关键词: 膨胀因果卷积, 注意力机制, 模型融合, 柱塞泵, 故障诊断

Abstract: The plunger pump is the core component of the hydraulic system in railway switch machines, and its performance is directly related to the stable operation of the switch machine. To further improve the accuracy of plunger pump fault diagnosis models, this paper enhances the TCN based on dilated causal convolution by introducing BiTCN and integrating BiLSTM modules with an attention mechanism, thereby constructing the BiTCN-BiLSTM-Attention (BBA) fault diagnosis model. The paper first describes and preprocesses the collected vibration signal dataset of the plunger pump. Then, model training is conducted on the processed dataset, followed by ablation experiments and comparative experiments. The experimental results demonstrate that the BBA model outperforms traditional machine learning methods and other deep learning models. Additionally, validation experiments were conducted on the 2023 Harbin Institute of Technology aerospace bearing dataset, and the results indicate that the proposed model exhibits good robustness and generalization capability. This research expands the application of TCN in the field of fault diagnosis and provides new insights for researchers in related domains.

Key words: Dilated Causal Convolution, Attention Mechanism;Model Fusion, Plunger Pump, Fault Diagnosis