In this paper a new genetic algorithm is developed for solving capacitated vehicle routing problem (CVRP) in situations where demand is unknown till the beginning of the trip. In these situations it is not possible normal metaheuristics due to time constraints. The new method proposed uses a new genetic algorithm based on modified sweep algorithm that produces a solution with the least number of vehicles, in a relatively short amount of time. The objective of having least number of vehicles is achieved by loading the vehicles nearly to their full capacity, by skipping some of the customers. The reduction in processing time is achieved by restricting the number of chromosomes to just one. This method is tested on 3 sets of standard benchmark instances (A, M, and G) found in the literature. The results are compared with the results from normal metaheuristic method which produces reasonably accurate results. The results indicate that whenever the number of customers and number of vehicles are large the new genetic algorithm provides a much better solution in terms of the CPU time without much increase in total distance traveled. If time permits the output from this method can be further improved by using normal established metaheuristics to get better solution
Development Of A New Genetic Algorithm For Solving Capacitated Vehicle Routing Problem In Short Time
