TFFC-RNN:A New RNN Based Approach for Bearing and Misalignment Compound Fault

This paper proposes a fault diagnosis method based on the speed signal of the servo motor for the bearing and misalignment compound fault. First, it explores the change of motor speed caused by compound fault excitation and analyzes the theoretical feasibility of the speed method to achieve compound fault diagnosis. Experiments show that the detection of weak fault signals such as bearings in compound faults is susceptible to interference from mismatched installation faults, which will make the traditional diagnosis algorithms invalid. The pre-processed speed signal and the signal obtained through fast fourier transform(fft) are passed through the recurrent neural network respectively, and the input time-domain features and frequency-domain features are fused together as the basis for fault classification. This Time-Frequency Feature Compound-RNN model (TFFC-RNN) can perfectly classify bearing faults and normal signals under the interference of misalignment faults. Finally, the research and experimental verification of which recurrent neural network to use as the best composition of the model is carried out.