Remaining Life Prediction of PEMFC Using FOGD-KELM Algorithm

Aiming at the difficulty of predicting the remaining service life of proton exchange membrane fuel cells (PEMFC), a novel prediction method combining fractional-order gradient learning and nuclear limit learning machine(is proposed. The method utilizes the long memory property of fractional-order calculus with the nonlinear mapping capability of the kernel-limit learning machine to capture the complex dynamics of the PEMFC decay process. Simulation results show that the prediction accuracy improvement of FOGD-KELM over BP neural network ranges from 65% to 75%, and that of GD-KELM ranges from 50% to 70%, further validating its advantages in data fitting and prediction.