Neurosimulation Modeling of a Scheduled Bus Route

Abstract

In a densely built-up urban society, operators of public bus services are faced with the recurrent problem of providing timely and reliable service. While they have no control over dynamically changing extraneous factors (such as passenger loads or road conditions) that may suddenly degrade the quality of the service provided, it is nonetheless desirable for management to study the extent to which these factors affect their business, and what measures, if any, can be adopted to neutralize them. This paper discusses how a simulation model of a bus route, embellished by a neural network, was created to model the historical pattern of the inputs (namely, passenger loads and road conditions) that affect the overall scheduled terminus-to-terminus time. Thus, in a case study of a bus route running from a suburb to the city center, it was found that the neurosimulation model could predict the cumulative terminus-to-terminus times better than a conventional simulation model could. A software module, embedded into the neurosimulation model for the purposes of speed regulation, was able to minimize the deviation of the bus service from schedule. When intentional delays were further introduced into the bus route, it was discovered that the speed regulator was more effective the longer the delay, and the further the bus traveled into the bus route. There is potential in applying neural computing in a dynamic bus scheduling problem such as the one discussed here.