基于神经网络的物流量预测

Prediction of logistics amount based on artificial neural networks

应用不同的人工神经网络模型进行物流量预测。分析影响物流发出量、吸引量和周转总量的各相关因素及神经网络预测的基本思想,研究神经网络静态前馈模型和简单动态模型预测物流量的局限性,认为采用二者结合的综合预测方法能对物流量进行准确的预测。给出动态反馈的实时递归法仿真计算步骤,仿真结果与实际结果比较,具有较高的可信度。

The logistics amount can be predicted in ANN (artificial neural networks) model. The influencing factors of the logistics amount and the basic thought of ANN prediction were analyzed. There are weakness existing in feedforward network and simple dynamic network prediction model. The method together with dynamic feedback and causality was put forward and real time recurrent learning was used in order to make up the lack of the traditional method in the regional logistics system .The theory can be tested in the simulation example . There are extensive practicability of the prediction method.