基于自动编码器和长短时记忆网络的智能汽车故障诊断方法研究  

Fault diagnosis research for intelligent vehicles based on autoencoder and LSTM

作　　者：闵海根 方煜坤 吴霞[1] 王武祺 宋晓鹏 MIN Hai-Gen;FANG Yu-Kun;WU Xia;WANG Wu-Qi;SONG Xiao-Peng(School of Information and Engineering,Chang’an University,Xi’an 710064,China;The Joint Laboratory for Internet of Vehicles,Ministry of Education-China Mobile Communications Corporation,Chang’an University,Xi’an 710064,China;Zhejiang Transportation Planning and Design Institute Company Limited,Hangzhou 310017,China)

机构地区：[1]长安大学信息工程学院,西安710064 [2]长安大学“车联网”教育部中国移动联合实验室,西安710064 [3]浙江省交通规划设计研究院有限公司,杭州310017

出　　处：《四川大学学报（自然科学版）》2021年第5期73-81,共9页Journal of Sichuan University(Natural Science Edition)

基　　金：国家自然科学基金青年基金(61903046);陕西省重点研发计划(2021GY-290);陕西省高校科协青年人才托举计划(20200106);浙江省重点研发计划(2020C01057);“车联网”教育部-中国移动联合实验室项目(教技司(2016)477号);中央高校基本科研业务费专项资金(300102240106)。

摘　　要：智能汽车故障诊断技术对于保障智能汽车安全行驶具有重要意义,针对智能汽车传感器数据异常检测和车辆运动的异常检测提出了一种故障诊断方法.针对非时序传感器数据,采用基于超限学习框架的自动编码器,对正常数据进行特征压缩学习其特征表示,再利用压缩的特征重构数据,根据重构误差的大小判断数据是否异常.针对时序传感器数据,采用多层长短时记忆网络学习时序数据之间的时间依赖关系来预测当下时刻的数据值,根据预测误差的大小判断数据是否异常.提出一种阈值随误差大小动态变化的自适应阈值确定方法,使得决策变量对于异常值相对敏感.进一步地,采用车辆自行车运动学模型和Kalman滤波,利用Jarque-Bera测试对预测值和量测值残差的正态性进行检验来检测车辆运动是否异常.实际场地测试验证了本文所提出的方法可以有效检测非时序或时序传感器数据的异常,并对车辆运动是否异常进行检测.Fault diagnosis for intelligent vehicles is of great significance to ensure the safe driving.This paper proposes a fault diagnosis method aiming at anomaly detection for sensor data and vehicle motion of intelligent vehicles.For the non-sequential sensor data,the extreme learning machine based on autoencoder is utilized to compress the normal data instances to learn the feature representation and reconstruct the data using the compressed feature.Whether an instance is normal or not is decided in accordance with the reconstruction error.To detect the anomaly in the sequential sensor data,multi-layer Long Short-Term Memory network is adopted to learn the time adherence of the sequential data to predict the current data value,and whether the data is normal or not is judged according to the prediction error.Besides,an adaptive threshold calculation method is proposed,in this method,the threshold would dynamically change with the reconstruction error or prediction error so that the decision variable is sensitive to the anomaly.Furthermore,to detect whether the vehicle motion is abnormal,the vehicle bicycle kinematic model and Kalman filter are adopted and the normality of the residuals between the estimated and measured values is checked using Jarque-Berra test.The experiment results show that the proposed methods in this paper can effectively detect the anomaly in the non-sequential or sequential sensor data,and detect the abnormality of the vehicle motion.

关 键 词：智能汽车 故障诊断 超限学习 自动编码器 长短时记忆网络 自适应阈值计算 

分 类 号：U495[交通运输工程—交通运输规划与管理]