数据稀缺场景下的配电网异常数据检测方法

doi:10.11930/j.issn.1004-9649.202510085

摘要/Abstract

摘要：

为了精准检测配电网电压、电流数据异常，并解决配电网正常运行状态下异常数据稀缺导致检测模型准确率较低等问题，提出一种基于改进的混沌优化算法（improved chaos optimization algorithm，ICEO）-双重注意力机制Transformer（dual attention mechanism-Transformer，DAM- Transformer）的异常数据检测方法。该方法首先利用强度可控的扩散异常合成方法（strength-controlled diffusion anomaly synthesis，SDAS）生成部分异常数据，以缓解真实异常样本稀缺导致模型识别准确率不足的问题；其次创新地提出了DAM-Transformer模型，通过融入双重注意力机制实现对不同时间尺度和特征空间中复杂模式的协同建模，有效提升配电网数据异常背景下多尺度特征耦合关系的辨识效果；最后采用ICEO对 DAM-Transformer 的超参数进行迭代优化，进一步改善模型的优化效率与复杂场景下的泛化性能。结果表明：该方法与传统模型对比，配电网异常电压识别准确率提升 12.81%、异常电流识别准确率提升 12.22%，在数据稀缺场景下的识别准确率显著优于传统模型。该方法有效解决了配电网异常数据识别中样本稀缺与多尺度特征建模难的核心瓶颈，提升了异常识别的精准性与模型运行稳定性，为智能配电网的数字化巡检、实时故障预警及运维决策优化提供了关键技术支撑，具有工程应用前景。

关键词: 双重注意力机制, 改进混沌优化算法, 异常数据检测

Abstract:

In order to accurately detect abnormal voltage and current data in the distribution network and solve the problem of low accuracy of the detection model caused by the scarcity of abnormal data under normal operation of the distribution network, a method for detecting abnormal data based on an improved chaos optimization algorithm (ICEO) - dual attention mechanism Transformer (DAM Transformer) is proposed. This method first utilizes the strength controlled diffusion anomaly synthesis (SDAS) method to generate partial anomaly data, in order to alleviate the problem of insufficient model recognition accuracy caused by the scarcity of real anomaly samples; Secondly, an innovative DAM Transformer model was proposed, which integrates a dual attention mechanism to achieve collaborative modeling of complex patterns in different time scales and feature spaces, effectively improving the identification of multi-scale feature coupling relationships in the context of abnormal distribution network data; Finally, ICEO was used to iteratively optimize the hyperparameters of DAM Transformer, further improving the optimization efficiency and generalization performance of the model in complex scenarios. The results show that compared with traditional models, this method improves the accuracy of identifying abnormal voltage in distribution networks by 12.81% and the accuracy of identifying abnormal current by 12.22%. In data scarcity scenarios, the recognition accuracy is significantly better than traditional models. This method effectively solves the core bottleneck of sample scarcity and difficulty in modeling multi-scale features in abnormal data recognition of distribution networks, improves the accuracy of abnormal recognition and the stability of model operation, and provides key technical support for digital inspection, real-time fault warning, and operation and maintenance decision optimization of intelligent distribution networks. It has engineering application prospects.

Key words: dual attention mechanism, improved chaos optimization algorithm, abnormal data detection

曾瑞江, 李志勇, 黄曙, 王伟光. 数据稀缺场景下的配电网异常数据检测方法[J]. 中国电力, 2026, 59(5): 67-75.

ZENG Ruijiang, LI Zhiyong, HUANG Shu, WANG Weiguang. Abnormal data detection method for distribution networks in data scarcity scenarios[J]. Electric Power, 2026, 59(5): 67-75.

导出引用管理器 EndNote|Ris|BibTeX

链接本文: https://www.electricpower.com.cn/CN/10.11930/j.issn.1004-9649.202510085

https://www.electricpower.com.cn/CN/Y2026/V59/I5/67

图/表 8

图 1 电压越限

Fig.1 Voltage out of limit

图 2 电流畸变

Fig.2 current distortion

图 3 SDAS异常数据生成示意

Fig.3 SDAS abnormal data generation

表 1 对比模型异常信号识别准确率

Table 1 Comparison of model abnormal signal recognition accuracy 单位：%

模型	电压异常识别率	电流异常识别率
DAM-SVM	48.72	56.41
DAM-LSTM	56.41	54.90
DAM-GRU	48.72	56.41
DAM-Transformer	51354.00	62.75

图 4 C相电压和电流检测区域识别情况

Fig.4 Identification of C-phase voltage and current detection area

图 5 不同模型三相电压异常信号识别个数

Fig.5 Number of abnormal three-phase voltage signals identified by different models

图 6 三相电流异常信号识别个数

Fig.6 Number of abnormal signals identified for three-phase current

表 2 对比模型异常信号识别准确率

Table 2 Comparison of model abnormal signal recognition accuracy 单位：%

模型	电压异常识别率	电流异常识别率
GWO-DAM-Transformer	66.67	56.86
WOA-DAM-Transformer	71.79	62.75
SSA-DAM-Transformer	64.10	58.82
CEO-DAM-Transformer	84.62	80.39
ICEO-DAM-Transformer	97.43	92.16

参考文献 37

1	于宗超, 文明, 李湘华, 等. 融合群体智慧的分布式智能电网高效发展管理策略[J]. 中国电力, 2024, 57 (10): 57- 68.
	YU Zongchao, WEN Ming, LI Xianghua, et al. Effective development and management strategy for distributed smart grids based on collective intelligence[J]. Electric Power, 2024, 57 (10): 57- 68.
2	程乐峰, 刘奕杭, 邹涛. 博弈论视角下智能电网需求响应研究综述[J]. 综合智慧能源, 2025, 47 (4): 1- 22.
	CHENG Lefeng, LIU Yihang, ZOU Tao. Review of demand response in smart grids from the perspective of game theory[J]. Integrated Intelligent Energy, 2025, 47 (4): 1- 22.
3	解国庆, 梁添, 周兵兵, 等. 基于差分隐私保护的智能电网分布式经济调度方案[J]. 山东电力技术, 2025, 52 (4): 20- 28.
	XIE Guoqing, LIANG Tian, ZHOU Bingbing, et al. Distributed economic dispatch scheme for smart grid based on differential privacy protection[J]. Shandong Electric Power, 2025, 52 (4): 20- 28.
4	胡嘉诚, 张宁, 曹雨桐, 等. 分布式能源接入下虚拟电厂负荷优化调度决策[J]. 综合智慧能源, 2025, 47 (1): 62- 69.
	HU Jiacheng, ZHANG Ning, CAO Yutong, et al. Load optimization scheduling decision for virtual power plants with distributed energy accessed[J]. Integrated Intelligent Energy, 2025, 47 (1): 62- 69.
5	王新宇, 王相杰, 张明月, 等. 基于改进自适应免疫遗传算法的智能电网虚假数据攻击检测方法[J]. 浙江电力, 2023, 42 (10): 84- 89.
	WANG Xinyu, WANG Xiangjie, ZHANG Mingyue, et al. A false data attack detection method for power grid based on an improved AIGA[J]. Zhejiang Electric Power, 2023, 42 (10): 84- 89.
6	王安宁, 范荣奇, 张旸, 等. 基于多特征量判据的新型配电系统早期故障检测[J]. 中国电力, 2024, 57 (9): 181- 193.
	WANG Anning, FAN Rongqi, ZHANG Yang, et al. Multiple characteristics criterion based incipient fault detection of distribution systems[J]. Electric Power, 2024, 57 (9): 181- 193.
7	杨晓华, 费正源, 代盛国, 等. 基于FP-Growth数据挖掘的直流串联故障电弧特征提取及检测方法[J]. 广东电力, 2025, 38 (3): 104- 112.
	YANG Xiaohua, FEI Zhengyuan, DAI Shengguo, et al. Feature extraction and detection method of DC series fault arc based on FP-growth data mining[J]. Guangdong Electric Power, 2025, 38 (3): 104- 112.
8	李临风, 饶丹, 樊瑞, 等. 基于双向LSTM和注意力机制的输电线路故障判别方法[J]. 广东电力, 2022, 35 (11): 91- 98.
	LI Linfeng, RAO Dan, FAN Rui, et al. Fault discrimination method based on Bi-LSTM-attention for transmission lines[J]. Guangdong Electric Power, 2022, 35 (11): 91- 98.
9	黄翔. 配变负载三相不平衡判断方法的设计[J]. 自动化仪表, 2023, 44 (5): 78- 82.
	HUANG Xiang. Design of three-phase imbalance judgment method for distribution transformer load[J]. Process Automation Instrumentation, 2023, 44 (5): 78- 82.
10	蔡木良, 范瑞祥, 王皓天, 等. 一种基于数据驱动递归AdaBoost的三相不平衡配电系统拓扑重构方法研究[J]. 智慧电力, 2025, 53 (4): 38- 44.
	CAI Muliang, FAN Ruixiang, WANG Haotian, et al. A data-driven recursive AdaBoost-based topology reconfiguration method forThree-phase unbalanced distribution systems[J]. Smart Power, 2025, 53 (4): 38- 44.
11	苏斌, 张子夏, 李程, 等. 基于多种群遗传算法的三相不平衡判据及调节策略研究[J]. 上海节能, 2023 (4): 481- 487.
	SU Bin, ZHANG Zixia, LI Cheng, et al. Research on three phase imbalance criterion and regulation strategy based on IMPGA[J]. Shanghai Energy Saving, 2023 (4): 481- 487.
12	罗程, 鲁玕, 钟德龙, 等. 低碳背景下数字电网大数据安全匿名化处理技术研究[J]. 电测与仪表, 2026, 63 (1): 34- 44.
	LUO Cheng, LU Gan, ZHONG Delong, et al. Research on secure anonymization processing techniques for digital power grid big data under low carbon background[J]. Electrical Measurement & Instrumentation, 2026, 63 (1): 34- 44.
13	刘金波, 郭庆来, 张靖, 等. 机理数据融合驱动的图神经网络状态估计方法[J]. 电力系统自动化, 2026, 50 (5): 173- 183.
	LIU Jinbo, GUO Qinglai, ZHANG Jing, et al. Hybrid mechanism-data-driven state estimation method based on graph neural network[J]. Automation of Electric Power Systems, 2026, 50 (5): 173- 183.
14	YANG B, LIU S W, GAO H D. Energy e-commerce user portrait and multi-agent cooperative game price mechanism design[J]. Frontiers in Energy Research, 2024, 12, 1379612.
15	ZHAO Y, ZHENG L H, HAN S Y, et al. Abnormal intensity and polarization of Raman scattered light at edges of layered MoS₂[J]. Nano Research, 2022, 15 (7): 6416- 6421.
16	LI D L, ZHAO X X, XU W M, et al. A novel renewable energy scenario generation method based on multi-resolution denoising diffusion probabilistic models[J]. Energies, 2025, 18 (14): 3781.
17	JARRAYA A, BOUJELBÈNE M. Corporate social responsibility, CEO overconfidence and bank performance[J]. Asia-Pacific Journal of Business Administration, 2026, 18 (1): 190- 212.
18	朱海, 许道强, 邓君华, 等. 基于变异粒子群算法的智慧指挥系统故障抢修策略优化研究[J]. 自动化与仪器仪表, 2025 (4): 278- 282.
	ZHU Hai, XU Daoqiang, DENG Junhua, et al. Study on the optimization of fault repair strategy of intelligent command system based on variant particle swarm algorithm[J]. Automation & Instrumentation, 2025 (4): 278- 282.
19	Bala I, Karunarathne W, Mitchell L. Optimizing feature selection by enhancing particle swarm optimization with orthogonal initialization and crossover operator[J]. Computers, Materials & Continua, 2025, 84 (1): 727- 744.
20	Zhang C, Fu H. LogDA: dual attention-based log anomaly detection addressing data imbalance[J]. Computers, Materials & Continua, 2025, 83 (1): 1291- 1306.
21	庄伟卿, 余晗彧. 考虑多头注意力机制的交通事故等级分类预测模型[J]. 福建理工大学学报, 2025, 23 (6): 599- 606.
	ZHUANG Weiqing, YU Hanyu. Traffic accident classification prediction model considering multi-head attention mechanism[J]. Journal of Fujian University of Technology, 2025, 23 (6): 599- 606.
22	黄晋, 赵隆懿, 高震, 等. 基于混合残差Transformer-GRU的飞行航迹预测[J]. 航空计算技术, 2025, 55 (6): 25- 30, 36.
	HUANG Jin, ZHAO Longyi, GAO Zhen, et al. Trajectory prediction of aircraft based on hybrid residual transformer-GRU[J]. Aeronautical Computing Technique, 2025, 55 (6): 25- 30, 36.
23	ZHANG E H, CAO H Y, WANG P, et al. Investigation on the mechanical response of a prefabricated underground pipe gallery with a flexible energy dissipation node: an experimental study[J]. Buildings, 2025, 15 (19): 3521.
24	AHMED A, SUN G M, SAADELDIN M, et al. Efficient melanoma detection using pixel intensity-based masking and intensity-weighted binary cross-entropy[J]. International Journal of Imaging Systems and Technology, 2025, 35 (5): e70179.
25	齐桓若, 陈晨, 郭放, 等. 考虑精细化充放电与碳效益的配电网储能多目标双层规划模型[J]. 中国电力, 2025, 58 (10): 121- 135.
	QI Huanruo, CHEN Chen, GUO Fang, et al. Multi-objective bi-level planning model for distribution network energy storage considering refined charging/discharging and carbon benefits[J]. Electric Power, 2025, 58 (10): 121- 135.
26	吕帅. 某电站厂用电400 V系统电压频繁越限分析及解决[J]. 电工材料, 2025 (6): 103- 106.
	LÜ Shuai. Analysis and resolution of frequent voltage limit exceedance in the 400 V auxiliary power system of a power plant[J]. Electrical Engineering Materials, 2025 (6): 103- 106.
27	胡胜龙, 刘星, 朱莲莉. 电流畸变对电机切向电磁力与气隙磁密的影响研究[J]. 微特电机, 2025, 53 (6): 42- 46.
	HU Shenglong, LIU Xing, ZHU Lianli. Study on the influence of abnormal currents on both tangential electromagnetic force and air-gap flux density of motor[J]. Small & Special Electrical Machines, 2025, 53 (6): 42- 46.
28	王德顺, 雷洁, 胡安平, 等. 基于多目标双层优化的台区分布式电源选址定容方法[J]. 中国电力, 2025, 58 (10): 136- 146.
	WANG Deshun, LEI Jie, HU Anping, et al. Siting and sizing of distributed generation in distribution transformer areas using multi-objective bi-level optimization[J]. Electric Power, 2025, 58 (10): 136- 146.
29	刘军会, 龚健, 佟炳绅, 等. 基于分布式储能与光伏的虚拟电厂与配电网协同优化方法[J]. 中国电力, 2025, 58 (6): 1- 9.
	LIU Junhui, GONG Jian, TONG Bingshen, et al. Coordinated optimization method for virtual power plants and distribution networks considering distributed energy storage and photovoltaics[J]. Electric Power, 2025, 58 (6): 1- 9.
30	SHEN L J, GE S B, ZHANG J, et al. Analyzing performance fluctuations in flat lenses due to feature structure distortion using a differential profile parsing method[J]. Optics Express, 2025, 33 (4): 8862.
31	程从金, 严浩, 谢明杨, 等. 600 MW汽轮发电机转子绕组故障机理诊断[J]. 机电信息, 2025 (24): 72- 75, 82.
32	吕鹏飞, 裘愉涛, 金盛, 等. 面向智能变电站保护的FMM测试及改进SVM校验诊断技术[J]. 中国电力, 2025, 58 (6): 76- 82.
	LV Pengfei, QIU Yutao, JIN Sheng, et al. FMM testing and improved SVM diagnostic verification technology for intelligent substation protection[J]. Electric Power, 2025, 58 (6): 76- 82.
33	杨国华, 祁鑫, 贾睿, 等. 基于CEEMD-SE的CNN&LSTM-GRU短期风电功率预测[J]. 中国电力, 2024, 57 (2): 55- 61.
	YANG Guohua, QI Xin, JIA Rui, et al. Short-term wind power forecast based on CNN & LSTM-GRU model integrated with CEEMD-SE algorithm[J]. Electric Power, 2024, 57 (2): 55- 61.
34	袁志涛, 李泽伟, 刘克中, 等. 基于GRU-Attention-BiLSTM的船舶轨迹预测模型[J]. 中国航海, 2025, 48 (4): 132- 140.
	YUAN Zhitao, LI Zewei, LIU Kezhong, et al. Research on ship trajectory prediction model based on GRU-Attention-BiLSTM[J]. Navigation of China, 2025, 48 (4): 132- 140.
35	董福贵, 夏美娟, 李婉莹. 基于PSO-GWO的省级能耗强度预测与碳减排潜力估算[J]. 中国电力, 2023, 56 (9): 226- 234.
	DONG Fugui, XIA Meijuan, LI Wanying. Prediction of provincial energy consumption intensity and estimation of carbon emission reduction potential based on PSO-GWO[J]. Electric Power, 2023, 56 (9): 226- 234.
36	陈湘东, 黎朝晖, 欧阳资铭, 等. 基于WOA优化的VMD-CNN-BiGRU储能电站负荷预测[J]. 现代电子技术, 2025, 48 (24): 81- 87.
	CHEN Xiangdong, LI Zhaohui, OUYANG Ziming, et al. WOA optimization based VMD-CNN-BiGRU load forecasting in energy storage power station[J]. Modern Electronic Technique, 2025, 48 (24): 81- 87.
37	范明辉, 李薇, 焦宇浩, 等. 基于SR-SSA-SVR的铁尾矿砂混凝土抗压强度预测[J]. 材料导报, 2025, 39 (24): 113- 120.
	FAN Minghui, LI Wei, JIAO Yuhao, et al. Compressive strength prediction of iron tailing sand concrete based on SR-SSA-SVR[J]. Materials Reports, 2025, 39 (24): 113- 120.