基于BP神经网络和二次指数平滑法组合预测模型的安徽省物流需求预测

徐健，桂海霞

安徽理工大学经济与管理学院，安徽 淮南  232001

摘要：为准确预测安徽省的物流需求，从经济发展、产量结构、地区贸易和消费水平4方面选取安徽省的地区生产总值，第一、二、三产业产值，社会消费品零售总额，固定资产投资，人均消费性支出7个影响因素作为安徽省物流需求评价指标，以安徽省货运量作为物流需求规模输出指标，采用灰色关联分析计算安徽省物流需求评价指标与物流需求规模间的关联度，判断评价指标的合理性。通过夏普利值法将BP神经网络预测模型和二次指数平滑法预测模型组合，预测2017—2021年安徽省物流需求。结果表明：BP神经网络预测模型、二次指数平滑法预测模型及二者的组合预测模型预测结果的平均相对误差分别为4.58%、6.70%、3.99%，组合预测模型的平均相对误差最小。通过组合预测模型预测2022—2024年安徽省物流需求分别为405 004.96万t、407 142.09万t、409 108.95万t，安徽省货运量呈持续增长趋势，但增幅降低。安徽省应加快传统物流向智慧物流的转移速度，扩大内需，加强物流枢纽城市间的联系，加速区域一体化发展步伐，确保物流高质量发展。

关键词：组合预测模型；BP神经网络模型；二次指数平滑法模型；物流需求；预测

Logistics demand forecast in Anhui Province based on combination forecasting model of BP neural network and second exponential smoothing method

XU Jian, GUI Haixia

School of Economics and Management, Anhui University of Science and Technology, Huainan 232001, China

Abstract: In order to accurately predict the logistics demand in Anhui Province, the regional gross domestic product, the output value of the primary, secondary, and tertiary industries, the total retail sales of social consumer goods, the fixed asset investment, and the per capita consumer expenditure of Anhui Province are selected as the evaluation indicators for Anhui Province′s logistics demand from four aspects: economic development, output structure, regional trade, and consumption level. The freight volume of Anhui Province is used as the output indicator of logistics demand scale. The grey correlation analysis is adopted to calculate the correlation between the evaluation indicators of logistics demand and the logistics demand scale, and to judge the rationality of the evaluation indicators. By combining the back propagation(BP) neural network prediction model with the second exponential smoothing method prediction model using the Shapley value method, the logistics demand of Anhui Province from 2017 to 2021 is predicted. The results show that the average relative errors of the BP neural network prediction model, the second exponential smoothing prediction model, and their combination prediction model are 4.58%, 6.70%, and 3.99% respectively, with the combination prediction model having the smallest average relative error. The combination prediction model predicts the logistics demand of Anhui Province from 2022 to 2024 to be 405 004.96 thousand tons, 407 142.09 thousand tons, and 409 108.95 thousand tons respectively. The freight volume of Anhui Province shows a continuous growth trend, but the growth rate is decreasing. Anhui Province should accelerate the transfer speed from traditional logistics to intelligent logistics, expand domestic demand, strengthen the connection between logistics hub cities, accelerate the pace of regional integration development, and ensure the high-quality development of logistics.

Keywords: combination prediction model; BP neural network model; second exponential smoothing method model; logistics demand; forecost