See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/379597452

Harmonic Control of voltage-source inverters using Particle Swarm

Optimization (PSO)-tuned Proportional Resonant (PR) Controllers

Conference Paper · April 2024

CITATIONS READS

0 23

3 authors, including:

Mayaka Moses
Jomo Kenyatta University of Agriculture and Technology
3 PUBLICATIONS 0 CITATIONS

SEE PROFILE

All content following this page was uploaded by Mayaka Moses on 05 April 2024.

The user has requested enhancement of the downloaded file.

 Harmonic Control of voltage-source inverters using
Particle Swarm Optimization (PSO)-tuned
Proportional Resonant (PR) Controllers
Michael J. Saulo
Mayaka N. Moses Peter K Kihato
Dept of Electrical Engineering
Dept of electrical and Electronics Dept of electrical and Electronics
Technical Umiversty of Mombasa
Engineering, JKUA Engineering, JKUAT
Mombasa, Kenya
Nairobi, Kenya Nairobi, Kenya
[email protected]
[email protected] [email protected]

Abstract – The concept of microgrids has gained popularity control to compensate current harmonics due to non-linear
in the distribution of electricity to the final consumer. loads. In the research by Dash et al. [6], active power filters
Microgrids integrating energy storage devices, combined heat were proposed to mitigate the harmonics produced at the
and power system (CHP) and renewable energy generation are PCC. However, such filters can affect the stability of the
especially becoming attractive to industrial consumers –such as
microgrids when operating on the islanded mode [4].
industrial parks – due to environmental and economic benefits.
However, the intermittence of renewable energy sources and Furthermore, the filters add additional costs to the operation
sensitivity of industrial loads makes the operation of of the microgrid without any operational benefits [7].
interconnected industrial microgrids challenging. Particularly,
power quality challenges – harmonic control and voltage Selective harmonic techniques have been proposed in
unbalance – are common in the operation of the industrial the literature – harmonic conductance-harmonic VAr droop
microgrids due to these challenges. The main objective of this and capacitive virtual admittance and impedance loops – to
research was to develop an adaptive particle swarm address the harmonic issues in inverter-based systems [4],
optimization (PSO)-tuned proportional resonant (PR)
[8]. The goal of such controllers is to provide selective
controller to improve the power quality of microgrids. The
proposed configuration was implemented using
compensation through the PR controller. In this research, a
MATLAB/SIMULINK software. The results obtained show PR controller with selective harmonic compensation is
that the PSO-tuned PR controller could perform better than a proposed to address harmonics in the microgrid
SISO-tuned PR controller with harmonic contents of 2.33% configuration. The implemented controller is also designed
and 2.38% respectively in the presence of nonlinear loads. to ensure balanced voltage in the three phases even when
single phase loads are connected. To ensure optimal
Keywords—NPC inverter, PSO, style, PR controller performance, the PR controller and its selective harmonic
compensators are tuned using the PSO technique.
I. INTRODUCTION
II. SYSTEM MODELLING
Microgrids integrating energy storage devices,
combined heat and power system (CHP) and renewable In this paper, interfacing of the renewable energy sources
energy generation are becoming attractive to industrial and microgrids is achieved through a neutral point clamped
consumers –such as industrial parks – due to environmental (NPC) inverter. The NPC configuration – as shown in figure
1 below – is associated with various benefits such as low
and economic benefits [1, 2, 3]. Through such microgrids,
switching losses, low harmonic content, and reduced
consumers can adopt cheaper alternative energy such as
electromagnetic interference (each switching device
solar energy, eliminate inconsistencies in energy supply and withstands half of the DC link voltage). Harmonics control
reduce greenhouse gas emissions through the integration of of the NPC inverter is achieved through the LCL filter.
renewable energy sources. Interfacing of renewable energy
sources is often achieved through power electronic
converters, which generate AC power from DC sources.

Regardless of the benefits of the industrial microgrids,

the presence of the main grid significantly affects harmonic
current flows in the microgrid network since the
characteristics of the stiff grid – voltage and frequency –are
not affected by harmonic disturbances [4]. There is also an
issue of harmonics during the islanded operation because of
the presence of non-linear and sensitive loads. The
electronic converters used in the interconnection of the
microgrids inject additional current harmonics due to local
Figure 1. NPC inverter configuration and its equivalent circuit
non-linear loads or voltage harmonics at the PCC. Due to
these issues, harmonics are a key concerns for industrial Using Kirchhoff’s voltage law, the current control
microgrids. model of the AC side can be derived as follows:
Traditionally, passive and active filters have been used
to mitigate harmonics. For instance, Vechiu et al. [5] applied
active power filters through PI controllers and hysteresis

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

 (II)
To control harmonics of the LCL filter, a PR
controller is used as shown in figure 3 below. The transfer
function of the PR controller as well as associated
harmonics is presented in equation (7) [9].
The model equation can then be transformed to the
(7)
dq/alpha-beta reference frame and linearized to obtain
equation (ii) when only DC terms are considered.
(2)

III. TUNING OF THE PR CONTROLLER

Equation (2) corresponds to the positive sequence
component. Negative sequence components rotate in the PSO is an optimization algorithm, which was inspired by
opposite direction of the positive sequence components. The the cooperative behavior of flocks of birds. The flock –
negative sequence components should be reduced to zero to referred to as particles – interact with each other with some
achieve voltage balance. As such, sort of updating the best performance in the current flight
(3) [11]. In common applications, the PSO algorithm is
associated with a set of parameters, which is a vector. The
vector has an equal number of elements for all particles and
The outer voltage loop control structure is obtained the size of the vector is the same as the vector to be
through the Kirchhoff’s current law and substitution of the optimized. Using arbitrary values, the performance of each
voltage through the capacitor as shown in equations (4) and particle is calculated. The PSO technique is then used to
(5) below. locate a multidimensional space using the heuristic search
(4) technique where all the particles satisfy an objective
function. In each iteration, one particle produces the best
result, which becomes the global best particle. All the other
(5) particles are updated with the vector giving the best results.
Thus, after each iteration, all particles assume the best
vectors so far and one becomes the globally best particle.
Equation (5) is used to develop the controller structure
shown in figures 2 and 3 below. The performance of the PSO algorithm depends on an
optimization function, which monitors optimization search.
The objective function is selected to ensure optimal error
tracking. In a general problem, the constrained optimization
problem minimizes a scalar function (J) of some decision
variable (D) in a universe (U). It is assumed that there is an
optimal set of parameters in U, which satisfies all constraints.
For the PR compensators, a six-dimensional search space of
the PSO algorithm is used to estimate the fundamental gains
of the controller and the gains (integral and proportional) of
each of the selected harmonics. For a given transfer function,
the gains are used to track the total harmonic distortion
Figure 2. Outer voltage Control loop of the system
(THD) of the inverter voltage output. The objective function
of the PSO algorithm is shown in equation (8) below. The
integral square error (ISE) is used in the assessment of the
error. The goal of the PSO algorithm is to minimize THD
through an online simulation.

(8)
Initially, the required parameters are defined as well as
the limits of the target parameters. The PSO algorithm is
applied as follows:
Figure 3. Inner current controller with harmonics compensators
STEP 1: Assign initial gain values to the PR controller (ki,
For the LCL filter with a damping resistor to reduce kp, ki3, ki5, ki7).
resonance and improve stability, the transfer function of the STEP 2: Run a simulation of the system to obtain the THD.
filter is shown in equation (6) [9]. The transfer function
links output current to the voltage of the NPC inverter. The STEP 3: Generate the objective function (ISE) using
parameters of the LCL filter can be obtained using the equation (8).
guidelines provided in [10]. STEP 4: Calculate personal best of each particle and the
(6) global best.
STEP 5: While iterations<=maximum iterations
 STEP A: Update inertia weight as well as velocity and compensators are as follows: kp=6.8, ki=1498.72,
position of the particles. ki3=211.2, ki5=83.867, and ki7=40.834. The bode diagram
STEP B: Run a simulation of the system using updated of the final controller is shown in figure 5 below. In this
values. case, the gain margin is infinite while the phase margin of
1700 is obtained at a frequency of 6280 rad/s, which shows
STEP C: Update personal best, global best, and ISE. that the system is stable. The step plot of the system also
END shows a rise time is 7.3367e-05 s, a settling time of 3.2475e-
04, and an overshoot of 2.6097.
STEP 6: Maximum iterations reached?
YES: Stop the algorithm and display particles with
minimum ISE.
NO: Return to STEP A.

IV. SIMULATION AND RESULTS

To verify the performance of the PSO-based PR Figure 5. Closed-loop bode plot of SISO-tuned PR controller and
controller, the NPC inverter was designed as described in the step response of the controller
[12]. The overall model and associated control structures are
shown in figure 3 and 4 below. The DC-link voltage was set In this case, the performance of PSO and SISO
to 393 V with a target rms Voltage of 220 V. The gains of approaches are comparable. It is however, important to note
the PI controllers were tuned through the Nickola-Ziegler that the SISO approach requires the tuning of individual
method. Harmonics were then injected into the system harmonic orders before they are combined with the resonant
through the addition of non-linear loads. For the PR current controller. Such an approach is time consuming considering
controller, a resonant frequency of 311 rad/s and a cut-off the availability of advanced tuning methods. For the PSO
frequency of 0.1 rad/s were used. The cut off frequencies for method, all gain parameters are tuned simultaneously, which
the 3rd, 5th, and 7th harmonic controllers were 2.5 rad/s, 4.5 provides an optimal solution within a short time.
rad/s, and 9 rad/s as proposed in [9]. An online simulation of
A. Case A
the proposed model was done to obtain the gain parameters.
The parameters of the PSO algorithm are as follows: In the first scenario, a simulation was carried out using
 Population size =100 only PI controllers to verify the presence of harmonics in the
 Maximum inertia weight =0.9 model as shown in figure 6 below. From the results, the
THD of 23.15% is very high in relation to the IEEE
 Minimum inertia weight =0.4
standards. The fifth and seventh order harmonics also
 Acceleration factor =2
violate the specified limits. As such, an effective controller
 Acceleration factor =2 is essential to minimize THD as well as the individual
 Maximum number of iteration =30 harmonics.
 Maximum number of runs=10
After the simulation, the values obtained for the PR
gains as well as compensators are as follows: kp=9.1243,
ki=2788.9, ki3=258.1245, ki5=174.71, and ki7=15.5390.
The bode diagram of the final controller is shown in figure 4
below. In this case, the gain margin is infinite while the
phase margin is 1700 at a frequency of 7450 rad/s, which
shows that the system is stable. The step plot of the system
also shows a rise time is 5.4286e-05 s, a settling time of
1.8381e-04, and an overshoot of 2.8115.
Figure 6. Performance of conventional PI controller

B. Case B
In this scenario, the performance of both the SISO-tuned
and PSO-tuned PR controllers were evaluated in the absence
of the nonlinear loads. The output voltage waves of the
controllers as well as their harmonic spectrums are
presented in figures 7 and 8 below. From the figures, it can
Figure 4. Closed-loop bode plot of PSO-tuned PR controller and be seen that both controllers produce smooth AC voltage
the step response of the controller signals with minimal harmonic content. The SISO-tuned
controller also results in less THD as compared to the PSO-
Similarly, the PR gains were tuned through the tuned controller.
MATLAB SISO tool to obtain comparative results. The
gains of the SISO tuned PR controller with harmonic
 Figure 7. Performance of the PSO-tuned PR controller Figure 10. Performance of the SISO-tuned controller in the
presence of nonlinear load

V. DISCUSSION AND CONCLUSIONS

From the results obtained, it is evident that the proposed
PSO-tuned PR controller can fulfill important power quality
objectives. First, the PSO algorithm can improve the
performance of the PR controller so that the overshoot is
below 5% and settling/rise time are below 0.001 seconds. In
this case, the voltage and current waveforms reach the
steady state condition within a short time (less than 0.1
Figure 8. Performance of the SISO-tuned PR controller seconds). Second, the sequence-component control scheme
C. Case C balances the voltages of the distribution line. Particularly,
the voltage unbalance is below 3% (less than 0.1%) as
In this case, single-phase non-linear loads were added to specified in the American National Standard (ANS) for
each of the three phases. The results obtained for the PSO- Electric Power Systems and Equipment ANSI C84.1 [44]. In
tuned PR controller show that the THD is 2.33% (third, relation to harmonics, the PSO-tuned PR compensator
fifth, and seventh-order harmonics are 0.25%, 0.1%, and provide for efficient harmonics compensation. The IEEE
0.2% respectively) as compared to the SISO-tuned PR 519 standard specifies that total harmonic distortion (THD)
compensator, which has a THD of 2.38% (third, fifth, and in systems below 69kV should be limited to 5% with each
seventh-order harmonics are 0.25%, 0.15%, and 0.15% individual harmonic limited to 3% [45]. In this research, the
respectively). Furthermore, observation of the voltage THD of the PSO-tuned PR controller is 2.33% while the
signals show that the PSO-tuned controller produce a third, fifth, and seventh-order harmonics are 0.25%, 0.1%,
consistent wave as compared to the SISO-tuned PR and 0.2% respectively. Since the PSO-tuned controller
controller, which produces significant distortions after the performs better than the SISO-tuned controller in the
31st second. These results show that the PSO-tuned presence of nonlinear loads, it can be concluded that
controller has good stability as compared to the other artificial intelligence optimization algorithms provide a
controller. The PSO-tuned controller also ensures that the viable alternative to conventional PR tuning methods.
voltage harmonics for each harmonic remain within the
desired limits. ACKNOWLEDGMENT

REFERENCES
[1] [1] Shi, Junyi, Pengcheng Cui, Fushuan Wen, Linrun Guo, and
Yusheng Xue. "Economic operation of industrial microgrids with
multiple kinds of flexible loads." In 2017 IEEE Innovative Smart Grid
Technologies-Asia (ISGT-Asia), pp. 1-6. IEEE, 2017.
Figure 9. Performance of the PSO-tuned controller in the presence [2] [2] Kennedy, J. Patrick, Chuck Wells, and L. L. C. OSIsoft.
of nonlinear load "Construction of a microgrid for industrial parks." San Leandro:
OSIsoft LLC (2009).
[3] [3] Naderi, M., S. Bahramara, Y. Khayat, and Hassan Bevrani.
"Optimal planning in a developing industrial microgrid with sensitive
loads." Energy Reports 3 (2017): 124-134.
[4] [4] Micallef A, Apap M, Spiteri-Staines C, Guerrero JM. Mitigation
of harmonics in grid-connected and islanded microgrids via virtual
admittances and impedances. IEEE Transactions on Smart Grid. 2015
Nov 13; 8(2):651-61.
[5] [5] Vechiu, Ionel, Gelu Gurguiatu, and Emil Rosu. "Advanced active
power conditioner to improve power quality in microgrids." In 2010
Conference Proceedings IPEC, pp. 728-733. IEEE, 2010.
[6] [6] Dash R, Paikray P, Swain SC. Active power filter for harmonic
mitigation in a distributed power generation system. In2017
 Innovations in Power and Advanced Computing Technologies (i- [11] [11] Song, Z., Liu, B. and Cheng, H., 2019. Adaptive particle swarm
PACT) 2017 Apr 21 (pp. 1-6). IEEE. optimization with population diversity control and its application in
[7] [7] Guerrero, J.M., Loh, P.C., Lee, T.L. and Chandorkar, M., 2012. tandem blade optimization. Proceedings of the Institution of
Advanced control architectures for intelligent microgrids—Part II: Mechanical Engineers, Part C: Journal of Mechanical Engineering
Power quality, energy storage, and AC/DC microgrids. IEEE Science, 233(6), pp.1859-1875.
Transactions on industrial electronics, 60(4), pp.1263-1270. [12] [12] Zhao, Tiefu. Design and control of a cascaded H-bridge
[8] [8] Savaghebi, M., Jalilian, A., Vasquez, J.C. and Guerrero, J.M., converter based solid state transformer (SST). North Carolina State
2012. Secondary control for voltage quality enhancement in University, 2010.
microgrids. IEEE Transactions on Smart Grid, 3(4), pp.1893-1902. [13] [13] Arghavani, H. and Peyravi, M., 2017. Unbalanced current-based
[9] [9] Zammit, Daniel, Cyril Spiteri Staines, Maurice Apap, and John tariff. CIRED-Open Access Proceedings Journal, 2017(1), pp.883-
Licari. "Design of PR current control with selective harmonic 887.
compensators using Matlab." Journal of Electrical Systems and [14] [14] Blooming, T.M. and Carnovale, D.J., 2006, June. Application of
Information Technology 4, no. 3 (2017): 347-358. IEEE Std 519-1992 harmonic limits. In Conference Record of 2006
[10] [10] Reznik, A., M. Godoy Simões, Ahmed Al-Durra, and S. M. Annual Pulp and Paper Industry Technical Conference (pp. 1-9).
Muyeen. "LCL filter design and performance analysis for small wind IEEE.
turbine systems." In 2012 IEEE Power Electronics and Machines in
Wind Applications, pp. 1-7. IEEE, 2012.

View publication stats