COMBINATION PREDICTION METHOD FOR PORT LOGISTICS DEMAND BASED ON GENETIC ALGORITHM

Port logistics is an important component of the modem Logistics system, and the Logistics demand prediction is an important fundamental work for the port logistics system planning. This paper analyzes the limitations of current different prediction methods for demand of port logistics, based on this, a grey - neural network prediction combination method based on genetic algorithm is introduced. With improved genetic algorithm, this paper discusses and resolves the several problems related to the weight optimization of the grey neural network combination prediction method, such as coding, fitness function, selection of genetic operator, ending condition of algorithm and test of effectiveness. First, finds the results of the grey prediction method and the neural network prediction method separately, and then, calculates the weights and prediction values of combination model based on genetic algorithm. According to the comparative analysis between the actual value and the predicted value in Dalian port for past 10 years, this combination prediction method is proved to be more effective, and it has higher accuracy than a single method such as grey or neural network method.