A GIS-based framework for bus network optimization using genetic algorithm

Abstract

A bus network needs to be dynamically adjusted to cope with increasing mobility in large cities. Although urban bus network optimization has been explored from many different perspectives, it remains a challenging problem. This article presents a framework of bus network optimization based on Geographical Information Systems (GIS) and genetic algorithms (GA). Given land use and population distribution, the amount of trip generation and attraction at bus stops can be estimated with accessibility models. With these demands, stops are employed in a k-shortest path algorithm to create candidate bus routes between each terminal pairs. A candidate route (CR) is valid only if it meets the criteria of route length, balance of generation and attraction, and minimum number of stops. Bus network individual and fitness function are carefully designed to meet the general requirements of GA. By applying different parameter values in CR generation, four scenarios are created and compared.