Computational intelligence approach to real-world cooperative vehicle dispatching problem

摘要

The vehicle dispatching problem for cooperative deliveries from multiple depots (VDP/CD/MD) is an important and difficult problem in the transport industries. In order to solve the VDP/CD/MD, A hierarchical multiplex structure (HIMS/sup ++/) calculation model is proposed. The HIMS/sup ++/ model take advantage of object-oriented modeling, heuristic method, and fuzzy inference in (atomic, molecular, individual) 3 layers, so it can find a utility decision (vehicles plan) close to expert dispatcher. Furthermore, since the vital input parameters are few and the computational engine is packaged into a software component, the HIMS/sup ++/ model is a convenient tool for the VDP/CD/MD. The HIMS/sup ++/ model and its optimization algorithm are implemented as a software component using object-oriented paradigm, and through meta-programming. The performance of the HIMS/sup ++/ model is evaluated through experiments using 3 days oil delivery data taken from an actual dispatching center in Tokyo area. A total of 27 tank lorries are available for daily cooperative deliveries from 3 depots to about 30-60 destinations with different owners are used to save a running cost. The experimental results and the evaluations by human experts confirm that the HIMS/sup ++/ model are better than the results of experienced dispatchers in six evaluation objectives, and can be applied to the planning support system for the VDP/CD/MD. The HIMS/sup ++/ model will be able to cover similar transportation problems in the real-world.