OPTIMIZING MAXIMUM POWER POINT TRACKING (MPPT) USING DEEP REINFORCEMENT LEARNING TO IMPROVE SOLAR PANEL EFFICIENCY UNDER DYNAMIC WEATHER CONDITIONS

Ahmad Fawzy Muntasir, Nabiyl

Journal of Moeslim Research Technik
Vol. 3 No. 1 (2026)

Solar photovoltaic (PV) systems play an important role in the transition toward sustainable energy. Variations in solar irradiance, temperature, and partial shading caused by dynamic weather conditions often reduce the efficiency of photovoltaic power generation. Conventional Maximum Power Point Tracking (MPPT) algorithms such as Perturb and Observe and Incremental Conductance frequently experience difficulties in maintaining the global maximum power point when environmental conditions change rapidly. Intelligent control approaches are therefore required to improve the adaptability and performance of MPPT systems. This study aims to develop and evaluate an MPPT optimization method based on Deep Reinforcement Learning (DRL) to enhance solar panel efficiency under dynamic weather conditions. The proposed method is designed to enable the controller to learn optimal operating strategies through continuous interaction with the photovoltaic system environment. A quantitative experimental design was implemented using a photovoltaic simulation model integrated with a DC–DC boost converter and a DRL-based control framework. Environmental scenarios including fluctuating irradiance and temperature variations were simulated to evaluate system performance. The DRL-based MPPT algorithm was compared with conventional MPPT techniques using metrics such as tracking efficiency, convergence speed, and power stability. Results show that the proposed DRL-based MPPT method achieved higher tracking efficiency (98.3%), faster convergence time, and improved power stability under dynamic weather conditions compared with traditional algorithms. These findings indicate that reinforcement learning provides a robust and adaptive solution for optimizing photovoltaic power extraction. The study concludes that Deep Reinforcement Learning can significantly enhance MPPT performance and support the development of intelligent photovoltaic energy systems capable of operating efficiently in highly variable environmental conditions.