A Data-based Expert System for Aero-Engine Gas Path Fault Diagnosis

Abstract

In response to the daily maintenance and repair requirements of domestic military aviation engines, a data-based expert system for aero-engine gas path fault diagnosis is proposed. The system is dedicated to the unified management and processing of flight data of various types of domestic engines. In addition to the basic functions of the general aero-engine condition monitoring system, the system has also done in-depth research on issues such as aero-engine parameter denoising, aero-engine parameter prediction, and automatic acquisition of expert knowledge base. Additionally, wavelet domain denoising, LSTM network prediction and rough set attribute reduction are investigated and applied in the proposed system. Finally, the normalized real flight data is applied to simulation verification, and good denizations and prediction results are obtained.