基于SVDD与D-S证据理论的发动机故障诊断研究  

A Research on Engine Fault Diagnosis Based on SVDD and D-S Evidence Theory

作　　者：张玲玲[1,2] 廖红云[2] 贾继德[2] 乔龙[2] 程利军 

机构地区：[1]军械工程学院火炮工程系 [2]军事交通学院汽车工程系 [3]济南军区72502部队

出　　处：《汽车工程》2013年第1期23-26,50,共5页Automotive Engineering

摘　　要：提出了一种基于支持向量数据描述(SVDD)和D-S证据理论相结合的故障诊断新方法,通过SVDD算法分别判断来自单个传感器的数据,然后运用D-S证据理论对各传感器的诊断结果进行信息融合,最终实现对发动机的故障诊断。试验表明,该方法充分利用各信号源的冗余互补信息,降低了诊断的不确定性,提高了故障诊断的可靠性。A novel fault diagnostic method based on the combination of support vector data description （SVDD） and D-S evidence theory is proposed. Firstly SVDD algorithm is used to judge the data from individual sensor. Then diagnostic results of all sensors are fused by applying D-S evidence theory, and the fault diagnosis of engine is finally completed. The test results show that the method proposed can fully utilize the redundant and complementary information of all signal sources, and hence reduce the uncertainty and enhance the reliability of fault diagnosis.

关 键 词：发动机 故障诊断 支持向量数据描述 D—S证据理论 信息融合 

分 类 号：U472.43[机械工程—车辆工程]