Fault Diagnosis of Vessel Motor Bearing Based on Multi-feature Fusion in Time Domain and Frequency Domain

The vibration signal is not suitable for application of induction motor bearing fault diagnosis in vessel due to vessel environment noise interference. Current signals are usually used for bearing fault detection. During the fault diagnosis process, the bearing fault information is very weak in the sampling current signal, leading to relatively low diagnostic accuracy. To solve this problem, a multi-feature fusion feature (Time domain and Frequency domain-Fusion Feature, TFFF) is proposed in this paper. The information of TFFF feature comes from 16 time domain features and 4 frequency domain features. By means of BP neural network, Pearson correlation analysis and entropy power fusion optimization, the TFFF feature of relatively simple structures with 20 feature fusion information is obtained. With the help of the random forest classifier, tests of motor bearing outer race fault diagnosis are made. The experimental results show that the problem of motor bearing fault diagnostic accuracy by current signal can be solved effectively with the assistance of the proposed fusion feature TFFF.