Residual Gated Dynamic Sparse Network for Gearbox Fault Diagnosis Using Multi-Sensor Data

This article proposed a new multi-sensor fusion fault diagnosis method for gearbox, namely residual gated dynamic sparse network, to improve multi-sensor feature learning and fusion ability. Considering that the fault sensitivity of the sensor varies with mounted location and complex transfer path modulation causes information from multi-sensor redundant, the lightweight channel attention unit was designed to strengthen the feature extraction ability of the network. The developed gated dynamic sparse unit was inserted into the deep architecture to eliminate ineffective components caused by high noise interference. Besides, the loss function was improved with multiple activation criteria to enhance convergence ability. The results of experiments and the engineering application showed that the proposed method was more effective than other methods under varying degrees of noise interference.