基于大数据和优化神经网络短期电力负荷预测

doi:10.11959/j.issn.1000-436x.2016245

摘要/Abstract

摘要：

随着电力数据采集成本降低及大规模电网互联等因素，电网中可获取的数据类型日益丰富。以往的集中式预测方法对海量电力数据的分析能力有限。提出基于大数据和粒子群优化BP神经网络短期电力负荷预测，建立短期电力负荷预测模型。利用国家电网的实际负荷数据，采用所提方法进行预测，与实际负荷数据及集中式负荷预测结果进行比较，结果证明，所提方法预测精度较高，降低了负荷预测时间，在实际应用中具有可行性。

关键词: 电力大数据, 粒子群算法, 并行PSO优化神经网络, 电力负荷预测, 电力负荷影响因素

Abstract:

With the reduction of the cost of power data acquisition and the interconnection of large scale power systems,the types of data available in the power network are becoming more and more abundant.In the past,the centralized fore-casting method was limited to the analysis of the massive power data.Therefore,a short-term power load forecasting based on large data and particle swarm optimization BP neural network was proposed,and short-term power load fore-casting model was established.The actual load data of the national grid,using the method of prediction,compared with the actual load data and centralized load forecasting results prove that this method is accurate enough,reduce the load forecasting time with feasibility in practical application.

Key words: electric power data, particle swarm algorithm, parallel PSO to optimize the neural network, power load fore-casting, power load factor

金鑫,李龙威,季佳男,李祉歧,胡宇,赵永彬. 基于大数据和优化神经网络短期电力负荷预测[J]. 通信学报, 2016, 37(Z1): 36-42.

Xin JIN,Long-wei LI,Jia-nan JI,Zhi-qi LI,Yu HU,Yong-bin ZHAO. Power short-term load forecasting based on big data and optimization neural network[J]. Journal on Communications, 2016, 37(Z1): 36-42.

图/表 9

图1

图2

表1

图3

图4

图5

图6

表2

表3

参考文献 28

[1]	刘玲 . 基于粒子群BP神经网络的短期负荷预测研究[[D]. 江苏：河海大学， 2006. LIU L . Short-term load forecasting based on particle swarm neural network research[D]. Jiangsu: HHU, 2006.
[2]	陆宁，周建中，何耀耀，等. 粒子群优化的神经网络模型在短期负荷预测中的应用[J]. 电力系统保护与控制， 2010,38(12):2010,38(12):65-68. LU N , ZHOU J Z , HE Y Y , et al. Particle swarm optimization-based neural network model for short-term load forecasting[J]. Power System Protection and Control, 2010,38(12):65-68.
[3]	HIPPER H S , PERFREIRA C E , SOUZA R C . Neural network for short-term load forecasting:a review and evaluation[J]. IEEE Transac-tions on Power Systems, 2001,16(2):44-54.
[4]	宋亚奇，周国亮，朱永利 . 智能电网大数据处理技术现状和挑战[J]. 电网技术， 2013,37(4):927-935. SONG Y Q , ZHOU G L , ZHU Y L . Present status and challenges of big data processing in smart grid[J]. Fudan International Studies Re-view, 2013,37(4):927-935.
[5]	陈健，刘明波，樊亚亮，等. 广州电网负荷特性分析[J]. 电力系统及其自动化学报， 2009,21(6):78-83. CHEN J , LIU M B , FAN Y L , et al. Studies on load characteristics of Guangzhou power network[J]. Proceedings of the Chinese Society of Universities, 2009,21(6):78-83.
[6]	廖峰，徐聪颖，姚建刚，等. 常德地区负荷特性及其影响因素分析[J]. 电网技术， 2012,36(7):117-125. LIAO F , XU C Y , YAO J G , et al. oad characteristics of Changde re-gion and analysis on its influencing factors[J]. Power System Tech-nology, 2012,36(7):117-125.
[7]	李明，王智灵，杨晓宇，等. 突变期电力负荷预测方法及其应用[J]. 电力系统自动化， 2006,30(10):93-96. LI M , WANG Z L , YANG X Y , et al. Short-term load forecasting for abnormal changes and its application[J]. Automation o Electric Power Systems, 2006,30(10):93-96.
[8]	康重庆，固安石，王鹏，等. 短期负荷预测中实时气象因素的影响分析及其处理策略[J]. 电网技术， 2006,30(7):5-10. KANG C Q , ZHOU A S , WANG P , et al. Impact analysis of hourly weather factors in short-term load forecasting and its processing strat-egy[J]. Power System Technology, 2006,30(7):5-10.
[9]	BENAOUDA D , MURTAGH F , STARCK J L , et al. Wavelet-based nonlinear multiscale decomposition model for electricity load fore-casting[J]. Neurocomputing, 2006,70(1):139-154.
[10]	李小燕，文福栓，卢恩，等. 基于相似日负荷修正的台风期间短期负荷预测[J]. 电力系统及其自动化学报， 2013,25(3):82-89. LI X Y , WEN F S , LU E , et al. Short-term load forecasting in Typhoon periods based on load modification of similar days[J]. Proceedings of the CSU-EPSA, 2013,25(3):82-89.
[11]	SONG K B , YONG S B , HONG D H , et al. Short-term load foreasting for the holidays using fuzzy liner regression method[J]. IEEE Transac-tions Power Systems, 2005,20(1):96-101.
[12]	李玲玲，朱博 . 基于混沌时间序列的短期电力负荷预测[J]. 信息技术， 2009(3):44-46. LI L L , ZHU B . Short-time power load forecast based on chaotic time series[J]. Information Technology, 2009(3):44-46.
[13]	朱陶业，李应求，张颖，等. 提高时间序列气象适应性的短期电力负荷预测算法[J]. 中国电机工程学报， 2006,26(23):14-19. ZHU T Y , LI Y Q , ZHANG Y , et al. A new algorithm of advancing weather adaptability based on arima model for day-ahead power load forecasting[J]. Proceedings of the Chinese Society for Electrical En-gineering, 2006,26(23):14-19.
[14]	张思远，何光宇，梅生伟，等. 基于相似时间序列检索的超短期负荷预测[J]. 电网技术， 2008,32(12):6-59. ZHANG S Y , HE G Y , MEI S W , et al. Ultra-short term load forecast-ing based on similarity search in time-series[J]. Power System Tech-nology, 2008,32(12):56-59.
[15]	程其云 . 基于数据挖掘的电力短期负荷预测模型及方法的研究[D]. 重庆：重庆大学， 2004:5-30. CHENG Q Y . Power short-term load forecasting model based on data mining,and methods of research[D]. Chongqing: Chongqing Univer-sity, 2004:5-30.
[16]	赵宏伟，任震，黄雯莹，等. 基于周期自回归模型的短期负荷预测[J]. 中国电机工程学报， 1997,17(5):348-351. ZHAO H W , REN Z , HUANG W Y , et al. A short term load forecast-ing method based on PAR model[J]. Proceedings of the CSEE, 1997,17(5):348-351.
[17]	BAHNAM K . Up to year 2020 load fprecasting using neural nets[J]. Electric Power and Energy Systems, 2002,24(9):789-797.
[18]	PING F , PAI W C H . Support vector machines with simulated anneal-ing algorithms in electricity load forecasting[J]. Energy Conversion and Management, 2005,46(17):2669-2688.
[19]	PING F , PAI W C H . Forecasting regional electricity load based on recurrent support vector machines with genetic algorithms[J]. Electric Power Systems Research, 2005,74(3):417-425.
[20]	牛东晓，曹树华，卢建昌，等. 电力负荷预测技术及其应用[M]. 北京：中国电力出版社， 2009. NIU D X , CAO S H , LU J C , et al. Power load forecasting technique and its application[M]. Beijing: China Electric Power Press, 2009.
[21]	LIU K . Comparion of very short-term lord forecasting technique[J]. IEEE Trans on Power System, 1996,11(2):877-822.
[22]	刘明 . 基于小波和神经网络理论的电力系统负荷预测研究[D]. 南京理工大学， 2012. LIU M . Based on wavelet and neural network theory of power system load forecasting research[D]. NUST, 2012.
[23]	谢勃 . 中长期电力负荷预测的不确定性模型研究与应用[D]. 河海大学， 2007. XIE B . The uncertainty of the medium and long term power load forecasting model research and application[D]. Jiangsu:HHU, 2007.
[24]	吕昌国 . 基于BP算法的网格资源调度研究[D]. 哈尔滨：哈尔滨理工大学， 2007. LV C G . rid resources scheduling research based on BP algorithm[D]. Harbin: HUST, 2007.
[25]	袁瑛 . 粒子群算法在电力系统经济调度中的应用研究[D]. 西安：西安理工大学， 2010. YUAN Y . Particle swarm algorithm application in power system eco-nomic operation[D]. Xi'an: Xi'an University of Technology, 2010.
[26]	LEE K Y , ELSHARKAWI M A . Modern heuristic optimization tech-niques with applications to power systems. IEEE Power Engineering Society, 2002,10(6):45-51.
[27]	KENNEDY J , EBERHART C . A discrete binary version of the particle swarm algorithm[J]. IEEE Trans on Power Systems, 2007,12(5):4104-4109.
[28]	李澄非 . 计算智能方法研究及其在流程工业中应用[D]. 北京：北京化工大学， 2007. LI C F . Computational intelligence methods and its application in process industry[D]. Beijing: Beijing University of Chemical Tech-nology, 2007.

\|R_1?x\|	线性相关程度
[0,0.3）		低度线性相关
[0.3,0.5）		中低度线性相关
[0.5,0.8）		中度相性相关
[0.8,1]	高度线性相关

时间	预测值/W	实际值/mW	误差	相对误差
1:00	0.224 25	0.212 1	?0.012 15	?5.41%
2:00	0.133 88	0.131 5	?0.002 38	?1.77%
3:00	0.141 14	0.112 1	?0.029 04	?20.57%
4:00	0.218 80	0.228 7	0.009 9	4.52%
5:00	0.578 75	0.616 2	0.037 42	6.47%
6:00	0.624 15	0.632 1	0.007 95	1.27%
7:00	0.655 24	0.655 2	0.000 24	0.06%
8:00	0.714 50	0.725 4	0.010 9	1.52%
9:00	0.715 47	0.726 1	0.010 63	1.48%
10:00	0.784 56	0.788 0	0.003 44	0.43%
11:00	0.835 42	0.843 1	0.007 68	0.91%
12:00	0.845 56	0.856 4	0.010 84	0.12%

模型	平均相对误差	训练步数	泛化能力
基本BP神经网络	1.88%	216	差
改进的神经网络算法	1.68%	7	一般
粒子群优化神经网络模型	1.20%	4	强