Online Control Using ANN-PI Controller for Single Stage Transformerless Grid Tied PV System

This paper investigates the design and effectiveness of a control based on an adaptive Proportional Integral (PI) controller using an Artificial Neural Network (ANN). The strategy of control is applied to a single phase single stage transformerless grid connected photovoltaic system. In order to overcome its sensitivity to parameter changes, a Feed Forward Neural Network (FFNN) is used to allow renewing the PI parameters to ensure the best performances when the operating point varies. In order to assess the performances of the suggested controller, a comparison has been established between the simple PI and the adaptive one using simulation results under solar irradiance variation. The results show that the proposed controller can provide an improved quality of power and gives good performances in terms of grid synchronization and reference tracking. Moreover, the supremacy of the proposed method is that the Total Harmonic Distortion (THD) of the injected current is enhanced especially under low solar irradiance.