| 基于电压频域特征和异常系数的动力电池故障诊断方法 |
| |
| 引用本文： | 刘鹏,吴志强,张照生,孙振宇.基于电压频域特征和异常系数的动力电池故障诊断方法[J].中国公路学报,2022,35(8):89-104. |
| |
| 作者姓名： | 刘鹏 吴志强 张照生 孙振宇 |
| |
| 作者单位： | 1. 北京理工大学 电动车辆国家工程研究中心, 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 |
| |
| 基金项目： | 国家重点研发计划项目(2019YFB1600800) |
| |
| 摘 要： | 动力电池系统是电动汽车(EV)的关键部件和主要故障源,因而提高动力电池故障诊断的效率和准确率显得尤为重要。基于此提出一种基于快速傅里叶变换(FFT)和异常系数评估(ACE)的动力电池电压不一致性故障诊断方法。针对6辆发生故障或热失控事故的电动汽车和1辆电压一致性良好的电动汽车,基于其在新能源汽车国家监管平台的全生命周期运行数据,经过电压数据的数据清洗、数据变换等大数据预处理后,利用FFT技术时频变换,提取频域中的幅值作为故障诊断的特征参数;然后,引进基于Z分数理论的异常系数对故障程度进行定量评估,以实现故障单体的检测和定位;此外,针对存在多个故障单体的情况,基于单体异常率的计算,实现单体故障程度的判定和排序;在此基础上,详细分析电压数据长度及采样间隔、FFT采样点数对模型的影响;最后,与基于熵和Z分数的电压故障诊断方法进行比较。研究结果表明:在上述研究条件下,该诊断方法对于电压一致性良好的车辆未产生误报警,且可以有效地检测出事故车辆动力电池系统存在的电压不一致性故障;相比之下,模型平均计算准确率提高了3.25%,模型平均耗时仅为熵值模型的0.55%;验证了该方法故障单体定位更精准、数据适用性更好及计算速度更快的优点。该研究成果能有效实现动力电池电压不一致性故障诊断,具有较高的工程应用价值。
|
| 关 键 词： | 汽车工程 故障诊断 快速傅里叶变换 电压不一致性 异常系数 大数据 |
| 收稿时间： | 2021-02-26 |
Fault Diagnosis for Battery Systems Based on Voltage Frequency-domain Indicator and Abnormal Coefficient |
| |
| LIU Peng,WU Zhi-qiang,ZHANG Zhao-sheng,SUN Zhen-yu.Fault Diagnosis for Battery Systems Based on Voltage Frequency-domain Indicator and Abnormal Coefficient[J].China Journal of Highway and Transport,2022,35(8):89-104. |
| |
| Authors: | LIU Peng WU Zhi-qiang ZHANG Zhao-sheng SUN Zhen-yu |
| |
| Affiliation: | 1. National Engineering Research Center of Electric Vehicles, Beijing Institute of Technology, Beijing 100081, China;2. Collaborative Innovation Center for Electric Vehicles, Beijing Institute of Technology, Beijing 100081, China;3. Beijing Institute of Technology Chongqing Innovation Center, Chongqing 401120, China |
| |
| Abstract: | Power battery systems are the key component and the main source of faults in electric vehicles.Therefore,it is of great importance to improve the efficiency and accuracy of battery fault diagnosis.Accordingly,a fault diagnosis method was proposed based on the fast Fourier transform (FFT) and abnormal coefficient evaluation for voltage inconsistency faults of a battery system.Six accident vehicles and one normal vehicle were selected from the National Monitoring and Management Center,and big-data preprocessing techniques,such as data cleaning and data transformation,were adopted for the full life-cycle operating voltage data.Then,the data were transformed in the frequency domain by using FFT,and the amplitude in the frequency domain was proposed as the characteristic indicator of fault diagnosis.Furthermore,the abnormal coefficient based on the Z-score was introduced to quantitatively evaluate the fault degree so that faulty cells may be detected and located.In addition,in the case of multiple faulty cells,the fault degree was determined and sorted by calculating the abnormal cell rate.Thereby,the influence of the voltage data length,date sampling time,and number of FFT sampling points on the model was analyzed in detail.Finally,a comparison with the voltage fault diagnosis method based on entropy and Z-score indicates that the proposed diagnosis method do not produce false alarms for normal vehicles and can effectively detect severe voltage inconsistency faults in accident vehicles under the above research conditions.Specifically,the accuracy of the model increases by 3.25%,whereas its time consumption is only 0.55% of the entropy model,verifying the advantages of the proposed method,namely,more accurate fault location,better applicability,and faster calculation.The proposed method can effectively diagnose voltage inconsistency faults,and thus it has high engineering application value. |
| |
| Keywords: | automotive engineering fault diagnosis fast Fourier transform voltage inconsistency abnormal coefficient big data |
|
| 点击此处可从《中国公路学报》浏览原始摘要信息 |
|
点击此处可从《中国公路学报》下载全文 |
|
