Fault diagnosis on the bearing of traction motor in high-speed trains based on deep learning

Abstract

The bearing health in the traction motor is the prerequisite and guarantee for the safe operation of high-speed trains. The vibration signals of the bearing in traction motor feature high nonlinearity, non-stationarity, and background noise. Therefore, the features of the vibration signals are diverse and complex, making it hard to diagnose the faults of the bearing effectively and accurately. To overcome the difficulty, this paper puts forward a novel fault diagnosis method for the bearing of traction motor in high speed trains based on discrete wavelet transform (DWT) and improved deep belief network (DBN). Firstly, the vibration signals were extracted from various faulty bearings, and used to generate a two-dimensional (2D) time–frequency map. Then, the time–frequency map was preprocessed, and subjected to the deep learning (DL) by the improved DBN, aiming to identify the correlation between fault features and fault types. In this way, the fault state of the bearing in the traction motor was diagnosed and identified in a semi-supervised manner. To verify its effectiveness, the proposed method was applied to diagnose the bearing faults of traction motor in high-speed trains through comparative experiments. The results show that our method achieved better diagnosis accuracy than contrastive methods like backpropagation neural network (BPNN) and support vector machine (SVM).