Sensitivity analysis of railway balise information transmission module with both degradation and model uncertainty

Balises are key safety-critical components in many railway systems. However, as the usage time of balises increases, the performance of the physical devices within a balise information transmitting module will gradually deteriorate. This deterioration leads to a decline in the output performance of a balise information transmitting circuit. It is crucial to identify the key physical devices that contribute to the degradation of the balise information transmitting circuit's output performance, as this can provide a foundation for assessing the health status and predicting the residual life of a balise. Consequently, this paper proposes a sensitivity analysis method for balise information transmitting module circuits that considers performance degradation and model uncertainty. First, a surrogate model for balise information transmitting circuits is established based on a deep neural network that integrates both physical knowledge and experimental data. Random sampling-high-dimensional model representation method is employed to analyse the sensitivity of the degraded input and scalar output of the balise information transmitting circuit surrogate model, thereby clarifying the key physical devices that impact performance degradation. Finally, the model uncertainty introduced by the establishment of the balise information transmitting circuit surrogate model is analysed. The experimental results indicate that the proposed method effectively identifies the key physical devices that influence the degradation of output performance in the balise information transmitting circuit.