Safety Prediction and Conflict Prevention for High-speed Train Operations in the Railway Network

Abstract

The architecture of safety prediction and conflict prevention is proposed for high-speed train operations in the railway network. Train movement prediction models are employed to predict the potential conflicts based on the interval numbers. The dependent parameters such as accelerations and decelerations are attained through off-line statistics of historical data and on-line feedback adjustments of the weights of multiple models. The predictive movement authorities are formulated in the moving prediction horizons to guarantee the train operation safety, different from the current reactive movement authorities. The energy-saving planning is performed for the scheduled trains. The measure of safety level is proposed using various braking distances. The evidence theory is utilized to deal with the inconsistent information flow and assign the belief degrees of prediction models and operation conditions. The discernment of inconsistent information flow and equipment failure is implemented centering on the main clue of train movements.