Ann-assisted MPPT for a novel voltage-stacked multilevel boost-sepic converter for PV systems under dynamic conditions

Growing demand for high-efficiency photovoltaic interfaces requires power converters with high step-up gain, stable operation, and accurate maximum power point tracking under varying operating conditions. This research introduces a hybrid neural network-based MPPT-controlled novel voltage-stacked multilevel boost-sepic converter. The proposed MPPT framework generates optimal duty cycles using NN-Fuzzy and NN-GWO models for controlled switching. The converter operational modes validate energy transfer and voltage stacking through a diode-capacitor multilevel network, cascaded boost and sepic energy transfer stages. Simulation and hardware results demonstrate stable converter operation with output voltages of 262 V in simulation and 257.9 V in experimental validation at 65% duty cycle and 10 kHz switching frequency. NN-GWO provides smoother steady-state performance under fixed loads, while NN-Fuzzy provides better dynamic robustness under rapid fluctuations, making the proposed approach suitable for high-gain PV applications.