Maximum power point tracking using neural networks for grid-connected photovoltaic system

Abstract

This paper proposes a method of maximum power point tracking (MPPT) using neural networks for gridconnected photovoltaic systems. The system is composed of a boost converter and a single-phase inverter connected to a utility grid. The maximum power point tracking control is based on output from neural networks to control a switch of a boost converter. Back-propagation neural networks is utilized as pattern classifier. Back-propagation neural networks is an example of nonlinear layered feed-forward networks. The single phase inverter uses hysteresis current control which provides current with sinusoidal waveform. Therefore, the system is able to deliver energy with low harmonics and high power factor. MPPT using neural networks are simulated and implemented to evaluate performance. Simulation and experimental results are provided for neural networks and fixed duty ratio under the same atmospheric condition. From the simulation and experimental results, neural networks can deliver more power than the conventional controller.