The second-life utilization of retired power batteries presents promising opportunities; however, the substantial upfront testing costs and subsequent maintenance expenses have hindered its widespread adoption. A key challenge is achieving a balance between accurate and comprehensive sorting outcomes and minimizing the costs associated with the detection process. Existing methods have primarily focused on exploring cost-effective hidden indicators with stronger representativeness or leveraging algorithms to extract low-cost latent features that correlate with the required performance parameters for sorting. Hence, we developed a rapid sorting method based on extracting multiple features from partial charge intervals. Through a comprehensive analysis of correlations and cost considerations, five features that can be extracted from the same partial charging segment are selected as classification criteria. Meanwhile, these features exhibit different tendencies toward the three main battery characteristics: capacity, internal resistance, and voltage. The proposed method utilizes a self-organizing map algorithm and subtractive clustering to achieve efficient sorting. In comparison to traditional sorting methods, the detection time is reduced to 19.96%, and energy consumption is decreased to 20.35%. By improving sorting efficiency and reducing costs while maintaining sorting effectiveness, the proposed method enhances the economic feasibility of hierarchical utilization, which may contribute to the advancement of second-life utilization of retired batteries.
Read full abstract