Fault Diagnostic Of Aircraft Electronic Systems Based On Rough Set And Neural Network

As a main transportation, civil aviation develop faster and faster. The modern civil aviation is expanding by leaps and bounds, and airplane model get aging gradually, various malfunction may happened in the process of flying inevitably. In order to ensure flight safety and passengers' interests, the aircraft fault diagnosis, exclusion, maintenance become an essential task.The plane is a complex system, the fault type is also complicated. In the process of aircraft operations, some failure may have disastrous consequences. How to diagnose the faults of aircraft accurately and timely is very important to ensure the safely aircraft operation.The accuracy and rapidity of aircraft fault detection has become a necessary research direction as a result of the inefficiency and sharp increase of aircraft maintenance. At present,it has come forth a research trend to use artificial intelligence to improve maintenance efficiency.Rough Set is a method of describing vagueness and uncertainty. It has powerful data processing ability and easy to implement. In This paper we collect the airlines B737 AFS fault information and use rough set to analysis, then find the key condition attributes and get the simple rules of fault diagnosis through attribute reduction and value reduction. Finally the problem caused by indecisive or overmuch malfunction reasons is resolved and this method has practical significance for improving the efficiency of aircraft troubleshooting.This paper, use the rough set to deal with the original data, delete redundant attributes and obtain the core content of the data. Processed data as input of neural network, reduce the sample size, simplify the process of neural network training and overcome the problems caused by too large simples. So as to realize the organic combination of neural network and rough set theory, make the training of neural network more quickly and precisely.