Research on Fault Diagnosis of Rotating Machinery Based on Multi-Scale Causal Dilation Convolution

Diagnosing faults in the rolling bearing of rotating machinery equipment is vital to ensuring the stability and safety of industrial production. Traditional convolution-based methods are limited by insufficient extraction of global temporal features. Therefore, this paper proposes a fault diagnosis model for rolling bearings based on multi-scale causal dilated convolution LSTM. This model adopts causal dilated convolution to expand the receptive field while keeping the number of parameters constant, effectively extracting global temporal information from signals, and introduces a multi-scale processing strategy to capture features at different time scales. Furthermore, the LSTM network is utilized to extract temporal dependencies on time series data, enhancing the diagnostic capability of the model for complex vibration signals. Experimental results on the HIT bearing fault dataset show that the proposed model has a significant advantage in fault diagnosis accuracy with an accuracy rate of 98.44%, validating its effectiveness and superiority. Through ablation experiments and comparative experiments, it further confirms the rationality of the model and provides a referable approach for fault diagnosis in rotating machinery equipment.