HYBRID APPROACHES TO MAXIMUM POWER POINT TRACKING (MPPT) BASED ON ARTIFICIAL INTELLIGENCE: SYNERGY OF METHODS TO INCREASE EFFICIENCY

This study investigates the performance of hybrid Maximum Power Point Tracking (MPPT) algorithms for photovoltaic systems using neural networks (NN), genetic algorithms (GA), and particle swarm optimization (PSO). Three hybrid models—NN-GA, PSO-GA, and NN-PSO—were evaluated based on response time, maximum power, and robustness to noise. Results revealed distinct trade-offs among the models. The NN-GA model exhibited the shortest response time of 0.0036 s/prediction, but its accuracy significantly degraded with increasing noise levels. The PSO-GA model demonstrated the highest maximum power output of 100.24 W and superior robustness to noise, with a mean absolute error (MAE) of 0.2062 at 30% noise. The NN-PSO model provided balanced performance, achieving a maximum power of 45.90 W and demonstrating the lowest MAE of 0.1537 at 20% noise. These findings highlight the transformative potential of hybrid MPPT methods in enhancing both the adaptability and performance of solar energy systems across diverse and challenging operational conditions.