With the deepening of the networked operation of urban rail transit, the signal system, as the "nerve center" of driving safety, has a direct impact on operational efficiency and safety in terms of the intelligence level of fault diagnosis and early warning. In response to the shortcomings of traditional methods in extracting complex temporal features and fusing multimodal data, this paper proposes an intelligent diagnosis and warning model that integrates multi-source information and deep learning. Firstly, by constructing a signal system knowledge graph, the device topology relationship and fault propagation logic are integrated into the prior knowledge of the model; Secondly, a hybrid deep learning architecture combining Convolutional Neural Network (CNN) and Bidirectional Long Short Term Memory Network (BiLSTM) was designed. CNN was used to extract local spatial features of sensor data such as current and voltage, while BiLSTM was used to capture long-term temporal dependencies of fault evolution; Finally, an attention mechanism is introduced to weight the key fault features and enhance the sensitivity of the model to weak fault signals. The experimental results show that the accuracy of fault diagnosis for key equipment such as switch machines and track circuits reached 98.47% and 96.09%, respectively, which is about 5% and 15% higher than the traditional LSTM model and Random Forest (RF) model, respectively. The warning time was advanced by about 30 minutes, effectively achieving the transition from "passive maintenance" to "active warning".
Wei Wang (Thu,) studied this question.