A Comparative Study of Parameter Identification Fusion Methods With a High-Precision Supercapacitor Model for Electric Vehicles

Abstract

Abstract To ensure energy storage system operates reliably for electric vehicles, it is vital to accurately identify supercapacitor model parameters in applications. In recent years, most of the algorithms focus on lithium-ion batteries, but few are reported to be used for supercapacitors. To fill this research gap, many algorithms and corresponding fusion methods for supercapacitors are designed in this study. First, seven popular intelligent optimization algorithms are selected to identify the supercapacitor model parameters, and the identification results are discussed in detail. Then, considering a single algorithm cannot guarantee convergence to all global optimal model parameters over state-of-charge (SOC) intervals, five fusion methods for supercapacitor parameter identification have been developed by combining information fusion technology. Finally, voltage errors are statistically analyzed to validate the effectiveness of the five proposed fusion methods. The results show that the five fusion methods can further enhance the global prediction performance of the supercapacitor model, particularly the reverse search-based parameter identification fusion (PIF-RS) method, which has better accuracy and reliability with respect to the maximum (Max) error, mean error, and root mean square (RMS) error decreasing by at least 10.1191%, 17.0024%, and 17.0989%, respectively.