SVM-Guided Improved Love Evolution Algorithm for Global Maximum Power Point Tracking of Photovoltaic Arrays Under Partial Shading and Temperature Disturbances

In a PV array, mismatch among modules changes the shape of the P–V curve and may create several power peaks. This makes maximum power point tracking (MPPT) more difficult, especially when the tracker needs to distinguish the global maximum power point (GMPP) from local peaks. This paper studies this problem with SVM-ILEA, a hybrid MPPT method that combines support vector machine (SVM) regression and an improved love evolution algorithm (ILEA). The SVM model takes module irradiance and temperature as inputs and predicts a voltage close to the GMPP. ILEA uses this voltage as the search center and avoids scanning the full voltage range. The modified convergence factor and adaptive distance factor further adjust the voltage movement during iteration, giving wider search steps at the early stage and smaller corrections near the optimum to reduce steady-state power oscillations. The simulation setup in MATLAB/Simulink R2019b includes standard test conditions (STC) and static partial shading with non-uniform irradiance and temperature distributions, as well as dynamic operating conditions. Across the four static conditions, SVM-ILEA achieves mean tracking times of 0.0233–0.0303 s and mean steady-state power fluctuations of 0.0111–0.0500 W. Across the three dynamic tests, the mean MPPT efficiency ranges from 97.9057% to 98.2991%. The results obtained demonstrate fast GMPP tracking, small power fluctuation, and stable re-tracking under complex PV operating conditions.