Performance Benchmark of Multi-Layer Neural Network Based Solar MPPT for PV Applications

Abstract

A multi-layer feed-forward ANN-based approach is discussed in this paper to achieve maximum power point tracking (MPPT). 70% of the data-set is trained with ANN algorithm, 15% is used for validation and 15% is utilized in testing of neural network. The photo generated output power after MPPT is interfaced with photovoltaic (PV) converters and subsequently to the solar-driven lead-acid battery charge controllers. The research effort aims to have a comparative performance analysis of the ANN approach against the well-known Perturb and Observe algorithm (P&O) and incremental inductance considering MPPT effectiveness for the PV modules and the charging and discharging response of the lead-acid based battery. The simulation results reveal that the neural network algorithm-based approach is more effective than the conventional algorithms for solar MPPT to mitigate the mismatch losses due to partial shading condition as well as for achieving the optimal efficiency of PV battery charge controllers and the overall PV system.