Data-Driven Fault Diagnosis for Rolling Bearing Based on DIT-FFT and XGBoost

Abstract

The rolling bearing is an extremely important basic mechanical device. The diagnosis of its fault play an important role in the safe and stable operation of the mechanical system. This study proposed an approach, based on the Fast Fourier Transform (FFT) with Decimation-In-Time (DIT) and XGBoost algorithm, to identify the fault type of bearing quickly and accurately. Firstly, the original vibration signal of rolling bearing was transformed by DIT-FFT and divided into the training set and test set. Next, the training set was used to train the fault diagnosis XGBoost model, and the test set was used to validate the well-trained XGBoost model. Finally, the proposed approach was compared with some common methods. It is demonstrated that the proposed approach is able to diagnose and identify the fault type of bearing quickly with almost 99% accuracy. It is more accurate than Machine Learning (89.88%), Ensemble Learning (93.25%), and Deep Learning (95%). This approach is suitable for the fault diagnosis of rolling bearing.