Transit Network Frequencies-Setting Problem Solved Using a New Multi-Objective Global-Best Harmony Search Algorithm and Discrete Event Simulation

Abstract

The rise of Bus Rapid Transit Systems (BRTS) in urban centers involves complex problems of design and scheduling including the scheduling of route intervals across the bus network. The difficulty stems from the fact that transport systems keep to established routes and must set frequencies for each route to minimize costs (measured in terms of transport capacity wasted) and maximize the quality of service (minimizing the total time of users in the system). All this depends on the maximum number of buses available in the system. In an effort to find an alternative solution to the Transit Network Frequencies Setting Problem (TNFSP) on BRTS, this paper proposes using Multi-Objective Global Best Harmony Search (MOGBHS), a multi-objective heuristic algorithm based on three main components: (1) Global-Best Harmony Search, as a heuristic optimization strategy, (2) Non-Dominated Sorting, as a multi-objective optimization strategy, and (3) Discrete Event Simulation, for obtaining quality measures in the solutions found. To test the proposed approach, a simulation model was implemented for Megabus, a BRTS located in Pereira (Colombia), for which the frequency of the buses assigned to routes previously defined in the system was optimized so that operating costs were reduced to a minimum, while user satisfaction was maximized. The MOGBHS algorithm was compared with NSGA-II. It was concluded that MOGBHS outperformed NSGA-II in the number of optimal solutions found (Pareto front points), from 175% in 3,000 fitness function evaluations to 488% in 27,000 evaluations.