基于数据挖掘的风电功率预测特征选择方法
|
摘要
输入特征向量的选择是建立风电功率预测模型中至关重要的第一步,但由于风电机组的待选监测量项目过多、部分监测量与风电功率的相关性不明显甚至不相关、信息冗余量大等因素造成输入向量集的选取不够合理,进一步影响功率预测模型的准确性。针对这一问题,在综合对比研究了邻域粗糙集、随机森林和互信息这三种较为有效的用于特征选择的数据挖掘算法的基础上,提出了一种综合性能较好的基于随机森林筛选风电功率预测模型输入向量的方法,并分析了另两种方法的特点和适用范围,最后使用风机的实际运行数据,基于最小二乘支持向量回归算法对文中所提出的方法进行了验证。仿真结果表明,该方法能够通过减少功率预测模型的输入向量有效地降低模型复杂度,不仅加快了模型的预测速度而且提高了预测的精度。
The selection of input feature vector is the first important step in the establishment of wind power prediction model,but due to the excessive monitor items,the correlation between partial monitor items and wind power is not obvious or even irrelevant,and the redundancy information causes the selection of input vector set is not reasonable,so the accuracy of the power prediction model is affected. In order to solve this problem,three effective data mining algorithms for feature selection,namely neighborhood rough set,random forest and mutual information,are studied synthetically. And a new method based on random forest for selecting input vectors of wind power prediction model with better comprehensive performance is proposed,and the characteristics and application range of other methods are analyzed. Finally,the proposed method is validated based on the least squares support vector regression algorithm through using the actual operation data of the turbine. The simulation results show that this method can effectively reduce the complexity of the model by reducing the input vectors,which not only speeds up the prediction speed,but also improves the prediction accuracy of the model.
The selection of input feature vector is the first important step in the establishment of wind power prediction model,but due to the excessive monitor items,the correlation between partial monitor items and wind power is not obvious or even irrelevant,and the redundancy information causes the selection of input vector set is not reasonable,so the accuracy of the power prediction model is affected. In order to solve this problem,three effective data mining algorithms for feature selection,namely neighborhood rough set,random forest and mutual information,are studied synthetically. And a new method based on random forest for selecting input vectors of wind power prediction model with better comprehensive performance is proposed,and the characteristics and application range of other methods are analyzed. Finally,the proposed method is validated based on the least squares support vector regression algorithm through using the actual operation data of the turbine. The simulation results show that this method can effectively reduce the complexity of the model by reducing the input vectors,which not only speeds up the prediction speed,but also improves the prediction accuracy of the model.
引文
[1]智研咨询集团. 2017-2023年中国风电产业竞争格局及发展趋势研究报告[R].北京:智研咨询集团,2017.
[2]牛东晓,范磊磊.风电功率预测方法综述及发展研究[J].现代电力,2013,30(4):24-28.Niu Dongxiao,Fan Leilei. Review and Development Study on Wind Power Prediction Methods[J]. Modern Electric Power,2013,30(4):24-28.
[3]朱乔木,李弘毅,王子琪,等.基于长短期记忆网络的风电场发电功率超短期预测[J].电网技术,2017,41(12):3797-3802.Zhu Qiaomu,Li Hongyi,Wang Ziqi,et al. Short-term Wind Power Forecasting Based on LSTM[J]. Power System Technology,2017,41(12):3797-3802.
[4]姚晟,徐风,赵鹏,等.基于邻域量化容差关系粗糙集模型的特征选择算法[J].模式识别与人工智能,2017,30(5):416-428.Yao Sheng,Xu Feng,Zhao Peng,et al. Feature Selection Algorithm Based on Neighborhood Valued Tolerance Relation Rough Set Model[J]. Pattern Recognition and Artificial Intelligence,2017,30(5):416-428.
[5]胡清华,赵辉,于达仁.基于邻域粗糙集的符号与数值属性快速约简算法[J].模式识别与人工智能,2008,21(6):732-738.Hu Qinghua,Zhao Hui,Yu Daren. Efficient Symbolic and Numerical Attribute Reduction with Neighborhood Rough Sets[J]. Pattern Recognition and Artificial Intelligence,2008,21(6):732-738.
[6]郭晓利,温延立.基于随机森林的风机状态监测数据可视化研究[J].电测与仪表,2016,53(22):12-15.Guo Xiaoli,Wen Yanli. Data visualization research on monitoring data of wind turbines based on random forest[J]. Electrical Measurement&Instrumentation,2016,53(22):12-15.
[7]杨茂,张强.基于最大相关最小冗余相关向量机的风电功率缺失数据补齐研究[J].太阳能学报,2017,38(04):938-944.Yang Mao,Zhang Qiang. Polishing Missinng Data For Wind Power Based on Mrmr of Relevance Vector Machine[J]. Acta Energiae Solaris Sinica,2017,38(4):938-944.
[8]张文修,吴伟志,梁吉业,等.粗糙集理论与方法[M].北京:科学出版社,2001:12-16.
[9]王栋,蒋亮,江陈桢,等.基于混合GEP的配电低压台区综合评价算法[J].电测与仪表,2017,54(9):18-23.Wang Dong,Jiang Liang,Jiang Chenzhen,et al. Comprehensive evaluation algorithm of distribution low voltage station area based on hybrid GEP[J]. Electrical Measurement&Instrumentation,2017,54(9):18-23.
[10]刘兴杰,岑添云,郑文书,等.基于模糊粗糙集与改进聚类的神经网络风速预测[J].中国电机工程学报,2014,34(19):3162-3169.Liu Xingjie,Cen Tianyun,Zheng Wenshu,et al. Neural Network Wind Speed Prediction Based on Fuzzy Rough Set and Improved Clustering[J]. Proceedings of the CSEE,2014,34(19):3162-3169.
[11]尹东阳,盛义发,蒋明洁,等.基于粗糙集理论-主成分分析的Elman神经网络短期风速预测[J].电力系统保护与控制,2014,42(11):46-51.
[12]蒋亚坤,李文云,赵莹,等.粗糙集与证据理论结合的电网运行优质性综合评价[J].电力系统保护与控制,2015,43(13):1-7.
[13]Breiman L. Random forest. Machine Learning[J]. 2001,45(1):5-32.
[14]陈华友.组合预测方法有效性理论及其应用[M].北京:科学出版社,2008.
[15]苗长新,冯学俊,马士亮,等.基于独立分量分析和互信息的多谐波源定位[J].电测与仪表,2014,51(11):60-64.
[16]刘吉臻,秦天牧,杨婷婷,等.基于偏互信息的变量选择方法及其在火电厂SCR系统建模中的应用[J].中国电机工程学报,2016,36(9):2438-2442.
[17]王书舟,伞冶.支持向量机的训练算法综述[J].智能系统学报,2008,2(6):467-475.
[18]Murray J F,Hughes G F,Kreutz-Delgado K. Machine Learning Methods for Predicting Failures in Hard Drives:A Multiple-Instance Application[J]. Journal of Machine Learning Research, 2005, 6(1):783-816.
[19]唐杰明,刘俊勇,刘友波.基于最优FCM聚类和最小二乘支持向量回归的短期电力负荷预测[J].现代电力,2008,25(2):76-81.
[2]牛东晓,范磊磊.风电功率预测方法综述及发展研究[J].现代电力,2013,30(4):24-28.Niu Dongxiao,Fan Leilei. Review and Development Study on Wind Power Prediction Methods[J]. Modern Electric Power,2013,30(4):24-28.
[3]朱乔木,李弘毅,王子琪,等.基于长短期记忆网络的风电场发电功率超短期预测[J].电网技术,2017,41(12):3797-3802.Zhu Qiaomu,Li Hongyi,Wang Ziqi,et al. Short-term Wind Power Forecasting Based on LSTM[J]. Power System Technology,2017,41(12):3797-3802.
[4]姚晟,徐风,赵鹏,等.基于邻域量化容差关系粗糙集模型的特征选择算法[J].模式识别与人工智能,2017,30(5):416-428.Yao Sheng,Xu Feng,Zhao Peng,et al. Feature Selection Algorithm Based on Neighborhood Valued Tolerance Relation Rough Set Model[J]. Pattern Recognition and Artificial Intelligence,2017,30(5):416-428.
[5]胡清华,赵辉,于达仁.基于邻域粗糙集的符号与数值属性快速约简算法[J].模式识别与人工智能,2008,21(6):732-738.Hu Qinghua,Zhao Hui,Yu Daren. Efficient Symbolic and Numerical Attribute Reduction with Neighborhood Rough Sets[J]. Pattern Recognition and Artificial Intelligence,2008,21(6):732-738.
[6]郭晓利,温延立.基于随机森林的风机状态监测数据可视化研究[J].电测与仪表,2016,53(22):12-15.Guo Xiaoli,Wen Yanli. Data visualization research on monitoring data of wind turbines based on random forest[J]. Electrical Measurement&Instrumentation,2016,53(22):12-15.
[7]杨茂,张强.基于最大相关最小冗余相关向量机的风电功率缺失数据补齐研究[J].太阳能学报,2017,38(04):938-944.Yang Mao,Zhang Qiang. Polishing Missinng Data For Wind Power Based on Mrmr of Relevance Vector Machine[J]. Acta Energiae Solaris Sinica,2017,38(4):938-944.
[8]张文修,吴伟志,梁吉业,等.粗糙集理论与方法[M].北京:科学出版社,2001:12-16.
[9]王栋,蒋亮,江陈桢,等.基于混合GEP的配电低压台区综合评价算法[J].电测与仪表,2017,54(9):18-23.Wang Dong,Jiang Liang,Jiang Chenzhen,et al. Comprehensive evaluation algorithm of distribution low voltage station area based on hybrid GEP[J]. Electrical Measurement&Instrumentation,2017,54(9):18-23.
[10]刘兴杰,岑添云,郑文书,等.基于模糊粗糙集与改进聚类的神经网络风速预测[J].中国电机工程学报,2014,34(19):3162-3169.Liu Xingjie,Cen Tianyun,Zheng Wenshu,et al. Neural Network Wind Speed Prediction Based on Fuzzy Rough Set and Improved Clustering[J]. Proceedings of the CSEE,2014,34(19):3162-3169.
[11]尹东阳,盛义发,蒋明洁,等.基于粗糙集理论-主成分分析的Elman神经网络短期风速预测[J].电力系统保护与控制,2014,42(11):46-51.
[12]蒋亚坤,李文云,赵莹,等.粗糙集与证据理论结合的电网运行优质性综合评价[J].电力系统保护与控制,2015,43(13):1-7.
[13]Breiman L. Random forest. Machine Learning[J]. 2001,45(1):5-32.
[14]陈华友.组合预测方法有效性理论及其应用[M].北京:科学出版社,2008.
[15]苗长新,冯学俊,马士亮,等.基于独立分量分析和互信息的多谐波源定位[J].电测与仪表,2014,51(11):60-64.
[16]刘吉臻,秦天牧,杨婷婷,等.基于偏互信息的变量选择方法及其在火电厂SCR系统建模中的应用[J].中国电机工程学报,2016,36(9):2438-2442.
[17]王书舟,伞冶.支持向量机的训练算法综述[J].智能系统学报,2008,2(6):467-475.
[18]Murray J F,Hughes G F,Kreutz-Delgado K. Machine Learning Methods for Predicting Failures in Hard Drives:A Multiple-Instance Application[J]. Journal of Machine Learning Research, 2005, 6(1):783-816.
[19]唐杰明,刘俊勇,刘友波.基于最优FCM聚类和最小二乘支持向量回归的短期电力负荷预测[J].现代电力,2008,25(2):76-81.