Intelligent Photovoltaic Maximum Power Point Tracking Controller for Energy Enhancement in Renewable Energy System

Abstract

Photovoltaic (PV) system is one of the promising renewable energy technologies. Although the energy conversion efficiency of the system is still low, but it has the advantage that the operating cost is free, very low maintenance and pollution-free. Maximum power point tracking (MPPT) is a significant part of PV systems. This paper presents a novel intelligent MPPT controller for PV systems. For the MPPT algorithm, an optimized fuzzy logic controller (FLC) using the Hopfield neural network is proposed. It utilizes an automatically tuned FLC membership function instead of the trial-and-error approach. The MPPT algorithm is implemented in a new variant of coupled inductor soft switching boost converter with high voltage gain to increase the converter output from the PV panel. The applied switching technique, which includes passive and active regenerative snubber circuits, reduces the insulated gate bipolar transistor switching losses. The proposed MPPT algorithm is implemented using the dSPACE DS1104 platform software on a DS1104 board controller. The prototype MPPT controller is tested using an agilent solar array simulator together with a 3 kW real PV panel. Experimental test results show that the proposed boost converter produces higher output voltages and gives better efficiency (90%) than the conventional boost converter with an RCD snubber, which gives 81% efficiency. The prototype MPPT controller is also found to be capable of tracking power from the 3 kW PV array about 2.4 times more than that without using the MPPT controller.

1. Introduction

Energy generation based on renewable energy resources has received a great attention. Photovoltaic (PV) system is one of the promising renewable energy technologies. The photovoltaic (PV) generation system is a promising renewable, clean, and environmentally friendly source of energy [1]. Each solar cell produces only about one-half volt of electricity, and dozens of individual solar cells are interconnected in a sealed, weatherproof package called a PV module [2]. PV modules can be connected in series, parallel, or both into what is called a PV array. This PV array is connected to a power conditioner and then to a load or grid tie as a grid connected PV system.

In general, the two main problems with PV power generation systems are the low conversion efficiency and that electrical power generated by a typical PV panel varies with weather conditions. Thus, many studies on enhancing the energy-generation efficiency of PV applications have been carried out [3]. A PV generation system should operate at its maximum power point (MPP) to increase system efficiency [4]. Therefore, MPP tracking (MPPT) is very crucial for PV power generation systems to operate at the maximum point as much as possible at any time. However, the MPP also changes with the irradiation level and temperature due to the nonlinear characteristics of PV modules  [5]. To overcome this problem, many MPPT algorithms have been developed [6].

Recently, fuzzy logic has been applied in tracking the MPP of PV systems because it has the advantages of being robust, simple in design, and minimal requirement for accurate mathematical modeling [7, 8]. However, fuzzy logic methods depend on a careful selection of parameters, definition of membership functions, and fuzzy rules. Developing fuzzy logic methods also requires expert knowledge and experimentation in selecting parameters and membership functions. For this reason, adaptive fuzzy logic control has been developed by adding an inverse fuzzy knowledge base [9], parameter tuning of fuzzy logic control by optimization techniques such as genetic algorithms [10], and particle swarm optimization [11]. A number of studies on MPPT have also concentrated on the application of artificial neural networks (ANN) [12]. In most ANN-based MPPT methods, large amounts of field data considering atmospheric conditions are required to train the ANN. The main drawback of ANN-based MPPT methods is that it is system dependent and requires time-consuming implementation of PV arrays with different characteristics. Thus, MPPT algorithms are not robust against the rapidly changing temperature and irradiation, as well as partial shading. For the purpose of developing a more robust MPPT algorithm, a new type of intelligent technique based on the Hopfield neural network (HNN) is proposed and used together with a fuzzy logic-based MPPT controller in a PV system. Here, the fuzzy logic MPPT controller is integrated with the HNN to optimize the membership function of the fuzzy system.

In any PV system, the MPP of a PV module is tracked by the MPPT algorithm and a DC-DC boost converter, which is controlled by a triggering signal with a specific duty cycle to trigger the switch of the converter and locate its operating point as closely as possible to its MPP. Boost converter with MPPT plays an important role in PV power systems because it maximizes the power output from a PV system under various conditions, thereby maximizing the PV array efficiency. To improve the boost conversion efficiency, many modified boost converter topologies have been developed. Conventional boost converters operate in the hard-switching mode, thus rendering it inefficient when voltages and currents in semiconductor switching devices are changed abruptly from high values to zero and vice versa at turn-on and turn-off times. Consequently, switching losses and electromagnetic interference occur [13]. An improved boost converter topology using a coupled inductor together with a passive snubber has been developed and applied in standalone PV [14] and grid-connected PV systems [15]. Considering these facts, a new variant of a high gain, soft-switching DC-DC converter is proposed in to reduce the number of series-connected PV modules and improve the conversion efficiency.

2. New Boost Converter for MPPT Controller in a PV System

2.1. System Description

The MPPT controller together with a boost converter is connected between the PV panel and load, as shown in Figure 1. Here, an MPPT algorithm to draw current controls the boost converter or voltage at the MPP thus maximum power available from the PV is delivered to the load. The intelligent MPPT algorithm is used to enhance the MPPT controller performance.

Figure 1 

Schematic diagram of the proposed intelligent MPPT controller.

2.2. High-Performance Boost Converter

The circuit diagram of the proposed boost converter for the MPPT of a PV system is depicted in Figure 2. It comprises six parts: a source with an input filter (part 1), a primary side circuit (part 2), a secondary side circuit (part 3), a snubber circuit (parts 4a and 4b), an output filter with load (part 5), and a pulse width modulation (PWM) control mechanism (part 6). The detailed components of the converter represented by the respective symbols are described according to the various parts of the circuit.

Figure 2 

Schematic circuit of the high-performance boost converter.

The input filter and primary circuit with source power are represented as follows: DC input voltage from PV panel (𝑉_in), DC input current from PV panel (𝐼_in), capacitor of input filter (𝐶_in), coupled inductor of the primary side (𝐿₁), and an IGBT as a switching device (𝑍₁) with a diode (𝐷_𝑍1). The secondary side circuit consists of the following components: coupled inductor of the secondary side (𝐿₂), connecting capacitor (𝐶₃), rectifier diode (𝐷₄), and rectifier output diode (𝐷_𝑜). There are two parts in the snubber circuit, namely, the active regenerative snubber circuit (part 4a) and the passive regenerative snubber circuit (part 4b). The components of the active regenerative snubber circuit are as follows: IGBT as a switching device (𝑍₂) with a diode (𝐷_𝑍2), input capacitor (𝐶₁), output capacitor (𝐶₂), input diode (𝐷₁) for input capacitor (𝐶₁), input diode (𝐷₂) for switching device (𝑍₂), snubber inductor (𝐿_𝑆), and rectifier diode (𝐷₃). The passive regenerative snubber circuit has the following components: diodes (𝐷₅ and 𝐷₆) and snubber capacitor (𝐶_𝑠). The output filter and output side are represented by the following components: filter capacitor (𝐶_𝑜) with output or load voltage (𝑉_𝑜) and output or load current (𝐼_𝑜).

Based on the circuit and detailed mathematical analyses in [16], the voltage gain (𝐺_𝑖) of the high-performance boost converter is given as following equation: 𝐺_𝑖=𝑉_𝑜𝑉_in=𝑛𝑘+1+𝑑⋅𝑛⋅𝑘+11−𝑑(1−𝑑)², (1) where 𝑛 is the turn ratio of the coupled inductor that can be regarded as an ideal transformer, 𝑘 is the coupling coefficient (0≤𝑘≤1), and 𝑑 is the duty cycle of the triggering switching signal (𝑑=𝑡_on/𝑡_off).

Equation (1) reveals that the voltage gain of the proposed boost converter is much greater than the voltage gain of the conventional boost converter and higher than the previously reported coupled inductor-based boost converters [15, 17].

3. MPPT Control Algorithm Using FLC and HNN

3.1. Fuzzy Logic Controller (FLC) for MPPT of PV Systems

A typical FLC includes three major parts: the fuzzification module at the input terminal, the inference engine in the core, and the defuzzification module at the output terminal. The basic structure of an FLC is shown in Figure 3. In this case, there are two input signals, that is, the change in PV power (Δ𝑃^𝑘_PV) and change in direction of duty cycle (Δ𝑑^𝑘), and one output signal, which is the duty cycle that must be applied to control DC-DC converter switching (Δ𝑑^𝑘_ref) such that all signals are crisp values.

Figure 3 

Structure of a two input-one output FLC for MPPT application.

The universe of discourse for the first input variable (Δ𝑃^𝑘_PV) is assigned in terms of its linguistic variable using seven fuzzy subsets denoted by negative large (NL), negative medium (NM), negative small (NS), zero (𝑍), positive small (PS), positive medium (PM), and positive large (PL). The universe of discourse for the second input variable (Δ𝑑^𝑘) is classified into three fuzzy sets, namely, negative (𝑁), zero (𝑍), and positive (𝑃). The output variable (Δ𝑑^𝑘_ref) is assigned in terms of its linguistic variable using nine fuzzy subsets denoted by negative double large (NL)L, negative large (NL), negative medium (NM), negative small (NS), zero (𝑍), positive small (PS), positive medium (PM), positive large (PL), and positive double large (PLL).

The fuzzy rules are generated as shown in Table 1 with Δ𝑃^𝑘_PV and Δ𝑑^𝑘 as inputs and Δ𝑑^𝑘_ref as the output. This table is also known as the fuzzy associative matrix. The fuzzy inference of the FLC is based on Mamdani’s method, which is associated with the max-min composition. The defuzzification technique is based on the centroid method, which is used to compute the crisp output, Δ𝑑^𝑘_ref.

Table 1 

Fuzzy rules of the FLC for MPPT.

3.2. Optimized FLC Using the HNN for MPPT of PV Systems

The HNN is useful for associative memory and optimization in a symmetrical structure [18]. The HNN uses a two-state threshold “neuron” that follows a stochastic algorithm where each neuron or processing element, 𝑁_𝑖, has two states, with values of either zerp or one in discrete, and between zero until one in continuous application. The inputs of each neuron come from two sources, namely, external inputs (𝐼_𝑖) and inputs from other neurons (𝑁_𝑗). The total input to neuron 𝑁_𝑖 is given by 𝑢_𝑖=𝐼_𝑖+_𝑗𝑤_𝑖𝑗⋅𝑜_𝑁_𝑗, (2) where 𝑢_𝑖 is the input of HNN, 𝑤_𝑖𝑗 is the synaptic interconnection strength from neuron 𝑁_𝑖 to neuron 𝑁_𝑗, 𝐼_𝑖 is the external input to neuron 𝑁_𝑖, and 𝑜_𝑁_𝑗 is the output of neuron 𝑁_𝑗.

A typical output of neuron 𝑁_𝑖 is a sigmoid function. Mathematically, it is given by [19] 𝑜_𝑁_𝑖=𝑔𝜆𝑢_𝑖=11+𝑒^{−𝜆𝑢𝑖}, (3) where 𝜆 is the gain that determines the shape of the sigmoid function.

The energy function 𝐸 of the continuous HNN is similarly defined as [19] 1𝐸=−2^𝑛𝑛_𝑖=1𝑗=1𝑤_𝑖𝑗𝑜_𝑁_𝑖⋅𝑜_𝑁_𝑗+^𝑛_𝑖=1𝐼_𝑖𝑜_𝑁_𝑖 (4) and its change in energy is given by Δ𝐸=−_𝑗𝑜_𝑁_𝑗⋅𝑤_𝑖𝑗+𝐼_𝑖Δ𝑜_𝑁_𝑖 (5)

The change in energy is always less than zero because 𝑔 is a monotonically increasing function. Therefore, the network solution moves in the same direction as the decrease in energy. The solution seeks out a minimum of 𝐸 and comes to a stop at the stability point.

In the design of the proposed optimal FLC, two inputs, Δ𝑃^𝑘_PV and direction Δ𝑑^𝑘, and one output, Δ𝑑^𝑘_ref, are used as described before. For simplicity, the design is only based on membership functions of Δ𝑃^𝑘_PV and Δ𝑑^𝑘_ref,. Δ𝑃^𝑘_PV is described by seven membership functions, as illustrated in Figure 4, and Δ𝑑^𝑘_ref is described by nine membership functions, as illustrated in Figure 5. In Figure 4, the centers of the Δ𝑃^𝑘_PV membership function are 𝑥₁, 𝑥₂, 𝑥₃, 𝑥₄, 𝑥₅, 𝑥₆, and 𝑥₇. In Figure 5, the centers of Δ𝑑^𝑘_ref, membership functions are 𝑧₁, 𝑧₂, 𝑧₃, 𝑧₄, 𝑧₅, 𝑧₆, 𝑧₇, 𝑧₈, and 𝑧₉. Based on the number of centers of Δ𝑃^𝑘_PV and Δ𝑑^𝑘_ref, the proposed HNN consists of 16 neurons with variables given as 𝑥₁=−𝑆_𝑃1,  𝑥₂=−𝑆_𝑃2,  𝑥₃=−𝑆_𝑃3,  𝑥₄=𝑆_𝑃4,  𝑥₅=𝑆_𝑃5,  𝑥₆=𝑆_𝑃6,  𝑥₇=𝑆_𝑃7,  𝑧₁=−𝑆_𝐷1,  𝑧₂=−𝑆_𝐷2,  𝑧₃=−𝑆_𝐷3,  𝑧₄=−𝑆_𝐷4,  𝑍₅=𝑆_𝐷5,  𝑍₆=𝑆_𝐷6,  𝑍₇=𝑆_𝐷7,  𝑍₈=𝑆_𝐷8,  and 𝑍₉=𝑆_𝐷9. 𝑆_𝑃1 until 𝑆_𝑃7 are output values of neurons 𝑁₁  until  𝑁₇(𝑜_𝑁₁--𝑜_𝑁₇), whereas 𝑆_𝐷1  until 𝑆_𝐷9  are output values of neurons 𝑁₈ until 𝑁₁₆  (𝑜_𝑁₈--𝑜_𝑁₁₆).

Figure 4 

Mapping neurons on the membership functions coding for Δ𝑃PV.

Figure 5 

Mapping neurons on the membership functions coding for Δ𝑑ref.

When calculating the values of neurons in Figures 4 and 5, the following constraints should be satisfied: 𝑆_𝑃4𝑆=0,_𝑃1𝑆=SPmax_left,_𝑃7=SPmax_right,0≤𝑆_𝑃3≤𝑆_𝑃2≤𝑆_𝑃1,0≤𝑆_𝑃6≤𝑆_𝑃5≤𝑆_𝑃7,𝑆_𝐷5𝑆=0,_𝐷1𝑆=SDmax_left,_𝐷9=SDmax_right,0≤𝑆_𝐷4≤𝑆_𝐷3≤𝑆_𝐷2≤𝑆_𝐷1,0≤𝑆_𝐷6≤𝑆_𝐷7≤𝑆_𝐷8≤𝑆_𝐷9. (6)

According to the implementation of FLC for MPPT, the control system converges at Δ𝑃^𝑘_PV=0 and Δ𝑑^𝑘_ref,=0. The goal of MPPT is to achieve Δ𝑃^𝑘_PV=0 and Δ𝑑^𝑘_ref,=0. Therefore, the objective function is assumed by minimizing 𝐸=𝐸₁+𝐸₂=12𝐴Δ𝑃^𝑘_PV²+12𝐵Δ𝑑^𝑘_ref², (7) where 𝐸 is the energy function to be minimized and 𝐴 and 𝐵 are constants.

From (7), the first part of 𝐸, which is 𝐸₁, only depends on the universe of Δ𝑃^𝑘_PV, which is the first input of FLC. Δ𝑃^𝑘_PV is defined by the defuzzification of the universe of Δ𝑃^𝑘_PV using a centroid function [20], which is written as Δ𝑃^𝑘_PV=∑⁷_𝑖=1𝜇Δ𝑃^𝑘_PV𝑖𝑥_𝑖∑⁷_𝑖=1𝜇Δ𝑃^𝑘_PV𝑖, (8) where 𝜇(Δ𝑃^𝑘_PV)_𝑖 is the membership value of 𝑥_𝑖.

𝐸₁ depends only on neurons 𝑁_𝑖  (𝑖=1,2,3,4,5,6,7). Knowing that the left side of Δ𝑃^𝑘_PV in the membership function of the first input is 𝜇(Δ𝑃^𝑘_PV)_𝑖≠0 for 𝑥₁, 𝑥₂, and 𝑥₃, then solving 𝐸₁ yields 𝐸₁=12⁷⁷_𝑖=4𝑗=4𝐴𝜇Δ𝑃^𝑘_PV𝑖𝜇Δ𝑃^𝑘_PV𝑗∑⁷_𝑖=1∑⁷_𝑗=1𝜇Δ𝑃^𝑘_PV𝑖𝜇Δ𝑃^𝑘_PV𝑗×𝑜_𝑁_𝑖⋅𝑜_𝑁_𝑗. (9)

The second part of 𝐸 in (7) is related to the output of FLC and depends only on neurons 𝑁_𝑖(𝑖=8,9,10,11,12,13,14,15,  and  16). Δ𝑑^𝑘_ref can be defined by defuzzification using the centroid method and 𝐸₂ can be obtained as 𝐸₂=12¹⁶_𝑖=8¹⁶_𝑗=8𝐵𝜇Δ𝑑^𝑘_PV𝑖𝜇Δ𝑑^𝑘_PV𝑗∑¹⁶_𝑖=8∑¹⁶_𝑗=8𝜇Δ𝑑^𝑘_PV𝑖𝜇Δ𝑑^𝑘_PV𝑗×𝑜_𝑁_𝑖⋅𝑜_𝑁_𝑗, (10) where 𝑁=0 for 𝑖=12 and 𝑗=12,  and 𝜇(Δ𝑑^𝑘_ref) is the membership value of 𝑧_𝑛. Finally, the total energy function 𝐸=𝐸₁+𝐸₂ defined in (7) becomes 1𝐸=2⁷⁷_𝑖=1𝑗=1𝐴𝜇Δ𝑃^𝑘_PV𝑖𝜇Δ𝑃^𝑘_PV𝑗∑⁷_𝑖=1∑⁷_𝑗=1𝜇Δ𝑃^𝑘_PV𝑖𝜇Δ𝑃^𝑘_PV𝑗×𝑜_𝑁_𝑖⋅𝑜_𝑁_𝑗+12¹⁶_𝑖=8¹⁶_𝑗=8𝐵𝜇Δ𝑑^𝑘_ref𝑖𝜇Δ𝑑^𝑘_ref𝑗∑¹⁶_𝑖=8∑¹⁶_𝑗=8𝜇Δ𝑑^𝑘_ref𝑖𝜇Δ𝑑^𝑘_ref𝑗×𝑜_𝑁_𝑖⋅𝑜_𝑁_𝑗, (11) where 𝑁=0 for 𝑖 = 4,12 and 𝑗 = 4,12.

Comparing (4) with (11), the weight matrix of neurons 𝑁_𝑖 to 𝑁_𝑗 in the HNN is derived and given as 𝑤_𝑖𝑗=𝑤₁₁𝑤₁₂𝑤₁₃𝑤⋯0₂₁𝑤₂₂𝑤₃₁𝑤₃₂⋮000𝑤_𝑛𝑛, (12) where 𝑤_𝑖𝑗=−𝐴𝜇Δ𝑃^𝑘_PV𝑖𝜇Δ𝑃^𝑘_PV𝑗∑⁷_𝑖=1∑⁷_𝑗=1𝜇Δ𝑃^𝑘_PV𝑖𝜇Δ𝑃^𝑘_PV𝑗,𝑤for𝑖and𝑗=1,2,3,4,5,6,7_𝑖𝑗=−𝐵𝜇Δ𝑑^𝑘_ref𝑖𝜑𝜇Δ𝑑^𝑘_ref𝑗∑¹⁶_𝑖=8∑¹⁶_𝑗=8𝜇Δ𝑑^𝑘_ref𝑖𝜇Δ^𝑘_ref𝑗,𝑤for𝑖and𝑗=8,9,10,11,12,13,14,15,16_𝑖𝑗=0forother. (13)

4. Hardware Implementation

Figure 6 shows a photograph of the prototype of the proposed boost converter circuit implemented on the PCB. The proposed boost converter is designed to provide 700–900 𝑉_dc for the input requirement of the three-phase inverter of 3 kW rating. Therefore, the IGBTs as switching devices with protection diodes 𝐷_𝑍 and 𝐷₁, 𝐷₂, 𝐷₃, 𝐷₄, 𝐷₅, 𝐷₆, and 𝐷_𝑜 are limited to an output voltage of 1000 V. The capacitors selected are able to withstand 1000 V. The connecting capacitor 𝐶₃ is selected as a high AC voltage capacitor at the secondary side. Based on the mathematical analysis and all the above considerations, the various components and devices in the proposed boost converter circuit are calculated at a switching frequency of 50 kHz as shown in Table 2.

Table 2 

Boost converter circuit components values.

Figure 6 

Prototype boost converter circuit implemented on the PCB.

For the implementation of the MPPT algorithm in dSPACE, the control design for the dSPACE real-time interrupt (RTI) control system method is constructed based on the Simulink model. Figure 7 shows the implementation of the MPPT algorithm using the RTI library in the form of a block diagram model. The MPPT block model is embedded with the C codes for implementing the MPPT algorithms using the Hopfield FLC (HFLC) technique. The MUXADC and ADC blocks read the current and voltage sensor signals, respectively. The PWM signal is generated by the DS1104 SL_DSP_PWM block. The signal is transferred to I/O pin number 21 of the P1B sub-D connector. The RTI control system is downloaded to the DS1104 board to generate PWM to trigger the IGBTs on the prototype.

Figure 7 

Control system model using the RTI library for PWM signal generation.

The MPPT controller prototype is tested using a 1 kW Agilent E4360 A PV solar array simulator (SAS), as shown in Figure 8. It is also connected to a 3 kW PV panel consisting of 25 PV modules of the SolarTIF STF-120P6PV type.

Figure 8 

Laboratory experimental setup for testing the MPPT controller using the agilent PV simulator.

5. Experimental Results

The performance verification of the new boost converter and intelligent MPPT controller for a PV system are discussed in the following experimental results.

5.1. High-Performance Boost Converter

To evaluate the performance of the new boost converter, the effect of the duty cycle on the boost converter output voltage is investigated. The result is compared with that of the conventional boost converter topology. Figure 9 depicts the output voltages obtained from the proposed and conventional boost converters. The output voltages of the proposed boost converter (140 V) are approximately three times greater than those of the conventional boost converter (45 V). The effects of duty cycle and various input voltages on the output voltages of both proposed and conventional boost converters are presented in Table 3. Figure 10 compares the output voltages of the new and conventional boost converters. The former gives a voltage gain much higher than the latter.

Table 3 

Performance comparison of the experimental output voltages of the various proposed and conventional boost converters.

(a)

(b)

Figure 9 

Output voltages obtained experimentally: (a) proposed boost converter and (b) conventional boost converter.

Figure 10 

Proposed and conventional boost converter output voltages.

From the currents and voltages shown in Table 3, the input and output powers, as well as the efficiency of the proposed and conventional boost converters are calculated. Figure 11 shows the plot of the efficiencies of the proposed and conventional boost converters against the input power. The maximum efficiency of the proposed boost converter is 90% at an input power of 3.6 kW. The proposed boost converter is able to enhance the efficiency of the conventional boost converter by approximately 9%.

Figure 11 

Proposed and conventional boost converter efficiencies.

5.2. Experimental Testing of the MPPT Controller

To assess the performance of the prototype MPPT controller, initial experiments were conducted using an agilent PV SAS simulator at 54 W, 200 W, and 848.6 W of power. The results of testing the MPPT controller for tracking the 54 W power from the PV simulator are captured in terms of the voltage-current (𝑉-𝐼) curve and PV MPP, as shown in Figures 12(a) and 12(b) for channels 1 and 2, respectively. The theoretical MPP is 25.0 and 1.08 A. The results of the MPP and 𝑉-𝐼 curve of the PV array channel 1 as shown in Figure 12(a) indicate that the MPP is at a voltage of 25.0616 V and a current of 1.0758 A. However, the results of the MPP and 𝑉-𝐼 curve of the PV array channel 2 shown in Figure 12(b) indicate that the MPP is at 25.1585 V and 1.0706 A. Given that the MPP experimental results are close to the theoretical MPP (with an error of 0.19%), the prototype MPPT controller is considered to give the accurate MPPT of the PV system. The slight differences in the MPP results at channels 1 and 2 may be have been caused by oscillation at around the MPP.

(a)

(b)

Figure 12 

Experimental result of the MPPT controller at (a) channel 1 and (b) channel 2 of the agilent SAS simulator for 54 W power tracking.

The prototype MPPT controller is also tested by connecting it to a 3 kW PV array and a resistive load of 210 Ω. The boost converter was tested by the controlled PWM with and without MPPT, that is, using triggering signals at duty cycles of 0.05 and 0.1. For further comparison, testing was done by directly connecting the PV panel to the load. From the measured current and voltage values, the PV output powers are calculated and plotted versus time, as shown in Figure 13. The maximum power can be obtained from the PV array using the HFLC-based MPPT controller compared with using only the constant duty cycle boost converter and without the MPPT controller. The prototype MPPT controller can also efficiently track the maximum power from the PV array, drawing 2.11 and 2.98 kW power from the rated 3 kW PV array for samples 1 and 2, respectively. If the PV array is directly connected to a resistive load bank without installing the MPPT controller, only 0.65 and 1.46 kW powers are drawn for samples 1 and 2, respectively. Hence, the prototype MPPT controller can produce 2.4 times more PV power than that without using the MPPT controller. The results prove the effectiveness of using the MPPT controller in PV system operation so that the maximum PV output power can be obtained.

(a)

(b)

Figure 13 

Comparison hourly PV powers with and without MPPT controller on samples (a) 1 and (b) 2.

6. Conclusion

This paper has presented a new MPPT controller for PV generation systems using an improved boost converter design and a novel MPPT algorithm. This novel MPPT algorithm based on the HFLC is developed for extracting the PV array power at its maximum. A hardware prototype of the proposed boost converter is developed and tested for verification. The proposed boost converter gives better energy conversion efficiency (90%) than conventional boost converters (81%). Laboratory testing of the MPPT controller is implemented using a 1 kW agilent PV solar array simulator. For the actual field testing of the MPPT controller, it is connected to a 3 kW SolarTIF STF-120P6PV PV array.