A Bearing Fault Diagnosis Method based on Improved LSTM-cascade CatBoost

Abstract

Bearing faults are widely concerned in the field of fault diagnosis, it has numerous excellent detection algorithms currently. In this paper, a new model called LSTM-Cascade CatBoost is applied, which can directly classify bearing vibration signals in the case of multiple granularities and high dimensions without signal processing. The model is based on gcForest, which can automatically adjust its complexity to the size of the dataset and it uses LSTM to improve its feature extraction ability. CatBoost is used as the base classifier of cascade forest to improve classification accuracy. Experimental results show that this model has high accuracy in CWRU and XJTU-SY datasets. Besides, it not only demonstrates that the feature extraction ability of LSTM is significantly better than that of multi-grained scanning, but CatBoost as a base classifier can further improve the accuracy of a cascade forest.