基于极限梯度提升算法的梯次电池热失控预警技术研究OA

Due to the complex thermal runaway process of cascaded utilization battery packs,the temperature prediction process is influenced by multiple factors,resulting in biased warning results.Therefore,a local thermal runaway warning method combining the extreme gradient boosting(XGBoost)algorithm and infrared thermal imaging technology is proposed.The thermal radiation energy of the battery pack is obtained through instruments and converted into electrical signals to generate thermal images,the infrared thermal imaging gradient is calculated based on this image,and the pixel gradient information of the battery pack thermal runaway area is extracted.The XGBoost algorithm is used to construct a prediction function,where each decision tree learns from the objective function and weights are calculated by using the inverse error method,combined with the regularization term to obtain local temperature prediction results.The pixel gradient information of the thermal runaway area of the battery pack is used as the input of the prediction function,the sliding window method is used to process the prediction residual,and the battery pack thermal runaway warning threshold is set to determine the warning result.Thus,the hierarchical utilization of local thermal runaway warning of the battery pack is achieved.The experimental results show that the error between the temperature values obtained by the proposed method and the true values remains within 0.3℃,the temperature prediction deviation is small,and the root mean square error consistently maintains below 0.7℃.The error distribution is relatively conce ntrated,which can meet the requirements of thermal runaway warning for cascaded utilization of battery packs.