Bus Network Optimization Through Time-Dependent Hybrid Algorithm

This paper focuses on a new hybrid technique that combines a genetic algorithm with simulation to solve the bus-network scheduling problem (BNSP). The BNSP has several factors that complicate both the problem formulation and the selection of efficient algorithms for its resolution. This problem is challenging because not only the BNSP is NP-complete, but also the existing methods fail to contemplate environment dependent dynamic variables. The hybrid algorithm proposed in this article comprises two stages: a modified GRASP (greedy randomized adaptive search procedures) as an initialization method, and the genetic algorithm with simulation to find the values of the environment- dependent dynamic variables. The final goal consisted in designing a meta-heuristic technique that yields an adequate scheduling to solve this general problem. The BNSP, chosen as case study, satisfies both the demand and the offer of transport. The method was applied to a solution of experimental examples with good results.