基于数据挖掘技术的电动汽车充电安全监测与故障预警方法研究  

Research on Electric Vehicle Charging Safety Early Warning Based on Outlier Detection Method

作　　者：顾兵[1] 李泽昊 王小林 赵梓潼 金圣权 GU Bing;LI Zehao;WANG Xiaolin;ZHAO Zitong;JIN Shengquan(Research Center for Jilin Energy Development(Northeast Electric Power University),Jilin Jilin 132012;North China Electric Power University College of International Education,Baoding Hebei 071000)

机构地区：[1]吉林省能源发展研究基地(东北电力大学),吉林吉林132012 [2]华北电力大学国际教育学院,河北保定071000

出　　处：《东北电力大学学报》2024年第5期87-93,共7页Journal of Northeast Electric Power University

基　　金：吉林省科技发展计划项目(20220203069SF)。

摘　　要：在实现碳达峰与碳中和的目标过程中,为实现环境友好型、能源节约型发展,电动汽车开始登上舞台。伴随市场上电动汽车保有量快速增长,由电动汽车所引发的安全问题也开始逐渐显露。电动汽车的自燃和火灾事故屡见不鲜,无论是对于汽车生产厂商、消费者、还是充电设备的经营者都造成了严重的经济损失,甚至会造成生命危险。电动汽车的充电安全问题已然开始制约着电动汽车行业的发展。文中从恒压恒流充电方法出发,通过对电动汽车充电故障原因的分析,提出了基于高斯分布的离散点检测数据挖掘预警方法。离群点检测法的特点是可以从一组数据集中有效的判别出具有差异显著的数据。电动汽车充电过程中会产生大量数据,通过获取正常充电下的数据,得到正常充电状态的高斯分布模型,运用离群点检测法,判断电动汽车充电是否处于危险状态,并及时对充电故障进行预警。实现对电动汽车充电过程实时监测与预警,实验证明了离群点检测法对于电动汽车充电安全预警具有良好的可行性与准确性。In the process of achieving the goal of carbon peak and carbon neutrality,in order to achieve environmentally friendly and energy-saving development,electric vehicles have begun to take the stage.With the rapid growth of electric vehicle ownership in the market,the safety problems caused by electric vehicles have gradually begun to emerge.Spontaneous combustion and fire accidents of electric vehicles are common.They have caused serious economic losses and even life hazards to automobile manufacturers,consumers,and operators of charging equipment.The charging safety of electric vehicles has begun to restrict the development of the electric vehicle industry.In this paper,starting from the constant voltage and constant current charging method,through the analysis of the causes of electric vehicle charging faults,a data mining early warning method based on discrete point detection of Gaussian distribution is proposed.The characteristic of outlier detection method is that it can effectively distinguish the data with significant difference from a set of data sets.A large amount of data will be generated during the charging process of electric vehicles.The Gaussian distribution model of normal charging state is obtained by obtaining the data under normal charging.The outlier detection method is used to judge whether the charging of electric vehicles is in a dangerous state,and the charging fault is warned in time.The real-time monitoring and early warning of electric vehicle charging process are realized.The experiment proves that the outlier detection method has good feasibility and accuracy for electric vehicle charging safety early warning.

关 键 词：电动汽车 充电安全 离群点检测 预警模型 

分 类 号：TM73[电气工程—电力系统及自动化]