基于数据的SecRPSO-SVM短期电力负荷预测

doi:10.16182/j.issn1004731x.joss.201708025

摘要/Abstract

摘要： 针对支持向量机在建模中的参数选取问题,提出一种二阶振荡和带斥力因子的粒子群优化算法优化支持向量机参数。采用非线性递减权重平衡算法的全局和局部搜索能力,二阶振荡因子保持种群多样性,提高全局搜索能力。斥力因子使粒子在搜索空间均匀分布,避免陷入局部最优。针对电力负荷的非线性、时变性、受多因素影响的复杂特点,提出一种基于数据的支持向量机预测模型,综合考虑天气、时间因素、历史负荷对预测结果的影响。仿真表明该方法可以建立短期电力负荷的有效高精度预测模型。

关键词: 支持向量机, 粒子群算法, 二阶振荡, 斥力因子, 数据分析, 短期电力负荷

Abstract: For the parameter selection of support vector machine in modeling, a particle swarm optimization algorithm based on second-order oscillation and repulsion factor was proposed to optimize the parameter of SVM. The algorithm employed the nonlinear decreasing weight to balance the global and local search ability. Second-order oscillation factor could maintain the population diversity. The repulsion factor was introduced to make the swarm even distribution in search space, which could avoid local optimum. For the complex characteristics of nonlinearity, time-varying and multifactorial of electric power load, a support vector machine forecasting model based on data was proposed, and the influence of weather, time and historical load on the forecast results was considered. Simulation results show that the proposed method can be used to build an effective and high precision short-term power load forecasting model.

Key words: support vector machine, particle swarm optimization, second-order oscillation, repulsion factor, data analysis, short-term load forecasting

中图分类号:

TP391

孙海蓉, 谢碧霞, 田瑶, 李卓群. 基于数据的SecRPSO-SVM短期电力负荷预测[J]. 系统仿真学报, 2017, 29(8): 1829-1836.

Sun Hairong, Xie Bixia, Tian Yao, Li Zhuoqun. Forecasting of Short-term Power Load of SecRPSO-SVM Based on Data-driven[J]. Journal of System Simulation, 2017, 29(8): 1829-1836.

参考文献

[1] 沈伟民, 黄蕾, 张晓闽. 大数据让负荷预测更精准[J]. 国家电网, 2016, 11(7): 64-65.
(Shen Weimin, Huang Lei, Zhang Xiaomin.Big data makes load forecasting more accurate[J]. National Grid, 2016, 11(7): 64-65.)
[2] 袁诚斌, 张涛, 朱建良. 基于MRA与回归分析法的短期电力负荷预测[J]. 信息技术, 2007, 46(10): 88-90.
(Yuan Chengbin, Zhang Tao, Zhu Jianliang.Short term power load forecasting based on MRA and regression analysis[J]. Information Technology, 2007 (10): 88-90.)
[3] 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, 33(3): 44-46.
(Li Lingling, Zhu Bo.Short term power load forecasting based on chaotic time series[J]. Information Technology, 2009(3): 44-46.)
[4] 周虎, 江岳春, 陈旭, 等. 模糊聚类分析和代数算法结合的短期负荷预测[J]. 电力系统及其自动化学报, 2011, 23(3): 101-105.
(Zhou Hu, Jiang Yue Chun, Chen Xu, et al.Combining fuzzy clustering analysis and algebraic algorithm for short-term load forecasting[J]. Power System And Automation, 2011, 23(3): 101-105.)
[5] 吴桂峰, 王轩, 陈东雷. 基于优化神经网络的短期电力负荷预测[J]. 计算机仿真, 2013, 30(11): 95-99, 111.(Wu Guifeng, Wang Xuan, Chen Donglei. Short term power load forecasting based on optimized neural network [J]. Computer Simulation, 2013, 30(11): 95-99, 111.)
[6] 王玉龙, 崔玉, 李鹏, 等. 基于小波分析改进的神经网络模型电力系统负荷预测[J]. 电网与清洁能源, 2015, 31(2): 16-20, 27.(Wang Yulong, Cui Yu, Li Peng, et al. Improved neural network model based on wavelet analysis, power system load forecasting [J]. Power Grid And Clean Energy, 2015, 31(2): 16-20, 27.)
[7] 段其昌, 曾勇, 黄大伟, 等. 基于扩展记忆粒子群-支持向量回归的短期电力负荷预测[J]. 电力系统保护与控制, 2012, 40(2): 40-44.
(Duan Qichang, Zeng Yong, Huang Dawei, et al.Short-term power load forecast based on particle swarm optimization with extended memory and support vector regression[J]. Power System Protection and Control, 2012, 40(2): 40-44.)
[8] 王义军, 李殿文, 高超, 等. 基于改进的PSO-SVM的短期电力负荷预测[J]. 电测与仪表, 2015, 52(3): 22-25.
(Wang Yijun, Li Dianwen, Gao Chao, et al.Short term power load forecasting based on improved[J]. PSO-SVM Electrical Measurement and Instrumentation, 2015, 52(3): 22-25.)
[9] 高春能, 张彪, 纪志成. 基于自适应动态搜索粒子群的SVM参数优化研究[J]. 系统仿真学报, 2015, 27(12): 2958-2964, 2972.
(Gao Chunneng, Zhang Biao, Ji Zhicheng.Research on Optimization of SVM parameters based on adaptive dynamic search particle swarm[J]. Journal of System Simulation (S1004-731X), 2015, 27(12): 2958-2964, 2972.)
[10] 祝晓燕, 张金会, 付士鹏, 等. 基于改进PSO的SVM参数优化及其在风速预测中的应用[J]. 中国电力, 2013, 46(11): 105-108.
(Zhu Xiaoyan, Zhang Jinhui, Fu Shipeng, et al.Optimization of SVM parameters based on improved PSO and its application in wind speed prediction[J]. China Electric Power, 2013, 46(11): 105-108.)
[11] 方伟, 孙俊, 须文波. 一种多样性控制的粒子群优化算法[J]. 控制与决策, 2008, 23(8): 863-868.
(Fang Wei, Sun Jun, Xu Wenbo.A particle swarm optimization algorithm with diversity control[J]. Control And Decision, 2008, 23(8): 863-868.)
[12] 高立群, 吴沛锋, 邹德旋. 基于变异策略的粒子群算法[J]. 东北大学学报(自然科学版), 2010, 31(11): 1530-1533.
(Gao Liqun, Wu Peifeng, Zou Dexuan.Particle swarm optimization algorithm based on mutation strategy[J]. Journal of Northeastern University (Natural Science edition), 2010, 31(11): 1530-1533.)
[13] 李丽, 牛奔. 粒子群优化算法 [M]. 北京: 冶金工业出版社, 2009.
(Li Li, Niu Ben.Particle swarm optimization algorithm [M]. Beijing, China: Metallurgical Industry Press, 2009.)
[14] 胡建秀, 曾建潮. 二阶振荡微粒群算法[J]. 系统仿真学报, 2007, 19(5): 997-999.
(Hu Jianxiu, Zeng Jianchao.Two order oscillating particle swarm algorithm[J]. Journal of System Simulation (S1004-731X), 2007, 19(5): 997-999.)
[15] 张成兴. 时变压缩因子粒子群算法[J]. 计算机工程与应用, 2015, 51(23): 59-64.
(Zhang Chengxing.Time varying compression factor particle swarm optimization[J]. Computer Engineering and Applications, 2015, 51(23): 59-64.)
[16] 魏俊, 周步祥, 林楠, 等. 基于蚁群支持向量机的短期负荷预测[J]. 电力系统保护与控制, 2009, 37(4): 36-40.
(Wei Jun, Zhou Buxiang, Lin Nan, et al.Short term load forecasting based on ant colony support vector machines[J]. Power System Protection and Control, 2009, 37(4): 36-40.)
[17] Eunite. World-wide European intelligence technology network competition load forecast sample data [DB/OL]. (2001-08-01) [2016-10-01]. http://neuron.tuke.sk/ competition/index.php
[18] 康重庆, 周安石, 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 Chongqing, Zhou Anshi, Wang Peng, et al.Analysis of the influence of real-time meteorological factors in short-term load forecasting and its treatment strategy[J]. Power System Technology, 2006, 30(7): 5-10.)
[19] Hinojosa V H, Hoese A.Short-Term Load Forecasting Using Fuzzy Inductive Reasoning and Evolutionary Algorithms[J]. IEEE Transaction on Power Systems (S0885-8950), 2010, 25(1): 565-574.