基于Matlab的神经网络预测模型研究

Research on the Forecast Model of Neutral Network Based on Matlab

本文针对物流系统多种因素相关的非线性特点,建立了人工神经网络模型并对物流系统历史数据进行拟合,抽取和逼近负荷曲线进行物流预测。它的优点在于它具有模拟多变量而不需要对输入变量做复杂的相关假定,不要求知道输入输出变量间的函数关系,只需通过对输入输出数据的训练,获得输入输出之间的映射关系,来进行负荷预测。该预测模型在Matlab软件平台上进行了仿真,结果反映了该预测模型的有效性和实用性。

Aiming at the non-linear characteristics of logistics system, this paper presented and established artificial neutral network model to simulate the history data of logistics system, and drew out to approach the burthen curve to forecast logistics. The advantage of this method is that it doesn＇t need to suppose the input attributes and know the function relatienship of input and output attributes, just needing to train the input and output attributes to get the mapped relation： between input and output and to forecast burthen. The forecast model was simulated on Matlab software platform , the reaults show the validity and practicability of this model.