基于自适应LTTB与DTW-DBA-Means的动力电池组不一致性评估方法OA北大核心CSTPCD

Aiming at the problem that the inconsistency of electric vehicle power battery pack is difficult to be effectively evaluated through external parameters,when analyzing the battery pack voltage data,the Silhouette Coefficient is introduced as the inconsistency evaluation index,and a new inconsistency evaluation method for power battery pack is proposed by integrating adaptive down-sampling(LTTB)and time-series clustering(DTW-DBA-Means)algorithms.Adaptive LTTB can adaptively adjust the compression ratio and sample point allocation in compression intervals according to the characteristics of the battery pack voltage sequence,which can improve the DTW-DBA-Means operation efficiency and ensure the clustering effect.Experiments is conducted based on the real vehicle data running for nine months,the results show that the adaptive LTTB down-sampling effect is better than dynamic LTTB and LTTB,and the DTW-DBA-Means time-series clustering effect is better than k-Shape,and the proposed method can save about 96.7％operation time while ensuring the accuracy of evaluation.