Adaptive Forgetting Factor Based Least Squares Method for Lithium Battery Parameter Identification

To achieve the state monitoring of the energy storage system, it is necessary to evaluate the battery health state in real time, and the accurate identification of the impedance parameters of lithium batteries is a prerequisite for the accurate nuclear capacity of the battery. Firstly, this paper selects the equivalent model, excitation signal and sampling method for battery parameter identification, establishes the second-order RC equivalent circuit model of lithium battery and discretises it. Then, a battery open-circuit voltage state observer is designed to identify the battery impedance spectral parameters segmentally based on the Variable Forgetting Factor Recursive Least-Squares (VFFRLS) in a multi-timescale manner. Then, the parameter identification results at different frequencies were clustered and analysed by K-means algorithm to obtain the optimal parameter identification results. Finally, a simulation model for battery parameter identification is constructed, and the simulation and experimental results verify the accuracy and effectiveness of the method.