Research Of The Logistics Park Cargo Flow Forecasting In The Supply Chain Management Environment

Logistics park, which is a concentrated logistics site with a certain scale and integrated service functions, will hold the balance to the future economy development under the influence of the environment of supply chain management. With the strict controlling of the land, the city land that can be used for the construction of the logistics park will be increasingly scarce. The cargo flow must be forecasted and evaluated accurately, it can strengthen the logistics park planning reasonable, and make the logistics park become the logistic hinge that can develop the modern logistics and promote economic growth. Therefore, it becomes the hotspot that to select prediction method and improve the accuracy of forecasts on the cargo flow forecasting. The key topic researched in this paper is to solve the problem of the cargo flow forecasting of the logistics park in the supply chain management environment.A detail contents and character of cargo flow analysis of the logistics park is proposed, and the main influence factors of the logistics park's cargo flow changes are analyzed deeply. Based on the compare and analysis about traditional and modern forecasting method of Regression Analysis (RA), Time Series (TS), Grey Model (GM) and Neural Network (NN) etc., hybrid grey theories and BP algorithm neural network forecast model (GBPNN) is established based on the season influence factor from the logistics park sides; and generalized regression neural network (GRNN) cargo flow forecast model Based on RBF is set up from regional economic level.Through a compare and analysis of 'traingd, trainlm, thaingdx' and so on the BP algorithm function, the 'trainlm' algorithm function is chose in GBPNN. A expert evaluation system is established for evaluating season influence factor of cargo flow in the logistics park, and the season influence factor is used as the GBPNN input layer. The counts of the network input layers and connotative layers are confirmed by trial calculation, the local extremum is avoided furthest and the forecasting precision is improved. Correlation analysis is used to determine correlation degree in GRNN, the GRNN model is tested by different smoothness factor, and the perfect forecasting effect is obtained.By using the TCL distribution cargo flow data in Gai Jiagou Logistics Park, and the Matlab programming is established to verify the GBPNN model and the GRNN model. The result is that a good forecasting approach effects and a high forecasting precision are gotten by GBPNN and GRNN in the application of the logistics park cargo flow forecasting.