A Variable Effective Capacity Model for <tex-math notation="TeX">LiFePO₄</tex-math> Traction Batteries Using Computational Intelligence Techniques

Computational intelligence techniques are used to approximate the nonlinear operation of LiFePO4batteries using rule-based systems. In this paper, rule-based systems are not directly fitted to data, but comprise constructive blocks in a differential-equation-based dynamical model that is numerically integrated to infer battery voltage, charge, and temperature. The design methodology has been validated with three different LiFePO4batteries, and the results were found to be more accurate than those of a selection of statistical models and state-of-the-art artificial intelligence techniques.