A parameter estimation method for modelling proton exchange membrane fuel cell based on enhanced meta evolutionary differential evolution algorithm

Accurate estimation of unknown parameters in proton exchange membrane fuel cells (PEMFCs) is crucial for developing a precise and reliable model, which is necessary for further investigation, simulation analysis, and optimal control. However, conventional approaches often fail to provide satisfactory results due to high non-linearity, multi-variable nature, and strong nonconvexity associated with these problems. The hybrid variants either combine two or more algorithms or adaptively vary the most sensitive and dominant system parameters, which are found to achieve more desirable results than individual algorithms in solving the PEMFC parameter estimation. With this motivation, an enhanced meta-evolutionary differential evolution algorithm (EMEDEA) is proposed in this study to determine the optimal values of specific parameters in PEMFC based energy systems. The idea of achieving peak performance through highly sensitive hyperparameters, such as the mutation factor, crossover probability, and the selection of specific Differential Evolution (DE) strategies, is suggested by adopting dynamic parameter tuning or adaptive mechanisms. To make the study more practical and feasible for application under real-time conditions, a mathematical model of the PEMFC, considering all its nonlinearity and complexity, with different control variables, is used in this study. The sum of squared deviations between the experimental and estimated (or simulated) stack voltages is considered to formulate the objective function for the parameter estimation of the PEMFC system. Additionally, the performance of the proposed method EMEDEA as a solution to this problem has been simulated and tested, considering three different PEMFC stacks. The justification for the accuracy, effectiveness, and reliability of the proposed approach has been provided through comparison tables between experimental and computed stack voltages at each experimental stack current. The convergence graphs demonstrate the swift attainment of optimal values with the proposed approach. A comparative analysis with other prominent methods further confirms the accuracy and superiority of EMEDEA in solving the problem of PEMFC parameter estimation.