The manta ray foraging optimization (MRFO) algorithm suffers from a fixed parameter \( S \) , limiting its adaptability in balancing search capability and convergence speed during different optimization stages. To address this limitation, a success-history-based parameter adaptation strategy is proposed to dynamically adjust \( S \) . Furthermore, to enhance population diversity and avoid premature convergence, a randomly selected individual from the top \( G \) high-quality solutions replaces the current best individual in the somersault foraging behavior. Based on these improvements, a parameter adaptive manta ray foraging optimization (PAMRFO) algorithm is developed. The experimental results demonstrate the effectiveness of PAMRFO. On the IEEE CEC2017 benchmark function set, PAMRFO achieved an average win rate of 82.39% across 29 functions compared to seven state-of-the-art algorithms. On 22 IEEE CEC2011 real-world optimization problems, PAMRFO achieved an average win rate of 55.91% compared to ten advanced algorithms. Sensitivity analysis identified optimal parameter settings, and further stability analysis revealed that PAMRFO exhibits higher success rates and computational efficiency among the four MRFO variants. Population diversity and exploration-exploitation analysis demonstrated the effectiveness of the proposed update mechanism in maintaining diversity and balancing exploration and exploitation. In solving parameter estimation problems for six multimodal solar photovoltaic models, PAMRFO outperformed other competing methods with a 100% success rate, highlighting its superior performance in the photovoltaic field. These findings validate the robustness, efficiency, and wide applicability of PAMRFO, providing advanced solutions for optimization problems in the new energy domain.
