Solving real-world vehicle routing problems with time windows using virus evolution strategy

This paper proposes a new solution to the vehicle routing problem with time windows using an evolution strategy adopting viral infection. The problem belongs to the NP-hard class and is very difficult to solve within practical time limits using systematic optimization techniques. In conventional evolution strategies, a schema with a high degree-of-fitness produced in the process of evolution may not be inherited when the fitness of the individual containing the schema is low. The proposed method preserves the schema as a virus and uses it by the infection operation in successive generations. Experimental results using extended Solomon's benchmark problems with 1000 customers proved that the proposed method is superior to conventional methods in both its rates of searches and the probability of obtaining solutions. Further experiments using the map of the central part of Tokyo with 20000 intersections and real traffic data also gave that the rate of search of the proposed method is higher than that of the conventional method.