Study of capacitated vehicle routing problem based on particle swarm optimization

Abstract

Vehicle Routing Problem (VRP) is one of the common problems that happen in human life. There are many applications of VRP such as garbage disposal, mail delivery, school bus routing, airline schedule and many more. The main objective of VRP is to minimize the distance of the route starting from a depot, serves all of customers demand, and return back to depot. VRP is one of the optimization problems that belong to NP- hard (Non-deterministic Polynomial-time hard) problem and difficult to solve. VRP has also becomes one of the important topic to discuss and analyze. There are many types of VRP; this research is focusing on capacitated VRP (CVRP). CVRP is defined as the problem of determining optimal routes to be used by vehicles starting from one or more depots to serve all customers’ demand, observing some constraints. Particle Swarm Optimization (PSO) method will be used to solve the VRP problems because there are lots of advantages of PSO. PSO is a population based stochastic optimization technique, inspired by social behavior of bird flocking or fish schooling. The experiment has been done to test this algorithm. Three variants of PSO have been used which are PSO with inertia weight, PSO without inertia weight, and PSO with constriction factor. The results show that the PSO with inertia weight strategy which include PSO with inertia weight and PSO with constriction factor have the best total distance. It can be concluded that PSO with inertia weight strategies have better performance because they take less iteration to arrive at the optimum value. The second comparison also showed that small range of inertia weight has the best total distance.