Polymer electrolyte membrane fuel cells degradation prediction using multi-kernel relevance vector regression and whale optimization algorithm

Abstract

Degradation and cost are the main factors affecting the commercial applications of Polymer Electrolyte Membrane Fuel Cells (PEMFC). This paper proposes a novel degradation prediction for PEMFC in various applications by using Multi-kernel Relevance Vector Regression (MRVR) and Whale Optimization Algorithm (WOA). This method uses data from a vehicle operating under real driving conditions and laboratory data to derive a robust model that covers a wide range of operation. In order to learn degradation trends better, MRVR is adopted to establish the PEMFC degradation prediction model. WOA is used to automatically adjust and optimize the weight and kernel parameters for improving the prediction precision. Proposed method is experimentally verified under different operational conditions. The test results show that compared with a single kernel function, the multi-kernel function can predict degradation of PEMFC more accurately. Compared with other metaheuristic methods, WOA greatly improves the precision of degradation prediction.