An Adaptive Variable Leaky Least Mean Square Control Scheme for Grid Integration of a PV System

This paper presents a new grid synchronization control scheme for integrating a 3-phase PV system to a grid. The Leaky Least Mean Square (LLMS) algorithm uses the fixed value for leak factor, which causes the weight vector to drift beyond its limit. Hence the proposed control scheme comprises of an adaptive Variable Leaky Least Mean Square (VLLMS) algorithm to generate Reference Inverter Current (RIC) which overcomes the aforesaid drawback, Reinforcement Learning (RL) algorithm for Maximum Power Point Tracking (MPPT) and a Sliding Mode approach to generate switching signals. The MPPT is designed with RL algorithm for extraction of maximum power from PV panels during varied solar insolations. Sliding Mode Controller (SMC) is designed for generation of switching signal to Voltage Source Inverter (VSI) during dynamic loading condition. The proposed VLLMS-RL-SMC control algorithm is first implemented in MATLAB/Simulink, then on a prototype PV experimental setup developed in the laboratory. The performance of the proposed VLLMS-RL-SMC algorithm is compared with that of other control schemes namely Leaky Least Mean Square-RL-SMC (LLMS-RL-SMC), Least Mean Fourth (LMF)-RL-SMC and Instantaneous Power Theory-RL-SMC (IPT-RL-SMC) in which RL is used for MPPT as in case of VLLMS-RL-SMC scheme for steady state and dynamic loading conditions. From the comparison, it is observed that the proposed VLLMS-RL-SMC algorithm outperforms amongst above three control schemes.