Improved clustering-multimodal feature decomposition for multi-time scale terminal voltage characterization, edge computing performance evaluation, and retirement prediction of large-capacity batteries in energy storage power stations

To address the critical problems of complex data dimensions of energy storage units in large-scale energy storage power stations, the lack of a performance evaluation system, and the excessive computational cost of traditional evaluation schemes, this paper proposes a comprehensive performance evaluation method and standard for large-capacity energy storage batteries based on a “cloud-large computing – local-small computing” collaborative architecture, namely the Clustering-Multimodal Feature Decomposition (C-MFD) algorithm. The C-MFD algorithm takes the easily collected battery terminal voltage as the core characteristic index. Through cluster analysis, it realizes the characteristic grouping of energy storage battery clusters, breaks through the application limitations of a single evaluation standard, and completes the targeted quantitative evaluation of battery packs with the same characteristic parameters. To verify the effectiveness of the proposed scheme, 500 Kirin batteries are selected, and a total of 15 operating conditions of cycle charge-discharge experiments are carried out under three scenarios: laboratory wide-temperature environment, large-scale energy storage application, and smart grid scenario. The experimental results show that the C-MFD algorithm can preliminarily determine the battery output power difference within 1 min, with an MRD of 0.68%; accurately diagnose the battery performance degradation degree within one discharge cycle, with an MRD of 0.19%; and complete the battery retirement grade judgment within one maintenance cycle, with an MRD of 0.14%. The C-MFD algorithm significantly reduces the computing load of the BMS and can intuitively and accurately quantitatively evaluate the battery output performance. It provides a scientific and quantitative basis for the safe application of large-scale energy storage batteries and key technical support for battery echelon utilization, which significantly improves the operational safety and resource utilization value of energy storage power stations.