Fault Diagnosis Of Overvoltage Protection Unit

Overvoltage protection unit (OPU) is an important equipment of aircraft power system， itsreliability has important implications for aircraft power systems and airborne equipment. Once themalfunction may cause irreparable damage, So OPU fault diagnosis has important and realisticsignificance. However, due to the special nature of aircraft and its complex structure, the measurementpoints of OPU system are not enough, and this makes us can not get a complete set of data samples. Inconclusion, the OPU fault diagnosis is about complex system fault diagnosis with uncertainty.To solve the uncertainty of OPU fault diagnosis, we proposed fault diagnosis method based onBayesian network. Firstly, we make a brief principle analysis principle of OPU, and then accordingto the functional structure block diagram and some problems of fault diagnosis, we select theappropriate variables to establish a model of the fault diagnosis of based on Bayesian network. Wecomplete the model parameters learning and structure learning under the complete and incompletedata. Secondly, input the discretized signal of test point in the Bayesian network which has completedparameter learning and structure learning to do the Fault Diagnosis experiment of OPU, Faultdiagnosis of the experiment demonstrate the effectiveness of Bayesian networks on OPU faultdiagnosis.Finally, we establish a failure prediction model of OPU based on Dynamic Bayesian network,and complete the parameters learning. After this we input the fault symptoms which is get from thefour consecutive time slice to Dynamic Bayesian network, and get the fault symptoms predictedprobability of other test point as well as the failure predicted probability. Simulation studies haveshown that the Dynamic Bayesian network good for complex system status and fault prediction...