# Cooperative Performance of ANFIS Controllers in Q-ZSI Topology for Deployment of Robust MPPT and Voltage Regulation in Grid-Tied Solar PV System

^{*}

**Corresponding Author:**Bilal Abdul Basit, Department of Electrical Engineering, University of Engineering and Technology, Lahore, Pakistan, Tel: +923335472177, Email: [email protected]

*
Received Date: Aug 19, 2018 /
Accepted Date: Aug 22, 2018 /
Published Date: Aug 29, 2018 *

### Abstract

In this paper, MPPT and voltage regulation related concerns occurring in low voltage distribution networks assimilated with solar photo-voltaic (PV) systems have been addressed and possible solutions have been established. This research suggests and gauges a combined cooperative performance of novel adaptive neurofuzzy inference system based energy management control (ANFIS-EMC) in quasi Z-source inverter (q-ZSI) topology and proposed ANFIS-based control for shoot-through intervals adjustment in the inverter PWM switching patterns. In the conventional q-ZSI topologies, Battery Storage Unit (BSU) has been controlled by a constant charge/discharge or state-based control strategy with an independent control of shoot-through periods for boost phenomenon. This synergy causes an improper usage of BSU which reduces battery life, increases expense and originates power quality concerns like voltage unbalance issues in grid-tied solar PV systems. Whereas an imprecise adjustment of shoot-through intervals causes loss of MPPT especially at low levels of solar irradiation and also increases switching losses in the inverter passive devices. The proposed ANFIS-based cooperative control scheme adjusts shootthrough intervals precisely and in constraint with an optimal charge/discharge of BSU to overcome the aforementioned limitations and to attain an efficient power regulation between dynamic load demand and solar PV array input. Moreover, this synergy makes our system capable to work under critical power system contingencies like sudden starting of large inductive loads and 3-phase faults. The trial model of the proposed control structure has been developed in MATLAB/Simulink environment. The operational behavior of the conventional control strategies has been compared with the proposed approach mainly focusing on the voltage regulation issues and robustness of MPPT control in grid-tied solar PV systems.

**Keywords:**
Cooperative performances of ANFIS controllers; Battery storage unit in grid-tied solar PV; ANFIS-based shoot-through intervals adjustment; Energy management system

#### Abbreviations

MPPT: Maximum Power Point Tracking; PV: Photo-Voltaic; ANN: Artificial Neural Network; FL: Fuzzy Logic; FIS: Fuzzy Inference System; ANFIS: Adaptive Neuro-Fuzzy Inference System; EMC: Energy Management Control; ZSI: Z-Source Inverter; q-ZSI: Quasi ZSource Inverter; BSU: Battery Storage Unit; VSI: Voltage Source Inverter; PID: Proportional-Integral-Differential; PI: Proportional- Integral; P&O: Perturb and Observe; CCM: Continuous Conduction Mode; DCM: Discontinuous Conduction Mode; PCC: Point of Common Coupling; IGBT: Integrated Gate Bipolar Transistor; VC1,VC2: Voltages across capacitors C1 and C2; ic1, ic2: Currents through Capacitors C1 and C2; ν_{0} Peak Value of DC-link Voltage Across Inverter Switching Devices; ν_{d }Peak Voltage Across Diode Element; Vpv: Terminal Voltage Across PV Array; Vmpp: Maximum Power Point Voltage; Pmpp: Maximum Power Point Power; Voltages Across L1 and L2; VL1,VL2: Average Voltages Across Inductors L1 and L2; iL1, iL2: Currents through Inductors L1 and L2; iB: Charging/Discharging Current of BSU; id: Current required by the 3-phase loads; id_max: Maximum current required by the 3-phase loads; T: Time period of sine PWM waveform; T0: Shoot-through time interval; Pload: Total active power demand of the 3-phase loads; T1: Non-shoot-through time interval; Dsh=T0/T1: Shoot-through Duty Ratio; SOC: State of Charge of Battery Storage Unit; Pgen: Power Generated by the Solar PV Array; ηbat: BSU Efficiency; ηboost: Boost Converter Efficiency; ηbuck: Buck Converter Efficiency; D: Duty-Cycle of Half-Bridge Converter for BSU; V0: Average Value of DC-link Voltage Across Inverter Switching Devices

#### Introduction

The generation of electrical energy through commercially practicable alternative natural resources is becoming essential due to the environmental degradation arising from the use of conventional sources of electricity generation. Solar PV systems, wind farms, biomass and hydroelectric power systems are major providers of green energy and they are collectively called as renewable energy generation systems [1]. One of the fastest growing renewable energy generation methods is solar photovoltaic (PV) generation technology due to its declining average costs worldwide. From 2010 to 2015, average costs of utility-scale photovoltaic systems lowered by 66.67% and an additional declination of 25% is being expected over 2015-20 [2]. However, the intermittent nature of solar irradiation mostly raises power quality concerns like fluctuations in the generated power and voltage flickering phenomena that reduce efficiency and reliability of the system preventing extensive use of the technology [3]. Moreover, to achieve maximum power from PV array under intermittent environmental conditions, it becomes essential to apply an appropriate control algorithm, known as maximum power point tracking (MPPT). Several algorithms have been devised in the literature for MPPT in solar PV systems under varying atmospheric conditions. A comprehensive review on various different MPPT control algorithms being employed in solar PV systems has been presented in [4].

The power regulation phenomenon in distributed renewable energy generation systems largely depends on the renewable interfacing inverter [5]. Several different topologies of interfacing inverters, for solar PV systems, have been proposed in the literature, differing in their buck/boost strategy for dc side voltage regulation. In case of solar PV systems the requirement of a galvanic isolation between the solar panels and utility grid divides the PV systems in two categories: 1) transformer-based topologies and 2) transformer-less topologies. In the transformer-based schemes, voltage boost and inversion phenomena take place in a single stage and are called single-stage inverter topologies. Whereas, transformer-less topologies take two stages for voltage boost and inversion and usually a boost converter is hired in such schemes instead of a transformer. In two-stage topologies, switching devices of dc-dc converters decrease the overall efficiency and make our system costly. Similarly, addition of isolations in the transformer-based single-stage structures can decrease the overall efficiency of the system and increase expense. Consequently, single stage inverter topologies with a transformer-less structure are mostly preferred to be used in solar PV systems due to their high efficiency, low cost and a small size [6].

Among a number of transformer-less single-stage inverter topologies, quasi Z-source inverter (q-ZSI) is somewhat more sophisticated and a reliable approach to be used in solar PV systems due to its following advantages: 1) A constant current drawn from the PV array helps to reduce the number of filter devices, 2) Requires lower component ratings than conventional Z-source inverter (ZSI) topologies, 3) A storage element could be connected in the control structure with a little modification, 4) Switching ripples flowing towards the PV array reduce considerably and 5) Offers higher reliability and simpler control strategy than conventional ZSI topologies [7-10].

The q-ZSI topology works under two modes of operation: nonshoot- through mode and shoot-through mode. During maximum input conditions, like solar irradiation, q-ZSI topology works in nonshoot- through mode of operation and behaves just like a conventional voltage source inverter (VSI). On the other hand, shoot-through mode requires PWM switching patterns to be modified by inserting shootthrough states. These shoot-through insertions let the upper and lower switches of one or more than one leg turn-on simultaneously to boost the DC-link voltage across the inverter switching devices. Under intermittent nature of input conditions, shoot-through intervals’ time span is required to be adjusted accordingly to regulate the ratio of shoot-through and non-shoot-through time periods known as shootthrough duty ratio. The duty-ratio ultimately adjusts the boost factor which revises the voltage across the bridge between PV array and inverter device known as DC-link voltage.

Connection schemes regarding utilization of battery storage unit (BSU) are critical issues in PV generation systems especially whenever we use battery-based q-ZSI topology as interfacing inverter. An improper energy management control (EMC) in q-ZSIs not only reduces the life span of BSU but can also disturb MPPT algorithm being deployed. In [11-13], several battery-based q-ZSI topologies have been proposed and discussed. In [14], author suggested the integration of bidirectional dc-dc converter in conventional q-ZSI structure to effectively control the power injection/absorption from/ into the BSU and also find a way to avoid its huge ratings. A detailed discussion on the bidirectional control of BSU has been performed in [15] and a mathematical model has also been provided. Moreover, in these works, charge/discharge of the battery element has been controlled through a state-based control or constant charge/discharge rate strategy by employing classical PI controllers.

Several different q-ZSI topologies have been addressed in the literature, in recent years, due to a robust and dynamic control they offer for MPPT phenomenon. Additionally, more effective schemes for charge/discharge of storage unit are also being addressed nowadays, to achieve an efficient energy management control. Literature survey also provides some works where ANFIS control has been employed for shoot-through adjustment in the q-ZSI network topology. In [16], ANFIS-based shoot-through intervals adjustment in q-ZSI topology has been employed. However, no control has been hired for optimal charge/discharge of battery element. To track maximum power point in grid-tied solar PV system a scheme based on q-ZSI topology has been presented in [2]. A model predictive control has been employed in this work to adjust the overlap duty-cycle in PWM patterns. Moreover, the modulation index has been controlled through PQ decoupling control from grid-side. The proposed control provides better dynamic performance than the conventional MPPT controls. However, the q-ZSI topology used in this work is a battery-less structure and the simulation results do not depict the performance of the given control in case of much lower values of solar irradiance conditions.

Similarly, there are some works that describe ANFIS-assisted charge/discharge of battery element in grid-tied PV systems. For example, the work presented in [17] proposes a cooperative performance of ANFIS-based charge/discharge of battery element with a novel PV inverter control scheme. In this scheme, a dc-dc converter based two-stage inverter topology has been hired which is not a tempting interfacing inverter topology especially from power quality point of view. Moreover, in this work, MPPT is being performed through perturb and observe (P&O) algorithm which has been proved to be an outdated method for MPPT after origination of intelligent techniques.

In this work, the authors first found that such types of conventional q-ZSI topologies remain inept in handling long range and sudden variations in the atmospheric conditions like solar irradiance. Moreover, they stay incapable to perform under critical power system contingencies like sudden starting of large inductive loads or 3-phase faults and causes voltage and power unbalance. To overcome these shortcomings, our proposed control suggests an improved q-ZSI topology which is based on a cooperative working of shoot-through duty-cycle regulation and a charge/discharge rate strategy for battery element. Two ANFIS controllers have been trained and hired to adjust gains of PI controllers optimally and supportively for accomplishment of both the above mentioned purposes. These intelligently controlled conventional controllers like PI or PID enhance the performance of q- ZSI topology for maximum power point tracking (MPPT) in solar PV systems. Moreover, they make our system economically feasible by reducing expense and increase life span of the battery element through optimal charge/discharge rate under intermittent system operating conditions.

ANFIS is basically a combined performance of artificial neural networks (ANN) and fuzzy logic (FL). A control strategy based on ANN alone lacks the heuristic sense and performs as a black box while FL based computation is unable to achieve correct fuzzy rules and membership functions because the training process requires an inclusive prior knowledge of the whole system under discussion [18,19]. These limitations of ANN and FL can be mitigated by combining both algorithms together to realize a powerful intelligent control known as ANFIS [16]. To the best of our knowledge, this is the first time that this type of enriched q-ZSI topology with a cooperative performance of ANFIS controllers is being employed for grid-tied solar PV system to implement MPPT and to alleviate voltage regulation concerns especially under critical power system contingencies.

Following objectives have been achieved after employing the proposed enriched q-ZSI topology in grid-tied solar PV system:

• Capable to track point of maximum power, even at wide range of variations in input atmospheric conditions like incident solar irradiance and ambient temperature.

• Vigorous and quick performance under short term input/PV array terminal voltage fluctuations.

• Less valued and short range voltage stress across the inverter switching devices especially for an extended drop in the incident solar irradiation.

• A balanced enactment under critical power system contingencies like sudden starting of high inductive loads or in case of 3-phase faults on load side.

• Improves efficiency of storage element.

• Increases economic feasibility of the system.

The suggested control has been applied on a sample model of gridtied solar PV system in MATLAB/Simulink. The operational behavior of the proposed control has been compared with the conventional topologies of q-ZSI, mainly focusing on MPPT and voltage regulation phenomena.

#### Materials and Methods

**Conventional q-ZSI topology and control**

From the past several years quasi Z-source inverters (q-ZSI) have replaced simple Z-source inverter (ZSI) topologies due to their several attractive advantages like constant current flow from the PV panel, reduced component ratings, higher reliability and simpler control scheme [7-10].

Several works have been performed on the BSU connections with capacitor elements of the q-ZSI network. According to** Figure 1,** a battery bank has been connected with capacitor to control the power injection/absorption according to the requirements. However, the demerits of this stratagem of BSU connections are clamping of the voltage acrossand thus loss of a constant DC-link peak voltage across .

Moreover, it also makes difficult to track the point of maximum power during low incident irradiance conditions .

**Figure 2 **shows second stratagem of storage connections, where BSU has been connected across capacitor. This reduces effects of varying BSU terminal voltage on MPPT and also helps to withstand a constant DC-link voltage across inverter device

Both topologies operate in continuous conduction mode (CCM) during non-shoot through states of q-ZSI with a BSU in charging mode. However, during shoot-through states, the q-ZSI topology, proposed in scheme 1, starts working in discontinuous conduction mode (DCM).

Thus it offers a limited battery discharging ability, which remarkably impacts the inverter output power [20]. Due to the stated demerits of topology 1, depending on BSU connections, the topology presented in scheme 2 has been employed in this research.

In non-shoot-through condition depicted in **Figure 3**, the conventional q-ZSI structure works in its normal mode.

The mathematical equations of voltages across capacitors are given below,

(1)

Where V_{pv} is the terminal voltage of PV array, V_{c1} and V_{c2} is the voltages across capacitors C_{1} and C_{2}, v_{L1} and v_{L2} is the voltages across inductors L_{1} and L_{2}.

Also the mathematical equations of currents through capacitors during normal mode are given below,

(2)

Where i_{c1} and i_{c2} is the currents through capacitors C_{1} and C_{2}, i_{L1} and i_{L2} is the currents through inductors L_{1} and L_{2}, i_{B} is the charged in/discharged out current of battery, i_{d} is the current required by load.

The values of peak voltage across diode and peak voltage across dclink are given below.

(3)

From (2)

(4)

During non-shoot-through condition remains equal to Thus from equation 4, one can (5)

The maximum required current by the connected 3-phase load at

the ac side of dc-ac inverter can be represented from equation (2).

(6)

Therefore, as long as the output current leaving the inductorremains equal to , BSU connections remain isolated from the impedance network and remain zero. As soon asbecomes negative, and BSU starts recharging from the utility grid side. From boost mode of conventional q-ZSI depicted in **Figure 4**, we can say

(7)

(8)

(9)

(10)

In boost mode of q-ZSI, value of increases because of an increase occurring in the voltage across capacitor with each boost applied on DC-link voltage. Also voltage across capacitor

is kept constant by the current controller. From equation (10), by increasing and keeping *V _{C1}* constant causes a discharging current from BSU. During steady state operation average value of

*V*

_{L1}and *V _{L2}* remains zero over one switching period.

Using (1) and (7) one can get

( 1 1 )

(12)

Where T is the time period of sine PWM waveform, T_{0} is the shoot through time interval, T_{1} is the T-Shoot through time interval.

It can be seen from equation (13) that V_{C1}and V_{C2} depend on T_{0} and T_{1}respectively

(13)

Hence

(14)

(15)

Based on (12) to (15)

(16)

The average value of peak dc-link voltage is given below

(17)

From (12), (13), (16) and (17)

(18)

Where

**Proposed control system**

Grid-tied solar PV system having a PV array of 5KW has been hired in this research with a 3-phase load connected at the point of common coupling (PCC) between PV array and utility grid. BP SOLAR SX3190 has been selected as a solar PV generator in the proposed Simulink model. BP SOLAR is regarded as one of the very efficient solar PV modules usually manufactured using silicon nitride multi-crystalline silicon cells. The varying parameters like solar irradiance (W/m2) and ambient temperature (ºC) of the selected area have been taken as inputs to the PV array. Power system model parameters employed in the Simulink model have been summarized in** Table 1.**

**Figure 5** depicts complete connections of BSU in the q-ZSI network by employing half-bridge converter across.ANFIS controller adjusts gains of PI controllers which in turn tunes duty cycles of a half-bridge bidirectional converter. economically infeasible huge ratings of the battery element, to recharge it from the utility grid side and inserts more flexibility in its charge/discharge operation. When it is required to discharge the battery element, pulses are applied on gate terminal of IGBT-2, to boost the battery voltage of 200V up to 780 V across . However, during recharging from utility grid, gate pulses are applied on IGBT-1 to convert 780V across to 200V across the terminals of BSU. In this type of buck/boost processes, switch gates of IGBT-1/IGBT-2 respectively are tuned optimally by ANFIS controllers to adjust the values of charging /discharging currents of BSU taking into account the real time requirements.

Delta-Star connected 50KVA transformer has been employed to step-down 3-phase, 11KV (line-line) voltage to 220V (line-ground voltage) across the terminals of the connected loads as per residential phase voltage level being used in Pakistan.

**Recommended ANFIS controls**

Adaptive neuro-fuzzy inference system (ANFIS) is the combination of the learning capability of artificial neural network (ANN) and inference ability of fuzzy inference system (FIS) to generate optimized membership functions and rule-base for FIS design from the simple collected data

**Figure 7** shows an equivalent ANFIS architecture for a general five layered fuzzy inference system having two inputs (a, b) and one output (f). Both the circle (fixed) and square (adaptive) nodes reflect several adaptive capabilities of ANFIS. A brief introduction to node’s functionality in different layers is given below:

Layer L_{1}

For the first layer, the node outputs of the square nodes (A1, A2, B1 and B2 in **Figure 7**) are the membership functions for the fuzzification of input variables (a, b) and are represented as follows.

(19)

(20)

In this work, triangular membership functions have been used for the fuzzification process. The expressions of this type of membership functions are given below:

(21)

Where i is the 1,2,3…

Here, A_{i }and B_{i }are the linguistic labels and x_{i} and y_{i} are the tuning parameters of triangular membership functions. Tuning parameters are tuned adaptively in accordance with the variation in inputs of the adaptive controller.

**Layer L _{2}**

Each circle (fixed) node of L_{2} identify the corresponding rule’s firing strength and behave just like a generalized AND operator. All the incoming signals are multiplied in L_{2} and the product is forwarded to L_{3} as w_{n.}

(22)

**Layer L _{3}**

Each circle (fixed) node of this layer calculates the corresponding rule’s normalized firing strength

(23)

**Layer L _{4}**

This layer comprises of square (adaptive) nodes with the following node functions.

(24)

Here, is the set of consequent parameters andn=1,2,3…

**Layer L _{5}**

The single fixed node in this layer performs the overall summation of all the incoming signals with the following node function.

(25)

**ANFIS-based energy management controller:** A highly intermittent nature of solar energy due to the intermittent weather conditions (like irradiance and temperature) requires some storage element to store energy during its plentiful amount in the system. Similarly, a storage element is required during energy deficiency from the PV array for smooth fulfillment of the energy demand. Quasi Z-source inverter (q- ZSI) topology has this attractive capability of having storage element connected in parallel with one of its capacitors. An efficient charge/ discharge of battery element is always necessary in a power system especially in a structure like q-ZSI. This is because an improper control of charge/discharge increases overall economical expenses and reduces the life span of battery element. Moreover, it causes MPPT get disturbed to a large extent even for minute differences of charge/ discharge currents from requirements.

In this work, an intelligent energy management control (EMC), for battery storage unit (BSU), has been proposed that is based on an adaptive neuro-fuzzy inference system (ANFIS) control. This type of intelligent control optimally charges/discharges the battery element according to the power requirements in the power system under discussion. According to the **Figure 8**, the proposed ANFIS controller takes three inputs: 1) Amount of PV generated power (P_{gen}), 2) Power demand of the end user (P_{load}) and 3) state of charge (SOC-%) of BSU and produces reference signals to adjust gate-pulses/duty-cycles of the dc-dc half-bridge buck-boost converter. The proposed ANFIS-based supervisory EMC also prevents the battery element to be charged more than (80% of rated storage capacity in this study) and to be discharged less than (20% of rated storage capacity in this study). Keeping the charge level within an appropriate permissible range prevents battery element from over-charging/over-discharging phenomena which in turn increases life span of battery element [21,22].

Earlier works, like [23-25], present classic state-based control schemes based on Proportional-Integral (PI) controllers for charging/ discharging of storage element (BSU). In such type of controls, BSU charges and discharges more than the actual requirements which not only increases expense but also affects the performance of q-ZSI network topology in achieving maximum power point voltage (V_{mpp}) and power (P_{mpp}). Moreover the classical state-based or constant charge/discharge rate strategy may leave the storage volume unused [26]. **Figure 9** depicts control algorithm of a classic state-based EMC based on traditional PI controllers.

**ANFIS-based shoot-through intervals controller:** To extract maximum available power from the solar PV system, it is required to operate it at a specified maximum voltage [27]. In case of intermittent nature of weather conditions the terminal voltage of the solar array does not remain at a fixed value which in-turn disturbs voltage across the inverter switching devices known as DC-link voltage (V_{0}). Therefore, we need to boost the value of (V_{0}) with falling terminal voltage of solar PV array to make our system able to work at the point of maximum power. As explained in section 2, the DC-link voltage (V_{0}) could be boosted to the desired level by regulating the periods of added shoot-through intervals in the sine-PWM switching patterns. In case of q-ZSI topology shoot-through intervals are added in the PWM patterns to make both the switches of one or all legs of the conventional voltage source inverter (V_{SI}) turn-on simultaneously.

**Figure 10 **demonstrates the alteration of PWM patterns by introducing shoot-through states of time period (T_{0}). Two regular constant lines are introduced in the sine-PWM producing waveforms and set to alter their positions up and down in order to regulate shootthrough intervals optimally. In this work, ANFIS control has been hired to adjust regular constants which regulate the shoot-through time intervals optimally.

**Figure 11** depicts the control structure being employed for shootthrough states tuning by taking incident solar irradiance (W/m2) and ambient temperature (°C) as inputs to the ANFIS controller.

The parameter D_{sh }is the ratio of shoot-through time period (T_{0}) and non- shoot-through time period (T_{1}). **Figure 11** also displays the control structure for the proposed cooperative performance of ANFIS controllers in q-ZSI topology

In this work, a modified simple boost-control strategy has been employed to insert shoot-through states in the sine-PWM patterns. In this type of boost control, modulation index is kept fixed and shootthrough period (T_{0}) is changed to boost the value of (V_{0}) across the switching devices of inverter device. This type of modified simple boost-control strategy has been proposed in [28]. This boost-control strategy has ability to handle a wide range of variations in the atmospheric conditions, like incident solar irradiance, with a low voltage stress on the inverter switching devices. Since modulation index is another control variable for output voltage regulation, the conventional simple boost control scheme uses an inverse relation between shoot-through duty ratio and modulation index. This type of inverse relationship causes voltage stress across the inverter device to become higher than normal after a certain input irradiance conditions after a limited range of modulation index. However, in modified simple boost control, shoot-through duty ratio has no connection with modulation index and regular constants are made free having no relation with peaks of reference signal waveforms.

Maximum power point voltage and current vary in accordance with the irradiance of sun and the temperature of PV cell. Several techniques have been used for MPPT from which some may work on the bases of variations in both irradiation and temperature, but some are more suitable if temperature is kept constant [29]. For the temperature rise of PV cell up to 55°C, an insignificant change occurs in the maximum power of BP SOLAR SX3190 [30]. According to the weather conditions of the selected area for this work the maximum temperature range even in summer is 45ºC. Due to these reasons the temperature effect comes very less. Simulations have been performed to obtain the training data sets by varying irradiation from 100 to 1000 W/m^{2} in a step of 50 W/m^{2 }and ambient temperature from 5°C to 65°C in a step of 10°C. ANFIS controller has been trained from these training data sets for 30,000 epochs. Hybrid learning method has been employed for ANFIS training which combines the least squares estimation and backpropagation methods for the optimized tuning of membership functions. The overall structure of the neuro-fuzzy depicted in **Figure 12 **is a five-layered network, with three triangular membership functions for solar irradiance parameter and three others for ambient temperature. Triangular membership functions are employed because these are easy to handle and require a comparatively less computational burden [31].

**Proposed cooperative performance of ANFIS controllers**

During the highest/peak value of incident solar insolation at maximum connected load, the generated power from solar PV array (P_{gen}) is equal to the amount of power required by the connected loads (P_{load}). Moreover, the dc voltage at the terminals of PV array is enough to maintain 780V at the DC-link Bridge between q-ZSI network and inverter switching devices. This value of DC-link voltage (V_{0}) is also a necessary voltage level at the terminals of capacitor C_{1} known as V_{C1}. The value of is required to be kept constant here in this work. Therefore, during maximum solar insolation, q-ZSI topology works in its normal mode (without shoot-through) and inverter device acts as a conventional voltage source inverter (VSI). Generated PWM patterns contain traditional eight sates of switches from which six are active and two are non-active or zero states. No additional active states called shoot-through states are present in the PWM pattern and diodeof the q-ZSI network behaves as a short circuit. Voltage across capacitorremains zero and current in both L1 and L2 remains same. The ambient temperature is kept at 25oC during the normal mode of q-ZSI.

As soon as the irradiance starts decreasing from 1000 KW/m2 with an increase in the ambient temperature, the PV array’s terminal voltage starts reducing which decreases V_{0} and finally V_{C1}. At this point, current controller acts to maintain a constant voltage across capacitorby taking inputs from grid side as shown in **Figure 6**. Shootthrough active states are inserted in the traditional sine-PWM pattern by ANFIS-based shoot-through generation controller due to which the diode element (D) starts acting as an open circuit. The maximum time for to be kept in an open circuit condition is equal to the shootthrough time period (T_{0}) which is adjusted by the ANFIS controller, taking irradiance and temperature as inputs. **Figure 6** also presents an overall structure depicting positions of the proposed ANFIS controllers for a cooperative performance.

Optimal cooperative BSU discharging is required at the same time when shoot-through states are being generated. To discharge the battery element, ANFIS-based EMC starts inserting gate pulses to IGBT-2 of half-bridge dc-dc converter, which performs as a boost converter to enhance the battery voltage of 200V to 780V across. To discharge an optimized amount of the required current, duty cycles of the boost-pulses are adjusted precisely by the ANFIS based control. Three parameters are taken as inputs to the ANFIS controller including; 1) state of charge of BSU (SOC), 2) generated power from PV array (P_{gen}) and 3) Power required by the connected loads (P_{load}).

The instant when solar insolation becomes 100 W/m^{2} or lower, system becomes simply battery based system. Solar PV array is then disconnected from the system and power is supplied entirely by the battery storage unit (BSU). PV array could be disconnected by inserting additional switches in the Simulink model. As the charge level of BSU reaches 20% of its maximum capacity, it starts recharging from the available grid utility or disconnected from the system in load shedding conditions. To recharge the battery element from utility grid side, ANFIS controller has been trained to adjust the duty-cycles of the gate pulses for IGBT-1to let the half-bridge dc-dc converter to perform as a buck converter. It will be concluded in the next section that if BSU is optimally discharged/recharged keeping a cooperative performance with an optimal adjustment of shoot-through states, the q-ZSI based system works in its economically reliable condition with higher efficiency and longer life span of BSU. Also it has been shown in the next section that this type of proposed cooperative performance helps to make our system able to bear the critical power system contingencies like sudden starting of large inductive loads or 3-phase faults.

#### Results and Discussions

Simulations have been performed to assess the proposed cooperative performance of ANFIS-based optimal charge/discharge of battery storage with ANFIS-based shoot-through states fine-tuning in sine-PWM patterns. Three general cases have been considered in which initially the solar irradiation has been dropped gradually and the discharging behavior of battery element (BSU) has been observed. Secondly results have been achieved during sudden starting of high inductive load and finally system behavior has been recorded in case of 3-phase fault condition. The working of proposed control strategy for cooperative performance of ANFIS controllers has been compared with the conventional ANFIS-based MPPT scheme in which a constant charge/discharge rate strategy was being applied for BSU. The equations used to compare the BSU efficiency for both ANFIS-based and classic state based energy management controls are given below [32]:

(26)

Where η_{bat }is the battery efficiency, η_{conv} is the DC-DC converter efficiency,

= Discharged power from battery

= Power charged into battery

The efficiency of boost and buck converters depends on the duty cycle for the PWM for switches and it is shown by equation and

Boost converter efficiency, (27)

dgfsagshgd

Where D is the duty ratio of pulses for converter's buck-boost operation, V_{out }is the Voltage across dc-link, V_{in(min)} is the Minimum voltage across C_{1}, V_{in(max)} is the Maximum voltage across C_{1}.

The variables in these equations are ηconv (dc-dc converter efficiency) amount of power discharged out from BSU), (amount of power charged into BSU),(duty ratio of pulses for dc-dc converter’s buck-boost operation), Vout (voltage across DC-link), Vin(min) (minimum voltage across capacitor C1) and Vin(max) (maximum voltage across capacitor C1). Equation 27 and 28 depict that the boost and buck converter efficiency decreases with an increase in the value of D. It is clear from equation 26 that the battery efficiency will also decrease with a decrease in the converter efficiency.

**Case1: Drop of solar irradiance**

The value of solar irradiance has been dropped in steps from 1000W/m2 to 180 W/m2. This drop represents the situation of surrounding atmosphere during day-time when solar energy drops gradually with the time. The simulation results by employing statebased EMC and ANFIS-based EMC, to control charge/discharge of BSU, have been compiled for 1 second of simulation running time. **Figure 13 **depicts the state of charge (SOC %) and the amount of current discharged out from BSU (in amperes) in case of both control strategies.

The simulation results clearly depict that the BSU with state-based EMC discharges more than the latter. As explained in previous sections the additional amount of discharged current occurs due to a greater value of duty cycle adjusted for the boost converter in case of statebased EMC. From Equation 27 we can see that the efficiency of boost converter goes on reducing as long as an increase in the value of dutycyclekeeps occurring. With a drop in the efficiency of the boost converter the BSU efficiency also goes on decreasing significantly as can be depicted from equation 26.

**Case2: Sudden start of high inductive loads**

**Figure 14 (a) and (b)** shows the Line-Line and Line-Ground voltage waveforms during conventional control strategies.

Whereas, the Line-Line and Line-Ground voltage waveforms achieved in case of proposed cooperative control scheme have been given in **Figure 14 (c) and (d)**. It can be seen from **Figure 14 (a) and (b)** that Line-Line voltage waveforms across the three phase dynamic load get unbalanced after the start of an inductive load due to the flow of high inrush currents. Similarly, from **Figure 14 (b)** it can also be noted that notches have occurred in the Line-Ground voltage waveforms during starting interval of high inductive load. However, the waveforms of **Figure 14 (c) and (d) **clearly depict that there is no disturbance occurring during the whole period of inductive load starting even no sags and swells are occurring at the point of starting.

**Figure 15** depict the state of charge of BSU and current being discharged out of it during sudden starting of inductive loads phenomenon. **Figure 15 (a)** depicts that under the conventional q-ZSI based control strategies BSU discharges more than the requirements during and after the starting period of a highly inductive load. This causes a decrease in the efficiency of BSU based on equations 26 and 27.

During sudden load starting phenomenon, conventional q-ZSI based control schemes produce an extensive voltage boost, which causes voltage stress (V_{0}) exceed its limits.** Figure 16 (a) **clearly depicts the value of (V_{0}) getting a huge value instead of an endurable range tolerable by the inverter device. This is unsafe for the inverter switching devices and causes our MPPT algorithm disturbed to an extent. On the other hand, the proposed enriched q-ZSI topology keeps V_{0 }within an acceptable range which enhances the robustness of MPPT algorithm, and also protects the inverter switching devices from huge values of voltage stress.

The outcomes of the proposed control scheme have been given in **Figure 16 (b)**. According to equation (16), we can infer that the value of voltage stress V_{0 }across the inverter switching devices depends on the shoot-through interval T_{0}.

Consequently, a huge value of V_{0} occurring in case of conventional q-ZSI topologies is due to the shoot-through control failure under sudden start of inductive loads.

This proves the fact that MPPT will be disturbed in conventional q- ZSI topologies whenever a sudden start of inductive loads occurs.

In our work, the value of terminal voltage across capacitor C_{1} (V_{C1}) is of worth importance and is required to be kept constant. This is because, any variation in V_{C1} instigates disturbance in MPPT algorithm at a specific state of power system under consideration. In addition, equation (18) verifies that a long settling time of V_{C1} causes a long range voltage stress (V_{0}) across the inverter switching devices. **Figure 17 (a)** depicts the behavior of V_{C1}under sudden starting of large inductive load phenomenon in conventional q-ZSI topology.

A huge value of V_{C1} with an extended settling time disrupts MPPT and causesto stay at a higher value for extra time. However, according to **Figure 17 (b)**, a lower value of V_{C1} having a smaller range of settling time verifies the solution of above stated demerits of conventional techniques.

This also validates the ability of the proposed enriched q-ZSI topology in sustaining MPPT under sudden starting of high rated inductive loads phenomenon.

**Case3: Behavior during 3-phase fault**

In case 3 of the work under study a 3-phase fault has been imposed on the Simulink model under discussion. Both conventional and proposed frameworks of q-ZSI have been considered separately. **Figure 18** shows the amounts of SOC and discharged current of BSU under 3- phase fault condition using both conventional and proposed control strategies.** Figure 18 (a)** depicts the performance of conventional q-ZSI holding a state-based EMC for BSU during a 3-phase fault condition. It can be seen from **Figure 18 (a) **that under the conventional q-ZSI based control strategies BSU discharges more than the requirements during the period of fault. This large amount of battery discharging causes an increase in expense and reduces the life span of BSU. Similarly, based on equations 26 and 27, this type of failure of state based EMC causes a decrease in the efficiency of battery element.

Moreover, a high amount of discharged current flow assists heat generation and increases noise disturbances in the line-ground voltage waveforms. However, the proposed intelligent EMC for BSU prevents the battery element from any discharge under fault conditions, as shown in **Figure 18 (b)**.

This in a way becomes a protection switch for battery element preventing it from any discharge under system faults and contingency situations. Similarly, it reduces expense and increases life span and efficiency of battery element by providing a robust intelligent EMC using ANFIS controllers.

Analogous to the sudden load starting phenomenon, conventional q-ZSI based control schemes produce an extensive voltage boost, under 3-phase faults or other contingency situations. From **Figure 19 (a)** it can be seen that a voltage stress (V_{0}) is getting beyond its limits at the time of fault application. This uncontrolled huge value of V_{0} generates heat and becomes harmful for the inverter switching devices and passive elements of q-SZI itself.

However, the proposed enriched q-ZSI topology keeps V0 within an acceptable range under any fault condition as can be seen from **Figure 19 (b)**. Similar to case 3, it enhances the robustness of MPPT algorithm, and also protects the inverter switching devices from huge values of voltage stress. Moreover, the proposed control has a protection semblance, which not only senses a malfunctioning condition but maintains our system in an active state during fault and protects the equipment from large amount fault currents as well.

As explained in case 3 of this section, V_{C1} is required to be kept constant in order to bring about MPPT under intermittent solar irradiance conditions. According to **Figure 20 (a),** V_{C1} gets disturbed and attains a high value during 3 phase fault condition. This uncontrolled huge value of V_{C1 }is an indication of failure of the conventional q-ZSI topology in maintaining MPPT algorithm under any power system contingency conditions. Moreover, such unrestrained values of V_{C1} could become harmful for the capacitor C_{1} and require q-ZSI network of passive elements to be disengaged from the system during fault.

This in a sense makes conventional q-ZSI topology inept to preserve MPPT algorithm during power system emergencies like 3-phase faults. However, a constant value of VC1in **Figure 20 (b**) substantiates the ability of the proposed enriched q-ZSI topology, in performing MPPT, during 3 phase fault conditions

#### Conclusions

MPPT and voltage regulation some of the most substantive concerns that need to be addressed in grid-tied solar PV systems especially under various critical conditions of power system operation like sudden starting of high inductive loads or three-phase faults. An improved q-ZSI topology, based on two-level cooperative performance of ANFIS controllers has been introduced and implemented in this paper. In this approach, ANFIS-based EMC for battery element (BSU) works in collaboration with the proposed ANFIS-based mechanism for shoot-through intervals adjustment in the inverter PWM switching pattern to enrich the performance of q-ZSI network topology. ANFISbased cooperative control scheme adjusts shoot-through intervals precisely with a cooperative optimal charge/discharge of BSU for an efficient power regulation between dynamic load demand and PV array input. The proposed implementation is able to track point of maximum power, even at a wide range of variations in the incident solar irradiance. Similarly, a vigorous and quick performance could be achieved under short term irradiance fluctuations too. Secondly, a stable performance has been realized under critical power system contingencies like sudden starting of high inductive loads and in case of 3-phase faults. Simulation results depict that no disturbance occurs during the whole period of load starting even no sags/swells could arise at the moment when starting occurs. Moreover, a compact settling time of ensures less valued and short range voltage stress across the inverter switching devices. In addition, simulation results depict that the proposed scheme controls charge/discharge of BSU accurately and optimally that increases economic viability of the whole system and ensures a longer life span of BSU with improved efficiency. To evaluate the performances of the proposed cooperative control strategy in real scenarios, practical load profile statistics, actual solar irradiance variations and ambient temperature measurements of the selected area has been used in the Simulink model of the proposed control structure based on cooperative performance of ANFIS controllers.

#### References

- Liang X (2017) Emerging power quality challenges due to integration of renewable energy sources. IEEE Trans Ind Appl 53: 855-866.
- Morcos M, Yushan L, Robert SB, Haitham AR (2017) Model predictive control for maximum power point tracking of quasi-Z- source inverter based grid-tied photovoltaic power system. IEEE 26th Int Symp Ind Electron.
- Caamaño-Martín E, Laukamp H, Jantsch M, Erge T, Thornycroft J, et al. (2008) Interaction between photovoltaic distributed generation and electricity networks. Prog Photovolt Res App l6: 629-643.
- Prakash S, Rajathy R, Kumar H (2016) A review of artificial intelligence algorithms for extracting maximum power from the PV system under partially shaded condition. J Electr Eng 16.
- Azzopardi B, Mutale J (2009) Smart integration of future grid-connected PV systems. IEEE Photovoltaic Spec Conf 002364-002369
- Patel H, Agarwal V (2009) A single-stage single-phase transformer-less doubly grounded grid-connected PV interface. IEEE Trans Energy Convers 24: 93-101.
- Anderson J, Peng F Z (2008) A class of quasi-Z-source inverters. IEEE Ind Appl Soc Annu Meet.
- Li Y, Anderson J, Peng F Z, Liu D (2009) Quasi-Z-source inverter for photovoltaic power generation systems. Appl Power Electron Conf Expo IEEE: 918-924.
- Park JH, Kim HG, Nho EC, Chun TW, Choi J (2009) Grid-connected PV system using a quasi-z-source inverter. IEEE Appl Power Electron Conf Expo APEC: 925-929.
- Park JH, Kim HG, Nho EC, Chun TW (2010) Power conditioning system for a grid connected PV power generation using a quasi-Z-source inverter. J Power Electron 10: 79-84.
- Cintron-Rivera JG, Li Y, Jiang S, Peng FZ (2011) Quasi-Z-source inverter with energy storage for photovoltaic power generation systems. IEEE Appl Power Electron Conf Expo APEC: 401-406.
- Li F, Ge B, Sun D, Bi D, Peng FZ, Haitham AR (2011) Quasi-Z-source inverter with battery based PV power generation system. Int Conf Electr Mach and Syst 1-5.
- Sun D, Ge B, Rub HA, Peng FZ, de Almeida, AT (2011) Power flow control for quasi-Z source inverter with battery based PV power generation system. IEEE Energy Convers Congr Expo ECCE: 1051-1056.
- Rasin Z, Rahman MF (2012) Design and simulation of quasi-Z source grid-connected PV inverter with battery storage. IEEE Int Conf Power Energy 303-308.
- Rasin Z, Rahman MF (2015) Control of bidirectional dc-dc converter for battery storage system in grid-connected quasi-z- source PV inverter. IEEE Conf Energy Convers CENCON: 205-210.
- Abu-Rub H, Iqbal A, Ahmed SM, Peng FZ, Li Y, et al. (2013) Quasi-z-source inverter-based photovoltaic generation system with maximum power tracking control using ANFIS. IEEE Trans Sustain Energy 4: 11-20.
- Mahmud N, Zahedi A, Mahmud A (2017) A cooperative operation of novel PV inverter control scheme and storage energy management system based on ANFIS for voltage regulation of grid-tied PV system. IEEE Trans Ind Inform 13: 2657-2668.
- Haykin S (1999) Neural networks: A comprehensive foundation. (2ndedn), Pearson Education, New York, USA.
- Chen G, Pham TT (2000) Introduction to fuzzy sets, fuzzy logic, and fuzzy control systems. (1stedn), CRC Press, Boca Raton, USA.
- Ge B, Peng FZ, Abu-Rub H, Ferreira FJTE, DeAlmeida AT (2014) Novel energy stored single-stage photovoltaic power system with constant DC-link peak voltage. IEEE Trans Sustain Energy 5: 28-36.
- Hara R, Kita H, Tanabe T, Sugihara H, Kuwayama A, et al. (2009) Testing the technologies. IEEE Power Energy Mag 7: 77-85.
- Yoshimoto K, Toshiya N, Koshimizu G, Uchida Y (2006) New control method for regulating state-of-charge of a battery in hybrid wind power/battery energy storage system. IEEE PES Power Systems Conf Expo 1244-1251.
- Rasin Z, Ahsanullah K, Rahman MFF (2013) Design and simulation of quasi-z-source grid-connected PV inverter with bidirectional power flow for battery storage management. 39th Annu Conf IEEE Ind Electron Soc 1587-1592.
- Rahman MS, Hossain MJ, Lu J (2015) Utilization of parked EV-ESS for power management in a grid-Tied hybrid AC/DC microgrid. Australas Univ Power Eng Conf.
- Rahman MS, Hossain MJ, Lu J (2016) Coordinated control of three-phase AC and DC type EV-ESSs for efficient hybrid microgrid operations. Energ Convers ManageA 122: 488-503.
- Alam MJE, Muttaqi KM, Sutanto D (2012) Distributed energy storage for mitigation of voltage-rise impact caused by rooftop solar PV. IEEE Power and Energy Soc Gen Meet.
- Roger M, Jerry V (2003) Photovoltaic systems engineering. (2ndedn), CRC Press, New York, USA.
- Husodo BY, MdAyob S, Anwari M, Taufik (2013) Simulation of modified simple boost control for z‐source inverter. Int J Autom Power Eng 2: 57-64.
- Esram T, Chapman PL (2007) Comparison of photovoltaic array maximum power point tracking techniques. IEEE Trans Energy Convers 22: 439-449.
- Sharma C, Jain A (2014) Simulink based multi variable solar panel modeling. TELKOMNIKA Indones J Electr Eng 5784-5792.
- Singh M, Chandra A (2013) Real-time implementation of ANFIS control for renewable interfacing inverter in 3P4W distribution network. IEEE Trans Ind Electron 60: 121-128.
- Castañeda M, Cano A, Jurado F, Sánchez H, Fernández LM (2013) Sizing optimization, dynamic modeling and energy management strategies of a stand-alone PV/hydrogen/battery-based hybrid system. Int J Hydrogen Energy 38: 3830-3845.

Citation: Basit BA, Kamran M (2018) Cooperative Performance of ANFIS Controllers in Q-ZSI Topology for Deployment of Robust MPPT and Voltage Regulation in Grid-Tied Solar PV System. J Fundam Renewable Energy Appl 8: 267. DOI: 10.4172/2090-4541.1000267

Copyright: © 2018 Basit BA, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.