Research on electro-mechanical actuator fault diagnosis based on ensemble learning method

With the rapid development of the aviation industry, people have increasingly higher requirements for the performance of aircraft. Therefore, effective health management of the airborne electro-mechanical actuator (EMA) is particularly critical. Aiming at the problem of aircraft health management, this paper first establishes the simulation model of EMA, and chooses the three-phase current as the characteristic quantity of subsequent fault diagnosis through the analysis of the model. Then an EMAs fault diagnosis framework based on ensemble learning method is proposed. The study compares the advantages and disadvantages of different ensemble learning strategies and proposes a fault diagnosis framework based on the Boosting ensemble learning method, which is based on XGBoost, LightGBM, and CatBoost models. Compared with popular deep learning frameworks (CNN), this method requires fewer computing resources and has stronger interpretability of the model. The test results indicate that the proposed framework has higher diagnosis accuracy compared to traditional machine learning methods and shorter training time and lower memory usage compared to deep learning methods (CNN), making it a valuable tool for engineering applications.