Improvements in identification of fuel cell temperature model

This paper presents the temperature model identification of a fuel cell using a polynomial NARMAX model. The NARMAX model identification using orthogonal least means square is affected by the input and output signal sample rate, since the regressors matrix can become ill conditioned. As a new approach, an evolutionary algorithm is proposed to determine the input and output signal sample rate, as to select the model structure as the parameter estimation. The algorithm performance is verified based on Bayesian information criterion and the mean square error.