基于RBF神经网络的物流业能源需求预测

doi:10.18402/resci.2016.03.08

摘要/Abstract

摘要：

随着中国经济的快速发展,物流业需求快速增长,规模不断扩大,也带来了能源消耗的增长.研究中国物流业能源消费水平以及能源需求,有利于物流业节能工作的开展,缓解能源压力.本文选取了影响物流业能源需求的11个主要因素,基于径向基神经网络对2001-2012年间中国物流业能源需求相关数据进行模拟与仿真,在此基础上对2016年和2020年物流业能源需求量进行了预测,并分析了11个影响因素的重要性和测算了物流业的能源效率.研究结果表明:①2001-2012年间中国物流业能源消耗总量在不断增加,随着物流业的进一步发展,到2020年物流业能源消费总量将达到51261.92万t标准煤;②在解决物流业能源需求预测问题时,RBF神经网络比GM(1,1)预测模型,BP神经网络方法有更高的预测精度;③通过RBF神经网络变量重要性分析发现固定资产投资对物流业能源消费量的影响程度最大;④目前物流业能源效率明显低于全国能源效率,为节约能源,提高能源利用效率,物流业需要转变能源利用方式和发展模式.

关键词: 物流业, 能源需求预测, 能源消费, 能源效率, 径向基神经网络

Abstract:

As economy of China grows rapidly,the logistics sector,by the tremendous needs of the market,is developing quickly and the scale and energy consumption are exploding. Studying energy consumption and demand in the logistics sector is significant in the implementation of energy conservation and ease energy pressure. We screened 11 main factors affecting energy demand in the logistics sector,and then established a model of prediction and simulation of energy demand from 2001 to 2012 on the basis of the radial basis function (RBF)neural network whereby energy demand in the logistics sector from 2016 and 2020 is predicted. We propose some recommendations to improve energy consumption efficiency based on the independent variable important analysis and measure energy efficiency in the logistics sector. We found that total energy consumption of the logistics sector increased continuously from 2001 to 2012. With further development of China's logistics sector,energy demand will keep increasing for years to come and energy consumption will arrive at 51 261.92 million tons in 2020. Compared with a GM (1,1)model and back propagation (BP)neural network,the RBF neural network is better than both in terms of forecast accuracy for the logistics sector. The variable of investment in fixed assets has a deeper impact on energy consumption in the logistics sector than other variables. The energy intensity of the logistics sector is significantly higher than China's GDP,to save energy and improve energy consumption efficiency the logistics sector needs to change energy utilization and development modes.

Key words: logistics sector, energy demand prediction, energy consumption, energy efficiency, radial basis function (RBF)neural network

李瑞, 张悟移. 基于RBF神经网络的物流业能源需求预测[J]. 资源科学, 2016, 38(3): 450-460.

LI Rui,ZHANG Wuyi. Energy demand forecast in the logistics sector based on RBF neural networks[J]. Resources Science, 2016, 38(3): 450-460.

图/表 8

图1

表1

表2

表3

图2

表4

表5

表6

参考文献 30

[1]	陈诗一,严法善,吴若沉. 资本深化,生产率提高与中国二氧化碳排放变化-产业,区域,能源三维结构调整视角的因素分解分析[J]. 财贸经济,2010,(12):111-119.
	[Chen S Y,Yan F S,Wu R C.Capital deepening,productivity promotion and CO2 emission in China[J]. Finance & Trade Economics,2010,(12):111-119.]
[2]	国家统计局能源统计司. 中国能源统计年鉴(2001-2012)[M]. 北京:中国统计出版社,2002-2013.
	[Department of Energy Statistics,National Bureau of Statistics of the People's Republic of China. China Energy Statistical Yearbook (2001-2012)[M]. Beijing:China Statistics Press,2002-2013.]
[3]	王安建,王高尚,陈其慎,等. 能源与国家经济发展[M]. 北京:地质出版社,2008.
	[Wang A J,Wang G S,Chen Q S,et al.Energy and National Economic Development[M]. Beijing:Geological Publishing House,2008.]
[4]	Anable J,Brand C,Tran M,et al.Modelling transport energy demand:A socio-technical approach[J]. Energy Policy,2012,41(4):125-138.
[5]	Forouzanfar M,Doustmohammadi A,Hasanzadeh S,et al.Transport energy demand forecast using multi-level genetic programming[J]. Applied Energy,2012,91(1):496-503.
[6]	张诚,周安,张志坚. 基于灰色预测模型的物流低碳效应分析[J]. 统计与决策,2014,(16):89-91.
	[Zhang C,Zhou A,Zhang Z J.Analysis of low carbon logistics effect based on the gray forecast model[J]. Statistics and Decision,2014,(16):89-91.]
[7]	Halldorsson A,Kovacs G.The sustainable agenda and energy efficiency[J]. International Journal of Physical Distribution & Logistics Management,2010,40(1-2):5-13.
[8]	Zajac P.The idea of the model of evaluation of logistics warehouse systems with taking their energy consumption under consideration[J]. Archives of Civil and Mechanical Engineering,2011,11(2):479-492.
[9]	刘勇. 物流业全要素能源效率评价及其影响因素分析[J]. 统计与决策,2014,(1):66-68.
	[Liu Y.Energy efficiency evaluation and its influencing factors in the logistics sector[J]. Statistics and Decision,2014,(1):66-68.]
[10]	张立国,李东,龚爱清. 中国物流业全要素能源效率动态变动及区域差异分析[J]. 资源科学,2015,37(4):754-763.
	[Zhang L G,Li D,Gong A Q.Dynamic changes in total factor energy efficiency and regional disparity in the logistics sector in China[J]. Resources Science,2015,37(4):754-763.]
[11]	Wu Z,Pagell M.Balancing priorities:Decision-making in sustainable supply chain management[J]. Journal of Operations Management,2011,29(6):577-590.
[12]	Halldorsson A,Svanberg M.Energy resources:Trajectories for supply chain management[J]. Supply Management:An Inter-national Journal,2013,18(1):66-73.
[13]	Pudleiner D,Colton J.Using sensitivity analysis to improve the efficiency of a net-zero energy vaccine warehouse design[J]. Building and Environment,2015,87:302-314.
[14]	Lioyd J,Mc Carney S,Ouhichi R,et al.Optimizing energy for a 'green' vaccine supply chain[J]. Vaccine,2015,33(7):908-913.
[15]	邓聚龙. 社会经济灰色系统的理论与方法[J]. 中国社会科学,1984,(6):47-60.
	[Deng J L.Grey system theory and method of social economy[J]. Social Sciences in China,1984,(6):47-60.]
[16]	邓聚龙. 本征性灰色系统的主要方法[J]. 系统工程理论与实践,1986,(1):60-65.
	[Deng J L.The main method of intrinsic gray system[J]. System Engineering-Theory & Practice,1986,(1):60-65.]
[17]	陈飞. 灰色预测模型及应用[J]. 教育经济研究,1998,19(4):33-34.
	[Chen F.Gray forecast model and application[J]. Educ-ational Economy Research,1998,19(4):33-34.]
[18]	崔强,武春友,匡海波. BP-DEMATEL在空港竞争力影响因素识别中的应用[J]. 系统工程理论与实践,2013,33(6):1471-1478.
	[Cui Q,Wu C Y,Kuang H B.Influencing factors research of airports competitiveness based BP-DEMATEL model[J]. System Engineering -Theory & Practice,2013,33(6):1471-1478.]
[19]	刘冰,郭海霞. Matlab神经网络超级学习手册[M]. 北京:人民邮电出版社,2014.
	[Liu B,Guo H X.Neural Network Super Learning Manual Based on MATLAB [M]. Beijing:Post & Telecom Press,2014.]
[20]	Birol F,Keppler J H.Prices,technology development and the rebound effect[J]. Energy Policy,2000,28(1-2):457-469.
[21]	刘龙政,潘照安. 中国物流产业碳排放驱动因素研究[J]. 商业研究,2012,(7):189-196.
	[Liu L Z,Pan Z A.A study on the influence factors of carbon emissions in China's logistics industry[J]. Commercial Research,2012,(7):189-196.]
[22]	Maudos J,Pastor J M,Serrano L.Total factor productivity measurement and human capital in OECD countries[J]. Economics Letters,1999,63(1):39-44.
[23]	Chang T,Hu J L.Total factor energy productivity growth,tech-nical progress,and efficiency change:An empirical study of China[J]. Applied Energy,2010,87(10):3262-3270.
[24]	Apergis N,Aye G C,Barros C P,et al.Energy efficiency of selected OECD countries:A slacks based model with undesirable outputs[J]. Energy Economics,2015,51(9):45-53.
[25]	中华人民共和国国家统计局. 国民经济行业分类(GB/T4754-2011)[EB/OL].(2013-10-23)[2015-09-08]. .
	[National Bureau of Statistics of the People's Republic of China. Standard Industrial Classification of all Economic Activities(GB/T4754-2011)[EB/OL].(2013-10-23)[2015-09-08]. .]
[26]	中华人民共和国国家统计局. 中国第三产业统计年鉴(2006-2012)[M]. 北京:中国统计出版社,2007-2013.
	[National Bureau of Statistics of the People's Republic of China. China Statistical Yearbook of the Tertiary Industry (2006-2012)[M]. Beijing:China Statistics Press,2007-2013.]
[27]	中华人民共和国国家统计局. 中国统计年鉴(2001-2012)[M]. 北京:中国统计出版社,2002-2013.
	[National Bureau of Statistics of the People's Republic of China. China Statistical Yearbook (2001-2012)[M]. Beijing:China Statistics Press,2002-2013.]
[28]	国家统计局,科学技术部. 中国科技统计年鉴(2001-2012)[M]. 北京:中国统计出版社,2002-2013.
	[National Bureau of Statistics,Ministry of Science and Technology of the People's Republic of China. China Statistical Yearbook on Science and Technology (2001-2012)[M]. Beijing:China Statistics Press,2002-2013.]
[29]	孙涵,杨普容,成金华. 基于Matlab支持向量回归机的能源需求预测模型[J]. 系统工程理论与实践,2011,31(10):2001-2007.
	[Sun H,Yang P R,Cheng J H.Forecasting model of energy demand based on Matlab support vector regression[J]. System Engineering -Theory & Practice,2011,31(10):2001-2007.]
[30]	曹翠珍,赵国浩. 区域物流发展,经济增长与能源消费-基于中国省际面板数据的实证研究分析[J]. 财贸研究,2015,(2):44-52.
	[Cao C Z,Zhao G H.Regional logistics development,economic growth and energy consumption:Provincial panel data analysis in China[J]. Finance and Trade Research,2015,(2):44-52.]

年份	产业增加值 /亿元	固定资产投资 /亿元	能源效率 /(t标准煤/万元)	R&D 经费 /万元	授权专利数 /件	交通运输设备拥有量/辆	货运量 /万t	货物周转量 /亿tkm	汽油,柴油,煤油,电力消费比例	从业人员数 /万人	社会消费品零售总额 /亿元	消费总量 /万t 标准煤
2001	6 870.30	5 642.87	0.663	99 453	7 829	8 433 459	1 401 786	47 710	0.693 4	593.77	43 055.40	10 363.00
2002	7 492.90	5 297.85	0.671	20 595	7 537	8 885 904	1 483 447	50 686	0.703 2	576.94	48 135.90	11 171.00
2003	7 913.20	6 289.40	0.617	9 326	9 386	9 363 595	1 564 492	53 859	0.719 2	573.07	52 516.30	12 818.80
2004	9 304.40	7 646.20	0.616	15 120	9 309	9 777 549	1 706 412	69 445	0.744 8	565.00	59 501.00	15 104.00
2005	10 666.20	9 614.00	0.580	11 480	9 698	10 406 342	1 862 066	80 258	0.766 1	443.34	67 176.60	18 391.01
2006	12 813.00	12 138.10	0.632	20 030	11 845	10 684 749	2 037 060	88 840	0.760 0	555.31	76 410.00	20 284.23
2007	14 601.00	14 154.00	0.665	31 480	15 986	11 374 679	2 275 822	101 419	0.757 3	560.64	89 210.00	21 959.18
2008	16 362.50	17 024.40	0.714	39 873	17 842	12 095 664	2 585 937	110 300	0.790 8	560.26	114 830.10	22 917.25
2009	16 727.10	24 974.70	0.706	46 051	22 914	14 463 501	2 825 222	122 133	0.777 9	609.46	132 678.40	23 691.84
2010	19 132.20	30 074.50	0.734	58 162	34 195	16 783 096	3 241 807	141 837	0.782 0	650.68	156 998.40	26 068.47
2011	22 432.80	28 291.70	0.786	34 302	42 172	18 707 110	3 696 961	159 324	0.779 7	637.88	183 918.60	28 535.50
2012	24 959.80	31 444.90	0.792	53 595	59 548	19 796 513	4 099 400	173 771	0.790 1	667.52	210 307.00	31 524.71

	实际值	RBF神经网络预测值	GM(1,1) 预测值	BP神经网络预测值
2011年	28 535.50	29 108.25	30 979.84	26 079.87
2012年	31 524.71	31 785.65	34 238.02	29 868.41

预测方法	平均绝对误差(MAE)	误差均方根(RMSE)	平均绝对百分误差(MAPE)
RBF神经网络	138.70	480.47	0.85%
GM(1,1)	1 257.64	4 356.58	5.83%
BP神经网络	1 550.66	5 371.63	8.77%

变量名	重要性	标准化的重要性/%
固定资产投资(X₂)	0.179	100.0
交通运输设备拥有量(X₆)	0.175	97.8
货运量(X₇)	0.160	89.4
社会消费品零售总额(X₁₁)	0.154	86.0
货物周转量(X₈)	0.122	68.2
产业增加值(X₁)	0.114	63.7
汽油,柴油,煤油,电力消费比例 (X₉)	0.083	46.4
能源效率(X₃)	0.025	14.0
R&D经费(X₄)	0.002	1.1
授权专利数(X₅)	0.002	1.1
从业人员数(X₁₀)	0.002	1.1

年份	物流业增加值 /亿元	物流业能源效率 /(t标煤/万元)	全国能源效率 /(t标煤/万元)	年份	物流业增加值 /亿元	物流业能源效率 /(t标煤/万元)	全国能源效率 /(t标煤/万元)
2001	6 870.3	0.663	0.766	2007	14 601.0	0.665	0.948
2002	7 492.9	0.671	0.793	2008	16 362.5	0.714	1.078
2003	7 913.2	0.617	0.776	2009	16 727.1	0.706	1.112
2004	9 304.4	0.616	0.787	2010	19 132.2	0.734	1.236
2005	10 666.2	0.580	0.784	2011	22 432.8	0.786	1.359
2006	12 183.0	0.632	0.836	2012	24 660.0	0.792	1.434