Research On Fault Diagnosis Of Aero-engine Gas Path Based On Data Drive

The aero-engine surge will bring a great potential rise for the flight safety and overall engine performance.Moreover,rotating stall is considered as the symptom of the aero-engine surge.Therefore,the precise forecasting of aero-engine rotating stall development process under complex working conditions is an effective method for the detection and diagnosis of aero-engine surge failure.In order to avoid the roughness result of the binary classification and the difficulty of feature extraction under high dimensional data in traditional machine learning methods,the SDAE combined with RVM is developed to provide an accurate stall fault prediction and a practical warning window.Firstly,the SDAE is implemented to extract the implicit feature beneath the high dimensional data.Then,the RVM is carried out to calculate the trigger probability under the reconstructed vector input.Finally,the stall alert window is identified according to the surge probability.In the end,the result of SDAE-RVM,SVM,and softmax algorithm is compared and analysis with the on service aero-engine data.In this paper,the real flight data of aero-engine provided by a research institute affiliated to AVIC is used for simulation and compared with other algorithms.The result demonstrates that the SDAE-RVM approach is an effective method to detect the rotation stall state and can provide a valid warning window.The paper also introduces the three software modules developed in the graduate stage-the aviation performance parameter prediction module,the gas path fault diagnosis module and the bearing fault diagnosis module.The main functions and operation procedures of the software module are explained.