Multi-strategy alpha evolution optimization for constrained parameter estimation in Proton Exchange Membrane Fuel Cells

Abstract

PEMFCs (Proton Exchange Membrane Fuel Cells) are devices widely used today in hydrogen power generation and energy storage systems. PEMFC parameter estimation is crucial for optimizing fuel cell performance, reducing costs, and ensuring system reliability. Accurate estimation allows for better modeling and simulation, and minimizes the need for expensive and time-consuming experiments. The study focuses on a multistrategy alpha evolution algorithm (MSAE) aimed at improving the accuracy of parameter estimation in PEMFCs. The MSAE features enhancements over the traditional alpha evolution method, such as employing a Halton sequence to create the initial population and using a fitness-distance balance technique for selecting appropriate candidate solutions. To assess the coherence and reliability of MSAE, a comparison is made with existing techniques in the literature in three distinct cases. In Case I, there are no parameter restrictions, reflecting conventional parameter estimation approaches. Case II introduces restrictions among the parameters to evaluate consistency, while Case III investigates consistency with varying limits. The results are presented using the sum of squared error (SSE) for comparison with other upcoming algorithms. Considering that SSE differences may be very small in some cases, additional error measures are also used for the evaluation. The results demonstrate that MSAE exceeds other competitive metaheuristic algorithms by achieving lower error rates, including SSE, mean absolute error (MAE), mean absolute percentage error (MAPE), root mean square error (RMSE) and relative error (RE), while also ensuring highly compatible parameter estimations.