Abstract

The primary objective of any transit system is to provide better level of service to its passengers. One of the good measures of level of service is the waiting time of passengers during their journey. The waiting time can be minimized through proper scheduling of the bus transit system. A good schedule minimizes both the overall transfer time (TT) of passengers transferring between different routes and the initial waiting time (IWT) of the passengers waiting to board a bus at their point of origin. In this study, a mathematical programming (MP) formulation of the scheduling problem for a transit system network is presented. The MP problem is large and non-linear in terms of the decision variables which are real as well as integers. This makes it difficult for traditional methods to solve the problem. Hence, Genetic algorithms (GAs) which are powerful search and optimization methods are used to solve the problem. Function based declarations and coding of variables in GAs allow an efficient reformulation of the original problem. The modified problem is computationally much simpler than the original problem. The results from a number of test problems show that the GAs are able to find optimal schedules with minimal computational resource.