An optimal adaptive neuro-fuzzy inference system for photovoltaic power system optimization under partial shading conditions

Different maximum power point tracking (MPPT) techniques have been evaluated based on fluctuation around maximum power point (MPP), tracking speed, and accuracy to maximize and enhance the photovoltaic solar panels' output power response. Also, the efficient tracking of the global maximum power point (GMPP) under partial shading conditions (PSCs) is one of the most significant goals of the MPPT algorithms. Under PSCs, the conventional adaptive neuro-fuzzy inference system (ANFIS) based on the MPPT algorithm vastly suffered from a lack of training data, which caused tracking the wrong zone on the power/voltage (P-V) curve and selecting a local maximum power point (LMPP) instead of the GMPP. To overcome these problems, this paper introduces an improved version of ANFIS controller-based MPPT supported with an automatic, fast, and accurate analysis method to collect accurate and comprehensive training data and acquire sufficient knowledge to enhance the performance of photovoltaic solar systems under uniform irradiance conditions (UICs) and PSCs. The proposed algorithm covers all potential atmospheric circumstances under UICs and PSCs. Comparison with the conventional MPPT algorithm and the normal ANFIS controller indicates the proposed ANFIS controller's superiority in tracking the GMPP under partial shading conditions. Under PSCs, the proposed controller tracks the GMPP (point A) faster than the conventional incremental conductance (InCond) technique (triangle t = 0.07 s), with fewer fluctuations at the steady-state response. In contrast, the conventional ANFIS controller cannot track the GMPP (point A) with a high-power loss (triangle P = 4906 watts).