Fuzzy induced controller for optimal power quality improvement with PVA connected UPQC

1 Introduction

There are many issues in power systems which includes voltage sags and voltage swells caused by sudden load changes on the grid (Ahmed et al. 2022). Along with grid voltage fluctuations, current harmonics generation introduced when operating non-linear loads is one of the major issues (Madhaiyan and Subramaniam 2015). These harmonics generated by the non-linear load may damage other loads and even the main source as they are connected in parallel to the non-linear load. To mitigate the voltage fluctuations series compensation devices like Dynamic Voltage Restorer (DVR) or Static Synchronous Series Compensator (SSSC) and for harmonics compensation, Active power filter (APF) or Static Synchronous Compensator (STATCOM) is used (Mansor et al. 2020; Sarita et al. 2020). In order to achieve both voltage fluctuation compensation and harmonic mitigation hybrid device like UPQC need to be connected to the grid with back-to-back connected VSCs (Kabra and Donepudi 2022; Ray et al. 2022). At the common DC link of the back-to-back VSCs, a parallel capacitor is connected along with the PV source.

Along with voltage fluctuation compensation and harmonics mitigation, the PV source injects active power into the grid through series VSC and reactive power through shunt VSC (Patnaik and Panda 2017). The series VSC is controlled using a voltage reference generation controller and the shunt VSC is controlled using a current reference generation controller (Malekpour et al. 2018). These controllers take feedback from the Point of Common Coupling (PCC) voltages and source currents to generate synchronized pulses for both the VSC modules. The test system with PV source integrated UPQC compensates for the voltage fluctuations and source current harmonics. The VSC connected in the UPQC module comprises six switches connected in a 3-legged format creating three-phase AC terminals on one side and DC terminals on the other side. One VSC is connected in shunt to the test system on the load.

The load side and the other VSC is connected in series to the transmission line on the source side through three individual series transformers. The shunt VSC is operated with current as the reference component and the series VSC is operated with voltage as the reference component. The series VSC module acts as a voltage compensation device that injects or absorbs the voltage as per sag or swell created in the transmission line due to fault. The shunt module compensates for the current harmonics generated by the non-linear load mitigating the injection of these harmonics into the source side. DC-link link capacitor is the element used for support of UPQC which is an inter-link between the series and shunt VSC.

In a conventional UPQC system, the DC link is only connected with a capacitor which has restrictions to compensate for certain voltage levels and reactive power. The rating of the capacitor decides the compensation capability of UPQC to the faults in the grid (Rauf and Khadkikar 2015; Samal et al. 2020). The module has no capability to inject active power as there will be no source that generates active power. The controllers of conventional UPQC are very simple with feedback taken from source voltages, current, and DC link voltage. Simple voltage control of the series module and current control of the shunt module is done with a reduction of harmonics and voltage fluctuations (Venkatachalam and Saravanan 2020). However, in modern technology, the DC link capacitor is updated with renewable sources like PVA for the injection of active power into the grid along with reactive power from the shunt module (Taherian et al. 2020). Along with the PV source, any other DC source like a battery or fuel cell can also be connected for more active power injection capability (Yahiya and Uzair 2016).

For a better response to the controllers, the DC voltage regulator is updated with the MPPT algorithm for maximum power extraction the from PV source (Meng et al. 2022). Along with this modification, the PI controller can be replaced by different complex controllers like sliding mode or ANFIS controllers (Chennai and Benchouia 2014). This improves the response of the control module making the UPQC operate at improved performance. The system will be more stable to the faults caused in the grid and the voltage will be stabilized at a faster response rate.

In the Figure 1. Three phase source with phase voltages V _Mabc is connected to three phase non-linear load which is RL branch connected to diode bridge rectifier. This non-linear load introduced injects harmonics in the grid which is one of the power quality issues which needs to be mitigated. The voltage fluctuations are introduced by the supply source causing sags and swells in the voltage magnitude (Malekpour et al. 2018). The PV source integrated UPQC is connected at the inter section of source and load, so as to mitigate harmonics and voltage fluctuations introduced in the grid (Yao et al. 2016). The series VSC is connected in series to the line through three individual transformers and the shunt VSC is connected directly to the grid in parallel (Samanta and Chanda 2021; Yao et al. 2016). Both modules are connected with series inductors for current harmonics mitigation generated by the VSCs. These VSCs are controlled with feedback loop control schematics integrated with PI and Hybrid Fuzzy-PI controllers.

Figure 1:

PV source integrated UPQC test system.

This paper introduces a PV source with boost converter controlled by modified P&O MPPT connected to UPQC in Section 1. The configuration of the PV source UPQC module in Section 2, the control system modification with hybrid Fuzzy-PI controller is discussed in Section 3, and the comparative simulation analysis results are presented in Section 4. The paper’s conclusion is given in Section 5, which defines the system improvements and results in discussion, followed by references used in this paper.

2 PV source UPQC configuration

The PV is a solar irradiation dependent source which is unstable as per the variable irradiation in a day. The voltage output of the PV module may impact the performance of UPQC and induce voltage instability into the grid. Hence the PV source voltage need to be stabilized even during variable irradiation conditions using a boost converter. The boost converter is controlled by modified P&O MPPT control which is included with a DC voltage reference. The circuit structure of PV module connected at DC link of UPQC can be seen in Figure 2.

Figure 2:

PV module internal circuit.

The modified P&O MPPT takes feedback of V _err along with V _pv and I _pv for DC link voltage stabilization. The duty cycle (D) of MOSFET in the boost converter is controlled with respect to the considered parameters of the PV and reference voltage (V _ref). The internal structure of modified P&O MPPT algorithm can be seen in Figure 3.

Figure 3:

Modified P&O MPPT algorithm.

The new updated D is varied as per the V _err generated by the comparison of measured DC link voltage V _dc and reference voltage V _ref. The change in D is given as per the below given conditions.

Here, P _pv(t) and V _pv(t) are present values of PV power and voltage, P _pv(t − 1) and V _pv(t − 1) are previous values of PV power and voltage. As per the given conditions in the algorithm, the previous duty cycle D(t − 1) is either incremented or decremented with ∆D. The ∆D value can be updated as per the response of the controller and tuned as per the DC voltage stabilization. For the PV module to operate at specified DC link voltage for the UPQC the reference values calculated as given below.

Here, V _L is the line voltage of the three-phase supply source and ‘m’ is the modulation index considered as 1.

For DC link capacitance C _dc value the expression is given as

Here, K is energy variation dynamic factor taken as 0.1, A is overloading factor taken as 1.5, V _p is phase voltage of the source, I _sh is the phase shunt compensation current of shunt VSC, ‘t’ required steady state time (Meng et al. 2022), V _dc is DC bus voltage, V _dc1 possible lower DC voltage.

The shunt ripple filter inductor L _f expression is given as

Here, f _sh is shunt converter switching frequency, ∆I is the ripple current of shunt connected inductor taken as 20% of grid current.

The series ripple filter inductor in (Simões et al. 2016), L _se expression is given as

Here, K _se is the turn’s ratio of the series transformer taken as 3, and ∆I _se is the ripple current of series connected inductor taken as 20% (Malekpour et al. 2018) grid current, f _se is the series VSC switching frequency.

The configuration of the system is done as per the values given by the above expressions and the modeling is done with control modules modeled as in Section 3.

3 Controller modeling

The control structures for operating VSC modules of the PV source UPQC are slightly modified from the conventional UPQC control structures (Jayakumar and Alexander Stonier 2020; Meng et al. 2022). Both the controllers take feedback from grid system voltage and current measurement so as to make the VSC modules to operate in synchronization to the grid (Lu et al. 2016). The series VSC control structure only consider voltage measurements as it is a voltage compensation module (Malekpour et al. 2018). However, the shunt VSC control structure includes voltage measurement and also current measurement as the shunt module needs to compensate current harmonics (Saggu et al. 2018). The shunt VSC also need to be operated in synchronization to the grid voltage (Devassy and Singh 2018).

3.1 Series VSC control structure

The series VSC control is an in phase compensation and pre sag compensation strategy. The series VSC needs to inject or absorb voltage from the grid as per the sag and swell created respectively (Patjoshi et al. 2017). So for this the controller has to be operated in synchronization to the grid (Prajapati and Prasad 2020). Therefore, the controller needs feedback from the source voltage with PLL (Yao et al. 2016) (Phase Locked Loop) module used to determine the phase and frequency of the main source supply (Devassy and Singh 2018; Poongothai and Srinath 2020). The complete schematic of series VSC controller is shown in Figure 4.

Figure 4:

Series VSC control structure.

As seen in Figure 2 the controller takes feedback from source phase voltages V _Sabc and load phase voltages V _Labc . The source voltages are considered for PLL so as to determine the phase angle θ at which the grid voltages are operating (Abdoli et al. 2018). The PLL module is used to calculate the stationary reference frame d–q–0 components of both source voltages and load voltages using Park’s transformation equation given as

(7) [ V d V q V 0 ] = [ sin   θ − cos   θ 0 cos   θ sin   θ 0 0 0 1 ] [ V a V b V c ]

From the above expression the dq components of the source voltages V _Sd V _Sq and load voltages V _Ld V _Lq are considered which are compared to each other (Da Silva et al. 2020). The comparison expressions are given as

(8) V E d = V L d − V S d

(9) V E q = V L q − V S q

Here, V _Ed V _Eq are the error d–q components generated after comparison. Now the reference error component is generated by below expressions.

(10) V E d * = V L d * − V S d

(11) V E q * = V L q * − V S q

Here, V L d * is considered as 1 pu and V L q * is considered as 0 because no reactive component is required for voltage injection or absorption.

Now for the final reference voltage components the expression is given as

(12) V d * = V E d * − V E d

(13) V q * = V E q * − V E q

These final reference values generated by the comparison of error components are fed to individual PI controller to generated reference signals (Boukhechem et al. 2020).

(14) V S E d * = V d * ( K p + ∫ K i ∙ d t )

(15) V S E q * = V q * ( K p + ∫ K i ∙ d t )

Form these final reference signals V S E d * V S E q * the three phase voltage signals components are generated which are fed to SPWM (Sinusoidal pulse width modulation) techniquefor generation of pulses to the six switch series VSC module. The inverse Park’s transformation expression is given as

(16) [ V a V b V c ] = [ sin   θ cos   θ 1 sin ( θ − 2 π 3 ) cos ( θ − 2 π 3 ) 1 sin ( θ + 2 π 3 ) cos ( θ + 2 π 3 ) 1 ] [ V d V q V 0 ]

As per the reference sinusoidal signals the pulses to the inverter are generated by SPWM technique controlling the inverter output voltage (Patjoshi et al. 2017). With respect to this the voltage at the load side is compensated and tends to maintain at given reference values are per the user (Zahariah and Pon Symon 2021).

3.2 Shunt VSC control structure

To control the shunt VSC the shunt controller takes feedback from source voltages V _Sabc , Load currents I _Labc , PV source voltage and current V _pv I _pv and DC link voltage V _dc (Dash and Ray 2018). The load current d–q axis components I _Ld I _Lq are generated by using the expression as in (7). For loss current component I _loss generation a PI controller fed DC voltage comparison is considered where the reference DC voltage is generated using MPPT algorithm (Sangwongwanich and Blaabjerg 2019).

A P&O MPPT algorithm is adopted in the controller which takes feedback of PV source voltage and current ( V p v   I p v ) (Dai et al. 2021). The complete control structure of shunt VSC module is shown in Figure 5. The loss current component expression is given as

(17) I l o s s = V d c * − V d c ( K p + ∫ K i ∙ d t )

Figure 5:

Shunt VSC control.

Here, the reference DC voltage V d c * is generated from P&O MPPT algorithm which is also known as voltage at maximum power point V _mpp. Along with the loss component, in the (Cheung et al. 2017), average PV power current is also calculated and is expressed as

(18) I p a v g = 2 3 ( P p v V s )

Here, P _pv is the power generated by PV source taken as P p v = V p v * I p v , and Vs is the magnitude of maximum value of the source voltage (Xu et al. 2016). The final reference direct axis component generated for shunt VSC control is expressed as

(19) I S d * = I L d f + I l o s s − I a v g

Here, I _Ldf is the filtered load current d-axis component with reduced disturbances. For conversion of d–q–0 to abc reference signals the expression (16) is considered where I q * and I 0 * components are considered as 0. Now the final reference signals I s a * I s b * I s c * are compared to measured source current I _sa I _sb I _sc with error current generation used in hysteresis current loop controller for generation of pulses for the shunt VSC (Vigneysh and Kumarappan 2017). The hysteresis loop control generates pulses as per the upper band and lower band limits (Yap et al. 2021).

3.3 Fuzzy-PI control module

In the above shunt VSC control the PI controller used for calculation of loss current component I _loss is a conventional controller (Campanhol et al. 2019). This controller has high disturbances and oscillations because of the preciseness of the controller with respect to K _p and K _i gains. This leads to disturbed current reference signal generation because of which the shunt VSC does not mitigate the harmonics optimally. To overcome this drawback the PI controller in Dimitroulis and Alamaniotis (2021), is replaced with Fuzzy-PI (Song et al. 2020; Saha and Biswas 2021). Which generates varible K _p and K _i values as per the error generated improving the performance of the controller. With reduced disturbances in the reference signal the shunt VSC mitigates the harmonics more optimally reducing further impact on the grid.

The Fuzzy-PI module is modeled with two input components error (e) and difference in error (de), one output component K _p or K _i (Campanhol et al. 2019; Vigneysh and Kumarappan 2017; Yap et al. 2021). Each input component is included with five membership functions type taken as triangular (Dimitroulis and Alamaniotis 2021; Song et al. 2020). The membership functions for the same Fuzzy-PI design is shown in Figures 6 and 7.

Figure 6:

Fuzzy-PI controller.

Figure 7:

FIS membership functions of components ‘e’ ‘ce’ and ‘K _p or K _i ’.

Here component ‘e’ is generated by comparison of reference DC voltage V d c * with measured DC voltage V _dc . The difference is error ‘de’ component generated by comparison of present value with previous value of ‘e’ (Dimitroulis and Alamaniotis 2021; Saha and Biswas 2021). The expression is given as

(20) d e = e ( k )   –   e ( k − 1 )

The membership functions are represented as Big Negative (BN), Medium Negative (MN), Zero (Z), Medium Positive (MP), Big Positive (BP) (Arcos-Aviles et al. 2018). To generate the output from the controller a 25 rule base is taken with IF-AND-IF-THEN rule structure (de Carne et al. 2019). A 5 × 5 rule base is taken for better resolution and to generate stable results with more accuracy. Lesser number of membership functions lead to reduced accuracy and response time of the controller. The rule base for the given Fuzzy module is given in Table 1.

Table 1:

Rule base for K _p or K _i .

Kp						Ki
ce	e					ce	e
	BBN	MMN	ZZ	MMP	BBP		BBN	MMN	ZZ	MMP	BBP
BN	BB	BB	SS	BB	BB	BBN	BB	MM	SS	MM	BB
MN	BB	BB	SS	BB	BB	MMN	BB	MM	MM	MM	BB
ZZ	BB	MM	MM	MM	BB	ZZ	BB	BB	MM	BB	BB
MMP	BB	BB	SS	BB	BB	MMP	BB	MM	MM	MM	BB
BBP	BB	BB	SS	SS	BB	BBP	BB	MM	SS	MM	BB

With the above updated of the current loss component a comparative analysis is carried out with conventional PI controller generation results using simulation modeling in next section.

4 Simulation results comparative analysis

With all the above modules of the test system which include three phase main source feeding non-linear load connected with PV-UPQC device at the intersection, modeling is done in MATLAB Simulink environment. The below table parameters are used for the modeling of the given system (Table 2).

The controllers are modeled as per the given structures in section III and the simulations are run for 1 s with different operating conditions. As per the given case in the total simulation time of 1.2 s voltage sags are created from 0.4 to 0.6 s and voltage swells are created from 0.8 to 1 s. However the harmonics are continuously generated as the non-linear load is connected to the source throughout the simulation time. The resultant graphs generated with respect to above conditions are shown Figure 8.

Figure 8:

Source and load voltages without PV-UPQC.

In the above Figure 8. Graph the voltage sag and swell created by the source are impacting the load side voltages also creating sag and swell in the load also as there is no PV-UPQC device connected. For same the three phase source current graph with huge harmonics generation due to non-linear load power feeding is shown below in Figure 9.

Figure 9:

Source currents without PV-UPQC.

The THD of the above source current graph is shown in Figure 10. With very high value of 28% which is very much greater than the IEEE standard of 5%.

Figure 10:

THD of source current without PV-UPQC.

A comparative graph of source voltage and load voltage magnitude is shown in Figure 11. Without any changes representing sag and swell impact on the load.

Figure 11:

Source and load voltage magnitude comparison without PV-UPQC.

Now, the test system is updated with PV-UPQC at the intersection with and the simulation is run for same sag and swell timing with same simulation time. The PV module is operated with variable irradiation conditions which changes with respect to time. The irradiation, voltage, current and power of PV source can be seen in Figure 12.

Figure 12:

Solar irradiation, voltage, current and power of PV source.

Even with the given variable solar irradiation the voltage output of the boost converter controlled by modified P&O MPPT technique is maintained at 700 V. The output voltage of the converter can be observed in Figure 13.

Figure 13:

Voltage at the DC link or output of boost converter.

Due to stable voltage magnitude generation at the DC link, load voltage and current harmonics compensation is achieved by UPQC. The below Figure 14 is the source and load 3-ph voltage comparison.

Figure 14:

Source and load voltages with PV-UPQC.

As it can be observed that the load voltages are intact and are maintained at 1 pu even with sag and swell created on the source side. This is achieved by the injection and absorption of voltages by the series VSC connected through series transformers on the source side. The compensation voltages from the series VSC as per the sag and swell conditions are shown in Figure 15.

Figure 15:

Series VSC compensation voltages.

For harmonics generated by the non-linear load after the PV-UPQC shunt VSC connected at the load side the source currents recorded are shown in Figure 16.

Figure 16:

Source currents with PV-UPQC.

As observed the harmonics are mitigated generating nearly sinusoidal waveforms from the source current as the shunt VSC is compensating the non-linear load current harmonics. The compensation shunt VSC current can be seen in Figure 17.

Figure 17:

Shunt VSC compensation currents.

The voltage magnitude comparison of source and load voltages can be seen in Figure 18. With load voltage magnitude maintained between 0.95 and 1 pu even during sag and swell conditions created by the source.

Figure 18:

Source and load voltage magnitude comparison with PV-UPQC.

The shunt VSC controller is updated with Fuzzy-PI controller replacing PI controller for further reduction in harmonics and stabilizing the system with better load voltage magnitude. Figure 19 is the rule viewer of the Fuzzy-PI controller K _p and K _i gains for the given error and difference in error values.

Figure 19:

FIS rule viewer of K _p and K _i gains.

The voltage magnitude of the load voltage is improved with reduced disturbances when the PV-UPQC is operated with Fuzzy-PI controller (Yu and Khan 2022). Therefore, the PV-UPQC has better performance when operated with Fuzzy-PI controller as illustrated Figure 20.

Figure 20:

Load voltage magnitude comparison with PI and Fuzzy-PI controllers.

Along with voltage magnitude improvement, the THD of the source current is also reduced to a lower value when the PV-UPQC is operated with the Fuzzy-PI controller. The FFT analysis comparison of source current with PI and Fuzzy-PI controller determining the THD is shown in Figure 21.

Figure 21:

THD comparison of source currents using FFT analysis tool.

A parametric comparative analysis table without and with PV-UPQC integrated with PI and Fuzzy-PI controllers is given in Table 3.

5 Conclusions

In this paper the complete modeling of the test system with PV source integrated UPQC is modeled impacting the system when introduced with different power quality issues. The stabilization of voltage magnitudes is shown when the supply source is introducing sag and swells condition in the system along with improvement in THD of the source current. A comparative analysis is discussed with simulation results shown when the system is operated without and with PV-UPQC device connected at the intersection. Further improvement in load voltage magnitude is achieved and also further mitigation of THD of source current is also shown when the PV-UPQC is operated with Fuzzy-PI controller. The voltage magnitude is stabilized at 1 pu and the THD of the source current is maintained as low as 1.16% with the updated control structure.