A Fault Diagnosis Method Based on GWO-DELM for Aeroengine Gas Path

Abstract

Aiming at the problem of complex working mechanism of aeroengine gas path system and difficulty in effective fault diagnosis in actual work, a new fault diagnosis method of aeroengine gas path based on Gray Wolf Optimization Deep Extreme Learning Machine (GWO-DELM) is proposed. Firstly, analyze a large amount of monitoring data of a certain type of aeroengine gas path components, and sort out the health data and fault sample data sets. Secondly, create the DELM fault diagnosis model by the health data and fault sample data set of the aeroengine gas circuit system. To reduce the influence of artificially setting network parameters on the diagnosis results, the Gray Wolf Optimization algorithm (GWO) is used to optimize the DELM network parameters, and the optimal DELM fault diagnosis model GWO-DELM is created. Finally, the GWO-DELM fault diagnosis model is used to study the fault diagnosis verification technology of the aeroengine air circuit system, and the diagnosis results of the ELM, DELM and Multilayer Kernel Extreme Learning Machine (ML-KELM) fault diagnosis models are compared. The result shows that the fault diagnosis accuracy of the proposed GWO-DELM fault diagnosis model is 96.0%, which is significantly higher than that of the ELM model of 88.0%, the DELM model of 92.0% and the ML-KELM model of 94.0%, the effectiveness of the proposed method is verified, and it has a good application prospect.