仲赞,邢翼,俞伟,李健,刘广生
ZHONG Zan,XING Yi,YU Wei,LI Jian,LIU Guangsheng

• 1:国网浙江省电力有限公司湖州供电公司
• 2:国网浙江省电力有限公司
• 3:四川大学电气工程学院

摘要(Abstract)：

售电企业对电力用户进行合理评估是开展售电业务的关键。针对当前评估方法存在的评估不全面、应用性不强等问题，提出一种基于WRSR（加权秩和比）和Catboost算法的电力用户分类模型。首先使用WRSR对现有电力用户进行分档并标记；接着使用Catboost算法学习分类规律，构建分类器，同时采用BOA（贝叶斯优化算法）优化Catboost的超参数，提升分类效果；最后根据模型分析每个特征的重要程度，并按重要性分数对用户特征进行筛选。实验结果表明：该方法能实现电力用户的合理分类；所提分类模型与其他机器学习模型相比准确性更高，可解释性更好。
Power sales business stands on reasonable assessment of power users by power sales enterprises. The present assessment methods are incomprehensive and less applied. Therefore, a power user classification model is proposed based on WRSR(weighted rank-sum ratio) and Catboost algorithm. Firstly, the existing power users are graded and labeled using WRSR. Then, the Catboost algorithm is used to learn the classification law and construct a classifier. The hyperparameters of Catboost are optimized using BOA(Bayesian optimization algorithm) to improve the classification effect. Finally, the importance of each feature is analyzed according to the model, and the user features are filtered by the importance scores. The experimental results show that the method can reasonably classify power users, and the proposed classification model is superior to other machine learning models in accuracy and interpretability.

关键词(KeyWords)： 电力用户标签;加权秩和比;用户分类;Catboost;贝叶斯优化
power user labels;weighted rank-sum ratio;user classification;Catboost;Bayesian optimization

Abstract:

Keywords:

基金项目(Foundation): 国网浙江省电力有限公司科技项目（5211UZ2000K2）

作者(Author): 仲赞,邢翼,俞伟,李健,刘广生
ZHONG Zan,XING Yi,YU Wei,LI Jian,LIU Guangsheng

DOI: 10.19585/j.zjdl.202305009

参考文献(References)：

• [1]中共中央国务院.关于进一步深化电力体制改革的若干意见（中发[2015]9号）[Z].2015.
• [2]王星华，刘升伟，陈豪君，等.考虑用户差异性的售电公司需求响应电价模型[J].电力建设，2019,40(9):116-123.WANG Xinghua,LIU Shengwei,CHEN Haojun,et al.Demand response pricing model for power sales companies considering user differences[J].Electric Power Construction,2019,40(9):116-123.
• [3]胡晨，杜松怀，苏娟，等.新电改背景下我国售电公司的购售电途径与经营模式探讨[J].电网技术，2016,40(11):3293-3299.HU Chen,DU Songhuai,SU Juan,et al. Preliminary research of trading approach and management modes of Chinese electricity retail companies under new electricity market reform[J].Power System Technology,2016,40(11):3293-3299.
• [4]李泓泽，郭森，王宝.基于遗传改进蚁群聚类算法的电力客户价值评价[J].电网技术，2012,36(12):256-261.LI Hongze,GUO Sen,WANG Bao.Evaluation on power customer value based on ants colony clustering algorithm optimized by genetic algorithm[J]. Power System Technology,2012,36(12):256-261.
• [5]杨捷，段明明，洪锋.基于层次分析和熵值法的电力用户信用综合评价模型研究[J].云南大学学报（自然科学版），2020,42（增刊2）:13-17.YANG Jie,DUAN Mingming,HONG Feng. The research on comprehensive credit evaluation model of electric power users based on Analytic Hierarchy Process and Entropy Weight Method[J].Journal of Yunnan University(Natural Sciences Edition),2020,42(S2):13-17.
• [6]余培，刘其辉，石城.新电改背景下电力用户信用风险预警模型与评价方法[J].电力需求侧管理，2020,22(2):34-38.YU Pei,LIU Qihui,SHI Cheng.Warning model and user credit risk evaluation method under the background of new power grid reform[J].Power Demand Side Management,2020,22(2):34-38.
• [7]王玉萍，刘磊，朱明，等.贵州电力交易市场主体信用评级模型研究[J].电力需求侧管理，2018,20(5):52-55.WANG Yuping,LIU Lei,ZHU Ming,et al.Research on credit rating model of players in Guizhou power trading market[J]. Power Demand Side Management,2018,20(5):52-55.
• [8]李蕊，李跃，徐浩，等.基于层次分析法和专家经验的重要电力用户典型供电模式评估[J].电网技术，2014,38(9):2336-2341.LI Rui,LI Yue,XU Hao,et al. Assessment on typical power supply mode for important power consumers based on analytical hierarchy process and expert experience[J].Power System Technology,2014,38(9):2336-2341.
• [9]宋美琦，陈烨，张瑞.用户画像研究述评[J].情报科学，2019,37(4):171-177.SONG Meiqi,CHEN Ye,ZHANG Rui.A review of user profile research[J]. Information Science,2019,37(4):171-177.
• [10]徐涛，黄莉，李敏蕾，等.基于多维细粒度行为数据的居民用户画像方法研究[J].电力需求侧管理，2019,21(3):47-52.XU Tao,HUANG Li,LI Minlei,et al.Research on portrait method of residential users based on multidimensional fine-grained behavior data[J].Power Demand Side Management,2019,21(3):47-52.
• [11]王利利，张琳娟，许长清，等.能源互联网背景下园区用户画像及成熟度评价模型研究[J].中国电力，2020,53(8):19-28.WANG Lili,ZHANG Linjuan,XU Changqing,et al.Research on park users portrait and maturity evaluation model under the background of energy Internet[J]. Electric Power,2020,53(8):19-28.
• [12]赵晋泉，夏雪，刘子文，等.电力用户用电特征选择与行为画像[J].电网技术，2020,44(9):3488-3496.ZHAO Jinquan,XIA Xue,LIU Ziwen,et al.User electricity consumption feature selection and behavioral portrait[J].Power System Technology,2020,44(9):3488-3496.
• [13]曹宏宇，刘惠颖，殷鑫，等.一种基于环境特征的智能电能表初值优选型K-means聚类算法[J].电测与仪表，2022,59(7):170-174.CAO Hongyu,LIU Huiying,YIN Xin,et al. An initial value optimization K-means clustering algorithm of smart meters based on environmental features[J].Electrical Measurement&Instrumentation,2022,59(7):170-174.
• [14]谢伟，赵琦，郭乃网，等.改进的并行模糊核聚类算法在电力负荷预测的应用[J].电测与仪表，2019,56(11):49-54.XIE Wei,ZHAO Qi,GUO Naiwang,et al.Application of the improved parallel fuzzy kernel C-means clustering algorithm in power load forecasting[J]. Electrical Measurement&Instrumentation,2019,56(11):49-54.
• [15]徐冰涵，孙云莲，易仕敏，等.基于模糊聚类分类与Elman神经网络算法的居民用户短期用电量预测及修正方法[J].电测与仪表，2020,57(5):1-7.XU Binghan,SUN Yunlian,YI Shimin,et al.Short-term electricity consumption forecasting and correcting method for residential users based on fuzzy clustering classification and Elman neural network algorithm[J]. Electrical Measurement&Instrumentation,2020,57(5):1-7.
• [16]徐碧霞，姚卫光.基于秩和比法评价粤港澳大湾区卫生资源配置现状[J].现代预防医学，2021,48(3):473-476.XU Bixia,YAO Weiguang. Evaluation of health resource allocation in Guangdong-HongKong-Macao greater bay area based on rank-sum ratio method[J].Modern Preventive Medicine,2021,48(3):473-476.
• [17] FU X Q,CHEN H Y,CAI R Q,et al.Optimal allocation and adaptive VAR control of PV-DG in distribution networks[J].Applied Energy,2015,137:173-182.
• [18] DOROGUSH A V,ERSHOV V,GULIN A.CatBoost:gradient boosting with categorical features support[EB/OL].[2022-08-12].https：//arxiv.org/abs/1810.11363.
• [19] LIU W R,DENG K Y,ZHANG X Y,et al. A semisupervised tri-CatBoost method for driving style recognition[J].Symmetry,2020,12(3):336.
• [20] SNOEK J,LAROCHELLE H,ADAMS R P. Practical Bayesian optimization of machine learning algorithms[C]//Proceedings of the 25th International Conference on Neural Information Processing Systems. New York:ACM,2012:2951-2959.
• [21]崔佳旭，杨博.贝叶斯优化方法和应用综述[J].软件学报，2018,29(10):3068-3090.CUI Jiaxu,YANG Bo. Survey on Bayesian optimization methodology and applications[J]. Journal of Software,2018,29(10):3068-3090.
• [22] GHAHRAMANI Z.Probabilistic machine learning and artificial intelligence[J].Nature,2015,521(7553):452-459.
• [23] SRINIVAS N,KRAUSE A,KAKADE S M,et al.Information-theoretic regret bounds for Gaussian process optimization in the bandit setting[J]. IEEE Transactions on Information Theory,2012,58(5):3250-3265.
• [24]许传龙，张粒子，陈大宇，等.基于预招标的月度偏差电量平衡机制及其多周期发电调度优化模型[J].中国电机工程学报，2019,39(17):5085-5094.XU Chuanlong,ZHANG Lizi,CHEN Dayu,et al.Monthly deviation power balance mechanism based on prebidding and its multi-cycle power generation scheduling optimization model[J].Proceedings of the CSEE,2019,39(17):5085-5094.
• [25]国家发展改革委，国家能源局.关于印发各省级行政区域2020年可再生能源电力消纳责任权重的通知关于建立健全可再生能源电力消纳保障机制的通知（发改能源[2020]767）[EB/OL].(2020-06-01)[2022-08-12].http：//www.nea.gov.cn/2020-06/01/c_139105253.htm.
• [26]李刚，李建平，孙晓蕾，等.兼顾序信息和强度信息的主客观组合赋权法研究[J].中国管理科学，2017,25(12):179-187.LI Gang,LI Jianping,SUN Xiaolei,et al. Research on a combined method of subjective-objective weighting based on the ordered information and intensity information[J].Chinese Journal of Management Science,2017,25(12):179-187.
• [27]吴力波，周阳，徐呈隽.上海市居民绿色电力支付意愿研究[J].中国人口·资源与环境，2018,28(2):86-93.WU Libo,ZHOU Yang,XU Chengjun. Research on household’s willingness to pay for green power in Shanghai[J]. China Population Resources and Environment,2018,28(2):86-93.