基于pccsAMOPSO的宏观物流量多目标变权组合预测模型  

Multi-objective variable weight combination forecasting model of macro logistics volume based on pccsAMOPSO

作　　者：樊东方 罗凯 靳志宏 FAN Dong-fang;LUO Kai;JIN Zhi-hong(Transportation Management College,Dalian Maritime University,Dalian 116026,China)

机构地区：[1]大连海事大学交通运输工程学院,辽宁大连116026

出　　处：《大连海事大学学报》2021年第4期19-29,共11页Journal of Dalian Maritime University

基　　金：国家自然科学基金面上项目(71572023,71702019);欧盟H2020项目(MSCA-RISE-777742-56);大连市领军人才项目(2018-573);中央高校基本科研业务费专项资金资助项目(3132020301)。

摘　　要：针对现有中长期宏观物流量预测模型的局限性,提出基于pccsAMOPSO算法的多目标变权组合预测模型(Mutil-Objective Variable Weight Combination Prediction Mode,MOVWCP)对我国宏观物流量进行预测分析.为提高多目标变权组合预测模型的稳定性,提出了误差熵的概念,并与MAPE同时作为权重规划模型的目标函数,设计了基于pccsAMOPSO的智能启发式算法求解拟合期变权的Pareto前沿,并采用灵敏度差选取了变权Pareto解.一系列数值试验结果验证了本文提出的多目标变权组合预测模型及其算法的优越性.Aiming at the limitations of the existing medium-to-long-term macro logistics volume forecasting models,a multi-objective variable weight combination prediction model(MOVWCP)based on the pccsAMOPSO algorithm was proposed to analyze and predict macro logistics volume.In order to improve the stability of the multi-objective variable weight combination prediction model,the concept of error entropy was proposed,which used as the objective function of the weight programming model together with MAPE.An intelligent heuristic algorithm based on pccsAMOPSO was designed to solve the Pareto front with variable weight in the fitting period,and the variable weight Pareto solution was selected by sensitivity difference.A series of numerical test results verify the superiority of the multi-objective variable weight combination prediction model and its algorithm proposed in this paper.

关 键 词：物流量预测 pccsAMOPSO 多目标变权组合预测模型(MOVWCP) 平均绝对百分比误差(MAPE) 误差熵 

分 类 号：F55[经济管理—产业经济] TP391[自动化与计算机技术—计算机应用技术]