基于多种群遗传BP的城市物流需求预测  

Study On Urban Logistics Demand BP Forecasting Model Based on MPGA Algorithms

作　　者：孙焰[1] 赵剑翔 郑文家[1] 

机构地区：[1]同济大学道路与交通工程教育部重点实验室,上海201804

出　　处：《综合运输》2016年第3期126-132,共7页China Transportation Review

基　　金：国家自然科学基金项目资助(71072027)

摘　　要：城市物流需求预测是城市物流规划的重要前提,为了提高物流需求预测的精度,本文首先通过分析影响城市物流需求的经济活动,选取城市物流需求预测的指标。随后,为了改善BP算法预测效果受到网络初始权值与阈值影响的缺陷,提出多种群遗传BP算法进行改进。通过仿真表明,该方法与BP预测模型以及单种群遗传BP预测模型相比,具有较高的预测精度。Indexes of logistics demand are chosen on the premise of analyzing economic activities to improve the forecasting accuracy of urban logistics demand, which in turn is an important precondition of urban logistics planning. Then the MPGA Algorithms based BP forecasting model is used to enhance the original model, which is obviously influenced by the initial weight of network. The simulation results present that the proposed algorithms are more accurate than BP model and single population genetic BP forecasting model.

关 键 词：多种群遗传算法 城市物流 BP神经网络 需求预测 

分 类 号：U492[交通运输工程—交通运输规划与管理]