基于模型的系统级故障诊断与预测推理技术  

A model based system level fault diagnosis and prognosis technology

作　　者：张士刚[1] 刘颖[1] 宋立军[1] 胡政[1] 杨拥民[1] 

机构地区：[1]国防科技大学机电工程与自动化学院装备综合保障技术重点实验室,湖南长沙410073

出　　处：《国防科技》2016年第2期9-13,共5页National Defense Technology

基　　金：国家自然科学基金资助项目(61503398)

摘　　要：针对装备状态监控和故障预测推理的需求,提出不同的诊断推理模型。首先,提出了基于二值相关性模型的快速推理算法,采用Vx Works平台对方法进行了验证评估。然后,针对应用过程中存在的时延、故障耦合等现象,提出了动态诊断模型。在此基础上,提出了考虑故障预测的诊断和预测一体化诊断推理模型,对推理运算方法进行了阐述,开发了仿真验证软件。以某卫星电源系统为例进行了验证,结果显示模型具有故障诊断和预测能力,能实现故障诊断和预测的一体化推理。In this paper, several diagnostic models are discussed according to different needs of equipment status monitoring and fault prognosis. Firstly, a method based on binary dependency matrix is proposed and tested on a system with Vx Works. Then, for the problem of test delay and coupled faults, a dynamic diagnostic model is developed. After that, an integrated diagnostic model and inference algorithm for fault prediction is proposed. A simulation software is developed based on the method. The application result on a satellite shows that the model has the ability to realize unified fault diagnosis and prognosis.

关 键 词：相关性矩阵 测试性 故障诊断 故障预测 PHM 

分 类 号：E920[军事—军事装备学]