Fault diagnosis of piston pump based on global attention residual shrinkage network

Objectives Aiming at the problem of insufficient feature extraction ability of traditional neural network under strong noise interference, a new global attention residual shrinkage network is proposed to realize the accurate diagnosis of piston pump faults under the complex external environment. Methods Firstly, data slicing is performed on the original signals; then, a new global feature extractor with an attention mechanism is established to extract fault-related features from the monitoring signals, while a threshold softening mechanism is introduced to reduce the influence of noise interference in the signals; then, back propagation optimization is performed on the network model to reduce the loss and improve the diagnostic performance of the model. Finally, the feature extraction results are input into the fault classifier for fault identification. Results The effectiveness of the proposed method is verified based on a piston pump fault simulation test bed. The results show that compared with other models, the established global attention residual shrinkage network model has higher diagnostic accuracy and stronger anti-interference ability. Conclusions It is shown that the proposed global attention residual shrinkage network-based fault diagnosis method for piston pumps is able to realize accurate fault diagnosis in complex and harsh environments.