Adaptive Neuro-Fuzzy Inference System-Genetic Algorithm approach for global maximum power point tracking in PV systems under different shading conditions

Abstract

Photovoltaic (PV) panels have become popular energy sources; nonetheless, when under partial shading, their efficiency greatly reduces because of the occurrence of several power peaks or spikes arising because of the variation in irradiance. An inherent problem with most conventional global maximum power point tracking (GMPPT) algorithms is that they do not distinguish local and global peaks, and thus energy extraction may not be optimal. To eliminate this, this paper proposes an Adaptive Neuro-Fuzzy Inference System-Genetic Algorithm (ANFIS-GA) solution in improving the GMPPT performance under diverse shading conditions. The ANFIS is ideal in modeling the nonlinearity and uncertain nature of PV systems; the Genetic Algorithm (GA) then dynamically determines the optimal Proportional–Integral–Derivative (PID) controller parameters by minimizing the error between the actual power output and the maximum power output calculated in ANFIS. This means that this controller can respond to nonlinear and time-varying nature of a DC–DC SEPIC converter, enhancing response time and stability in various operating conditions. The proposed ANFIS-GA method, simulated with MATLAB/Simulink, was evaluated in three environments, i.e., no shading, partial shading, and severe shading, and performances were 99.98%, 98.5%, and 97.67% (GMPPT efficiencies), respectively. Besides, a comparative test to ANFIS-only systems at 400 W/m², 700 W/m², and 1000 W/m² also revealed that the hybrid methodology better tracked the global peak particularly in an oscillating environment. Furthermore, to test the strength of the controller, time-varying irradiance profiles that would simulate the real-world cloud motions using randomized step-wise changes are applied to it. The system sustained efficiencies better than 98.5% in such difficult conditions using convergence times that were below 0.25 s. These findings support that the ANFIS-GA approach has a high degree of flexibility, accuracy, and consistency, and it is thus an effective method in improving PV performance in a complex and dynamically varied systems.