Demand Forecasting is regarded as a basis for pre-phase planning of underground logistics system, since the rationality of underground logistics system planning can only be guaranteed under the principle that the demand can be matched with supply. Underground logistics meets the development characteristics of urban logistics in the future. Nowadays, national new districts, represented by Xiongan New Area of Hebei Province and Jinan Prior Zone For Replacing Old Growth Drivers With New Ones of Shandong Province, have put forward the planning and research task of underground logistics. In practical work, the authors find that the current logistics industry in new urban areas has not been developed, and the historical sample data needed for logistics forecasting are lacking, so traditional forecasting methods such as time series and regression analysis can not be used. In addition, the demand of underground logistics is closely related to the orientation of urban development, industrial structure and service type, which should be considered in the process of forecasting. By considering the above two problems, this paper puts forward a prediction method of underground logistics demand based on two-stage combination forecasting. In the first stage, by using the clustering analysis, we select several target cities according to the scale of the city and the structure of economic and industrial development, and use the historical data of the target cities to obtain the basic value of demand forecasting. In the second stage, support vector regression, an intelligent prediction method, is used to adjust the basic value of prediction, and finally the demand of underground logistics in the planning year is obtained. This paper takes Xiongan New Area as the background to apply the forecasting method, and verifies the feasibility and validity of this method.

CICTP