DEEP NEURAL NETWORK BASED MPPT MODELLING AND SIMULATION FOR PHOTOVOLTAIC SYSTEM

Abstract

With the growing demand of electrical power, the deployment of renewable energy resources within the electric power network has become essential. Among these renewable sources, solar is one of the most common renewable resource used nowadays for environment friendly power production. To extract maximum power output from photovoltaic system a techniques known as maximum power point tracking (MPPT) is employed. Perturb and observe (P&O) method is one among the multiple techniques integrated with solar panels for obtaining maximum power output out of them. Besides having simple and practical structure it still lacks in efficiency due to the presence of oscillations near the maximum power point, Slow tracking in changing conditions and performance degradation under varying irradiation. This paper aims to develop a MPPT controller based on deep neural network (DNN) for photovoltaic system. Furthermore, this paper also provide comparison between conventional perturb and observe method and deep neural network based MPPT under standard test condition (STC). The results have showed that the power output of deep neural network based MPPT has reduced oscillations and show better performance. Simulations and deep neural network are developed in SIMULINK and MATLAB environment.