基于CNN-AE-MAML的低压配电网自适应分类方法

doi:10.12204/j.issn.1000-7229.2024.05.006

电力建设 ›› 2024, Vol. 45 ›› Issue (5): 48-58.doi: 10.12204/j.issn.1000-7229.2024.05.006

• 大模型小样本条件下新能源规划设计与优化运行技术·栏目主持葛磊蛟副教授、孙铭阳教授、郑锋副教授、黄文焘副教授· • 上一篇下一篇

基于CNN-AE-MAML的低压配电网自适应分类方法

陈子靖¹, 蒋金琦¹, 赵健¹(), 杨德格², 胡陈晨², 张凯³

1.上海电力大学电气工程学院,上海市 200090
2.国网浙江省电力有限公司温州供电公司,浙江省温州市 325000
3.上海电力大学计算机科学与技术学院,上海市 200090

收稿日期:2023-11-24 出版日期:2024-05-01 发布日期:2024-04-29
通讯作者: 赵健(1990),男,博士,教授,主要研究方向为智能配用电、灵活性资源调度、分布式发电管控等,E-mail:zhaojianee@foxmail.com。
作者简介:陈子靖(1999),男,硕士研究生,主要研究方向为低压配电网负荷建模;
蒋金琦(2002),女,本科生,主要研究方向为微电网优化运行;
杨德格(1991),男,本科,主要研究方向为配电网规划、运维管理;
胡陈晨(1990),女,硕士,主要研究方向为配电网监测管理;
张凯(1990),男,博士,副教授,主要研究方向为应用密码学、网络安全、可信人工智能、区块链。
基金资助:
国家自然科学基金项目(51907114)

Adaptive Classification Method of Low Voltage Distribution Network Based on CNN-AE-MAML

CHEN Zijing¹, JIANG Jinqi¹, ZHAO Jian¹(), YANG Dege², HU Chenchen², ZHANG Kai³

1. School of Electric Power Engineering, Shanghai University of Electric Power, Shanghai 200090, China
2. Wenzhou Power Supply Company of State Grid Zhejiang Electric Power Co., Ltd., Wenzhou 325000, Zhejiang Province, China
3. School of Computer Science and Technology, Shanghai University of Electric Power, Shanghai 200090, China

Received:2023-11-24 Published:2024-05-01 Online:2024-04-29
Supported by:
National Natural Science Foundation of China(51907114)

摘要/Abstract

摘要：

低压配电网分类有利于提高低压配电网经济运行管理措施及新能源规划运行方案制定的效率。随着各类新能源、充电桩等新型源荷的不断接入,低压配电网原有负荷特征发生变化,一方面导致配电网负荷特征复杂,另一方面导致变化后可用的新负荷特征数据较少,给配电网分类带来挑战。针对以上挑战,提出一种基于卷积自编码器和模型不可知元学习(convolutional neural network-auto encoder-model agnostic meta learning, CNN-AE-MAML)的低压配电网自适应分类方法。首先,利用卷积自编码器(convolutional neural network auto encoder, CNN-AE)提取可表征低压配电网的配变负荷、光伏发电特征,采用谱聚类(spectral clustering, SC)对低压配电网进行分类;然后,构建基于softmax配电网类型识别方法,利用低压配电网实际数据的降维特征识别配电网类型;此外,利用模型不可知元学习(model-agnostic meta-learning, MAML)方法训练CNN-AE特征提取模型,使CNN-AE模型在少量数据下能自适应提取配电网新负荷特征,最终实现低压配电网准确、快速自适应分类。最后,利用低压配电网实际数据验证了所提方法的有效性。

关键词: 低压配电网, 自适应分类, 卷积自编码器, 谱聚类, 模型不可知元学习

Abstract:

The classification of low-voltage distribution network is conducive to improving the efficiency of formulating economic operation management measures and new energy planning operation schemes of low-voltage distribution network. With the continuous access of various new sources of energy, charging piles and other new sources, the original load characteristics of the low-voltage distribution network have changed, which on the one hand leads to complex load characteristics of the distribution network, and on the other hand leads to less available load characteristic data after the change, which brings challenges to the classification of the distribution network. Aiming at the above challenges, this paper proposes an adaptive classification method of low-voltage distribution network based on CNN-AE-MAML. Firstly, convolutional neural network auto encoder (CNN-AE) is used to extract the dimensionality reduction features of the distribution load of low-voltage distribution network and the photovoltaic power generation. Spectral clustering (SC) was used to classify low-voltage distribution networks. Then, the distribution network type identification method based on softmax is constructed to identify the distribution network type by using the dimensional-reduction features of the actual data of low-voltage distribution network. In addition, the model agnostic meta-learning (MAML) method is used to train the CNN-AE feature extraction model, so that the CNN-AE model can adaptively extract the new load features of the distribution network under a small amount of data, and finally achieve accurate and fast adaptive classification of the low-voltage distribution network.

Key words: low-voltage distribution network, adaptive classification, convolutional autoencoder, spectral clustering, model-agnostic meta-learning

中图分类号:

TM714

陈子靖, 蒋金琦, 赵健, 杨德格, 胡陈晨, 张凯. 基于CNN-AE-MAML的低压配电网自适应分类方法[J]. 电力建设, 2024, 45(5): 48-58.

CHEN Zijing, JIANG Jinqi, ZHAO Jian, YANG Dege, HU Chenchen, ZHANG Kai. Adaptive Classification Method of Low Voltage Distribution Network Based on CNN-AE-MAML[J]. ELECTRIC POWER CONSTRUCTION, 2024, 45(5): 48-58.

图/表 19

图1 自动编码器示意图

Fig.1 Diagram of auto encoder

图2 SC算法示意图

Fig.2 Diagram of spectral clustering

图3 CNN-AE模型结构示意图

Fig.3 Structure diagram of CNN-AE model

图4 基于MAML的模型参数更新

Fig.4 Model parameter updating based on MAML

图5 本文分类方法整体流程图

Fig.5 Overall flow chart of classification method in this paper

图6 不同聚类数k与参数σ的CH指数

Fig.6 The CH index of different k and σ

图7 基于t-SNE的聚类二维可视化

Fig.7 Two-dimensional visualization of clustering based on t-SNE

图8 8类典型低压配电网周负荷曲线图

Fig.8 Weekly load curves of 8 typical low-voltage distribution networks

图9 8类典型低压配电网周光伏曲线图

Fig.9 Weekly photovoltaic curves of 8 typical low-voltage distribution networks

表1 Softmax识别模型参数结构

Table 1 The parameter structure of the softmax identification model

图10 softmax识别模型训练损失下降曲线

Fig.10 Training loss descent curve of softmax identification model

图11 softmax识别模型测试集准确率曲线

Fig.11 Test set accuracy curves of softmax identification model

表2 不同降维-聚类方法效果对比

Table 2 Comparison of effects of different dimensionality reduction and clustering methods

表3 softmax输出结果的pi-pj差值分布

Table 3 Distribution of the pi-pj difference of the softmax output

表4 softmax识别模型输出结果

Table 4 The maximum and second values of softmax identification model

图12 新特征台区聚类在不同k与σ的CH指数

Fig.12 CH index of new feature distribution substation area clustering under different k and σ

图13 基于t-SNE的新特征台区聚类二维可视化

Fig.13 Two-dimensional visualization of new feature distribution substation clustering based on t-SNE

图14 新典型低压配电网周负荷曲线图

Fig.14 Weekly load curves of new typical low-voltage distribution networks

图15 三类模型的5个指标随数据量的变化

Fig.15 The five indexes of the three types of models change with the amount of data

参考文献 33

[1]	蔡黎, 葛棚丹, 代妮娜, 等. 电动汽车入网负荷预测及其与电网互动研究进展综述[J]. 智慧电力, 2022, 50(7): 96-103.
	CAI Li, GE Pengdan, DAI Nina, et al. Review of research progress on load prediction and grid interaction of electric vehicles[J]. Smart Power, 2022, 50(7): 96-103.
[2]	唐巍, 张起铭, 张璐, 等. 新型配电系统多层级交直流互联理念、关键技术与发展方向[J]. 电力系统自动化, 2023, 47(6): 2-17.
	TANG Wei, ZHANG Qiming, ZHANG Lu, et al. Concept, key technologies and development direction of multilevel AC/DC interconnection in new distribution system[J]. Automation of Electric Power Systems, 2023, 47(6): 2-17.
[3]	惠慧, 李蕊, 朱逸镝, 等. 含高比例分布式光伏的配电网多目标概率规划方法[J]. 电测与仪表, 2023, 60(11): 2-10.
	HUI Hui, LI Rui, ZHU Yidi, et al. Multi-objective probabilistic planning method for distribution network with high proportion of distributed photovoltaic[J]. Electrical Measurement & Instrumentation, 2023, 60(11): 2-10.
[4]	曲大鹏, 范晋衡, 刘琦颖, 等. 考虑配电网综合运行风险的充电桩接纳能力评估与优化[J]. 电力系统保护与控制, 2022, 50(3): 131-139.
	QU Dapeng, FAN Jinheng, LIU Qiying, et al. Assessment and optimization of charging pile acceptance capacity considering the comprehensive operational risk of a distribution network[J]. Power System Protection and Control, 2022, 50(3): 131-139.
[5]	韦善阳, 黎静华, 黄乾, 等. 考虑多重因素耦合的广义负荷特征曲线的模式分析[J]. 电力系统自动化, 2021, 45(1): 114-122.
	WEI Shanyang, LI Jinghua, HUANG Qian, et al. Pattern analysis of generalized load characteristic curve considering coupling of multiple factors[J]. Automation of Electric Power Systems, 2021, 45(1): 114-122.
[6]	谢敏, 张世平, 李弋升, 等. 基于多模式柔性互联的交直流低压配电网优化调度[J]. 电力系统自动化, 2023, 47(6): 79-89.
	XIE Min, ZHANG Shiping, LI Yisheng, et al. Optimal dispatch of AC/DC hybrid low-voltage distribution network based on multi-mode flexible interconnection[J]. Automation of Electric Power Systems, 2023, 47(6): 79-89.
[7]	胡维昊, 曹迪, 黄琦, 等. 深度强化学习在配电网优化运行中的应用[J]. 电力系统自动化, 2023, 47(14): 174-191.
	HU Weihao, CAO Di, HUANG Qi, et al. Application of deep reinforcement learning in optimal operation of distribution network[J]. Automation of Electric Power Systems, 2023, 47(14): 174-191.
[8]	张琳娟, 张平, 卢丹, 等. 基于“云-群-端”架构分布式光伏群调群控策略研究[J/OL]. 电测与仪表: 1-13(2023-02-07)[2023-06-19]. https://kns.cnki.net/kcms/detail//23.1202.TH.20230207.1129.002.html.
	ZHANG Linjuan, ZHANG Ping, LU Dan, et al. Research on group control strategy of distributed photovoltaic based on "cloud-group-end” architecture[J/OL]. Electrical Measurement & Instrumentation, 1-13(2023-02-07)[2023-06-19]. https://kns.cnki.net/kcms/detail//23.1202.TH.20230207.1129.002.html.
[9]	陈晓乾, 秦文萍, 杨乐, 等. 逆变式电源接入下的配电网自适应距离保护方案[J]. 电力系统保护与控制, 2023, 51(11): 27-38.
	CHEN Xiaoqian, QIN Wenping, YANG Le, et al. Adaptive distance protection scheme for a distribution network given access of inverter power supply[J]. Power System Protection and Control, 2023, 51(11): 27-38.
[10]	张潇龙, 齐林海. 融合稀疏降噪自编码与聚类算法的配电网台区分类研究[J]. 电力信息与通信技术, 2019, 17(12): 15-23.
	ZHANG Xiaolong, QI Linhai. Research on distribution transformer area classification based on sparse de-noising auto-encoder and clustering algorithm[J]. Electric Power Information and Communication Technology, 2019, 17(12): 15-23.
[11]	汤波, 杨鹏, 余光正, 等. 基于负荷峰谷耦合特性的中压配电网供区优化方法[J]. 中国电机工程学报, 2022, 42(19): 7051-7063.
	TANG Bo, YANG Peng, YU Guangzheng, et al. Power supply area optimization method of medium voltage distribution network based on load peak valley coupling characteristics[J]. Proceedings of the CSEE, 2022, 42(19): 7051-7063.
[12]	骆柏锋, 穆云飞, 贾宏杰, 等. 基于负荷特征库的大用户供电接入决策方法[J]. 电力系统自动化, 2018, 42(6): 66-72.
	LUO Bofeng, MU Yunfei, JIA Hongjie, et al. Decision method of power supply access for large consumers based on load feature library[J]. Automation of Electric Power Systems, 2018, 42(6): 66-72.
[13]	庞传军, 余建明, 冯长有, 等. 基于LSTM自动编码器的电力负荷聚类建模及特性分析[J]. 电力系统自动化, 2020, 44(23): 57-63.
	PANG Chuanjun, YU Jianming, FENG Changyou, et al. Clustering modeling and characteristic analysis of power load based on long-short-term memory auto-encoder[J]. Automation of Electric Power Systems, 2020, 44(23): 57-63.
[14]	RYU S, CHOI H, LEE H, et al. Convolutional autoencoder based feature extraction and clustering for customer load analysis[J]. IEEE Transactions on Power Systems, 2020, 35(2): 1048-1060. doi: 10.1109/TPWRS.59 URL
[15]	何哲楠, 吴浩, 程祥, 等. 基于变电站-用户双层结构的变电站负荷聚类研究[J]. 电网技术, 2019, 43(8): 2983-2991.
	HE Zhenan, WU Hao, CHENG Xiang, et al. Substation load clustering based on substation-consumer two-level structure[J]. Power System Technology, 2019, 43(8): 2983-2991.
[16]	姚黄金, 雷霞, 付鑫权, 等. 基于改进自适应密度峰值算法的日负荷曲线聚类分析[J]. 电力系统保护与控制, 2022, 50(3): 121-130.
	YAO Huangjin, LEI Xia, FU Xinquan, et al. Cluster analysis of daily load curves based on an improved self-adaptive density peak clustering algorithm[J]. Power System Protection and Control, 2022, 50(3): 121-130.
[17]	祁梦雪, 梅林珏昊, 李知艺, 等. 基于网络结构约简的配电网多日负荷曲线聚类方法[J]. 电力系统自动化, 2023, 47(5): 35-43.
	QI Mengxue, MEI Linjuehao, LI Zhiyi, et al. Multi-day load curve clustering method for distribution network based on network structure reduction[J]. Automation of Electric Power Systems, 2023, 47(5): 35-43.
[18]	徐胜蓝, 司曹明哲, 万灿, 等. 考虑双尺度相似性的负荷曲线集成谱聚类算法[J]. 电力系统自动化, 2020, 44(22): 152-160.
	XU Shenglan, SI Caomingzhe, WAN Can, et al. Ensemble spectral clustering algorithm for load profiles considering dual-scale similarities[J]. Automation of Electric Power Systems, 2020, 44(22): 152-160.
[19]	仲赞, 邢翼, 俞伟, 等. 基于WRSR和BOA-Catboost的电力用户分类模型研究[J]. 浙江电力, 2023, 42(5): 76-84.
	ZHONG Zan, XING Yi, YU Wei, et al. Research on a power user classification model based on WRSR and BOA-Catboost[J]. Zhejiang Electric Power, 2023, 42(5): 76-84.
[20]	徐嘉杰, 陈光宇, 袁飞, 等. 基于IBBO和K-means++融合的多属性台区聚类研究[J]. 电气自动化, 2022, 44(1): 44-46.
	XU Jiajie, CHEN Guangyu, YUAN Fei, et al. Research on multi-attribute station clustering based on fusion of IBBO and K-means++[J]. Electrical Automation, 2022, 44(1): 44-46.
[21]	时珊珊, 田英杰, 金瑞杨, 等. 基于聚类的电网低压台区用电画像方法研究[J]. 电工电能新技术, 2020, 39(8): 66-74. doi: 10.12067/ATEEE2006020
	SHI Shanshan, TIAN Yingjie, JIN Ruiyang, et al. Research on power consumption portrait method of low-voltage substation areas based on clustering[J]. Advanced Technology of Electrical Engineering and Energy, 2020, 39(8): 66-74. doi: 10.12067/ATEEE2006020
[22]	徐嘉杰, 卢兆军, 袁飞, 等. 基于深度神经网络和SoftMax分类器的台区负荷分类识别方法[J]. 电气自动化, 2021, 43(6): 102-104, 114.
	XU Jiajie, LU Zhaojun, YUAN Fei, et al. Classification and identification method of station load based on deep neural network and softmax classifier[J]. Electrical Automation, 2021, 43(6): 102-104, 114.
[23]	张静, 尚光伟, 龚正国, 等. 计及负荷分类及其需求响应的配电网设备利用率提高方法[J]. 电网与清洁能源, 2021, 37(2): 50-56, 63.
	ZHANG Jing, SHANG Guangwei, GONG Zhengguo, et al. An optimization method of distribution network equipment utilization considering load classification and demand response[J]. Power System and Clean Energy, 2021, 37(2): 50-56, 63.
[24]	CHICCO G, NAPOLI R, PIGLIONE F. Comparisons among clustering techniques for electricity customer classification[J]. IEEE Transactions on Power Systems, 2006, 21(2): 933-940. doi: 10.1109/TPWRS.2006.873122 URL
[25]	LI R, GU C H, LI F R, et al. Development of low voltage network templates: part I: substation clustering and classification[J]. IEEE Transactions on Power Systems, 2015, 30(6): 3036-3044. doi: 10.1109/TPWRS.2014.2371474 URL
[26]	XU T S, CHIANG H D, LIU G Y, et al. Hierarchical K-means method for clustering large-scale advanced metering infrastructure data[J]. IEEE Transactions on Power Delivery, 2017, 32(2): 609-616. doi: 10.1109/TPWRD.2015.2479941 URL
[27]	邬浩泽, 朱晨烜, 张贻山, 等. 一种基于改进模糊聚类算法的自适应典型日选取方法[J]. 智慧电力, 2022, 50(1): 60-67.
	WU Haoze, ZHU Chenxuan, ZHANG Yishan, et al. Adaptive method for selecting typical days based on improved fuzzy clustering algorithm[J]. Smart Power, 2022, 50(1): 60-67.
[28]	卢锦玲, 郭鲁豫. 基于改进深度残差收缩网络的电力系统暂态稳定评估[J]. 电工技术学报, 2021, 36(11): 2233-2244.
	LU Jinling, GUO Luyu. Power system transient stability assessment based on improved deep residual shrinkage network[J]. Transactions of China Electrotechnical Society, 2021, 36(11): 2233-2244.
[29]	王哲, 万宝, 凌天晗, 等. 基于谱聚类和LSTM神经网络的电动公交车充电负荷预测方法[J]. 电力建设, 2021, 42(6): 58-66. doi: 10.12204/j.issn.1000-7229.2021.06.006
	WANG Zhe, WAN Bao, LING Tianhan, et al. Electric bus charging load forecasting method based on spectral clustering and LSTM neural network[J]. Electric Power Construction, 2021, 42(6): 58-66. doi: 10.12204/j.issn.1000-7229.2021.06.006
[30]	VON LUXBURG U. A tutorial on spectral clustering[J]. Statistics and Computing, 2007, 17(4): 395-416. doi: 10.1007/s11222-007-9033-z URL
[31]	NG A Y, JORD M I, WEISS Y. On spectral clustering: analysis and an algorithm[J]. Advances in Neural Information Processing Systems, 2002(11): 849-856.
[32]	FINN C, ABBEEL P, LEVINE S. Model-agnostic meta-learning for fast adaptation of deep networks[C]// Proceedings of the 34th International Conference on Machine Learning - Volume 70. ACM, 2017: 1126-1135.
[33]	LAURENS V D M, HINTON G. Visualizing data using t-SNE[J]. Journal of Machine Learning Research, 2008, 9(2605): 2579-2605.

基于CNN-AE-MAML的低压配电网自适应分类方法

Adaptive Classification Method of Low Voltage Distribution Network Based on CNN-AE-MAML

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 19

参考文献 33

相关文章 8

编辑推荐

Metrics

本文评价

[1]	刘艳茹, 刘洪, 谷毅, 韩柳, 滑雪娇. 考虑多种终端配置的中低压配电网供电可靠性协同评估[J]. 电力建设, 2022, 43(2): 54-62.
[2]	王哲, 万宝, 凌天晗, 董晓红, 穆云飞, 邓友均, 唐舒懿. 基于谱聚类和LSTM神经网络的电动公交车充电负荷预测方法[J]. 电力建设, 2021, 42(6): 58-66.
[3]	刘科研, 贾东梨, 王薇嘉, 耿光飞, 魏琛豪. 考虑分布式光伏电源接入模式的低压配电网不平衡线损计算方法[J]. 电力建设, 2021, 42(10): 129-138.
[4]	黄冬梅, 林孝镶, 胡安铎, 孙锦中. 基于卷积自编码器的日负荷深度嵌入聚类方法[J]. 电力建设, 2021, 42(1): 132-138.
[5]	徐杉杉，朱俊澎，袁越，吴涵. 基于马尔可夫模型的光伏出力聚类与模拟[J]. 电力建设, 2019, 40(2): 54-62.
[6]	董雷，张涛，陈乃仕，卫泽晨. 降低电压不平衡度的三相并网逆变器阻尼控制策略[J]. 电力建设, 2018, 39(10): 113-119.
[7]	刘扬，严正，马洪艳，徐潇源. 风电接入下基于半监督谱聚类的备用动态分区方法[J]. 电力建设, 2018, 39(1): 90-.
[8]	李力，傅子昊，孙磊，杨银国，谭嫣，林振智，文福拴. 基于半监督谱聚类的黑启动分区策略[J]. 电力建设, 2017, 38(4): 9-.