能源紧缺、环境污染日益严重等问题是人类发展所面临的一项巨大挑战。电动汽车无污染、低排放、节能环保等特点是实现绿色交通，促进可持续发展的战略选择。近年来，国家对新能源行业的大力支持及政策优惠，推动了我国电动汽车产业的迅速发展，我国已经成为全球最大的新能源汽车制造和消费国。电动汽车的高压电安全在其可靠性发展过程中占据着至关重要的作用。绝缘性能是衡量电动汽车高压电安全的重要指标，如何快速、准确的定位绝缘故障显得尤为重要。但目前针对电动汽车绝缘故障系统性的研究相对较少，鉴于此，本文对电动汽车绝缘故障进行了系统性的分析。基于数学模型搭建电动汽车绝缘故障仿真平台，然后构建绝缘故障数据集，使用SVM、RF模型进行单模块绝缘故障诊断，使用图卷积神经网络（GCN）进行整车绝缘故障诊断，并使用Pyqt5构建绝缘故障诊断系统，对电动汽车绝缘故障进行诊断。

本文对电动汽车各主要高压模块建立了绝缘故障数学模型，通过改变各模块模型参数设置不同类型的绝缘故障，以模拟实际使用中可能出现的问题。通过理论计算和分析，研究绝缘故障对模块性能的影响，对比正常运行状态和故障状态下的输出，得出了能够反应不同类型和不同位置绝缘故障的特征参数。该理论分析和计算结果为后续的故障诊断系统开发提供了重要的依据。根据构建的数学模型使用Simulink搭建各模块的仿真模型，并将各模型联合形成整车绝缘故障仿真平台。在整车平台上设置不同类型、不同位置的绝缘故障，得到仿真数据，对比仿真数据与实测结果，以此来验证仿真平台的有效性。

由于目前针对电动汽车绝缘故障的数据较少，暂无公开数据集，因此本文基于仿真平台构建电动汽车绝缘故障数据集。为了反映不同位置绝缘故障对模块级和整车级数据的影响，本文分别收集了单个模块绝缘故障数据和整车绝缘故障数据；为了确保数据的有效性，每种类型的数据均采集了20000个样本；这些数据为故障诊断模型提供了更加丰富的输入，增强了诊断的精确性和可靠性。然后将采集的数据划分为80%的训练集和20%的测试集，用于模型的训练。在单模块绝缘故障诊断中，本文使用两种广泛应用的机器学习算法，检验两者在故障诊断中的适用性。然而在整车系统中，单一模块的绝缘故障往往会联动影响其他模块数据，增加了整车绝缘故障诊断的复杂性。因此本文引入了图卷积神经网络（GCN），这是一种能够有效处理多模块间复杂关系的深度学习方法。GCN利用交叉验证来评估模型的泛化能力，防止过拟合，并通过学习图结构中的节点特征及其连接关系，更精准地捕捉各模块之间的相互作用，尤其适用于复杂系统的故障诊断。最后，为使故障诊断易于应用，本文制作了绝缘故障诊断系统软件。该软件实现了从数据输入、模型选择、故障诊断到结果输出的全流程自动化，极大简化了操作步骤，为电动汽车绝缘故障诊断提供了更方便、高效的解决方案，弥补了现有研究的不足，有助于提高电动汽车的可靠性和安全性。

Energy scarcity and increasingly serious environmental pollution are enormous challenges facing human development. The characteristics of electric vehicles, such as no pollution, low emissions, energy conservation and environmental protection, are strategic choices for achieving green transportation and promoting sustainable development. In recent years, the strong support and policy incentives provided by the government for the new energy industry have promoted the rapid development of China's electric vehicle industry. China has become the world's largest producer and consumer of new energy vehicles. The high-voltage safety of electric vehicles plays a crucial role in their reliability development process. Insulation performance is an important indicator for measuring the high-voltage safety of electric vehicles, and it is particularly important to quickly and accurately locate insulation faults. However, there are relatively few systematic studies on insulation faults in electric vehicles. In view of this, this paper conducts a systematic analysis of insulation faults in electric vehicles. Based on the mathematical model, an electric vehicle insulation fault simulation platform was built, and then an insulation fault dataset was constructed. SVM and RF models were used to diagnose single module insulation faults, and graph convolutional neural networks (GCN) were used to diagnose vehicle insulation faults. Insulation fault diagnosis system was built using Pyqt5 to diagnose electric vehicle insulation faults.

In this paper, mathematical models for insulation fault have been established for the major high-voltage modules of electric vehicles, By changing the model parameters of each module, different types of insulation faults are set to simulate possible problems in practical use. Through theoretical calculations and analysis, the impact of insulation faults on module performance is studied. The mathematical model outputs under normal operation and fault conditions are compared, and characteristic parameters that can reflect different types and positions of insulation faults are obtained. This theoretical analysis and calculation results provide important basis for the development of subsequent fault diagnosis systems. Based on the constructed mathematical model, simulation models of each module were built using Simulink, and the models were combined to form a simulation platform for insulation faults in the entire vehicle. Set insulation faults of different types and positions on the vehicle platform, obtain simulation data, compare simulation data with actual measurement results, in order to verify the effectiveness of the simulation platform.

Since there is currently little data on insulation faults in electric vehicles and no public datasets, this paper constructs an electric vehicle insulation faults dataset based on a simulation platform. In order to reflect the impact of insulation faults at different locations on module-level and vehicle-level data, this paper collects single module insulation fault data and vehicle insulation fault data respectively; to ensure the validity of the data, 20000 samples of each type of data are collected; these data provide richer input for the fault diagnosis model and enhance the accuracy and reliability of the diagnosis. The collected data is then divided into 80% training set and 20% test set for model training. In the single module insulation fault diagnosis, this paper uses two widely used machine learning algorithms to test their applicability in fault diagnosis. However, in the whole vehicle system, the insulation fault of a single module often affects the data of other modules, which increases the complexity of the insulation fault diagnosis of the whole vehicle. Therefore, this paper introduces graph convolutional neural network (GCN), a deep learning method that can effectively handle complex relationships between multiple modules. GCN uses cross-validation to evaluate the generalization ability of the model to prevent overfitting, and by learning the node features and their connection relationships in the graph structure, it can more accurately capture the interactions between modules, which is particularly suitable for fault diagnosis of complex systems. Finally, in order to make fault diagnosis easy to apply, this paper produced insulation fault diagnosis system software. The software realizes the automation of the whole process from data input, model selection, fault diagnosis to result output, greatly simplifies the operation steps, provides a more convenient and efficient solution for insulation fault diagnosis of electric vehicles, makes up for the shortcomings of existing research, and helps to improve the reliability and safety of electric vehicles.

[1]Muratori M, Alexander M, Arent D, et al. The rise of electric vehicles—2020 status and future expectations[J]. Progress in Energy, 2021, 3(2): 022002.

[2]Westbrook M H. The Electric Car: Development and future of battery, hybrid and fuel-cell cars[M]. London: The Institution of Electrical Engineers,2001.

[3]Zhiqiang W. Research on on-line monitoring method of insulation state of electric vehicle high voltage system[C]//2020 IEEE International Conference on High Voltage Engineering and Application (ICHVE). IEEE, 2020: 1-4.

[4]张强. 电动汽车绝缘监测设备研究与开发[D]. 合肥：中国科学技术大学, 2019.

[5]Choudhary A, Fatima S, Panigrahi B K. State-of-the-art technologies in fault diagnosis of electric vehicles: A component-based review[J]. IEEE Transactions on Transportation Electrification, 2022, 9(2): 2324-2347.

[6]Beckerle P, Schaede H, Rinderknecht S. Fault diagnosis and state detection in centrifugal pumps—A review of applications[C]//Proceedings of the 9th IFToMM International Conference on Rotor Dynamics. Springer International Publishing, 2015: 387-398.

[7]Vichare N M, Pecht M G. Prognostics and health management of electronics[J]. IEEE transactions on components and packaging technologies, 2006, 29(1): 222-229.

[8]Zhiqiang W. Research on on-line monitoring method of insulation state of electric vehicle high voltage system[C]//2020 IEEE International Conference on High Voltage Engineering and Application (ICHVE). IEEE, 2020: 1-4.

[9]吉嘉辉. 电动汽车电气高压系统绝缘故障定位方法研究[D]. 北京：北京交通大学, 2020.

[10]任宝森. 纯电动汽车绝缘电阻在线检测装置研制[D]. 青岛：青岛大学, 2018.

[11]代建才. 电动汽车电气系统的绝缘检测研究[D]. 天津：天津科技大学, 2019.

[12]于超. 电动汽车绝缘电阻监测方法研究[D]. 济南：齐鲁工业大学, 2019.

[13]GB 38031-2020. 电动汽车用动力蓄电池安全要求[S].北京：国家市场监督管理总局，2020.

[14]ISO 6469-1:2019 Electrically propelled road vehicles-Safely specifications-Par 1:0n-board rechargeable energy storage system (RESS)[S].2019.

[15]ISO 6469-2:2009 Electrically propelled road vehicles-Safely specifications-Part 2: Vehicle operational safety means and protection against failures[S].2009.

[16]ISO 6469-3:2018 Blecric road vehicles-Safety specinieations-Par 3. Protechon of bensons gainst electric hazards[S].2018.

[17]ECE R100:2023 Uniform provisions concerning the approval of vehicles with regard to specific requirements for the electric power train[S].2023.

[18]SAE J1766:2014 Recommended Practice for Electric and Hybrid Electric Vehicle Battery Systems Crash Integrity Testing[S].2014.

[19]GB 18384-2020. 电动汽车安全要求[S] .北京：国家市场监督管理总局&国家标准化管理委员会. 2020.

[20]GB/T 17045-2020. 电机防护装置和设备通用部分[S].北京：国家标准化管理委员会.2020.

[21]陈宁, 陈嘉瑶, 钟志平. 基于电压注入法的电动汽车绝缘电阻在线检测[J]. 电气传动, 2019, 49(2): 66-73.

[22]张超, 杜博超, 崔淑梅, 等. 电动汽车高压系统绝缘状态在线监测方法[J]. 电工技术学报, 2019, 34(12): 2657-2663.

[23]高峰, 陈宏伟, 张宗玲, 等. 电动汽车绝缘电阻要求的相关标准研究[J]. 汽车实用技术, 2022,47(07):161-165.

[24]卓文得, 孙贵斌, 许文丰, 等. 基于 Matlab 纯电动汽车匹配及仿真系统开发[J]. 汽车电器, 2022 (3): 18-21.

[25]王俊兰,吴义忠,熊会元.纯电动汽车整车建模仿真研究[J].计算机仿真,2015,32(10):190-195.

[26]尹会春. 基于电桥法的电动汽车动力电池组绝缘监测系统研究[D]. 长沙：湖南大学, 2018.

[27]陈照添, 刘成沛. 电动汽车电池组内部绝缘故障研究[J]. 邵阳学院学报: 自然科学版, 2020, 17(4): 23-30.

[28]兰昊,刘桂彬,王青贵.电动汽车碰撞后电安全分析与研究[J].中国汽车,2022,(01):18-24+50.

[29]Zhang Y, Chen Q, Liu X, et al. An insulation diagnosis method for battery pack based on battery model[C]//IOP Conference Series: Materials Science and Engineering. IOP Publishing, 2020, 793(1): 012061.

[30]黄雨龙, 陈振斌, 崔相雨, 等. 电阻绝缘检测中低压脉冲信号注入法的算法改进[J]. 海南大学学报: 自然科学版, 2020, 38(1): 13-19.

[31]何耀, 刘兴涛, 张陈斌, 等. 基于动力电池组内阻模型的绝缘检测算法[J]. 吉林大学学报 (工学版), 2013, 43(05): 1165-1170.

[32]Yao L, Xiao Y, Gong X, et al. A novel intelligent method for fault diagnosis of electric vehicle battery system based on wavelet neural network[J]. Journal of Power Sources, 2020, 453: 227870.

[33]段继强,陆洁,曾庆亮,等.电动汽车动力系统故障检测及诊断方法[J].时代汽车,2023,(18):178-180.

[34]梁永浩. 纯电动汽车高压系统故障诊断系统设计研究[D]. 广州：广东工业大学, 2019.

[35]赵忠杰.MATLAB软件在感应电动机仿真中的应用[J].电机技术,2022,(01):5-7.

[36]Moghadam D E, Herold C, Zbinden R. Electrical insulation at 800 V electric vehicles[C]//2020 International Symposium on Electrical Insulating Materials (ISEIM). IEEE, 2020: 115-119.

[37]王昕灿,李京鑫,翟羽,等.纯电动汽车绝缘检测方法及维修案例分析[J].时代汽车,2022,(10):187-189.

[38]施一新, 杨林, 杨坤. 电动汽车高压电连接电阻故障测试系统开发[J]. 汽车技术, 2016 (8): 48-52.

[39]周召召, 崔杰, 童兵亮. 电动汽车高压母线对电平台异常电压研究[J]. 2020 中国汽车工程学会年会论文集 (4), 2020.

[40]胡大满, 窦红伟, 曹晓鸽, 等. 电动汽车直流充电系统绝缘检测问题分析[J]. 第十七届河南省汽车工程科技学术研讨会论文集, 2020.

[41]付国良,蓝贤平,李军剑,等.电动汽车高压母线Y电容对漏电流的影响分析[J].安全与电磁兼容,2024,(01):83-87.

[42]张翔, 钱立军, 张炳力, 等. 电动汽车仿真软件进展[J]. 系统仿真学报, 2004, 16(8): 1621-1623.

[43]周荣宽.电动汽车电驱动系统建模仿真及硬件在环测试[D].重庆：重庆理工大学,2015.

[44]熊刚,章云区.基于充电桩的电动汽车电池非拆卸式绝缘检测方法[J].电气技术,2023,24(05):46-51+57.

[45]宋炳雨, 高松, 郎华, 等. 纯电动汽车高压电故障诊断与安全管理策略研究[D]. 重庆：重庆交通大学, 2010.

[46]邵玉龙. 电动汽车 BMS 关键技术研究及硬件在环测试系统构建[D]. 长春: 吉林大学, 2018.

[47]苏肖. 电动汽车动力电池绝缘电阻检测方法研究[D]. 北京: 华北电力大学, 2016.

[48]吴振军, 王丽芳. 电动汽车智能在线绝缘检测装置研究[J]. 低压电器, 2009 (5): 20-22.

[49]Li D, Zhang Z, Liu P, et al. Battery fault diagnosis for electric vehicles based on voltage abnormality by combining the long short-term memory neural network and the equivalent circuit model[J]. IEEE Transactions on Power Electronics, 2020, 36(2): 1303-1315.

[50]项彬. 电动客车绝缘故障的解决方案[J]. 客车技术与研究, 2020, 42(4): 30-32.

[51]袁雷. 现代永磁同步电机控制原理及 MATLAB 仿真[M]. 北京航空航天大学出版社, 2016.

[52]赵春明, 吴志新, 马宁, 等. 电动汽车高压电系统状态参数在线监测[J]. 吉林大学学报 (工学版), 2007, 37(01): 37-42.

[53]王嘉悦. 电动汽车有源在线绝缘监测装置的研究与开发[D]. 北京: 北京交通大学, 2011.

[54]周启晟, 杨林, 蔡亦山, 等. 电动汽车高压电连接与绝缘状态参数在线监测[J]. 电源技术, 2016, 40(3): 655-658.

[55]赵志刚,杨金波,吴春燕,等.电动汽车绝缘电阻检测系统设计与研究[J].工业仪表与自动化装置,2018,(04):54-57+64.

[56]牛飞飞.电动汽车绝缘电阻测试方法的研究[J].电子制作,2024,32(04):35-37.

[57]宋炳雨. 纯电动汽车高压电安全管理系统研究与设计[D]. 淄博：山东理工大学, 2011.

[58]Jiang J, Li T, Chang C, et al. Fault diagnosis method for lithium-ion batteries in electric vehicles based on isolated forest algorithm[J]. Journal of Energy Storage, 2022, 50: 104177.

[59]张天强,宋芳.电动汽车绝缘电阻计算方法研究[J].汽车技术,2021,(10):42-46.DOI:10.19620/j.cnki.1000-3703.20201160.

[60]Gu X, See K W, Li P, et al. A novel state-of-health estimation for the lithium-ion battery using a convolutional neural network and transformer model[J]. Energy, 2023, 262: 125501.

[61]Lv Z, Chen D, Lou R, et al. Industrial security solution for virtual reality[J]. IEEE Internet of Things Journal, 2020, 8(8): 6273-6281.

[62]袁玉萍, 钟萍, 安增龙, 等. 基于支持向量机的多类分类问题的一种新算法[J]. 数学的实践与认识, 2011, 41(1): 94-100.

[63]Pal M. Random forest classifier for remote sensing classification[J]. International journal of remote sensing, 2005, 26(1): 217-222.

[64]Alkan A, Günay M. Identification of EMG signals using discriminant analysis and SVM classifier[J]. Expert systems with Applications, 2012, 39(1): 44-47.

[65]Jackins V, Vimal S, Kaliappan M, et al. AI-based smart prediction of clinical disease using random forest classifier and Naive Bayes[J]. The Journal of Supercomputing, 2021, 77(5): 5198-5219.

[66]Kipf T N, Welling M. Semi-supervised classification with graph convolutional networks[J]. arXiv preprint arXiv:1609.02907, 2016.

[67]Zhang S, Tong H, Xu J, et al. Graph convolutional networks: a comprehensive review[J]. Computational Social Networks, 2019, 6(1): 1-23.