PSO-BP混合预测模型及在港口集装箱吞吐量预测中的应用  

Application of BP Network Model Based on PSO for Prediction of Container Throughput

作　　者：严武元[1] 王少梅[1] 

机构地区：[1]武汉理工大学物流工程学院,武汉430063

出　　处：《武汉理工大学学报（交通科学与工程版）》2007年第3期525-528,共4页Journal of Wuhan University of Technology（Transportation Science & Engineering）

基　　金：交通部重点科技项目资助(批准号:20033G0091)

摘　　要：运用粒子群优化算法代替BP神经网络的初始寻优,再用BP算法对优化的网络权值参数进一步精确优化,从而建立基于粒子群优化的BP神经网络模型.运用该模型对某港口集装箱吞吐量进行预测.应用结果表明,该预测模型不仅能较好地拟合港口集装箱吞吐量的历史数据,同时对港口集装箱吞吐量的远期预测也具有较好的效果.A BP neural network model based on particle swarm optimizer （PSO） is proposed in this paper. The basic idea of this model is： Firstly PSO is used to optimize the BP neural network＇s initialized weights, an optimized result is got; then based on the optimized result, the BP neural network is used for further optimization. The model is used to predict the container throughput of a port. The result of prediction shows that the PSO-BP model not only can fit the history data smoothly, but also can get good prediction of long term container throughput of the port.

关 键 词：粒子群优化算法(PSO) BP神经网络 集装箱 吞吐量 预测 

分 类 号：U695.22[交通运输工程—港口、海岸及近海工程] TP206[交通运输工程—船舶与海洋工程]