Logistic Demands and Forecasting of Agriculture Cold Chain Serving Beijing, Tianjin and Hebei Province
-
摘要: 采用定性分析和定量统计相结合的方法研究农产品冷链物流需求的影响因素，并在此基础上分别建立基于灰色模型、支持向量机、BP神经网络、RBF神经网络、遗传神经网络的农产品冷链物流需求预测模型。通过研究模型对变量之间相关关系的刻画能力及预测精度两方面因素，发现五类模型分析农产品冷链物流需求问题的能力排序为：遗传神经网络模型> RBF神经网络模型> BP神经网络模型>支持向量机模型>灰色模型，这一结果表明遗传神经网络用于农产品冷链物流需求分析具有优越性。Abstract: Understanding the demands on the logistics of an agricultural product cold chain is essential for appropriate planning, investment, and development of the system, which is unique and complex in operation. A specifically designed program is needed to accurately forecast the demands for an adequate and effective system management. This study applied a qualitative analysis and evaluated with statistical data on factors that might affect the logistics. Subsequently, forecasting models based on the grey model, support vector machine, BP neural network, RBF neural network, and genetic neural network were constructed. By challenging the models on the ability to characterize and correlate the variables as well as predict the outcomes, the following ranking was obtained:genetic neural network > RBF neural network > BP neural network > support vector machine > gray model.
-
Key words:
- agricultural products /
- cold chain /
- logistic demand /
- genetic neural network /
- forecast /
- Beijing-Tianjin-Hebei
-
图 1 冷链农产品需求量与其产量的时间序列
Figure 1. Time sequence schematic on demand and output of agricultural products in a cold chain
图 2 冷链农产品需求量与各个经济指标的时间序列
Figure 2. Time sequence schematic on demand and economic indicators in a cold chain for agricultural products
图 3 冷链农产品需求量与各个人文因素的时间序列
Figure 3. Time sequence schematic on demand and human factors in a cold chain for agricultural products
图 4 冷链农产品需求量与物流规模因素的时间序列
Figure 4. Time sequence schematic on demand and logistic scales in a cold chain for agricultural products
表 1 2017-2020年各指标回归方程预测值
Table 1. Predicted values of various indicators for 2017-2020 by regression equations
年份 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 x14 x15 x16 x17 2017 13851.25 105.10 33962.3 132076.2 79097.8 3875.2 27688.0 47412.0 22.72 430.37 11.98 35812.7 11435.6 3037.5 280295.0 7907.1 207.24 2018 14558.55 100.40 36592.5 140627.3 82134.5 3764.8 26765.0 51420.0 21.87 478.01 12.94 39587.4 11589.9 3133.2 270265.9 7604.5 221.81 2019 15400.24 110.30 39166.0 148934.8 84389.7 3562.7 25090.0 55477.0 21.31 532.29 13.95 43669.4 11744.1 3224.2 253574.1 6993.1 235.65 2020 16391.83 108.20 41658.2 156940.4 85757.2 3259.1 22584.0 59567.0 21.08 593.78 15.01 48072.2 11898.3 3309.7 229482.2 6036.3 248.53 
下载: 导出CSV
表 2 遗传神经网络模型预测值
Table 2. Predicted values by genetic neural network model
年份 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 实际值/万t 2082.59 2154.31 2245.43 2283.14 2329.63 2349.46 2408.79 2447.18 2451.16 2525.59 2553.06 2549.41 2625.73 2690.59 2774.10 2836.86 2961.48 预测值/万t 2102.23 2167.82 2249.87 2283.75 2326.90 2346.61 2406.90 2446.16 2451.63 2526.13 2552.35 2546.20 2620.61 2682.27 2764.29 2808.68 2909.21 绝对误差 19.64 13.51 4.44 0.65 2.70 2.89 1.90 1.04 0.43 0.53 -0.75 3.20 5.09 8.33 9.81 28.22 22.29 绝对百分误差/% 0.94 0.63 0.20 0.03 0.12 0.12 0.08 0.04 0.02 0.02 0.03 0.13 0.19 0.31 0.35 0.99 0.75
下载: 导出CSV
表 3 样本内数据预测值
Table 3. Predicted values on sampled specimens
(单位/万t) 年份 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 实际值 2082.59 2154.31 2245.43 2283.14 2329.63 2349.46 2408.79 2447.18 2451.16 2525.59 2553.06 2549.41 2625.73 2690.59 2774.10 灰色模型预测值 2082.60 2176.87 2217.68 2259.25 2301.60 2344.75 2388.70 2433.48 2479.10 2525.57 2572.92 2621.15 2670.28 2720.39 2771.33 支持向量机预测值 2113.68 2177.42 2227.87 2279.63 2331.44 2373.57 2384.91 2443.48 2481. 4 6 2521.99 2549.55 2587.19 2629.28 2694.20 2762.66 BP神经网络预测值 2070.49 2157.62 2234.27 2275.97 2312.07 2349.98 2411.66 2434.87 2396.80 2488.50 2541.15 2557.52 2663.20 2671.24 2766.56 RBF神经网络预测值 2105.16 2160.69 2211.61 2268.97 2323.97 2377.98 2393.62 2447.38 2478.87 2522.72 2528.05 2569.53 2617.68 2690.02 2773.93
下载: 导出CSV
表 4 样本外数据预测值
Table 4. Predicted values on specimens not sampled
(单位/万t) 年份 2015 2016 2017 2018 2019 2020 灰色模型预测值 2823.28 2876.21 2930.12 2985.05 3041.01 3098.01 支持向量机预测值 2816.86 2861.32 2959.99 3055.90 3133.52 3243.62 BP神经网络预测值 2875.34 2920.65 2959.41 3024.11 3115.05 3195.22 RBF神经网络预测值 2869.00 2920.43 2997.98 3124.26 3198.42 3329.14
下载: 导出CSV
表 5 各类模型的预测误差
Table 5. Prediction errors of various models
年份 灰色模型 支持向量机 BP神经网络 RBF神经网络 绝对误差 相对百分误差/% 绝对误差 相对百分误差/% 绝对误差 相对百分误差/% 绝对误差 相对百分误差/% 2000 0.01 0.00 31.09 1.49 12.1 0.58 22.57 1.08 2001 22.56 1.05 23.11 1.07 3.31 0.15 6.38 0.30 2002 27.75 1.24 17.56 0.78 11.16 0.50 33.82 1.51 2003 23.85 1.04 3.47 0.15 7.13 0.31 14.13 0.62 2004 28.00 1.20 1.84 0.08 17.53 0.75 5.63 0.24 2005 4.75 0.20 24.07 1.02 0.48 0.02 28.48 1.21 2006 20.10 0.83 23.89 0.99 2.86 0.12 15.18 0.63 2007 13.72 0.56 3.72 0.15 12.33 0.50 0.18 0.01 2008 27.90 1.14 30.26 1.23 54.40 2.22 27.67 1.13 2009 0.03 0.00 3.61 0.14 37.10 1.47 2.88 0.11 2010 19.82 0.78 3.55 0.14 11.95 0.47 25.05 0.98 2011 71.75 2.81 37.79 1.48 8.12 0.32 20.13 0.79 2012 44.58 1.70 3.58 0.14 37.50 1.43 8.02 0.31 2013 29.79 1.11 3.6 0.13 19.36 0.72 0.58 0.02 2014 2.77 0.10 11.44 0.41 7.54 0.27 0.17 0.01 2015 13.62 0.48 20.04 0.71 38.44 1.36 32.1 1.13 2016 85.29 2.88 100.18 3.38 40.85 1.38 41.07 1.39 样本外误差均值 49.46 1.68 60.11 2.05 39.65 1.37 36.59 1.26 样本内误差均值 22.49 0.92 14.84 0.63 16.19 0.66 14.06 0.60
下载: 导出CSV
-
[1] QI FANG, DAZHOTIG. The model of highway logistic demand forecasting based on gray neural network[J].Soft Science, 2010, 24(11):132-135. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=rkx200911028 [2] JOSEPH BERECHUNAN.Cold Chain Development and Challenges in the Developing World[J]. ActaHorticulturae, 2003, 18(8):127-134. http://www.cabdirect.org/abstracts/20103372258.html [3] TERRY MOORE.An Introduction to Supply Chain Management[M].Palgrave Macmillan Ltd., 2003, 6(17): 153-165. [4] 王之泰.冷链——从思考评述到定义[J].中国流通经济, 2010(9):15-17. doi: 10.3969/j.issn.1007-8266.2010.09.004 [5] 兰洪杰, 汝宜红.2008北京奥运食品冷链物流需求预测分析[J].中国流通经济, 2008(2):19-22. doi: 10.3969/j.issn.1007-8266.2008.02.005 [6] 朱坤萍, 江琳琳, 王赫男.河北省农产品冷链物流市场分析及对策[J].价格月刊, 2016(475):64-68. http://d.old.wanfangdata.com.cn/Periodical/jgyk201612014 [7] 陆芳, 由建勋.基于组合模型的陕西特色农产品物流需求预测[J].农业经济, 2017(12):140-141. doi: 10.3969/j.issn.1001-6139.2017.12.052 [8] 韩立民, 周海霞.我国水产品冷链物流需求分析及政策建议[J].中国渔业经济, 2012, 30(4):19-23. doi: 10.3969/j.issn.1009-590X.2012.04.003 [9] 李隽波.基于多元线性回归分析的冷链物流需求预测[J].安徽农业科学, 2011, 39(11):6519-6523. doi: 10.3969/j.issn.0517-6611.2011.11.087 [10] 海峰, 水璐, 矿玉玲, 等.湖北省农副产品发展现状及冷链物流需求趋势研究[J].物流工程与管理, 2012, 34(1):32-34. doi: 10.3969/j.issn.1674-4993.2012.01.012 [11] 原静.正向权重组合预测机制下的农产品冷链物流量需求预测[J].江苏农业科学, 2017, 45(19):341-346. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=jsnykx201719080 [12] 蒋智凯, 陈晓华.基于灰色模型的连云港市水产品冷链物流需求预测分析[J].淮海工学院学报, 2017, 15(8):101-103. doi: 10.3969/j.issn.2095-333X.2017.08.027 [13] 周宾.山西果品现代冷链物流需求的SD分析与对策[J].安徽农业科学, 2012, 40(21):11056-11058. doi: 10.3969/j.issn.0517-6611.2012.21.109 [14] 蔡自兴, 徐光祐.人工智能及其应用[M].北京:清华大学出版社, 2004. [15] AMJADY N.Day-aheaad Price Forecasting of Electricity Markets by a New Fuzzy Neural Network[J].IEEE Transaction on Power Systems, 2006, 21(2):887-896. doi: 10.1109/TPWRS.2006.873409 [16] QWG, CHEN G, ZHU L L Short-tem Marginal Price Forcasting Based on Genetic Algorithm and Radial Basis Function Neural Network[J].Power System Technology, 2006, 30(7):18-21. [17] 刘浩, 韩晶.MATLAB R2016a完全自学一本通[M].电子工业出版社, 2016. [18] 周伟祝, 宦婧, 孙媛.遗传神经网络在保障资源需求预测中的应用[J].火力与指挥控制, 2013, 38(8):72-75. doi: 10.3969/j.issn.1002-0640.2013.08.020 -
点击查看大图
计量
- 文章访问数: 1531
- HTML全文浏览量: 235
- PDF下载量: 29
- 被引次数: 0
