基于改进D-S决策融合的航电设备故障诊断  

Avionics equipment fault diagnosis based on improved Dempster-Shafe decision fusion method

作　　者：牛强军[1] 黄家成[1] 胡秀洁[2] 宋家友[2] 

机构地区：[1]空军第一航空学院航空电子工程系,河南信阳464000 [2]郑州大学信息工程学院,河南郑州450001

出　　处：《计算机工程与设计》2015年第8期2255-2259,共5页Computer Engineering and Design

基　　金：军内计划基金项目(KJ2012255)

摘　　要：为提高故障诊断的正确率和精确度,采用信息融合技术对航空电子设备故障诊断进行研究。利用多个子模糊神经网络进行故障局部诊断,获得彼此独立的证据,对各个证据应用Dempster-Shafe证据理论进行决策融合。针对D-S证据理论无法解决高冲突现象的问题,提出一种改进方法,计算各个证据对不同模式的正确率,将其作为可靠性系数加入概率赋值转化过程中,利用证据间的冲突系数进行分类,对系数高的证据,采用贴近度方法判定其可信度,利用可信度修改证据权重。通过在电台上的研究比较验证了该方法的有效性,避免了误诊现象。To improve the accuracy and precision of fault diagnosis, the information fusion technology was used to study aviation electronic equipment fault diagnosis. The sub fuzzy neural network was used to do primary diagnosis and obtain independent evi- dence, the Dempster-Shafer evidence theory was considered as the diagnosis conclusions. An improved information method based on D-S theory was proposed. The accuracy of all the evidence for different patterns was calculated to join the probability of as- signment conversion process. The conflict degree between evidences in the evidence was calculated, and the credibility of each evidence was determined using a nearness method of evidence bodies. Then its nearness was used to revise the weight value of each evidence. Through comparing studies on a radio, the validity of this method is verified, and the misdiagnosis phenomenon is avoided.

关 键 词：航电设备 证据理论 故障诊断 模糊神经网络 信息融合 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]