Modified particle swarm optimization algorithm to solve location problems on urban transportation networks (Case study: Locating traffic police kiosks)

Nowadays, traffic congestion is a big problem in metropolises all around the world. Traffic problems rise with the rise of population and slow growth of urban transportation systems. Car accidents or population concentration in particular places due to urban events can cause traffic congestions. Such traffic problems require the direct involvement of the traffic police, and it is urgent for them to be present at the scene as soon as possible. Due to the shortage of space, constructing traffic police centers in all areas is not possible. As a result, building traffic police kiosks with limited number of personnel and small cabins is a solution to solve this problem. Finding suitable places to build kiosks is a location optimization problem that can be solved by geospatial analyses. Artificial intelligent algorithms are suitable approaches to solve such problems. Particle Swarm Optimization (PSO) algorithm proved to be a fast and exact algorithm in solving continuous space problems. However, this algorithm cannot be used for discrete space problems without any modifications. In this paper, we modified PSO to solve problems in combinatorial space. Crossover and mutation operators from Genetic Algorithm were used to modify the behavior of particles. After conducting experiments on a part of Tehran’s transportation network, results were compared to the results of Artificial Bee Colony algorithm. In experiments with 2 and 4 kiosks, both algorithms are performing the same in accuracy, stability, convergence trend, and computation time. But in experiments with 10 kiosks on a bigger environment, results are in favor of the modified PSO algorithm in obtaining the optimum value; stability and better distribution in the area of interest. Results indicate that the proposed algorithm, is capable of solving combinatorial problems in a fast and accurate manner.