Fast and reliable parameterization of PEMFC models – sequential optimization for efficient parameter estimation

Abstract

A fast and reliable methodology for parameterizing non-linear fuel cell models is presented and validated using real-life measurement data. The approach separates the dynamic fuel cell model into multiple submodels by using output measurements, enabling a sequential optimization strategy that estimates subsets of the parameter vector in consecutive steps. By splitting the high-dimensional optimization problem into smaller, more manageable subproblems, the proposed algorithm significantly reduces computational time while increasing model fit compared to traditional optimization methods.