复杂工况下综合传动装置状态监测数据异常检测方法  

Abnormal detection method of integrated transmission device condition monitoring data under complex working conditions

作　　者：宁子俊 陈涛[1] 徐峰[2] 王立勇[1] 贾然[1] NING Zijun;CHEN Tao;XU Feng;WANG Liyong;JIA Ran(The Ministry of Education Key Laboratory of Modern Measurement and Control Technology,Beijing Information Science and Technology University,Beijing 100192,China;Unit 32184 of the People s Liberation Army,Beijing 100075,China)

机构地区：[1]北京信息科技大学现代测控技术教育部重点实验室,北京100192 [2]中国人民解放军32184部队,北京100075

出　　处：《机电工程》2023年第9期1387-1394,共8页Journal of Mechanical & Electrical Engineering

基　　金：国家重点基础研究发展计划项目(MKF20210009)。

摘　　要：针对综合传动装置运行过程中,工况变化及装置故障状态引起的数据异常、难以有效区分这一问题,提出了一种适用于复杂工况下综合传动装置状态监测数据异常检测的方法。首先,采用基于密度的聚类方法(DBSCAN)对状态监测数据进行了关联变量聚类,以排除非关联数据对数据重构准确度的干扰;然后,利用深度降噪自编码网络构建了状态监测数据重构模型,获取了对异常数据敏感的偏差特征;最后,利用支持向量数据描述(SVDD)算法构建了正常状态监测数据偏差特征的超球体,完成了复杂工况下对综合传动装置状态监测数据异常的检测;为了验证该方法对综合传动装置状态监测数据异常检测的有效性,以某型综合传动装置为研究对象,在多组综合传动装置漏油实验数据上进行异常检测验证分析。实验结果表明:该方法实现了在综合传动装置不同程度漏油故障条件下对状态监测数据异常进行检测的目的,且其准确度整体高于92%。研究结果表明:该方法可以有效检测出综合传动装置早期异常运行状态,为综合传动装置健康管理与劣化评估奠定基础。Aiming at the problem that it was difficult to effectively distinguish the abnormal data caused by the change of working condition and the fault state of the integrated transmission device during its operation,a method for the abnormal detection of the condition monitoring data of the integrated transmission device under complex working conditions was proposed.Firstly,density-based spatial clustering of applications with noise(DBSCAN)method was used to cluster the associated variables of the state monitoring data to eliminate the interference of non-associated data on the accuracy of data reconstruction.Then,a reconstruction model of the state monitoring data was constructed by deep denoising auto coding network to obtain the deviation features which was sensitive to abnormal data.Finally,support vector data description(SVDD)algorithm was used to construct a hypersphere with deviation characteristics of normal condition monitoring data,and the abnormal detection of the condition monitoring data of the integrated transmission device under complex working conditions was completed.In order to verify the effectiveness of the method for anomaly detection of comprehensive transmission device condition monitoring data,a comprehensive transmission device was taken as the research object,and the anomaly detection was verified and analyzed on several groups of comprehensive transmission device oil leakage experimental data.The experimental results show that the method realizes the purpose of detecting the abnormal condition monitoring data under the condition of different degrees of oil leakage failure of integrated transmission device,and the accuracy is higher than 92%.The results show that the proposed method can effectively detect the early abnormal operating state of the integrated transmission and lay a foundation for the health management and deterioration evaluation of the integrated transmission.

关 键 词：综合传动装置 机械传动 异常检测 数据重构 数据关联 基于密度的聚类方法 深度降噪自编码 

分 类 号：TH132[机械工程—机械制造及自动化] U463.2[机械工程—车辆工程]