Polar lights optimizer: A novel algorithm for accurate parameter estimation in proton exchange membrane fuel cells

• The new algorithm which takes inspiration from auroras manages robust optimization through a balance between local exploitation and global exploration. • The Polar Lights Optimization algorithm achieved better performance than nine leading algorithms in terms of accuracy, convergence speed and SSE when tested on six PEMFC models. • The optimization process reached lowest Absolute Error of 0.259293 and Relative Error of 1.185075% for PEMFC parameter estimation. • The algorithm reached full convergence speedily after completing 50 iterations while achieving top positions in efficiency and accuracy metrics (FR=1). • The technology shows potential for large-scale proton exchange membrane fuel cells while enabling dynamic operations and hybrid system integration. Proton Exchange Membrane Fuel Cells play a key role in sustainable power systems by delivering both superior energy conversion efficiency and eco-friendly operation. The complex multivariate structure together with nonlinear behavior of these systems makes it parameter estimation a challenging task, which negatively affects operational reliability and system lifespan. This research presents Polar Lights Optimization (PLO), a noval aurora-inspired method developed to address the current optimization challenges. PLO combines auroral gyration motion for local exploitation with oval walking dynamics for global exploration, ensuring a balanced search process. PLO undergoes performance testing against nine established optimization techniques including GSA, DE, PSO, MFO, ACOR, MVO, WOA, SCA, and JAYA through six PEMFC models which include BCS 500 W, Nedstack 600 W PS6, SR-12 W, Horizon H-12, Ballard Mark V, and STD 250 W Stack. The proposed algorithm consistently outperforms its competitors, achieving the lowest Sum of Squared Errors (SSE) and fastest convergence rates. The PLO algorithm generates SSE results of 0.025493 for BCS 500 W and 0.275211 for Nedstack 600 W PS6 and 0.283774 for STD 250 W Stack while keeping AE at 0.259293 and RE% at 1.185075 for the STD 250 W Stack. The algorithm completes 50 iterations in every test scenario and achieves the top Friedman Ranking score of 1. The experimental data shows that PLO application improves PEMFC system predictive accuracy and operational reliability and energy output performance.