ResNet-based Vibration Fault Diagnosis for Aero-engine Bearings

Abstract

As a key component of the aircraft, the operation of the aeroengine is very important. Traditional fault diagnosis methods face many challenges in dealing with high-dimensional data and complex pattern recognition. The purpose of this study is to compare the fault diagnosis performance of various machine learning and deep learning algorithms in bearing vibration data. By optimizing the algorithms and parameter settings, the classification accuracy and computational efficiency of the models are improved, thus enhancing the operational reliability and maintenance efficiency of the equipment. On the one hand, the time domain and frequency domain features of vibration data are extracted, and then the PCA method is used to reduce the dimension of high-dimensional feature data, and then the model is optimized by grid search and cross validation. On the other hand, the data is directly transmitted to the neural network model for learning prediction without feature extraction. The results show that the classification accuracy of the optimized SVM model on the experimental data set reaches 97 %, and the prediction accuracy of the deep learning algorithm based on Resent-CNN on the data set reaches 100 %. The research method and results can be applied to equipment fault prediction and maintenance in industrial production, and further improve the reliability and safety of equipment operation.