基于融合技术的电子产品PHM系统研究  

Study of Electronic PHM System Based on Information Fusion

作　　者：刘倩倩[1] 王红霞[1] 尹明[1] 

机构地区：[1]海军工程大学电子工程学院,武汉430033

出　　处：《计算机测量与控制》2014年第11期3454-3456,共3页Computer Measurement &Control

基　　金：国防预研基金(914A27020113JB11393)

摘　　要：针对故障预测与健康管理(PHM)的复杂性,对融合技术进行了深入的分析和研究,提出了一种基于神经网络的特征融合方法,融合结果最大限度的给出决策分析所需要的特征信息,提高了故障诊断的可靠性;在预测方法上,提出了一种基于故障预兆监控与推理和失效物理(PoF)模型方法相融合的预测方法,充分利用了每种预测方法的优势,故障预兆监控与推理的方法能够提供故障诊断功能,而失效物理(PoF)模型的方法则有助于确定故障根源,融合预测方法能更加及时准确的预测故障;融合技术丰富了PHM的理论体系,提高了其实用价值。Concerned with the complexity of prognostic and health management,this paper proposed an algorithm of fusing data based on neural network.The proposed method has a high accuracy,which contains plentiful information for fault diagnosis.Two theory methods that combined the method of fault prediction based on data and the method of Physics-of-Failure models,which take full advantage of each method.On the one hand,the method of fault prediction based on data can find out whether the fault is existed.On the other hand,the method of Physics-of- Failure models can provide us the root of fault.The efficiency and accuracy of fault prediction were both quite higher based on the approach of fusion prognostics,which enrich the theory system of prognostic and health management and strengthen the practical value.

关 键 词：故障预测与健康管理 特征融合 预测方法融合 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]