基于GM(1,1)-MLP神经网络模型的大宗货物运输需求预测  

Transportation Demand Forecast of Bulk Cargo Based on GM(1,1)-MLP Neural Network Model

作　　者：武慧荣[1] 陈少阳 崔淑华[1] WU Hui-rong;CHEN Shao-yang;CUI Shu-hua(School of Civil Engineering and Transportation,Northeast Forestry University,Harbin Heilongjiang 150040,China;Road Transport Comprehensive Law Enforcement Detachment of Shiyan,Shiyan Hubei 442000,China)

机构地区：[1]东北林业大学土木与交通学院,黑龙江哈尔滨150040 [2]十堰市道路交通运输综合执法支队,湖北十堰442000

出　　处：《公路交通科技》2023年第10期233-240,共8页Journal of Highway and Transportation Research and Development

基　　金：中央高校基本科研业务费专项资金项目(2572015CB16)。

摘　　要：针对大宗货物运输需求预测的复杂性,以货物产量为基础,提出了一种基于产运系数的大宗货物运输需求预测方法,并根据大宗货物的运输需求发展趋势确定了运输结构调整路径。以黑龙江省为例,综合考虑粮食产量的化肥施用量、农村用电量、农业机械总动力、粮食作物播种面积等影响因素,建立了GM(1,1)模型以及GM(1,1)-MLP神经网络模型进行粮食产量预测,并应用实际数据进行了验证。根据黑龙江省常住人口、常住人口城镇化率、粮食产量、中国城镇居民和农村居民人均粮食消费量等统计数据确定黑龙江省粮食产运系数,综合粮食产量预测值和产运系数预测了未来几年粮食运输量,以此分析黑龙江省粮食运输需求趋势,为黑龙江省大宗货物运输结构调整方案制订提供了依据。结果表明:与GM(1,1)模型相比,构建的GM(1,1)-MLP神经网络模型进行粮食产量预测,预测精度提高了1.68%;采用产运系数搭建粮食产量与运输量之间关系进行粮食运输需求预测具有可行性;根据预测结果,黑龙江省粮食运输需求将持续增长,仍是黑龙江省大宗货物运输对象的主要组成,积极调整粮食运输结构,推进中长距离的粮食运输转向铁路运输,公路运输作为铁路运输两端的短驳分拨,实现公铁联运,对于优化黑龙江省大宗货物运输结构具有重要作用。In view of the complexity of bulk cargo transportation demand forecast,a forecast method of bulk cargo transportation demand based on the production and transportation coefficient is put forward,and the route of transport structure adjustment is determined according to the development trend of bulk cargo transport demand.Taking Heilongjiang province as an example,taking into account factors such as the amount of chemical fertilizer applied,rural electricity consumption,total power of agricultural machinery and the area sown for grain crops,GM(1,1)model and GM(1,1)-MLP neural network model are established to forecast grain yield,and are verified by actual data.Heilongjiang’s grain production and transportation coefficient are determined based on statistics on Heilongjiang’s permanent population,urbanisation rate of permanent population,food production,per capita food consumption of urban and rural population,and the grain transport volume for the next few years is forecast combining with the forecast of grain output and the production and transportation coefficient to analyze the trend of grain transport demand in Heilongjiang and provide a basis for formulating the structural adjustment plan for bulk cargo transport in Heilongjiang.The result shows that(1)compared with GM(1,1)model,the precision of GM(1,1)-MLP neural network model is improved by 1.68%;(2)based on the forecast results,the demand for grain transport in Heilongjiang will continue to increase,and the transport demand will continue to increase,Heilongjiang is still the main part of the bulk cargo transport object,actively adjusting the grain transport structure,promoting the shift of medium-and long-distance grain transport to railway transport,and road transport as a short-barge distribution at both ends of railway transport,combined rail and public transport plays an important role in optimizing Heilongjiang’s bulk cargo transport structure.

关 键 词：物流工程 运输需求预测 GM(1 1)-MLP神经网络 大宗货物 产运系数 

分 类 号：U116.1[交通运输工程]