractional calculus, The rest of the paper is structured as follows: Firstly, the
Podlubny [2] expanded the integer-order (IO) PID controller to benchmark system, the FOPID controller, and its mathematical
fractional order, namely the PIλDµ controller, incorporating an basis, and the optimal design framework of the FOPID controller
integrator with order λ and a differentiator with order µ. And other by the LSA method are described in section II. Secondly,
extensive simulation investigations results are given to illustrate
†
Corresponding author. the proposed optimal design approach's validity in Section III.
Permission to make digital or hard copies of all or part of this work for personal or
classroom use is granted without fee provided that copies are not made or Finally, conclusions are drawn in Section IV.
distributed for profit or commercial advantage and that copies bear this notice and
the full citation on the first page. Copyrights for components of this work owned
by others than ACM must be honored. Abstracting with credit is permitted. To
copy otherwise, or republish, to post on servers or to redistribute to lists, requires 2 Problem Formulation
prior specific permission and/or a fee. Request permissions from
[email protected].
EITCE 2020, November 6–8, 2020, Xiamen, China
2.1 Position Servomechanism Model
© 2020 Association for Computing Machinery. The Benchmark position servomechanism system is a “DC motor
ACM ISBN 978-1-4503-8781-1/20/11…$15.00
https://doi.org/10.1145/3443467.3443851 with elastic shaft” from [12], [13]. It consists of a DC motor,

767
gearbox, elastic shaft, and load. The system schematic is showed
in Figure 1. The differential equations representing this system are
where Γ(ꞏ) is the Euler’s gamma function:

When we deal with dynamic systems, is usually a causal

function of , so the definition of fractional-order integral to be
used bellow is

Figure 1: DC Motor Schematic For the positive integer m, the Riemann-Liouville definition
for the fractional-order derivative of order has the
following form:

(1)
where, represents load angular velocity, where .
represents the motor shaft angular velocity, and T represents the Another important definition of fractional-order derivative is
torque acting on the load, and V represents the applied voltage. based on the Grunwald-Letnikov's definition of backward
The remaining terms are constant parameters. If we describe the difference. This definition takes the form of:
variables of the state as , then the
state-space model of the system can be described as follow: ,

where is integer and is the binomial coefficients.

The linear time invariant fractional-order differential equation
is

when the Laplace transform is applied to the case of zero

initial condition, the input-output representation of the fractional-
order system can be obtained. In the case of continuous situation,
the fractional order system will be given by the transfer function
in the following form:

A fractional-order PID controller is given by

(2) (5)
where . Applying Laplace transform to (5) with
(3)
zero initial conditions, the transfer function of the controller can
be expressed by
The controller must set the angular position of the load, , at
a desired value by adjusting the applied voltage V. However, due .
to the finite shear strength of the elastic shaft, the torque T must (6)
be kept within the range |T| ≤ 78.5Nm. In addition, the voltage
source physically limits the applied voltage to a range of |V|≤ 2.3 LSA based FOPID Design
220V. The optimization of FOPID controller design for DC motor
From the input/output point of view, the transfer function of position control by lightning search algorithm can be represented
the angular position control process can be written as follows: by the block diagram in Figure 2. The integral of time weighted
absolute error (ITAE) criteria is employed as an objective
(4) function J.
(7)
2.2 Fractional order PID Controller: A brief
The Riemann-Liouville fractional-order integral is defined as

768
 Kp, Ki, Kd, λ, μ J R(s) where is the distance between the lead
LSA
+
D(s) - projectile and the old space projectile .
R(s) E(s) + U(s) + C(s) • The lead projectile: represents the best solution projectile. It
+ Kp + Gp(s)
is modeled as a random number extracted from a standard normal
- +
+ Plant distribution with shape parameter and scale parameter . The
Ki/sλ
normal PDF is expressed as
+
μ
Kd*s
Gc(s)
FOPID Controller This projectile is updated as

where normrand is a random number generated by normal

Figure 2: LSA based FOPID controller optimization
distribution function. The new position of the pilot projectile
cannot guarantee the propagation of the step leader, unless the
Designing the FOPID is to search for optimal parameters
energy of the lead projectile is greater than of the old step leader
and in (6). The best J of each iteration will to extend to a satisfactory solution.
be fed back to the LSA tuning block to be minimized in order to The whole procedure of the Lighten Search Algorithm can be
find the optimal FOPID parameters satisfying to the inequality summarized as follow:
constraints stated in (7).
Algorithm 1. Procedure of LSA
1: Reset iterations and channel time;
2: Reset leader tips energies;
(7) 3: Randomly generate walkers (transition projectiles);
4: Evaluate projectiles energy;
2.4 Lightning Search Algorithm 5: while iteration ≤ Max iterations do
Lightning Search Algorithm is a new meta-heuristic algorithm 6: Update leader tips energies;
inspired by the natural phenomenon of lightning [14]. LSA 7: Update worst and best walkers;
performs the search process useing a set of fast particles called 8: if Max channel time then
projectiles, which move in the search space in the form of a step- 9: Move walker from worst position to best;
size binary tree structure. 10: Reset channel time;
LSA incorporates three types of projectiles, which are 11: end if
described as follows: 12: Update direction and kinetic energy;
• Transition projectile: projectiles that forms the original group 13: Update space and lead projectiles;
of walkers. These projectiles are generated using random numbers 14: Evaluate projectiles energies;
extracted from the standard uniform probability distribution on the 15: if fork then
open interval representing the solution space. The probability 16: Create two symmetrical channels at fork point;
17: Eliminate channel which has lower energy;
density function (PDF) of this distribution is given as
18: end if
19: Update positions;
20: iteration + 1;
21: end while
where a and b are the lower and upper bounds, respectively, of the 22: return best step leader.
solution space.
• Space projectiles: these projectiles are updated and evolved
so one of them becomes the leader. The position of the space
3 Numerical results
projectile can be partially modeled as a random number The benchmark position servomechanism system was modeled on
generated by an exponential distribution with a shaping parameter MATLAB/Simulink platform, and the FOPID controller was
D. The PDF of an exponential distribution is: implemented based on MATLAB toolbox fotf from [15].
Parameters of the system are summarized in the following Table I.

Table I. Parameters of the servomechanism system

Symbol Value(SI Units) Meanings

769
 1280.2 Torsional rigidity
10 Motor constant
0.5 Motor inertia
Load inertia
20 Gear ratio
0.1 Motor viscous friction coefficient
25 Load viscous friction coefficient
20 Armature resistance
The Lighten Search Algorithm is coded, and the numerical
simulation is implemented by MATLAB/Simulink platform. The
parameters of the algorithm are a priori set: the numbers of
projectiles N = 30, the lower bounds a = 0, the upper bounds b =
150, the dimension of search space, D = 5, the shape parameter ν
= 0 and the maximum number of iterations, MaxIter = 100. Figure 3: Convergence process of ITAE (J) towards optimum
The convergence of the search is shown in Figure 3. After the
search stops, as summarized in the following Table II, the best
FOPID parameters and their response index values are
successfully obtained. Where is rise time, is settling time
and is maximum overshoot.

Table II. Comparison of response performances

Methods Controllers (s) (s) (%)

Xue PID 4.10 39.50 47.70 67.70
Xue FOPID 5.50 5.20 1.39 2.20
LSA FOPID 3.70 2.70 0.66 0.57
From the results of LSA, the optimal FOPID controller is

(8)

Figure 4: Comparison of close-loop step responses

The unit step responses of the system with different controllers
are shown in Figure 4. Compared with the best results in reference
[13], the position servo system with the LSA FOPID controller
has better response performance, smaller overshoot, and faster rise
time than the position servo system with the IOPID controller.
The frequency responses of the controlled position
servomechanism system under different controllers are shown in
Figure 5 for further comparison. The results show that a better
gain margin and phase margin were created by the proposed LSA
tuned FOPID controller.

Figure 5: Comparison of frequency performances

770
4 Conclusion
In this investigation, the aim was to realize the optimal design of
the FOPID controller for a DC motor position servomechanism
system. The Lightening Search Algorithm was employed to
optimize the FOPID parameters. This investigation shows that the
controlled servomechanism system with the LSA tuned FOPID
controller provided better overall performance than the IOPID
controller and existing literature results.

ACKNOWLEDGMENTS
This work was partly funded by the National Natural Science
Foundation of China under Grant No.61773008.

REFERENCES
[1] T. Hagglund and K. J. Astrom, The Control Systems Handbook, Second
Edition, vol. 20103237. CRC Press, 2010.
DOI: https://doi.org/10.1201/b10384
[2] I. Podlubny, “Fractional-order systems and PIλDµ-controllers,” IEEE Trans.
Automat. Contr., vol. 44, no. 1, pp. 208–214, Jan. 1999.
DOI: https://doi.org/10.1109/9.739144
[3] R.Magin, M.D.Ortigueira, I.Podlubny, and J.Trujillo, “On the fractional signals
and systems,” Signal Processing, vol. 91, no. 3, pp. 350–371, Mar. 2011.
DOI: https://doi.org/ 10.1016/j.sigpro.2010.08.003
[4] J. Xu, C. C. Mi, B. Cao, and J. Cao, “A new method to estimate the state of
charge of lithium-ion batteries based on the battery impedance model,” J.
Power Sources, vol. 233, pp. 277–284, 2013.
DOI: https://doi.org/ 10.1016/J.JPOWSOUR.2013.01.094
[5] Y. F. Pu and X. Yuan, “Fracmemristor: Fractional-Order Memristor,” IEEE
Access, vol. 4, pp. 1872–1888, 2016.
DOI: https://doi.org/ 10.1109/ACCESS.2016.2557818
[6] J. Tian, R. Xiong, and Q. Yu, “Fractional-Order Model-Based Incremental
Capacity Analysis for Degradation State Recognition of Lithium-Ion Batteries,”
IEEE Trans. Ind. Electron., vol. 66, no. 2, pp. 1576–1584, 2019.
DOI: https://doi.org/ 10.1109/TIE.2018.2798606
[7] D.Y. Xue and Y.Q. Chen, “ A comparative introduction of four fractional order
controllers,” in Proceedings of the 4th World Congress on Intelligent Control
and Automation, pp. 3228–3235, 2002.
DOI: https://doi.org/ 10.1109/WCICA.2002.1020131
[8] R. S. Barbosa, J. A. T. Machado, and I. M. Ferreira, “Tuning of PID controllers
based on bode’s ideal transfer function,” Nonlinear Dyn., vol. 38, no. 1–4, pp.
305–321, 2004.
DOI: https://doi.org/ 10.1007/s11071-004-3763-7
[9] S. K. Damarla and M. Kundu, “Design of Robust Fractional PID Controller Using
Triangular Strip Operational Matrices,” Fract. Calc. Appl. Anal., vol. 18, no. 5,
pp. 1291–1326, Jan. 2015.
DOI: https://doi.org/ 10.1515/fca-2015-0074
[10] M. Zamani, M. Karimi-Ghartemani, N. Sadati, and M. Parniani, “Design of a
fractional order PID controller for an AVR using particle swarm optimization,”
Control Eng. Pract., vol. 17, no. 12, pp. 1380–1387, Dec. 2009.
DOI: https://doi.org/10.1016/j.conengprac.2009.07.005
[11] W. Zheng, Y. Luo, Y. Chen, Y. Pi, and W. Yu, “An Improved Frequency
Domain Method for the Fractional Order PIλDµ Controller Optimal Design,”
IFAC-PapersOnLine, vol. 51, no. 4, pp. 681–686, 2018.
DOI: https://doi.org/ 10.1016/j.ifacol.2018.06.180
[12] Mathworks Inc., Model Predictive Control Toolbox User’s Guide. [Online].
Available:https://ww2.mathworks.cn/help/mpc/ug/servomechanismcontroller.ht
ml
[13] Dingyu Xue, Chunna Zhao, and YangQuan Chen, “Fractional order PID control
of a DC-motor with elastic shaft: a case study,” in 2006 American Control
Conference, 2006, no. May 2014, p. 6 pp.
DOI: https://doi.org/ 10.1109/ACC.2006.1657207
[14] H. Shareef, A. A. Ibrahim, and A. H. Mutlag, “Lightning search algorithm,”
Appl. Soft Comput., vol. 36, no. August, pp. 315–333, Nov. 2015.
DOI: https://doi.org/10.1016/j.asoc.2015.07.028
[15] Dingyu Xue, A MATLAB toolbox – FOTF Toolbox. [On-line]. Available:
http://cn.mathworks.com/matlabcentral/fileexchange/60874fotf-toolbox

771