Single-stage Grid-connected PV System with Artificial Neural Network Controller

Abstract

This paper presents an artificial neural network (ANN)-based controller for the single-stage grid-connected photovoltaic (SSGC-PV) system with the unification of an improved maximum power point tracking (MPPT) algorithm. The proposed ANN controller is first trained with the Levenberg-Marquardt based backpropagation method by employing MPC as a supervisory controller before deploying it for the SSGC-PV system. Unlike the highly complex MPC scheme, the proposed controller only includes simple mathematical equations to compute the desired output modulation signals. The MPPT control and ANN controller combination assure optimal power extraction under varying environmental conditions, less harmonic content, and stable output signals under transient conditions by addressing the detriments of traditional control schemes. Comparative simulation results are presented to validate the effectiveness of the proposed controller by considering different dynamic test cases such as step irradiance change and partial shading conditions. Finally, experimental analyses based on a control hardware-in-the-loop (C-HIL) approach are provided to justify the real-time working of the proposed controller.