Objective Signaling system is the core for ensuring train safe operation, and the lifespan prediction of its equipment is of great significance for operation and maintenance decision-making. Traditional prediction methods are unable to handle the nonlinearity, high-dimensional features, and uncertainty in the equipment degradation process, and their prediction accuracy is limited. Therefore, it is necessary to conduct in-depth research on the prediction of the remaining useful life of signaling equipment.

Method The theoretical principles of RBM (restricted Boltzmann machine) and DBN (deep belief network) are briefly introduced, and a DBN-based lifespan prediction model for signaling equipment is established. A method for calculating equipment health indicator values is proposed. Based on the four-state classification theory, the entire lifecycle of equipment is divided into four states so as to achieve intuitive quantification of the equipment degradation degree. Finally, taking a single track circuit as a test case, the effectiveness of the established model is verified.

Result & Conclusion  Using the proposed model to predict the equipment remaining useful lifespan, the prediction accuracy reaches 94.7%. Compared with traditional prediction methods, the proposed model has better adaptability and accuracy with strong generalization capability, and therefore can be applied to life prediction of various signaling equipment, such as turnouts, track circuits, and signals.