A Hybrid Intelligent Method for Rolling Bearing Fault Diagnosis Integrated with Expert Knowledge and Deep Learning

The rolling bearing is essential for the rotating machinery and can be easily damaged in the real working conditions. It is very important to monitor the health status of rolling bearings. Aiming at this problem, fault diagnosis based on deep learning at present is popular, which automatically extracts features from raw data. However, the accuracy of fault diagnosis based on deep learning is dependent mostly on the quantity of data. In the real industries, a large amount of data may not be available, which largely deteriorates the performance of deep learning. To solve this problem, it is promising to exploit the features extracted with the expert knowledge for relaxing the limitations of deep learning. In this paper, a new hybrid intelligent method for rolling fault diagnosis is proposed, which is integrated with deep convolutional neural network and the expert knowledge. The features extracted with expert knowledge are used to improve the feature learning effect and efficiency of deep learning. The experiments on the Case Western Reserve University (CWRU) bearing data validate the effectiveness of the proposed hybrid rolling bearing fault diagnosis method.