Research on Rotor Fault Diagnosis Method Based on Multi-Sensor Fusion

Given the extremely harsh working environment of aero engines, relying on data from a single sensor is often insufficient due to its susceptibility to its own quality issues and environmental factors, making it challenging to comprehensively monitor the operational status of rotor components. To address this issue, this paper proposes a rotor fault diagnosis method based on a deep residual shrinkage network (DRSN_CS) utilizing multi-sensor data fusion. First, the method treats signals from sensors placed at different locations as multi-channel inputs to the fault diagnosis model. Inspired by RGB image processing techniques, these multi-channel signals are constructed into an n-dimensional channel signal set, effectively addressing the challenge of feature-level multi-channel data fusion. Finally, by employing the deep residual shrinkage network, the method effectively filters out noise from fault signals, thereby enhancing the capability to learn the features of the original vibration signals. Experimental results on a self-developed rotor fault simulation dataset demonstrate that the proposed method achieves superior fault diagnosis accuracy, meeting the demands of multi-sensor data fusion for fault diagnosis.