Optimal Locating of Urban Bus Stations by Using a Competitive Recurrent Neural Network and Genetic Algorithm

Provide an appropriate method to locate optimal urban bus stations in transportationnetwork is the principle aim of this study. In this study after determining thepoints of demand, identify candidate locations and locate p facilities with minimalcost in the network.In this paper a new formulation for the p-median problembased on two types of decision variables and with only linear equality constraintsis presented, where is the number of demand points or customers and is thenumber of facilities. Then for solving the problem, is proposed hybrid algorithmconsists of genetic algorithm and competitive recurrent neural network.In this method instead of the reproduction operator of genetic algorithm, competitiverecurrent neural network will use. Competitive recurrent neural network structureconsists of location layer related to decision variables and allocation layer relatedto decision variables. The process units constitute disjoint groups, where only oneprocess unit per group is active at the same time. In addition, the network energyfunction that is equivalent fitness function of genetic algorithm always decreases orremains constant according to the dynamical rule proposed. The effectiveness andefficiency of this algorithm for standard problems is analyzed. The results indicatethat the proposed algorithm generates fine quality and acceptable solutions.