Professional Documents
Culture Documents
A PID Control Scheme With Enhanced Non-Dominated Sorting Genetic Algorithm Applied To A Non-Inverting Buck-Boost Converter
A PID Control Scheme With Enhanced Non-Dominated Sorting Genetic Algorithm Applied To A Non-Inverting Buck-Boost Converter
https://doi.org/10.1007/s12046-022-02012-z Sadhana(0123456789().,-volV)FT3
](0123456789().,-volV)
Abstract. The non-inverting buck-boost converter is designed to regulate the output voltage of the non-linear
converter in real-time applications. Numerous studies have been carried out in recent years to improve the
dynamic behavior of converters using intelligent control techniques; still, the occurrence of large overshoots in
transient and reduction of ripple over voltage has not been effectively discussed. This can be efficiently handled
by selecting filter capacitance values effectively. This article proposes a design of a Proportional-integral-
derivative controller optimized by a non-dominated sorting genetic algorithm (NSGA) and enhanced non-
dominated sorting genetic algorithm (ENSGA) to choose the filter capacitance value and to maintain the
converter output voltage at the desired level. A dynamic crowding distance-based ENSGA has been employed to
attain the homogenous distribution of non-dominated solutions. The stability of the buck-boost mode is analyzed
using frequency response. Based on the simulation results obtained for various load conditions, the ENSGA
controller outperformed the NSGA controller in terms of better output voltage regulation, lesser transients, and
minimizing the ripple voltage.
Keywords. Non-inverting buck-boost converter; PID controller; NSGA; ISE; Performance specifications.
transfer in the BBC. The PWM control method has been better convergence, and robustness. This article uses NSGA
addressed in [10] to operate the controller at operation and ENSGA to optimize the PID controller parameters.
mode switching when the output voltage falls near the input NSGA-based algorithms are preferred because of their less
voltage range. computational effort and more independent distribution
In the works of literature, various control approaches for over the Pareto front.
converters have been developed [11–14]. Still, many The appropriate selection of filter capacitance can reduce
researchers concentrated on PID controllers with opti- the ripple voltage in DC to DC converters; the very high
mization techniques for the output voltage regulation, value of the DC filtering capacitor may lead to poor
components values optimization, and performance dynamic performance, and the small value of the DC link
improvement of the BBC. The PID controller is the stan- capacitor may increase the ripple voltage across the load.
dard control technique for many industrial applications [15] So, it is essential to select the optimal value of the filter
due to its simple structure with stable operating conditions capacitor. In this paper, two objectives are considered, such
[16–19]. The particle swarm optimization algorithm has as tuning the parameters of the PID controller and choosing
been presented in [20] with the PID controller for the filter capacitance which can be solved by framing a multi-
interleaved BBC converter; the simulation results are objective strategy. The two objectives of the algorithm are
demonstrated a fast and stable response under varying load to find optimal values for kp , ki and kd are selected to
conditions. A similar algorithm has been proposed in [21] achieve minimal ISE of the output voltage to the reference
for handling uncertain parameters in converters. The solu- DC output voltage and the mitigation of ripple in the DC
tion also relies on a particle swarm optimization algorithm voltage output. The NSGA algorithm has been proved to be
that minimizes an error, allowing better results than the a versatile multi-objective optimization algorithm in
conventional fixed gain PID controller. A proportional-in- numerous works of literature. The NSGA algorithm has
tegral controller with an artificial fish swarm optimization been employed to optimize the economic reactive power
algorithm is proposed for a BBC for voltage regulation dispatch problem [35]. The combined economic dispatch
[22]. The PID controller with the Levy flight distribution and emission minimization have been optimized in [36].
algorithm has been presented for the buck converter [23], This strategy in [37] has solved the reactive support plan-
and the effectiveness of the algorithm is proved through ning problem. In [38], the NSGA algorithm is adopted for
disturbance rejection, robustness, and transient response. A the speed regulation and minimization of torque ripple that
cuckoo search algorithm was introduced to find the optimal arises in the switched reluctance motor. The proposed
PID controller parameters for the BBC of the light-emitting article addresses the following objectives:
diode circuit; the time-domain specifications are presented
1. To select optimal filter capacitance value, the NSGA and
[24]. A hunger games search algorithm has been proposed
ENSGA have been realized in simulation and real-time
to optimize converters parameters and controller gains [25].
environments for tuning PID controller parameters.
In [26], the genetic algorithm optimized by a pre-calculated
2. To control the output voltage of the converter by
duty cycle is presented for a Non-Inverting Buck-Boost
minimizing the ISE of desired and the actual output
Converter (NIBBC). A solar-powered battery charging
voltage.
application, a genetic algorithm-based PID controller, has
3. To compare the performance of the proposed NSGA and
been used for the BBC [27]. The outcomes showed a better
ENSGA and to identify the better strategy to design the
system stability response and low steady-state error. Simi-
converter.
larly, a modified real-coded genetic algorithm has been
proposed for interleaved boost converter for efficiency
improvement [28]. In a similar contribution, [29] have
The article is presented as follows. The average state-
developed the design procedure for an isolated DC to DC
space modeling of energizing mode and de-energizing
power electronic converter. The stable and robust control
mode of the NIBBC is presented in section 2. The
scheme in [30, 31] manages transients and the steady-state
PID controller gains for NIBBC, multi-objective
performance of DC to DC converters. In [32], the NSGA II
function, review of optimization algorithms, and the
optimization algorithm has been proposed for power con-
proposed structure model are presented in section 3.
version and controller stages of interleaved boost convert-
Simulink model implementation for NIBBC, param-
ers in electric vehicle trains. Even though many
eters for algorithm development, and statistical per-
optimization controllers have been proposed for the BBC
formance of the NSGA and ENSGA are discussed in
control purpose and parameters optimization, a genetic
section 4. Section 5 presented the simulation results
algorithm-based optimization technique is prevalently
for various source voltage, load conditions, statistical
compared to other methods. Because most search opti-
performance, and stability analyses of the NIBBC.
mization issues are constrained in nature, this approach is
The experimental validation is described in section 6.
best suited for unconstrained optimization problems. Also,
The conclussins are provided in section 7.
many newly proposed modified genetic algorithms [33, 34]
have improved searching capacity, low computation time,
Sådhanå (2022)47:222 Page 3 of 18 222
Figure 3. Block Diagram of the proposed NIBBC with the PID controller tuned by NSGA.
Figure 4. PWM generated strategy for the buck and boost switches.
222 Page 6 of 18 Sådhanå (2022)47:222
Table 2. Specification of the NIBBC. Table 3. Parameter used for the optimization.
Switch 1
g m L
D S
D2
g
D
Vs
Switch 2 + v
D1 C R -
m
S
Vo
Sawtooth
vo
Generator
Gain To Workspace
PID Controller
1/setpoint
PID(s)
Constant
Parameters Best value Mean value Worst value Std. optimal value
NSGA
ISE desired output voltage (V) 43.52 47.53 55.65 6.75
ISE ripple output voltage (V) 0.218 0.352 0.581 0.318
ENSGA
ISE desired output voltage (V) 39.17 43.15 49.62 5.231
ISE ripple output voltage (V) 0.193 0.291 0.55 0.213
1X r
f k f k
CDi ¼ iþ1 i1 ð27Þ
r k¼1
Figure 8. Simulation result of NIBBC (boost mode, Vs =6V) for output voltage Vo = 12V and current Io=0.24A (a) NSGA and
(b) ENSGA.
Figure 9. Simulation result of NIBBC (boost mode, Vs = 24 V) for output voltage Vo = 12 V and current Io=0.24 A (a) NSGA and
(b) ENSGA.
Figure 10. Ripple voltage of NIBBC (a) NSGA and (b) ENSGA.
Figure 11. Dynamic response of NIBBC in boost operation for line variation (Vs changes from 6 V to 12 V) (a) NSGA and (b) ENSGA.
Figure 12. Dynamic response of NIBBC in buck operation for line variation (Vs changes from 12 V to 24 V) (a) NSGA and
(b) ENSGA.
Figure 13. Dynamic response of NIBBC in boost (Vs = 6V) operation for load variation (Io changes from 0.24A to 0.16A) (a) NSGA
and (b) ENSGA.
running at 3.90 GHz and 32 GB of RAM. figure 5 shows minimizing ISE, the optimal parameters listed in table 3. To
the NIBBC’s MATLAB/Simulink model. ensure the optimum result, each trial made 10 independent
The NSGA and ENSGA optimization algorithms are runs and average values are reported. The statistical per-
applied to NIBBC for the PID controller’s parameters formance of the proposed PID controller tuning is tabulated
tuning and to select optimum filter capacitance value by in table 4. Figures 6 and 7 represent the Pareto-optimal
Sådhanå (2022)47:222 Page 11 of 18 222
Figure 14. Dynamic response of NIBBC in boost (Vs = 24 V) operation for load variation (Io changes from 0.24A to 0.16 A) (a) NSGA
and (b) ENSGA.
Figure 15. The control signal of the PID controller for NIBBC.
front of the NSGA and ENSGA. The obtained optimal 5. Simulation results and analysis of NIBBC
parameters using the optimization algorithm are presented
in table 5, the optimal controller parameters of the PID The NIBBC has been tested under various conditions with
controller optimized by conventional ENSGA are Kp is the proposed optimization strategies with the PID controller
1.32, Ki is 63.42 and Kd is 0.0023, the filter capacitance is to analyze the proposed methodology’s performance. The
470 lF. simulation result for the PID controller tuned for boost
222 Page 12 of 18 Sådhanå (2022)47:222
Table 6. Performance Comparisons Proposed ENSGA controllers for NIBBC with Existing Works of Literature.
Figure 17. Performance Comparisons of NSGA and ENSGA-based controllers for NIBBC.
Sådhanå (2022)47:222 Page 13 of 18 222
Figure 18. Frequency response of NIBBC in buck mode. Figure 20. Computational analysis of the proposed controller.
resistance is varied in the range of 50 X to 75 X at t= 2 Figure 23. Experimental result of NIBBC (boost mode, Vs = 6
seconds and is restored at t= 4 seconds, and the input V) using ENSGA-based controller.
voltage is maintained at 6V for the boost converter. The
observed output current varies from 0.24 A to 0.16 A, as
shown in figure 13 the proposed controllers. Similarly,
been compared with existing works of literature and pre-
during buck mode, the load is varied at an input voltage of
sented in table 6. It is identified that the proposed ENGSA-
24 V, and the output current is varied load variations are
PID controller outperformed other techniques in literatures
also given to the buck mode, as shown in figure 14, with an
with better time domain specifications, also the bar
input voltage of 24 V the output current varies from 0.24 A
chart representation given in figure 17.
to 0.16 A. The input voltage to the PID controller and
For a dc-dc converter system to be stable and properly
output voltage from the PID controller is represented in
regulated, the frequency response tries to ensure the fol-
figure 15, and PWM generated signal for NIBBC is given in
lowing criteria. [51, 52].
figure 16. The performance of the proposed controller has
Sådhanå (2022)47:222 Page 15 of 18 222
Figure 24. Experimental result of NIBBC (boost mode, Vs = 6 Figure 26. Experimental dynamic response of NIBBC (buck
V) using ENSGA-based controller. mode) using ENSGA for line variation.
List of symbols
PID Proportional-Integral-Derivative Controller
MOO Multi-objective optimization
NSGA Non-dominated Sorting Genetic Algorithms
ENSGA Enhanced Non-dominated Sorting Genetic
Algorithms
ISE Integral Squared Error
NIBBC Non-Inverting Buck Boost Converter
BBC Buck Boost Converter
MOSFET Metal Oxide Semi-Conductor Field Effect
Transistor
PWM Pulse Width Modulation
GA Genetic Algorithm
CD Crowding Distance
DCD Dynamic crowding distance
Figure 28. Ripple voltage of ENSGA controller-based NIBBC. D1 and Power Diodes
D2
L Inductor
C filtering capacitor
varied from 6 V to 12 V, whereas, for buck mode, the line R Load Resistor
variation is done by varying from 24 V to 12 V. Similarly, S1 Buck switch of N-channel MOSFET
on varying the load resistance the proposed converter S2 Boost switch of N-channel MOSFE
illustrated better performance with less ripple voltage. fig- VH1 Maximum voltages of the carrier waveform of
ure 28 illustrates the ripple produced in NIBBC with a filter the buck region
capacitance of 470 microfarads. The proposed ENSGA VL1 Minimum voltages of the carrier waveform of
approach produces a ripple voltage of 0.4 whereas in sim- the buck region
ulation the produced ripple voltage is 0.32 V as a ripple VH2 Maximum voltages of the carrier waveform of
value. the boost region
VL2 Minimum voltages of the carrier waveform of
the boost region
7. Conclusion G1 Gate pulses for buck switch
G2 Gate pulses for boost switch
In this article, a multi-objective optimization strategy such Vs Source Voltage
as ENSGA has been proposed to optimize the performance V0 Output Voltage
of the PID controller by tuning its parameters to optimal fs Switching Frequency
value and identifying the filter capacitance so that the rL equivalent resistance of inductor
performance of the NIBBC is improved. The ENSGA rC equivalent resistance of capacitor
combines the NSGA algorithm with DCD to handle lateral
Sådhanå (2022)47:222 Page 17 of 18 222
[22] Chanjira P and Tunyasrirut S 2020 Intelligent control using [37] Ramesh S, Kannan S and Baskar S 2012 Application of
metaheuristic optimization for buck-boost converter. J. Eng. modified NSGA-II algorithm to multi-objective reactive
2: 744 power planning. Appl. Soft Comput. 12(2): 741–753
[23] Izci D, Ekinci S and Hekimoğlu B 2022 A novel modified Lévy [38] Kalaivani L, Subburaj P and Iruthayarajan M W 2013 Speed
flight distribution algorithm to tune proportional, integral, control of switched reluctance motor with torque ripple
derivative and acceleration controller on buck converter system. reduction using non-dominated sorting genetic algorithm
Trans. Inst. Measur. Control 44(2): 393–409 (NSGA-II). Int. J. Electr Power Energy Syst. 53: 69–77
[24] Verma P, Patel N, Nair N K C and Sikander A 2016. Design [39] Rajesh R and Baranilingasen I 2022 Modeling and analysis
of PID controller using cuckoo search algorithm for buck- of different tuning methodologies of PID controller for a
boost converter of LED driver circuit. In: 2016 IEEE 2nd linearly parameterized non linear system. Int. J. Sci. Adv.
Annual Southern Power Electronics Conference, pp. 1–4 Res. Tech. 3(5): 615–620
[25] Izci D and Ekinci S 2022 A novel improved version of [40] Rajamani M P E, Rajesh R and Willjuice Iruthayarajan M
hunger games search algorithm for function optimization and 2021 Design and experimental validation of PID controller
efficient controller design of buck converter system. e- for buck converter: a multi-objective evolutionary algorithms
Prime-Adv. Electr. Eng. Electr. Energy 2: 100039 based approach. IETE J. Res. 2: 1–12
[26] Ortatepe Z and Karaarslan A 2020 Pre-calculated duty cycle [41] Marler R T and Arora J S 2004 Survey of multi-objective
optimization method based on genetic algorithm imple- optimization methods for engineering. Struct. Multi. Optim.
mented in DSP for a non-inverting buck-boost converter. J. 26(6): 369–395
Power Electr. 20(1): 34–42 [42] Kou G, Xiao H, Cao M and Lee L H 2021 Optimal
[27] Sivamani D, Shyam D, Ali A N, Premkumar K, Narendiran S computing budget allocation for the vector evaluated genetic
and Alexander S A 2021 Solar Powered Battery Charging algorithm in multi-objective simulation optimization. Auto-
System Using Optimized PI Controller for Buck Boost matica 129: 109599
converter. In: IOP Conference Series: Materials Science and [43] Murata T and Ishibuchi H 1995 MOGA: multi-objective
Engineering, p. 012151 genetic algorithms. In: IEEE Inter. Conference on Evolu-
[28] Valarmathi K, Arundevi M and Mahendran R 2015 Real tionary Computation, pp. 289–294
coded genetic algorithm based improvement of efficiency in [44] Deb K and Goel T 2001 Controlled elitist non-dominated
interleaved boost converter. Int. J. Power Electron. Drive sorting genetic algorithms for better convergence. In:
Syst. 5(4): 529 Interernational Conference on Evolutionary Multi-Criterion
[29] Gu X, Wang X, Liu Z, Zha W, Xu X and Zheng M 2020 A Optimization, pp. 67–81
multi-objective optimization model using improved NSGA- [45] Horn J, Nafpliotis N and Goldberg D E 1994 A niched Pareto
II for optimizing metal mines production process. IEEE genetic algorithm for multiobjective optimization. In: Pro-
Access 8: 28847–28858 ceedings of the first IEEE Conference on Evolutionary
[30] Versele C, Deblecker O and Lobry J 2011 Multiobjective Computation. IEEE world congress on computational intel-
optimal choice and design of isolated dc-dc power convert- ligence, pp. 82–87
ers. In: Proceedings of the 2011 14th European Conference [46] Jiang S and Yang S 2017 A strength Pareto evolutionary
on Power Electronics and Applications, pp. 1–10 algorithm based on reference direction for multiobjective
[31] Dell’Isola D, Urbain M, Weber M, Pierfederici S and and many-objective optimization. IEEE Trans. Evolut.
Meibody-Tabar F 2019 Optimal design of a DC–DC boost Comput. 21(3): 329–346
converter in load transient conditions, including control [47] Knowles J and Corne D 1999 The pareto archived evolution
strategy and stability constraint. IEEE Trans. Transp. Electr. strategy: A new baseline algorithm for pareto multiobjective
5(4): 1214–1224 optimisation. In: Proceedings of the 1999 Congress on
[32] Dell’Isola D, Urbain M, Weber M, Pierfederici S and Evolutionary Computation, pp. 98–105
Meibody-Tabar F 2019 Optimal design of a DC–DC boost [48] Kuriakose S and Shunmugam M S 2005 Multi-objective
converter in load transient conditions, including control optimization of wire-electro discharge machining process by
strategy and stability constraint. IEEE Trans. Transp. Electr. non-dominated sorting genetic algorithm. J. Mater. Process.
5(4): 1214–1224 Technol. 170: 133–141
[33] Bharti M. and Jindal H 2019 Modified genetic algorithm for [49] Deb K, Pratap A, Agarwal S and Meyarivan T A M T 2002 A
resource selection on internet of things. In: International fast and elitist multiobjective genetic algorithm: NSGA-II.
Conference on Futuristic Trends in Networks and Computing IEEE Trans. Evol. Comput. 6(2): 182–197
Technologies, pp. 164–176 [50] Luo B, Zheng J, Xie J and Wu J 2008 Dynamic crowding
[34] Bagis A 2007 Determination of the PID controller param- distance? A new diversity maintenance strategy for MOEAs.
eters by modified genetic algorithm for improved perfor- In: 2008 Fourth International Conference on Natural
mance. J. Inform. Sci. Eng. 23(5): 1469–1480 Computation, pp. 580–585
[35] Jeyadevi S, Baskar S, Babulal C K and Iruthayarajan M W [51] Basso C 2008 The link between the phase margin and the
2011 Solving multiobjective optimal reactive power dispatch converter transient response. ON Semiconductor
using modified NSGA-II. Int. J. Electr. Power Energy Syst. [52] Erickson R W and Maksimovic D 2007 Fundamentals of
33(2): 219–228 power electronics. Springer, Berlin
[36] Dhanalakshmi S, Kannan S, Mahadevan K and Baskar S
2011 Application of modified NSGA-II algorithm to com-
bined economic and emission dispatch problem. Int. J. Elect.
Power Energy Syst. 33(4): 992–1002