Research Of Aero-engine Gas Path Fault Diagnosis Based On Neural Networks And Synergetic Theory

There are many different aero-engine failure types, and of which the gas path system failure takes a large proportion, thus the aero-engine gas path system fault diagnosis research has an important significance for reliable operation of the aircraft. The diagnosing method based on similar fault types was researched:(1) Using SOM to diagnose and classify aero-engine faults preliminarily, and analyzing diagnosis results to obtain the indistinguishable similar failures.(2) SVM was used to further verify above similar failures, firstly, choosing appropriate kernel function and optimized the parameters, and then training the training set to acquire accurate SVM model, finally testing the samples and the results show that this diagnosing model can get higher accuracy, and meanwhile the acquired similar fault types are same with above.(3) To further distinguish these similar failures, synergetic theory was adopted. Firstly,constructing synergetic neural networks of these similar fault types based on synergetic pattern recognition principle, and using the order parameters evolution to distinguish these similar types, the simulation results show that synergetic pattern recognition method can distinguish these similar types and it can further improve the gas system fault diagnosis accuracy.(4) The aero-engine gas path fault diagnosis system software was designed based on Matlab, and the experimental results show that for each test sample this software can be more accurately shows its corresponding fault type.