Research Analysis Of The Prediction To Serve The Third Party Automobile Logistics

The Third Party Automobile Logistics is a mode of Automobile Logistics separates partial function of logistics management from automobile manufacture and outsources to the third party logistics providers, which is an effective way of focusing on core business, improving efficiency, and reducing the cost by automobile manufactures. The logistics demand forecast is the precondition of effective programming transportation capacity, making object and plan, and improving enterprise core competition for the third party automobile logistics. Therefore, the prediction method of automobile logistics is becoming an important research direction of the third party automobile logistics. The key topic researched in this paper is to solve the logistics demand forecast problem of the third party automobile logistics, which has important theoretical meaning and practical application value.It detailedly analyzes the development course and present situation for the automobile logistics and the third party automobile logistics in China, and discusses and studies the operation mode of the automobile logistics, and demonstrates the feasibility and necessity for operating the third party automobile logistics mode in China. Then it briefly describes the general method and process of logistics system forecast, and compares and analyzes many kinds of common logistics forecast methods, and establishes a logistics demand forecasting models based on Grey Residual-Back Propagation Neural Network(GRBPNN) and a combination forecasting model based on Radia Basis Function Neural Network(RBFNN). According to the macroscopic development environment of the third party automobile logistics enterprises, it establishes a RBF continuation forecasting model based on Grey Correlation Analysis (GCA) to predicting the total production and sales volume of Chinese automobile in the future。This paper comparative analyses the fitting precision and prediction precision of forecasting models while selected different number of node and hidden neurons, and establishes a network to fitting and forecasting the grey residual sequence of logistics demand in GRBPNN. And it puts forward a seeking optimum parameters method by searching optimal solution for two-dimensional space (goal, spread) in a certain range, and realizes the combination forecasting of logistics demand, and improves the stability of network and the precision of prediction in RBFNN. It adopts a continuation forecasting model mixed temporal prediction and interaction prediction and obtains a reasonable result in the total production and sales volume of Chinese automobile forecasting model based on Grey Correlation Analysis (GCA).Finally, an instance is presented to realize the three models GRBPNN, RBFNN and continuation forecasting based on RBF by MATLAB, and some comparisons between them are made. The results showed that a good fitting precision and a high forecasting precision are reached in the application of the logistics demand forecasting by designed forecasting models.