Forecast of Line-level Logistics Cargo Volume Based on Hybrid Dual Neural Network Model

It has certain significance that accurate line-level logistics volume forecast provides strong support for the following vehicle scheduling and route planning. In this paper, the hybrid dual ARIMA&BP model is improved by forming a dual neural network architecture which divides the prediction value into linear and nonlinear parts. The results of the two parts are again fitted by another neural network to obtain the final prediction result. To demonstrates the superiority of the proposed algorithm, State-of the art algorithms are compared by using some practical data from some leading logistics company. It is found that there is about 21.5% relative improvement over the existing ARIMA&BP method[8].