多传感器数据融合的复杂系统退化模式挖掘  

Degradation mode mining for complex system based on multi-sensor data fusion

作　　者：彭宅铭 程龙生[2] 姚启峰 PENG Zhaiming;CHENG Longsheng;YAO Qifeng(School of Economics and Management,Jiangsu University of Science and Technology,Zhenjiang 212100,China;School of Economics and Management,Nanjing University of Science and Technology,Nanjing 210094,China)

机构地区：[1]江苏科技大学经济管理学院,江苏镇江212100 [2]南京理工大学经济管理学院,南京210094

出　　处：《振动与冲击》2022年第13期239-245,251,共8页Journal of Vibration and Shock

摘　　要：退化模式挖掘对复杂系统剩余寿命预测具有重要意义。为了解系统运行状态,掌握其退化规律,提出一种基于时间序列聚类的退化模式挖掘方法。首先,利用改进马田系统筛选并融合多传感器数据特征,构建健康指数来表征系统的退化趋势。然后,采用累积和算法将健康曲线进行分段处理,获取退化曲线,并利用基于动态时间弯曲距离度量的层次聚类算法将退化模式进行归类。最后,以相似度和退化时间为判别依据,对系统的退化模式进行有效识别。以航空发动机为对象的研究表明,该方法能够有效的挖掘和识别退化模式,为复杂系统剩余寿命预测提供依据。Degenerate mode mining is of great significance to prediction of residual life of a complex system.Here,to understand operating state of the system and master its degradation law,a degradation mode mining method based on time series clustering was proposed.Firstly,the improved Mahalanobis-Taguchi system(MTS)was used to screen and fuse multi-sensor data features,and construct the health index for characterizing degradation trend of the system.Then,the health curve was segmented by using the cumulative sum algorithm to obtain degradation curve,and the hierarchical clustering algorithm based on dynamic time warping distance measurement was used to classify degradation modes.Finally,based on the similarity and degradation time,the degradation mode of the system was effectively identified.The aeroengine study showed that the proposed method can effectively mine and identify degradation modes,and provide a basis for predicting residual life of complex system.

关 键 词：复杂系统 马田系统(MTS) 健康指数 退化模式 层次聚类 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]