Proceeding from the sharp contrast between the actual and planned station arrival-departure times in the dispatching database, this paper analyzes the shortages of the existing scale of train working diagram (TWD), and proposes a new TWD scale optimization method for high-speed railway based on data mining theory, aiming to improve the accuracy and intelligence of high-speed railway dispatching. The conclusions are drawn as follows: (1) through the analysis of the operation data on a high-speed railway line under a Chinese railway bureau, it is learned that the actual RTS is 0.326min shorter than the planned RTS in the TWD scale. The time gap helps to elevate capacity of a 30-section long distance high-speed railway (train interval: 5min; train type: 16-car; attendance rate: 70%) by over 1million person-time. (2) The data analysis shows that the difference between the actual RTS and the planned RTS is highly consistent across different sections. The max standard deviation is 0.535min, and the min is 0.267min. (3) This paper presents a TWD scale modification method based on probability, and gives the results under the probabilities of 50%, 69.15% and 84.13%, respectively.
Keywords: high-speed railway; scale of Train Working Diagram (TWD); data mining; probability