Study on the Cluster Selection Method of Echelon Utilization Power Battery Based on Confidence Interval Estimation

In order to improve the screening accuracy of decommissioned power batteries and reduce the regulation errors of batteries in the process of echelon utilization, a cascade batteries selection method based on confidence interval probability prediction is proposed. Firstly, according to the internal and external characteristic parameters, residual capacity, state of health (SOH) and other historical parameters, K-means clustering is used to divide static clusters of retired power batteries with similar parameters. Secondly, the bootstrap probability method is used to analyze the deviation subset of SOH estimation in echelon utilization process, and the probability interval density function of SOH subset is established. Finally, further optimization grouping is carried out according to the predicted SOH characteristics. By comparative analysis, the influence of confidence interval estimation on the screening accuracy is determined, and the effectiveness of the proposed screening method is determined. The consistency of dynamic characteristics is improved while the static parameters are similar in the screening process, which lays a foundation for accurate regulation of retired power battery echelon utilization.