基于多模型耦合的电动汽车三电系统安全性估计方法  

Safety Estimation Method of Electric System in Electric Vehicles Based on Multiple Model Coupling

作　　者：李达 张普琛 林倪 张照生[1,2,3] 王震坡 邓钧君[1,2] LI Da;ZHANG Puchen;LIN Ni;ZHANG Zhaosheng;WANG Zhenpo;DENG Junjun(National Engineering Laboratory for Electric Vehicles,Beijing Institute of Technology,Beijing 100081;Collaborative Innovation Center for Electric Vehicles in Beijing,Beijing Institute of Technology,Beijing 100081;Chongqing Innovation Center,Beijing Institute of Technology,Chongqing 401120)

机构地区：[1]北京理工大学电动车辆国家工程实验室,北京100081 [2]北京理工大学电动车辆协同创新中心,北京100081 [3]北京理工大学重庆创新中心,重庆401120

出　　处：《机械工程学报》2023年第12期354-363,共10页Journal of Mechanical Engineering

基　　金：国家重点研发计划(2019YFB1600800);国家自然科学基金(52172398);国家高技术船舶科研计划“船用电池动力系统工程化应用研究”资助项目。

摘　　要：电动汽车动力电池、驱动电机和电控系统的安全性对于车辆的正常运行和乘员的生命财产安全至关重要。提出一种多模型耦合的电动汽车三电系统安全性估计方法。该方法仅需要实车传感器采集的稀疏数据作为输入,识别同车型中故障车辆。首先,构建三电系统安全性估计体系,该体系采用“自顶向下”的多层结构。之后,提出一种多模型耦合的方法,该方法由高斯混合、熵权计算和安全分数计算三部分组成,高斯混合得到安全性估计体系中各指标分布规律并输出概率密度,避免熵权重主观划分区间导致的误差;所提出的熵权计算基于概率密度确定各指标的权重,并根据安全性体系计算各车辆/系统的安全总指标,避免主观确定各指标的重要性;基于统计学与数据归一化,得到各车辆/系统的安全分数。最后,采用10辆电动汽车的实车数据对方法进行验证,包括整车安全性估计、三电系统安全性估计和不同季节鲁棒性。结果表明,所提出的方法识别故障车辆/系统与正常车辆/系统的准确率比层次分析法高40%/26.7%,且在不同季节不会对正常汽车产生误判。The safety of power battery,drive motor and electronic control system is essential for the normal operation of electric vehicles and the safety of occupant's life and property.A novel safety estimation method for electric system in electric vehicles is proposed based on multiple model coupling.The method only needs sparse data collected by onboard sensors as input and can detect the fault vehicles with the same specification.Firstly,a safety estimation scheme of electric system is proposed,which is constructed by multiple layers“from top to bottom”.Then,a multi-model coupling method is proposed,consisting of gaussian mixture,entropy weight calculation and safety score computation.Gaussian mixture can obtain the distribution of safety indicators in safety estimation scheme and output the probability density.This can avoid the error caused by the subjective interval division of entropy weight;The proposed entropy weight calculation can determine the weight of each indicator based on probability density,and calculate the total safety indicator of each vehicle/system according to the safety estimation scheme.This can avoid the subjective determination of the importance of each indicator;The safety scores of each vehicle/system are then computed based on statistics and data normalization.Finally,the method is verified by the data of ten real-world electric vehicles,including vehicle safety estimation,electric system safety estimation and robustness in different seasons.The results show that the accuracies of the proposed method for normal and fault vehicle/electric system classification are 40%/26.7%higher than analytic hierarchy process,and it will not misjudge normal vehicles in different seasons.

关 键 词：电动汽车 动力电池 驱动电机 电控系统 故障诊断 

分 类 号：U469[机械工程—车辆工程]