Multi-level Models And Algorithms For Evaluating Operational Reliability Of Bus Network

The operational reliability of bus transit system is an important indicator in evaluating the quality of bus services, which plays a critical role in influencing passengers'choice of mode for travel. The study on the bus operational reliability is of much significance for capturing the operational characteristics of buses and improving the operational efficiency of the bus transit system. At present, however, the study on the bus operational reliability has suffered a number of shortcomings such as almost from passenger's view, only focusing on problems with known bus schedule, as well as evaluation models for limited stops and pure theoretical study on bus routes. This dissertation is intended to provide a more comprehensive and clearer definition of the bus operation reliability after synthesizing development course of bus system reliability and existing evaluation indicators which reflect bus operational reliability, models based on several indicators and bus operational characteristic. It set up multi-level bus operational reliability models and algorithms, which lie on stations, bus lines and bus networks but independent on bus schedules.The main contributions of the dissertation include:1. The distribution fitting of travel times between adjacent stations is developed, resulting that the bus travel time between stops follows the lognormal distribution. Then, the travel time thresholds for different bus service levels are determined using the mathematical statistics methods. Further, a model for the expected travel time reliability is developed on the basis of the accepted bus service levels.2. A thorough analysis of the stop time composition at bus stops is conducted, which contains intervals between the arriving and door open times, the door close and open timse, and the door close and departing times. The distributions of these three intervals are examined. It is shown that the intervals between the arriving time and door open time follows the Poisson distribution, the average time for the passenger to get on or off the bus follows the lognormal distribution, and the number of passengers getting on or off the bus follows the negative binomial distribution. The range of time thresholds under different bus service levels is determined using the cluster analysis and mathematical statistics methods. Finally, the stop time reliability models are developed based on the expected stop times.3. Two indicators are selected for evaluating the meso-level bus operational reliability, which are the sectional travel time reliability and the arriving headway reliability. Then, a meso-level bus operational reliability model is developed using the inverse of the coefficient of variations of these two indicators as the weights. The sectional travel time is the sume of the stop time at stations and the travel time between adjacent stations. So the sectional travel time reliability can be calculated by the probability when the travel time is equal to or less than the expected travel time threshold. The arriving headway reliability only depends on the current state, the stop time reliability and the travel time reliability of later state, which is the characteristic of Markov theory. Finally, the arrival headway reliability can be calculated using the C-K state transition equation.4. A macro-level network and regional bus operational reliability is established using the sectional bus operational reliability as the basis, based on the role that each secion plays in the network, and using the boarding and alighting passengers as the weights.