胡苏筠,曹瑛,张霞,吴震旦,胡军
HU Suyun,CAO Ying,ZHANG Xia,WU Zhendan,HU Jun

• 1:国网江苏省电力有限公司常州供电公司

摘要(Abstract)：

新型电力系统下配电网运行方式调整愈来愈频繁，配电网负荷-馈线匹配面临采样数据高维异构且价值密度低、现有匹配算法对负荷物理特征依赖度高、参数设置灵活性弱等难点，为此提出一种基于改进SAX（符号聚合近似）算法与贝叶斯超参数优化的配电网负荷-馈线智能匹配方法。首先，建立面向离散符号化时间数据序列的数据价值提升模型，将高维异构的数据近似表示为低维统一的符号，修正和填充异常数据、空白数据。其次，构建改进CNN-LSTM（卷积神经网络-长短期记忆）混合神经网络，对负荷数据进行所属馈线匹配分类训练，利用多头注意力机制深入挖掘负荷数据的潜在数学关系，降低对负荷物理特征的依赖度。然后，引入贝叶斯超参数优化算法对神经网络训练参数进行逐次更新，提高馈线拓扑变化时神经网络模型的灵活性与适应性。最后，对某地区100条馈线进行负荷匹配实验验证，结果证明所提方法较传统方法具有更高的匹配精度。
As the operation mode of distribution networks is adjusted more and more frequently in the new-type power system, the load-feeder matching of distribution networks meets with difficulties such as high-dimensional heterogeneity and low value density of sampled data, high dependence of existing matching algorithms on physical characteristics of load, and poor flexibility of parameter settings, etc. To this end, an intelligent load-feeder matching method of distribution networks based on an improved SAX(symbolic aggregation approximation) algorithm and Bayesian hyperparameter optimization is proposed. Firstly, a data value enhancement model for discrete symbolic time data series is established to approximate the high-dimensional heterogeneous data into low-dimensional uniform symbols to correct and fill the abnormal data and blank data. Secondly, an improved CNN-LSTM(convolutional neural network-long short-term memory) hybrid neural network is constructed to train the load data as per the matching, and the multi-headed attention is employed to explore the potential mathematical relationships between load and data to curtail dependency on physical features of load. Thirdly, a Bayesian hyperparameter optimization algorithm is introduced to update the neural network training parameters one by one to improve the flexibility and adaptability of the neural network model in the case of feeder topology changes. Finally, the proposed method is experimentally validated by matching between load and 100 feeders in a region. The results prove that the proposed method is superior to the traditional method in matching accuracy.

关键词(KeyWords)： 改进符号聚合近似算法;贝叶斯超参数优化;多头注意力机制;改进CNN-LSTM;负荷-馈线匹配
improved SAX;Bayesian hyperparameter optimization;multi-head attention;improved CNN-LSTM;load-feeder matching

Abstract:

Keywords:

基金项目(Foundation):

作者(Author): 胡苏筠,曹瑛,张霞,吴震旦,胡军
HU Suyun,CAO Ying,ZHANG Xia,WU Zhendan,HU Jun

DOI: 10.19585/j.zjdl.202307009

参考文献(References)：

• [1]牛耕，孔力，周龙，等.含分布式电源的配电网的供电恢复技术研究综述[J].电工电能新技术，2017,36(9):51-62.NIU Geng,KONG Li,ZHOU Long,et al.Review of power service restoration of distribution network with distributed generation[J].Advanced Technology of Electrical Engineering and Energy,2017,36(9):51-62.
• [2]刘佳男，李业行，张董，等.基于风光储一体化的新型农村配电网研究[J].山东电力技术，2021,48(11):42-49.LIU Jianan,LI Yehang,ZHANG Dong,et al.Study on new rural distribution network based on the integration of wind solar energy and storage[J].Shandong Electric Power,2021,48(11):42-49.
• [3]邓燕国，王冰，曹智杰，等.基于熵权法与GRA-ELM的配电网空间负荷预测[J].电力工程技术，2021,40(4):136-141.DENG Yanguo,WANG Bing,CAO Zhijie,et al.Spatial load forecasting of distribution network based on entropy weight method and GRA-ELM[J].Electric Power Engineering Technology,2021,40(4):136-141.
• [4]刁赢龙，盛万兴，刘科研，等.大规模配电网负荷数据在线清洗与修复方法研究[J].电网技术，2015,39(11):3134-3140.DIAO Yinglong,SHENG Wanxing,LIU Keyan,et al.Research on online cleaning and repair methods of large-scale distribution network load data[J].Power System Technology,2015,39(11):3134-3140.
• [5]罗韬，王志伟，赵海波.分布式光伏并网点电压控制调整方法研究[J].山西电力，2021(3):30-33.LUO Tao,WANG Zhiwei,ZHAO Haibo.Research on voltage control and adjustment method of distributed photovoltaic grid connection point[J].Shanxi Electric Power,2021(3):30-33.
• [6]李方舟，霍健，周博曦.基于有序树拓扑的配电网转供能力分析及网络重构方法[J].山东电力技术，2021,48(6):22-27.LI Fangzhou,HUO Jian,ZHOU Boxi.Analysis of switching capacity of distribution network and network reconfiguration method based on ordered tree topology[J].Shandong Electric Power,2021,48(6):22-27.
• [7]许鸣吉，沈磊，李胜，等.计及综合因素的光伏接入配电网优选研究[J].电力工程技术，2021,40(2):46-52.XU Mingji,SHEN Lei,LI Sheng,et al.Optimization of photovoltaic access distribution network considering comperhensive factors[J].Electric Power Engineering Technology,2021,40(2):46-52.
• [8]邵宇鹰，彭鹏，张秋桥，等.基于极限学习机与负荷密度指标法的空间负荷预测[J].电力工程技术，2021,40(1):86-91.SHAO Yuying,PENG Peng,ZHANG Qiuqiao,et al.Spatial load forecasting based on ELM and clustering algorithm[J].Electric Power Engineering Technology,2021,40(1):86-91.
• [9]乔博鑫，章坚民.基于历史负荷的配电网量测数据关联清洗方法研究[J].信息与电脑，2018(8):70-71.QIAO Boxin,ZHANG Jianmin.Research on correlative cleaning method of distribution network measurement data based on historical load[J].China Computer&Communication,2018(8):70-71.
• [10]顾民，葛良全，熊文贤.基于神经网络的电力负荷不良数据的清洗[J].微计算机信息，2007,23(21):265-267.GU Min,GE Liangquan,XIONG Wenxian.Cleaning for bad electric load data based on neural network[J].Microcomputer Information,2007,23(21):265-267.
• [11]张晓星，程其云，周湶，等.基于数据挖掘的电力负荷脏数据动态智能清洗[J].电力系统自动化，2005,29(8):60-64.ZHANG Xiaoxing,CHENG Qiyun,ZHOU Quan,et al.Dynamic intelligent cleaning for dirty electric load data based on data mining[J].Automation of Electric Power Systems,2005,29(8):60-64.
• [12]卜凡鹏，陈俊艺，张琪祁，等.一种基于双层迭代聚类分析的负荷模式可控精细化识别方法[J].电网技术，2018,42(3):903-913.BU Fanpeng,CHEN Junyi,ZHANG Qiqi,et al.A controllable refined recognition method of electrical load pattern based on bilayer iterative clustering analysis[J].Power System Technology,2018,42(3):903-913.
• [13]李培强，李欣然，陈辉华，等.基于模糊聚类的电力负荷特性的分类与综合[J].中国电机工程学报，2005,25(24):73-78.LI Peiqiang,LI Xinran,CHEN Huihua,et al.The characteristics classification and synthesis of power load based on fuzzy clustering[J].Proceedings of the CSEE,2005,25(24):73-78.
• [14]安琪，梁宇飞，王耀强，等.基于K-means聚类与PSO特征优选KNN的分级负荷识别方法[J].河北科技大学学报，2022,43(3):249-258.AN Qi,LIANG Yufei,WANG Yaoqiang,et al.Hierarchical load identification method based on K-means clustering and PSO feature optimization KNN[J].Journal of Hebei University of Science and Technology,2022,43(3):249-258.
• [15]唐毅，祁甘露，池坤鹏，等.基于三维时空特性数据融合的电力物联网用户侧负荷识别方法研究[J].山东师范大学学报（自然科学版），2022,37(2):120-127.TANG Yi,QI Ganlu,CHI Kunpeng,et al.User-side load identification method of power Internet of Things based on three-dimensional spatiotemporal feature data fusion[J].Journal of Shandong Normal University (Natural Science),2022,37(2):120-127.
• [16]李想，王鹏，刘洋，等.考虑类别不平衡的海量负荷用电模式辨识方法[J].中国电机工程学报，2020,40(1):128-137.LI Xiang,WANG Peng,LIU Yang,et al.Massive load pattern identification method considering class imbalance[J].Proceedings of the CSEE,2020,40(1):128-137.
• [17]林顺富，顾乡，汤继开，等.基于稀疏自动编码器神经网络的负荷曲线分类方法[J].电网技术，2020,44(9):3508-3515.LIN Shunfu,GU Xiang,TANG Jikai,et al.Power load profile classification method based on neural network of sparse automatic encoder[J].Power System Technology,2020,44(9):3508-3515.
• [18]马恒，任美学，高科.基于随机搜索优化XGBoost的瓦斯涌出量预测模型[J].中国安全生产科学技术，2022,18(5):129-134.MA Heng,REN Meixue,GAO Ke.Prediction model of gas emission amount based on XGBoost optimized with random search algorithm[J].Journal of Safety Science and Technology,2022,18(5):129-134.
• [19]季海娟，周从华，刘志锋.一种基于始末距离的时间序列符号聚合近似表示方法[J].计算机科学，2018,45(6):216-221.JI Haijuan,ZHOU Conghua,LIU Zhifeng.Symbolic aggregate approximation method of time series based on beginning and end distance[J].Computer Science,2018,45(6):216-221.
• [20]BERGSTRA J,BENGIO Y.Random search for hyperparameter optimization[J].Journal of Machine Learning Research,2012,13(1):281-305.
• [21]黄俊杰，徐兴华，崔小鹏，等.融合趋势信息的时间序列符号聚合近似方法[J].计算机应用研究，2023,40(1):86-90.HUANG Junjie,XU Xinghua,CUI Xiaopeng,et al.Time series symbolic aggregate approximation method for fusion trend information[J].Application Research of Computers,2023,40(1):86-90.
• [22]朱有产，梁玮轩，王英姿.基于时间序列的配电网数据清理和融合方法研究[J].电网技术，2021,45(7):2839-2846.ZHU Youchan,LIANG Weixuan,WANG Yingzi.Research on data cleaning and fusion in distribution power grid based on time series technology[J].Power System Technology,2021,45(7):2839-2846.
• [23]杨祎，崔其会，秦佳峰，等.改进BERT的故障案例智能匹配方法[J].山东电力技术，2022,49(2):47-53.YANG Yi,CUI Qihui,QIN Jiafeng,et al.Intelligent matching method for fault cases based on improve BERT[J].Shandong Electric Power,2022,49(2):47-53.
• [24]LECUN Y,BOTTOU L,BENGIO Y,et al.Gradientbased learning applied to document recognition[J].Proceedings of the IEEE,1998,86(11):2278-2324.
• [25]GIULIARI F,HASAN I,CRISTANI M,et al.Transformer networks for trajectory forecasting[C]//2020 25th International Conference on Pattern Recognition (ICPR).January 10-15,2021,Milan,Italy:IEEE,2021:10335-10342.