Resilience evaluation of train control on-board system based on multi-dimensional continuous-time Bayesian network

Abstract

The Bayesian network belongs to a commonly used model in the resilience research of Train Control on Board System (TCOBS). In the previous works, component failure is only considered to be affected by time, without showing the trend of resilience changing with time, and less attention paid to whether it can effectively recover to a usable state after failure. In this paper, we propose a multi-dimensional continuous-time Bayesian network to evaluate TCOBS resilience. Firstly, based on multi-dimensional influencing factors, the quantitative function of TCOBS component resilience is constructed. Secondly, the unit step function and impulse function are used to construct a multi-dimensional continuous-time Bayesian network, and the TCOBS resilience is derived and evaluated. A real-world case study is conducted by using the Chinese Train Control System-3 On-Board System as the background, two comparative experiments are conducted. The results show that, in addition to time factor, TCOBS resilience is also influenced by human and environmental factors. Therefore, to enhance the resilience of TCOBS, we should reduce failure rate and improve maintenance rate, and reduce the probability of human failure and environmental interference failure. The new model can find out the continuous trend of TCOBS resilience under the influence of multi-dimensional factors.