基于相似数据选取和改进梯度提升决策树的电力负荷预测
|
摘要
针对高精度电力负荷预测问题,构建了相似数据选取和改进梯度提升决策树的新预测方法。该方法借助灰色关联分析等方法计算历史日与待预测日在气象、时间和前趋势等类特征因子上的局部相似度,依据取小综合相似度选择相似历史日数据组成训练数据集;进而,引入相似度加权损失函数,改进梯度提升决策树算法。仿真结果表明,其预测平均绝对百分比误差小于2.2%,日最大误差不超过6%;与BP神经网络和梯度提升决策树相比,其日平均绝对误差、日最大误差及周平均误差均方差分别减少1.136%和0.316%、4.738%和1.324%以及1.062和0.822。
In consideration of the high-accuracy power load forecasting problem,a novel forecasting method based on similar-data selection and improved gradient boosting decision tree(GBDT)is proposed. In this method,the local similar degree for characteristic factors(e.g.,weather,time,and trend)on historical days and the day to be forecasted is calculated using methods including gray correlation analysis,thus a training data set is formed by selecting data on similar historical days according to the minimum synthetic similar degree. Then,a similar degree weighted loss function is introduced to improve the GBDT method. Simulation results show that the mean absolute percentage error(MAPE)is less than 2.2%,and the daily maximum absolute percentage error(MaxE)is less than 6%;compared with the values obtained using the BP neural network and GBDT methods,the values of daily MAPE,daily MaxE,and weekly mean square error(MSE)calculated using the novel method are reduced by 1.136% and 0.316%,4.738% and 1.324%,1.062 and 0.822,respectively.
In consideration of the high-accuracy power load forecasting problem,a novel forecasting method based on similar-data selection and improved gradient boosting decision tree(GBDT)is proposed. In this method,the local similar degree for characteristic factors(e.g.,weather,time,and trend)on historical days and the day to be forecasted is calculated using methods including gray correlation analysis,thus a training data set is formed by selecting data on similar historical days according to the minimum synthetic similar degree. Then,a similar degree weighted loss function is introduced to improve the GBDT method. Simulation results show that the mean absolute percentage error(MAPE)is less than 2.2%,and the daily maximum absolute percentage error(MaxE)is less than 6%;compared with the values obtained using the BP neural network and GBDT methods,the values of daily MAPE,daily MaxE,and weekly mean square error(MSE)calculated using the novel method are reduced by 1.136% and 0.316%,4.738% and 1.324%,1.062 and 0.822,respectively.
引文
[1]毛李帆,姚建刚,金永顺,等(Mao Lifan,Yao Jiangang,Jin Yongshun,et al).中长期电力组合预测模型的理论研究(Theoretical study of combination model for medium and long term load forecasting)[J].中国电机工程学报(Proceedings of the CSEE),2010,30(16):53-59.
[2] Jin Min,Zhou Xiang,Zhang Zhiming,et al. Short-term power load forecasting using grey correlation contest modeling[J]. Expert Systems with Applications,2012,39(1):773-779.
[3]中国电机工程学会信息化专委会.中国电力大数据发展白皮书[M].北京:中国电力出版社,2013.
[4] Ballini R,Yager R R. OWA filters and forecasting models applied to electric power load time series[J]. Evolving Systems,2014,5(3):159-173.
[5]崔和瑞,彭旭(Cui Herui,Peng Xu).基于ARIMAX模型的夏季短期电力负荷预测(Summer short-term load forecasting based on ARIMAX model)[J].电力系统保护与控制(Power System Protection and Control),2015,43(4):108-114.
[6] Javedani S H,Gadelha G F,Cidiney J da S,et al. Shortterm load forecasting method based on fuzzy time series,seasonality and long memory process[J]. International Journal of Approximate Reasoning,2017,83:196-217.
[7] Fan Guofeng,Shan Qing,Wang Hua,et al. Support vector regression model based on empirical mode decomposition and auto regression for electric load forecasting[J]. Energies,2013,6(4):1887-1901.
[8]张素香,赵丙镇,王风雨,等(Zhang Suxiang,Zhao Bingzhen,Wang Fengyu,et al).海量数据下的电力负荷短期预测(Short-term power load forecasting based on big data)[J].中国电机工程学报(Proceedings of the CSEE),2015,35(1):37-42.
[9] Dudek G. Pattern-based local linear regression models for short-term load forecasting[J]. Electric Power Systems Research,2016,130:139-147
[10] Niu Dongxiao,Wang Yongli,Dash Wu Desheng. Power load forecasting using support vector machine and ant colony optimization[J]. Expert Systems with Applications,2010,37(3):2531-2539.
[11]王大鹏,汪秉文(Wang Dapeng,Wang Bingwen).基于变权缓冲灰色模型的中长期负荷预测(Medium-and longterm load forecasting based on variable weights buffer grey model)[J].电网技术(Power System Technology),2013,37(1):167-171.
[12] Khadse C B,Chaudhari M A,Borghate V B. Conjugate gradient back-propagation based artificial neural network for real time power quality assessment[J]. International Journal of Electrical Power&Energy Systems,2016,82:197-206.
[13] He Yaoyao,Xu Qifa,Wan Jinhong,et al. Short-term power load probability density forecasting based on quantified regression neural network and triangle kernel function[J].Energy,2016,114:498-512.
[14]谷云东,张素杰,冯君淑(Gu Yundong,Zhang Sujie,Feng Junshu).大用户电力负荷的多模型模糊综合预测(Multi-model fuzzy synthesis forecasting of electric power loads for larger consumers)[J].电工技术学报(Transactions of China Electrotechnical Society),2015,30(23):110-115.
[15] Wang Y,Niu D,Ji L. Short-term power load forecasting based on IVL-BP neural network technology[J]. Systems Engineering Procedia,2012,4:168-174.
[16]张永峰,崔凯(Zhang Yongfeng,Cui Kai).应用共识PSO协同Trust-Tech方法的短期负荷预测(Consensus-based PSO-assisted Trust-Tech method for short-term load forecasting)[J].电力系统及其自动化学报(Proceedings of the CSU-EPSA),2017,29(7):7-12.
[17]王保义,王冬阳,张少敏(Wang Baoyi,Wang Dongyang,Zhang Shaomin).基于Spark和IPPSO_LSSVM的短期分布式电力负荷预测算法(Distributed short-term load forecasting algorithm based on Spark and IPPSO_LSSVM)[J].电力系统自动化设备(Electric Power Automation Equipment),2016,36(1):117-122.
[18]肖白,聂鹏,穆钢,等(Xiao Bai,Nie Peng,Mu Gang,et al).基于多级聚类分析和支持向量机的空间负荷预测方法(A spatial load forecasting method based on multilevel clustering analysis and support vector machine)[J].电力系统自动化(Automation of Electric Power Systems),2015,39(12):56-61.
[19] Zhang Xiaobo,Wang Jianzhou,Zhang Kequan. Short-term electric load forecasting based on singular spectrum analysis and support vector machine optimized by Cuckoo search algorithm[J]. Electric Power Systems Research,2017,146:270–285.
[20]吴潇雨,和敬涵,张沛,等(Wu Xiaoyu,He Jinghan,Zhang Pei,et al).基于灰色投影改进随机森林算法的电力系统短期负荷预测(Power system short-term load forecasting based on improved random forest with gray relation projection)[J].电力系统自动化(Automation of Electric Power Systems),2015,39(12):50-55.
[21]王德文,孙志伟(Wang Dewen,Sun Zhiwei).电力用户侧大数据分析与并行负荷预测(Big data analysis and parallel load forecasting of electric power user side)[J].中国电机工程学报(Proceedings of the CSEE),2015,35(3):527-537.
[22] Huang Nantian,Lu Guobo,Xu Dianguo. A permutation importance-based feature selection method for short-term electricity load forecasting using random forest[J]. Energies,2016,9(10):767-790.
[23] Ebrahimi A,Moshari A. Holidays short-term load forecasting using fuzzy improved similar day method[J]. International Transactions on Electrical Energy Systems,2013,23(8):1254–1271.
[24]李啸骢,李春涛,从兰美,等(Li Xiaocong,Li Chuntao,Cong Lanmei,et al).基于动态权值相似日选取算法的短期负荷预测(Short-term load forecasting based on dynamic weight similar day selection algorithm)[J].电力系统保护与控制(Power System Protection and Control),2017,45(6):1-8.
[25]王立平,邓芳明(Wang Liping,Deng Fangming).基于小波包和GBDT的瓦斯传感器故障诊断(Fault diagnosis for Gas sensor based on wavelet package and GBDT)[J].测控技术(Measurement&Control Technology),2016,35(12):31-37.
[26]李航.统计学习方法[M].北京:清华大学出版社,2012.
[27]周志华.机器学习[M].北京:清华大学出版社,2016.
[2] Jin Min,Zhou Xiang,Zhang Zhiming,et al. Short-term power load forecasting using grey correlation contest modeling[J]. Expert Systems with Applications,2012,39(1):773-779.
[3]中国电机工程学会信息化专委会.中国电力大数据发展白皮书[M].北京:中国电力出版社,2013.
[4] Ballini R,Yager R R. OWA filters and forecasting models applied to electric power load time series[J]. Evolving Systems,2014,5(3):159-173.
[5]崔和瑞,彭旭(Cui Herui,Peng Xu).基于ARIMAX模型的夏季短期电力负荷预测(Summer short-term load forecasting based on ARIMAX model)[J].电力系统保护与控制(Power System Protection and Control),2015,43(4):108-114.
[6] Javedani S H,Gadelha G F,Cidiney J da S,et al. Shortterm load forecasting method based on fuzzy time series,seasonality and long memory process[J]. International Journal of Approximate Reasoning,2017,83:196-217.
[7] Fan Guofeng,Shan Qing,Wang Hua,et al. Support vector regression model based on empirical mode decomposition and auto regression for electric load forecasting[J]. Energies,2013,6(4):1887-1901.
[8]张素香,赵丙镇,王风雨,等(Zhang Suxiang,Zhao Bingzhen,Wang Fengyu,et al).海量数据下的电力负荷短期预测(Short-term power load forecasting based on big data)[J].中国电机工程学报(Proceedings of the CSEE),2015,35(1):37-42.
[9] Dudek G. Pattern-based local linear regression models for short-term load forecasting[J]. Electric Power Systems Research,2016,130:139-147
[10] Niu Dongxiao,Wang Yongli,Dash Wu Desheng. Power load forecasting using support vector machine and ant colony optimization[J]. Expert Systems with Applications,2010,37(3):2531-2539.
[11]王大鹏,汪秉文(Wang Dapeng,Wang Bingwen).基于变权缓冲灰色模型的中长期负荷预测(Medium-and longterm load forecasting based on variable weights buffer grey model)[J].电网技术(Power System Technology),2013,37(1):167-171.
[12] Khadse C B,Chaudhari M A,Borghate V B. Conjugate gradient back-propagation based artificial neural network for real time power quality assessment[J]. International Journal of Electrical Power&Energy Systems,2016,82:197-206.
[13] He Yaoyao,Xu Qifa,Wan Jinhong,et al. Short-term power load probability density forecasting based on quantified regression neural network and triangle kernel function[J].Energy,2016,114:498-512.
[14]谷云东,张素杰,冯君淑(Gu Yundong,Zhang Sujie,Feng Junshu).大用户电力负荷的多模型模糊综合预测(Multi-model fuzzy synthesis forecasting of electric power loads for larger consumers)[J].电工技术学报(Transactions of China Electrotechnical Society),2015,30(23):110-115.
[15] Wang Y,Niu D,Ji L. Short-term power load forecasting based on IVL-BP neural network technology[J]. Systems Engineering Procedia,2012,4:168-174.
[16]张永峰,崔凯(Zhang Yongfeng,Cui Kai).应用共识PSO协同Trust-Tech方法的短期负荷预测(Consensus-based PSO-assisted Trust-Tech method for short-term load forecasting)[J].电力系统及其自动化学报(Proceedings of the CSU-EPSA),2017,29(7):7-12.
[17]王保义,王冬阳,张少敏(Wang Baoyi,Wang Dongyang,Zhang Shaomin).基于Spark和IPPSO_LSSVM的短期分布式电力负荷预测算法(Distributed short-term load forecasting algorithm based on Spark and IPPSO_LSSVM)[J].电力系统自动化设备(Electric Power Automation Equipment),2016,36(1):117-122.
[18]肖白,聂鹏,穆钢,等(Xiao Bai,Nie Peng,Mu Gang,et al).基于多级聚类分析和支持向量机的空间负荷预测方法(A spatial load forecasting method based on multilevel clustering analysis and support vector machine)[J].电力系统自动化(Automation of Electric Power Systems),2015,39(12):56-61.
[19] Zhang Xiaobo,Wang Jianzhou,Zhang Kequan. Short-term electric load forecasting based on singular spectrum analysis and support vector machine optimized by Cuckoo search algorithm[J]. Electric Power Systems Research,2017,146:270–285.
[20]吴潇雨,和敬涵,张沛,等(Wu Xiaoyu,He Jinghan,Zhang Pei,et al).基于灰色投影改进随机森林算法的电力系统短期负荷预测(Power system short-term load forecasting based on improved random forest with gray relation projection)[J].电力系统自动化(Automation of Electric Power Systems),2015,39(12):50-55.
[21]王德文,孙志伟(Wang Dewen,Sun Zhiwei).电力用户侧大数据分析与并行负荷预测(Big data analysis and parallel load forecasting of electric power user side)[J].中国电机工程学报(Proceedings of the CSEE),2015,35(3):527-537.
[22] Huang Nantian,Lu Guobo,Xu Dianguo. A permutation importance-based feature selection method for short-term electricity load forecasting using random forest[J]. Energies,2016,9(10):767-790.
[23] Ebrahimi A,Moshari A. Holidays short-term load forecasting using fuzzy improved similar day method[J]. International Transactions on Electrical Energy Systems,2013,23(8):1254–1271.
[24]李啸骢,李春涛,从兰美,等(Li Xiaocong,Li Chuntao,Cong Lanmei,et al).基于动态权值相似日选取算法的短期负荷预测(Short-term load forecasting based on dynamic weight similar day selection algorithm)[J].电力系统保护与控制(Power System Protection and Control),2017,45(6):1-8.
[25]王立平,邓芳明(Wang Liping,Deng Fangming).基于小波包和GBDT的瓦斯传感器故障诊断(Fault diagnosis for Gas sensor based on wavelet package and GBDT)[J].测控技术(Measurement&Control Technology),2016,35(12):31-37.
[26]李航.统计学习方法[M].北京:清华大学出版社,2012.
[27]周志华.机器学习[M].北京:清华大学出版社,2016.