An Intelligent Modeling Method Based On Hybrid Reasoning for Vehicle Routing Problems

In this paper an intelligent modeling method based on hybrid reasoning is presented that focuses on the weakness of the manual and off-line modeling in dealing with dynamic vehicle routing problems in a practical environment. The method uses Case-Based reasoning as a framework and adopts Rule-Based reasoning to accomplish the reuse of the case modeling experience and the revision of the modeling result. As the first step of the modeling method, the multilayer and weighted nearest neighbor retrieval approach presented in the paper improved the nearest neighbor retrieval approach which ignores the associations between VRP attributes. The retrieval process was equipped with the abilities of intelligent reasoning and knowledge discovery. Based on the method, a prototype system for modeling vehicle routing problems is implemented. The application of the system shows that the modeling method can satisfy the requirement of practical distribution on velocity and validity.