Research On Methods And Their Application For Forecasting Cargo Volume Of Waterway Transportation

Prediction of the cargo volume of Waterway transportation is the main basis for determining the scale of water transport infrastructures. A reasonable and accurate prediction result of cargo volume has the most significant effect on determining the investment and benefit of future waterway transportation engineering projects, making developing strategy for future waterway transportation and giving full play to benefit of waterway transport facilities.Based on the amount of reference literatures on cargo volume forecast, The paper expatiates on principle, method and steps of forecasting waterway transport cargo volume;it studies and discusses the application of theory of grey system to cargo volume forecast, analyzes the model building ideas, model checking methods and scope of model application on the basis of grey system theory, improves the exponentially weighting grey model, modified residual error correction model and modified equal dimension model with new information and grey information filling in successively, thus widening the scope of the grey forecasting models, strengthening the applicability of grey models and increasing the accuracy of forecasting.Then, on the basis of brief introduction of the basic structure and learning rules of artificial neural network, the paper analyzes the deficiency of standard BP algorithm, and introduce L-M algorithm to modify the network which improves the convergence of the BP network greatly. Accordingly, a regression forecasting model and nonlinear combined forecasting model based on ANN are built. With the comparison of nonlinear combined model with the linear one, the superiority of nonlinear combined forecasting method is then concluded.By the case analysis of the total transportation forecast volume of Jiangsu inland river, the paper describes the application of all the fore-mentioned models and make a analysis on the results with the forecasting models. The results show that the forecasting result of the nonlinear combined model based on ANN is the best to confirm to the practical situation.