A robust MPPT method based on optimizable Gaussian process regression and high order sliding mode control for solar systems under partial shading conditions

This study proposes a new hybrid approach for maximum power point tracking (MPPT) under partial shading conditions. The approach consists of a high order sliding mode controller (HOSMC) and a hyperparameters optimized Gaussian Process Regression (GPR). The proposed hybrid method consists of two stages. In the first stage, the model created from real-time data was trained using optimized hyperparameters GPR method, one of the machine learning methods. Moreover, at this phase, a model is constructed utilizing temperature and irradiance values as inputs, with the output being the voltage value. Using this model, the reference voltage values required for HOSMC in different scenarios were generated. In the second stage, the error expression was obtained by comparing the generated reference voltage value with the PV input voltage value. The control signal is generated to minimize the error value and MPPT is performed. The Incremental Conductance (INC) algorithm, a traditional optimization technique, and the Gray Wolf Optimization (GWO), a metaheuristic optimization algorithm, were used to compare the suggested hybrid method in order to show its efficacy. Data from different months and seasons (February, May, August and October) were used to determine the scenarios. The irradiance and temperature values used in the scenarios were obtained from real-time data. Simulation studies were carried out using Matlab/Simulink. When the proposed hybrid method was compared with the INC and GWO algorithms, it was seen that it had the fastest convergence time and the largest efficiency value in all scenarios.