AGENT-BASED MODEL FOR ASSESSING TECHNIQUES TO REDUCE KNOCK-ON URBAN TRAIN DELAY

Abstract

 The decreasing travel time reliability of urban railway services in the Tokyo Metropolitan Area is the result of train delays during the daily morning rush hour period. A proposed multi-agent simulation model is developed to analyze knock-on train delay over a 48.3-km track length from the suburban stations of the Tokyu Den-en-toshi Line to the downtown stations of the Tokyo Metro Line. The model consists of a train traffic simulator and a passenger flow simulator working in parallel. Compared with the actual train delay, the errors in the delay estimation of individual trains using the proposed simulator are within the allowable limits. Some strategies for reducing delay are assessed using the results of this forecasting model.