Simulation of assignment problem based on BP and RBF neural network

The paper is discussing the algorithm in assignment problem. The detail process is to loading, transporting and uploading the gravel by inland water transportation. The main task is to solve the assignment problem dispatching the load barges to pusher tugs for the planned period of one day. The target of the paper is to generate an effective system that is able to decrease the dispatchers´ work loading and work much faster than the dispatchers. Assume that the dispatchers mentioned in this paper are experienced enough to approach the optimum. In addition, The dispatchers don´t want their philosophy of the dispatching task to be changed by the system. We generate the approach neural network to adapt to or learn from the examples of the dispatcher´s decision process. This paper improves the algorithm proposed by Katarina et al. and compares different neural networks (Back Propagation neural network and Ratio Basis Function neural network) and summarizes advantages and disadvantages of different kind of neural networks in vessels dispatching problem.