基于改进聚类融合的办公型建筑用电行为分析
|
摘要
以上海市长宁区的大型办公建筑为研究对象,利用数据分析方法分析其用电行为与节能潜力。针对传统用电行为分析,采用单一聚类算法拓展性较差的问题,文章提出通过优选方法进行聚类融合以吸收不同算法优点,增强算法适应能力。首先进行方法优选,针对聚类效果评价指标的不一致问题,提出综合聚类评价指标并对R语言库中大量的单一聚类方法进行评价,采用基于簇的相似度划分算法(CSPA)进行聚类融合。试验集的结果表明该聚类融合方法具有更好的有效性。利用该改进聚类融合算法对用户负荷曲线进行聚类,提取用户用电模式,分析其用电构成与特征,并进行节能策略的分析。结果表明,该办公类建筑具有4类基本用电模式,且有一定节能潜力。
In this paper,a large office building in Changning District( Shanghai) is studied to analyze its electricity consumption behavior and energy-saving potential using data analysis methods. A cluster ensemble model using optimizing clustering algorithms is proposed to solve the problem of poor scalability of single clustering algorithms,which are used frequently in this field. Firstly,during the period of selecting algorithms,a comprehensive clustering evaluation index is proposed for the problem of the inconsistency of indicators. Then different clustering algorithms in R library are evaluated,and results are fused by cluster-based similarity partitioning algorithm( CSPA). The results show that the cluster ensemble model is more effective. Users' consumption patterns are extracted by this improved cluster ensemble algorithm. Then constitution and characteristics of different patterns and energy conservation strategies are analyzed. The results show that there are 4 different consumption patterns and certain energy saving potential of this large-scale office building.
In this paper,a large office building in Changning District( Shanghai) is studied to analyze its electricity consumption behavior and energy-saving potential using data analysis methods. A cluster ensemble model using optimizing clustering algorithms is proposed to solve the problem of poor scalability of single clustering algorithms,which are used frequently in this field. Firstly,during the period of selecting algorithms,a comprehensive clustering evaluation index is proposed for the problem of the inconsistency of indicators. Then different clustering algorithms in R library are evaluated,and results are fused by cluster-based similarity partitioning algorithm( CSPA). The results show that the cluster ensemble model is more effective. Users' consumption patterns are extracted by this improved cluster ensemble algorithm. Then constitution and characteristics of different patterns and energy conservation strategies are analyzed. The results show that there are 4 different consumption patterns and certain energy saving potential of this large-scale office building.
引文
[1]王继业,季知祥,史梦洁,等.智能配用电大数据需求分析与应用研究[J].中国电机工程学报,2015,35(8):1829-1836.WANG Jiye,JI Zhixiang,SHI M engjie,et. al. Scenario analysis and application research on big data in smart pow er distribution and consumption systems[J]. Proceedings of the CSEE,2015,35(8):1829-1836.
[2]徐磊,杨秀,张美霞.基于数据挖掘的工业用户用电行为分析[J].电测与仪表,2017,54(16):68-74.XU Lei,YANG Xiu,ZHANG M eixia. Industrial users of electricity behavior analysis based on data mining[J]. Electrical M easurement&Instrumentation,2017,54(16):68-74.
[3]胡长华.基于大用户用电行为分析的错峰管理系统研究与设计[[J].现代计算机:专业版,2014(14):42-47.HU Changhua. Research and design of peak load shifting management system based on analysis of electrical behavior[J].M odern Computer,2014(14):42-47.
[4] LI Kangping,WANG Fei, ZHEN Zhao, et. al, Analysis on residential electricity consumption behavior using improved k-means based on simulated annealing algorithm[C]//Pow er&Energy Conference at Illinois. IEEE,2016.
[5]林锦波.聚类融合与深度学习在用电负荷模式识别的应用研究[D].广州:华南理工大学,2014.LIN Jinbo. Application of recognition model for pow er load in cluster ensemble and deep learning[D]. Guangzhou:South China University of Technology,2014.
[6]彭研枫.基于聚类算法的大用户用电行为研究与应用[D].北京:华北电力大学,2016.PENG Yanfeng,Research and application of the behaviors of users demanding for large amounts of electricity based on the clustering algorithm[D]. Beijing:North China Electric Pow er University,2016.
[7]李智勇,吴晶莹,吴为麟,等.基于自组织映射神经网络的电力用户负荷曲线聚类[J].电力系统自动化,2008,32(15):66-70,78.LI Zhiyong,WU Jingying,WU Weilin,et. al,Customers load profile clustering using the som neural netw ork[J]. Automation of Electric Pow er Systems,2008,32(15):66-70,78.
[8]赵莉,候兴哲,胡君,等.基于改进k-means算法的海量智能用电数据分析[J].电网技术,2014,38(10):2715-2720.ZHAO Li, HOU Xingzhe, HU Jun, et al, Improved k-means algorithm based analysis on massive data of intelligent pow er utilization[J]. Pow er System Technology, 2014,38(10):2715-2720.
[9] HANSEN L K,SALAMON P. Neural network ensembles[J].IEEE Transactions on Pattern Analysis and M achine Intelligence,1990,12(10):993-1001.
[10]侯娟,费耀平,胡小霞,等.基于先验信息和谱分析的聚类融合算法[J].计算机应用研究,2010,27(6):2103-2105.HOU Juan,FEI Yaoping,HU Xiaoxia,et. al,Clustering ensemblealgorithm based on priorknow ledge and spectral analysis[J].Application Research of Computers,2010,27(6):2103-2105.
[11]刘丽敏.选择性聚类融合算法研究[D].长沙:中南大学,2013.LIU Limin,Study on clustering ensemble selection algorithm[D].Changsha:Central South University,2013.
[12]杨燕,靳蕃,KAMEL Mohamed.聚类有效性评价综述[J].计算机应用研究,2008(6):1630-1632,1638.YANG Yan,JIN Fan,KAM EL M ohamed,Survey of clustering validity evaluation[J]. Application Research of Computers,2008(6):1630-1632,1638.
[13]白亮,高锦.基于聚类准则融合的加权聚类集成算法[J].山西大学学报(自然科学版),2018,41(2):302-307.BAI Liang,GAO Jin. A w eighted cluster ensemble algorithm based on clustering criterion integration[J]. Journal of Shanxi University(Natural Science Edition),2018,41(2):302-307.
[14]赵一明.基于Metis图划分算法的图平衡划分方法[D].西安:西安电子科技大学,2014.ZHAO Yiming. A density balance aw are partition methodology based on graph partitioning program metis[D]. Xi'an:Xidian University,2014.
[15] IAMON N,BOONGOEN T,GARRETT S,et al. A link-based approach to the cluster ensemble problem[J]. IEEE Transactions on Pattern Analysis&M achine Intelligence, 2011, 33(12):2396-2409.
[2]徐磊,杨秀,张美霞.基于数据挖掘的工业用户用电行为分析[J].电测与仪表,2017,54(16):68-74.XU Lei,YANG Xiu,ZHANG M eixia. Industrial users of electricity behavior analysis based on data mining[J]. Electrical M easurement&Instrumentation,2017,54(16):68-74.
[3]胡长华.基于大用户用电行为分析的错峰管理系统研究与设计[[J].现代计算机:专业版,2014(14):42-47.HU Changhua. Research and design of peak load shifting management system based on analysis of electrical behavior[J].M odern Computer,2014(14):42-47.
[4] LI Kangping,WANG Fei, ZHEN Zhao, et. al, Analysis on residential electricity consumption behavior using improved k-means based on simulated annealing algorithm[C]//Pow er&Energy Conference at Illinois. IEEE,2016.
[5]林锦波.聚类融合与深度学习在用电负荷模式识别的应用研究[D].广州:华南理工大学,2014.LIN Jinbo. Application of recognition model for pow er load in cluster ensemble and deep learning[D]. Guangzhou:South China University of Technology,2014.
[6]彭研枫.基于聚类算法的大用户用电行为研究与应用[D].北京:华北电力大学,2016.PENG Yanfeng,Research and application of the behaviors of users demanding for large amounts of electricity based on the clustering algorithm[D]. Beijing:North China Electric Pow er University,2016.
[7]李智勇,吴晶莹,吴为麟,等.基于自组织映射神经网络的电力用户负荷曲线聚类[J].电力系统自动化,2008,32(15):66-70,78.LI Zhiyong,WU Jingying,WU Weilin,et. al,Customers load profile clustering using the som neural netw ork[J]. Automation of Electric Pow er Systems,2008,32(15):66-70,78.
[8]赵莉,候兴哲,胡君,等.基于改进k-means算法的海量智能用电数据分析[J].电网技术,2014,38(10):2715-2720.ZHAO Li, HOU Xingzhe, HU Jun, et al, Improved k-means algorithm based analysis on massive data of intelligent pow er utilization[J]. Pow er System Technology, 2014,38(10):2715-2720.
[9] HANSEN L K,SALAMON P. Neural network ensembles[J].IEEE Transactions on Pattern Analysis and M achine Intelligence,1990,12(10):993-1001.
[10]侯娟,费耀平,胡小霞,等.基于先验信息和谱分析的聚类融合算法[J].计算机应用研究,2010,27(6):2103-2105.HOU Juan,FEI Yaoping,HU Xiaoxia,et. al,Clustering ensemblealgorithm based on priorknow ledge and spectral analysis[J].Application Research of Computers,2010,27(6):2103-2105.
[11]刘丽敏.选择性聚类融合算法研究[D].长沙:中南大学,2013.LIU Limin,Study on clustering ensemble selection algorithm[D].Changsha:Central South University,2013.
[12]杨燕,靳蕃,KAMEL Mohamed.聚类有效性评价综述[J].计算机应用研究,2008(6):1630-1632,1638.YANG Yan,JIN Fan,KAM EL M ohamed,Survey of clustering validity evaluation[J]. Application Research of Computers,2008(6):1630-1632,1638.
[13]白亮,高锦.基于聚类准则融合的加权聚类集成算法[J].山西大学学报(自然科学版),2018,41(2):302-307.BAI Liang,GAO Jin. A w eighted cluster ensemble algorithm based on clustering criterion integration[J]. Journal of Shanxi University(Natural Science Edition),2018,41(2):302-307.
[14]赵一明.基于Metis图划分算法的图平衡划分方法[D].西安:西安电子科技大学,2014.ZHAO Yiming. A density balance aw are partition methodology based on graph partitioning program metis[D]. Xi'an:Xidian University,2014.
[15] IAMON N,BOONGOEN T,GARRETT S,et al. A link-based approach to the cluster ensemble problem[J]. IEEE Transactions on Pattern Analysis&M achine Intelligence, 2011, 33(12):2396-2409.