Integration of an optimized neural network in a photovoltaic system to improve maximum power point tracking efficiency

Indonesian Journal of Electrical Engineering and Computer Science Vol. 28, No. 3, December 2022, pp. 1276~1285 ISSN: 2502-4752, DOI: 10.11591/ijeecs.v28.i3.pp1276-1285  1276 Integration of an optimized neural network in a photovoltaic system to improve maximum power point tracking efficiency Ezzitouni Jarmouni1, Ahmed Mouhsen2, Mohamed Lamhamedi1, Hicham Ouldzira2, Ilias En-Naoui1 1 Laboratory of Radiation-Matter and Instrumentation (RMI), The Faculty of Sciences and Technology, Hassan First University of Settat, Settat, Morocco 2 Laboratory of Engineering, Industrial Management and Innovation (IMII), The Faculty of Sciences and Technology, Hassan First University of Settat, Settat, Morocco Article Info ABSTRACT Article history: Due to the variability of weather conditions and equipment properties the maximum power point tracking (MPPT) performance is influenced. MPPT controllers are widely used to improve photovoltaic (PV) efficiency because MPPT can produce maximum power under various weather conditions. Among the most used techniques and representing a satisfactory efficiency are those based on artificial intelligence. Since the use of neural networks requires resources at the implementation level, the optimization of these systems is an important phase. This work represents an optimized system for tracking the maximum power point, the latter based on a multi-layer neural network. The optimized multi layer perceptron (MLP) will ensure a fast convergence to the maximum power point with a low oscillation compared to the classical method. Received Jun 29, 2022 Revised Aug 27, 2022 Accepted Sep 7, 2022 Keywords: Artificial neural network DC/DC converter Multi layer perceptron Maximum power point tracking Photovoltaic system Perturbation and observation This is an open access article under the CC BY-SA license. Corresponding Author: Ezzitouni Jarmouni Laboratory of Radiation-Matter and Instrumentation (RMI) The Faculty of Sciences and Technology, Hassan First University of Settat BP: 577, Route de Casablanca. Settat, Morocco Email: 1. INTRODUCTION The recently, renewable energy systems have become very important and photovoltaic (PV) technology production has grown exponentially around the world. Among these renewable energy sources, the most known and widespread systems in the world are the solar photovoltaic energy and wind energy sources [1], [2]. Photovoltaic panels produce electricity by converting sunlight into electricity through the photovoltaic effect of semiconductors. Among the advantages and the strong points that encourage the use of these renewable sources other than the conventional or traditional energy sources (fossil energy) are the following: the renewable energy sources are nearly inexhaustible, clean, green and do not represent a danger to the environment [3]. Due to the availability of PV modules, solar PV has seen considerable growth compared to other renewable energy technologies. The PV modules are recognized by their non-linear behavior and the non-linear current versus voltage curve. This means that the production of energy with maximum efficiency is not an easy task, since the maximum power points are unique and reaching their specific techniques is important. All these techniques are collectively called maximum power point tracking (MPPT) techniques or algorithms. These algorithms make the PV system function at about its maximum power point by adapting the impedance of the load and the PV source. As a result of their nonlinear characteristics, MPPT techniques are fundamental to any PV system. Journal homepage: http://ijeecs.iaescore.com Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752  1277 There are dozens of methods that have been reported in the literature to track the maximum power point [4], [5]. Among the most widely used methods and techniques used by researchers cited by [6], [7], are the following: fractional open circuit voltage, perturbation and observation (P&O), fractional short circuit current and incremental conductance (IncCon). The research and development communities are continuously striving to improve the existing methods with the addition of artificial intelligence (AI) based systems such as fuzzy logic, neuronal networks and ANFIS [8], [9]. This paper represents a maximum power point tracking system of a photovoltaic system based on artificial neural networks (MLP). The objective of this work is to improve the efficiency of MPPT search by integrating the optimized MLP model. The Figure 1 illustrates the main components of the system under study. Figure 1. The studied system architecture 2. EQUIVALENT MODULE OF A PHOTOVOLTAIC SOLAR CELL A solar or photovoltaic cell is in fact a source of current that is produced when sunlight is incident on the surface of the cell. The process of transforming light into electricity is known as the "photovoltaic effect". In order to show the characteristics of voltage, current and power under different operating conditions, the mathematical model of the PV cell is necessary for the simulation. Figure 2 shows a simplified equivalent model of a PV device. As illustrated in Figure 2, photovoltaic cell model consists mainly of a series resistor (𝑅𝑠), this latter connected with a parallel shunt resistor combination (𝑅𝑠ℎ) in series, exponential diode (𝐷) and cell photo-current (𝐼𝑝ℎ) [10]. Vp𝑣, and Ip𝑣 are respectively corresponding to the current voltage of the PV cell. Figure 2. The equivalent model of a PV cell In (1) and (2) representing the current generated by the solar cell: Ipv = Iph − Id − Ish Ipv = Iph − I0 . (e q(Vpv +Ipv .Rs) nKT (1) V − 1) − pv +Ipv .Rs Rsh (2) Integration of an optimized neural network in a photovoltaic system to improve … (Ezzitouni Jarmouni) 1278  ISSN: 2502-4752 where 𝐼𝑝ℎ , 𝐼𝑠 , q, K, n and T represent respectively the solar current indued, the saturation current of the diode, the charge of the electrons (1.6𝑒 −19 C), the Boltzmann constant (1.38𝑒 −23 J/K), the Ideality factor of the PN junction (1~ 2) and the Temperature (K). Table 1, demonstrates the main electrical characteristics of the solar panel used in this study. As mentioned in the table, the maximum power that can be produced by this photovoltaic generator is 220 watts. Table 1. The electrical characteristics of the used photovoltaic array Parameters and symbol Rated power 𝑃𝑀𝑃 Open circuit voltage 𝑉𝑂𝐶 Voltage at maximum power 𝑉𝑀𝑃 Short circuit current 𝐼𝑆𝐶 Current at maximum power 𝐼𝑀𝑃 Value 220W 54V 44.63V 5.52A 4.94A 3. DC/DC CONVERTER A DC/DC converter is used to convert the DC voltage delivered by the PV array into a DC voltage that is suitable for supplying DC voltage to consumers. In this study, a DC-DC boost converter is used to realize the MPPT power stage due to its high reliability, reduced implementation costs and reduced number of components [11], [12]. The Figure 3 demonstrates the electrical model of boost converter used in this study. Figure 3. (...truncated)