FPGA-based neural network for simulation of photovoltaic array: application for estimating the output power generation

This paper introduces the preliminary results of the simulation and implementation of PV array based on neural network and Hardware Description language (VHDL). In fact, an experimental database of meteorological data (irradiation, temperature) and output electrical generation signals of the PV array (current and voltage) has been used in this study. The inputs of the ANN-PV array are the total solar irradiation and the temperature while the outputs are the current and voltage generated from the PV-array. Firstly, a dataset of 4x365 have been used for training the network. Subsequently, the neural network (MLP) corresponding to PV array is simulated using VHDL language based on the saved weights and bias of the network. Simulation results of the trained MLP-PV array based on Matlab and VHDL are presented. The proposed ANN-VHDL model permits the evaluation of the performance of the PV array using only the environmental factors and involves less computational efforts. It can also be used for predicting the output electrical energy from the PV array and for estimating the I–V characteristic of PV-array. The implementation of the PV-array based on reconfigurable Field Programming Gate Array (FPGA) is suggested in this paper for a real time PV simulator.