Risk Assessment and Cause Tracing of New Distribution System Based on Prior Feature Fitting Learning

MA Fuqi; MIN Jin; JIA Rong; WANG Bo; MA Hengrui; ZHANG Zhihua

doi:10.7500/AEPS20251106003

您当前的位置：

首页 >

文章列表页 >

Risk Assessment and Cause Tracing of New Distribution System Based on Prior Feature Fitting Learning

Basic Research | 更新时间：2026-09-01

- Risk Assessment and Cause Tracing of New Distribution System Based on Prior Feature Fitting Learning
- Automation of Electric Power Systems Vol. 50, Issue 17, Pages: 210-218(2026)
- 作者机构：
  
  1.西安理工大学电气工程学院，陕西省西安市 710054
  2.武汉大学电气与自动化学院，湖北省武汉市 430072
  3.国网陕西省电力公司电力科学研究院，陕西省西安市 710199
- 作者简介：
- 基金信息：
  
  National Natural Science Foundation of China(52407143)
- DOI：10.7500/AEPS20251106003
  CLC：
- Received：06 November 2025，
  
  Revised：2026-03-25，
  
  Online First：23 July 2026，
  
  Published：10 September 2026
- 稿件说明：
移动端阅览
马富齐,闵金,贾嵘等.基于先验特征拟合学习的新型配电系统风险评估及诱因溯源[J].电力系统自动化,2026,50(17):210-218.

MA Fuqi,MIN Jin,JIA Rong,et al.Risk Assessment and Cause Tracing of New Distribution System Based on Prior Feature Fitting Learning[J].Automation of Electric Power Systems,2026,50(17):210-218.
马富齐,闵金,贾嵘等.基于先验特征拟合学习的新型配电系统风险评估及诱因溯源[J].电力系统自动化,2026,50(17):210-218. DOI： 10.7500/AEPS20251106003.

MA Fuqi,MIN Jin,JIA Rong,et al.Risk Assessment and Cause Tracing of New Distribution System Based on Prior Feature Fitting Learning[J].Automation of Electric Power Systems,2026,50(17):210-218. DOI： 10.7500/AEPS20251106003.

摘要

新型配电系统在分布式资源高渗透接入下易出现反向重过载、电压越限等安全问题，现有风险评估方法对运行数据特征依赖性强、跨场景泛化能力弱、可解释性不足。为此，文中提出一种基于先验特征拟合学习的新型配电系统风险评估及诱因溯源方法，实现零训练样本条件下跨场景风险评估。首先，考虑分布式资源海量接入引发的安全风险问题，构建多维度风险评估指标体系；其次，提出一种基于先验任务驱动的风险特征拟合学习模型，将风险评估问题建模为先验任务约束下的贝叶斯条件推理问题，利用双向注意力机制刻画特征间交互关系与样本间统计依赖；然后，基于结构因果模型生成刻画风险特征依赖关系的先验合成数据，使模型能够从上下文样本中学习特征分布规律并执行条件推理；最后，引入合成少数类过采样技术优化类别不平衡问题，并融合Shapley值加性解释方法量化特征贡献度实现风险特征分析及诱因溯源。实验结果表明，所提方法在不同拓扑结构、上下文集规模变化及噪声干扰条件下均具有良好的泛化能力与鲁棒性。

Abstract

New distribution systems with high penetration of distributed resources are prone to security issues such as reverse heavy overload and voltage limit violations. Existing risk assessment methods suffer from strong dependence on operational data features

poor cross-scenario generalization capability

and insufficient interpretability. To address these problems

this paper proposes a risk assessment and cause tracing method for new distribution systems based on prior feature fitting learning

which enables cross-scenario risk assessment with zero training samples. Firstly

considering the security risks caused by the massive integration of distributed resources

a multi-dimensional risk assessment indicator system is constructed. Secondly

a prior task-driven risk feature fitting learning model is proposed. The risk assessment problem is formulated as a Bayesian conditional inference problem constrained by prior tasks

and a bidirectional attention mechanism is adopted to characterize the interactive relationships among features and statistical dependencies among samples. Then

prior synthetic data that characterize the dependency relationships of risk features are generated based on the structural causal model

enabling the model to learn the distribution law of features from in-context samples and perform conditional inference. Finally

the synthetic minority oversampling technique is introduced to optimize the class imbalance

and the Shapley additive explanation method is integrated to quantify feature contributions

so as to realize risk feature analysis and cause tracing. Experimental results demonstrate that the proposed method possesses excellent generalization capability and robustness under variable topological structures

changing in-context set scales

and noise interference conditions.

关键词

Keywords

references

国家发展改革委 . “十四五”可再生能源发展规划［EB/OL］.［ 2025-04-08 ］. https：//www.ndrc.gov.cn https://www.ndrc.gov.cn .

National Development and Reform Commission . Renewable energy development plan for the “14th Five-Year Plan ”［EB/OL］. ［ 2025-04-08 ］. https：//www.ndrc.gov.cn https://www.ndrc.gov.cn .

王守相，尹孜阳，赵倩宇 . 考虑多供电层级耦合的中低压配电网分布式光伏承载力一体化精细评估方法［J］. 电工技术学报， 2025 ， 40 （ 6 ）： 1930 - 1944 .

Wang Shouxiang ， Yin Ziyang ， Zhao Qianyu . A precise distributed PV hosting capability evaluation method for MV and LV distribution network considering the coupling of multiple power supply layers ［J］. Transactions of China Electrotechnical Society ， 2025 ， 40 （ 6 ）： 1930 - 1944 .

郭雪丽，胡志勇，王爽，等 . 考虑大规模风光分层接入的配电网多层协调无功优化方法［J］. 电力系统保护与控制， 2024 ， 52 （ 12 ）： 113 - 122 .

Guo Xueli ， Hu Zhiyong ， Wang Shuang ， et al . A multi-layer coordinated reactive power optimization method for a distribution network considering large-scale distributed wind-photovoltaic hierarchical access ［J］. Power System Protection and Control ， 2024 ， 52 （ 12 ）： 113 - 122 .

陈春，黄俊贤，孙辰昊，等 . 多源异构数据分布下的配电网风险状态多维评估方法［J］. 高电压技术， 2023 ， 49 （ 6 ）： 2297 - 2307 .

Chen Chun ， Huang Junxian ， Sun Chenhao ， et al . Multidimensional assessment method for the risk states of distribution networks under multi-source heterogeneous data distributions ［J］. High Voltage Engineering ， 2023 ， 49 （ 6 ）： 2297 - 2307 .

张稳，盛万兴，刘科研，等 . 高渗透率分布式电源按节点关键性接入配电网的运行风险评估［J］. 高电压技术， 2021 ， 47 （ 3 ）： 937 - 945 .

Zhang Wen ， Sheng Wanxing ， Liu Keyan ， et al . Operation risk assessment of high penetration distributed generation connected to distribution network according to node criticality ［J］. High Voltage Engineering ， 2021 ， 47 （ 3 ）： 937 - 945 .

University X J ， Lin C F . An online data-driven risk assessment method for resilient distribution systems ［J］. CPSS Transactions on Power Electronics and Applications ， 2021 ， 6 （ 2 ）： 136 - 144 .

黄煜，李忠行，胡松林，等 . 基于Lasso回归稀疏多项式混沌展开的配电网风险评估［J/OL］. 工程科学学报， 2025-12-23 . https：//link.cnki.net/urlid/10.1297.tf.20251223.1240.002 https://link.cnki.net/urlid/10.1297.tf.20251223.1240.002 .

Huang Yu ， Li Zhongxing ， Hu Songlin ， et al . Risk assessment of distribution network based on sparse polynomial chaos expansion with Lasso regression ［J/OL］. Chinese Journal of Engineering ， 2025-12-23 . https：//link.cnki.net/urlid/10.1297.tf.20251223.1240.002 https://link.cnki.net/urlid/10.1297.tf.20251223.1240.002 .

黄煜，李忠行，岳东，等 . 基于双稀疏化多项式混沌展开的高维配电网风险评估方法［J］. 电力系统自动化， 2026 ， 50 （ 4 ）： 101 - 110 .

Huang Yu ， Li Zhongxing ， Yue Dong ， et al . High-dimensional risk assessment method for distribution network based on dual-sparse polynomial chaos expansion ［J］. Automation of Electric Power Systems ， 2026 ， 50 （ 4 ）： 101 - 110 .

梁远升，程康，王钢，等 . 基于概率预测与随机响应面法的新能源孤岛配电网实时风险评估与调控策略［J］. 电网技术， 2023 ， 47 （ 12 ）： 4948 - 4957 .

Liang Yuansheng ， Cheng Kang ， Wang Gang ， et al . Real-time risk assessment and regulation strategy of new energy islanded distribution network based on probabilistic prediction and stochastic response surface methodology ［J］. Power System Technology ， 2023 ， 47 （ 12 ）： 4948 - 4957 .

Tan B D ， Zhao J B ， Chen Y S . Scalable risk assessment of rare events in power systems with uncertain wind generation and loads ［J］. IEEE Transactions on Power Systems ， 2025 ， 40 （ 2 ）： 1374 - 1388 .

滕堂勇，黄煜，王毅，等 . 计及网络动态重构的含分布式电源配电网风险评估方法［J/OL］. 电气工程学报， 2025-12-31 . https：//link.cnki.net/urlid/10.1289.TM.20251231.1019.002 https://link.cnki.net/urlid/10.1289.TM.20251231.1019.002 .

Teng Tangyong ， Huang Yu ， Wang Yi ， et al . Risk assessment method for distributed power distribution networks considering network dynamic reconstruction ［J/OL］. Journal of Electrical Engineering ， 2025-12-31 . https：//link.cnki.net/urlid/10.1289.TM.20251231.1019.002 https://link.cnki.net/urlid/10.1289.TM.20251231.1019.002 .

Aslam Ansari O ， Gong Y Z ， Liu W J ， et al . Data-driven operation risk assessment of wind-integrated power systems via mixture models and importance sampling ［J］. Journal of Modern Power Systems and Clean Energy ， 2020 ， 8 （ 3 ）： 437 - 445 .

王鹤，余中枢，李筱婧，等 . 基于主成分分析方法的多类型电动汽车接入配电网的综合风险评估［J］. 电力自动化设备， 2021 ， 41 （ 11 ）： 57 - 65 .

Wang He ， Yu Zhongshu ， Li Xiaojing ， et al . Comprehensive risk assessment of multiple types of electric vehicles connected to distribution network based on principal component analysis method ［J］. Electric Power Automation Equipment ， 2021 ， 41 （ 11 ）： 57 - 65 .

方文倩，王颖，王明深，等 . 计及分布式电源不确定性的新型配电网综合风险评估［J/OL］. 华北电力大学学报（自然科学版）， 2026-02-27 . https：//link.cnki.net/urlid/10.1999.TM.20260227.1150.002 https://link.cnki.net/urlid/10.1999.TM.20260227.1150.002 .

Fang Wenqian ， Wang Ying ， Wang Mingshen ， et al . Comprehensive risk assessment of new distribution network considering distributed generation uncertainty ［J/OL］. Journal of North China Electric Power University （Natural Science Edition）， 2026-02-27 . https：//link.cnki.net/urlid/10.1999.TM.20260227.1150.002 https://link.cnki.net/urlid/10.1999.TM.20260227.1150.002 .

Shi X ， Qiu R ， Mi T B ， et al . Adversarial feature learning of online monitoring data for operational risk assessment in distribution networks ［J］. IEEE Transactions on Power Systems ， 2020 ， 35 （ 2 ）： 975 - 985 .

Liu X M ， Liu J ， Zhao Y ， et al . A Bayesian deep learning-based probabilistic risk assessment and early-warning model for power systems considering meteorological conditions ［J］. IEEE Transactions on Industrial Informatics ， 2024 ， 20 （ 2 ）： 1516 - 1527 .

Hamilton R I ， Papadopoulos P N . Using SHAP values and machine learning to understand trends in the transient stability limit ［J］. IEEE Transactions on Power Systems ， 2024 ， 39 （ 1 ）： 1384 - 1397 .

GB/T 1094.12—2013 电力变压器第12部分：干式电力变压器负载导则［S］.

GB/T 1094.12—2013 Power transformers： Part 12： loading guide for dry-type power transformers ［S］.

GB/T 1094.7—2008 电力变压器第7部分：油浸式电力变压器负载导则［S］.

GB/T 1094.7—2008 Power transformers： Part 7： loading guide for oil-immersed power transformers ［S］.

GB 12325—1990 电能质量供电电压允许偏差［S］.

GB 12325—1990 Quality of electric energy supply—admissible deviation of supply voltage ［S］.

DL/T 2041—2019 分布式电源接入电网承载力评估导则［S］.

DL/T 2041—2019 Technical guideline for evaluating power grid hosting capacity of distributed resources connected to network ［S］.

于雨彤，王灿，李勇，等 . 高比例光伏接入的配电网多层级反向重过载风险评估［J］. 高电压技术， 2024 ， 50 （ 10 ）： 4540 - 4549 .

Yu Yutong ， Wang Can ， Li Yong ， et al . Multi-level reverse heavy overload risk assessment for distribution system with high percentage of photovoltaic access ［J］. High Voltage Engineering ， 2024 ， 50 （ 10 ）： 4540 - 4549 .

张海峰，任洲洋，冯健冰，等 . 基于内驱进化预测模型和载荷能力动态评估的变压器周前重过载预警方法［J］. 电网技术， 2024 ， 48 （ 10 ）： 4349 - 4357 .

Zhang Haifeng ， Ren Zhouyang ， Feng Jianbing ， et al . A week-ahead heavy and overload warning strategy based on an endogenous evolutionary forecasting model and a transformer rating dynamic evaluation method ［J］. Power System Technology ， 2024 ， 48 （ 10 ）： 4349 - 4357 .

曹璞璘，申忠友，束洪春，等 . 基于波形突变特征的海量行波数据中雷击干扰波形辨识［J］. 电力系统自动化， 2023 ， 47 （ 17 ）： 178 - 186 .

Cao Pulin ， Shen Zhongyou ， Shu Hongchun ， et al . Identification of lightning interference waveforms in massive traveling wave data based on waveform mutation characteristics ［J］. Automation of Electric Power Systems ， 2023 ， 47 （ 17 ）： 178 - 186 .

Chen T P ， Zhang C ， Jing W Z ， et al . State of health estimation for lithium-ion batteries using separable LogSparse self-attention transformer ［J］. IEEE Transactions on Instrumentation and Measurement ， 2025 ， 74 ： 3522513 .

王小君，窦嘉铭，刘曌，等 . 可解释人工智能在电力系统中的应用综述与展望［J］. 电力系统自动化， 2024 ， 48 （ 4 ）： 169 - 191 .

Wang Xiaojun ， Dou Jiaming ， Liu Zhao ， et al . Review and prospect of explainable artificial intelligence and its application in power systems ［J］. Automation of Electric Power Systems ， 2024 ， 48 （ 4 ）： 169 - 191 .

顾菊平，蒋凌，张新松，等 . 基于特征提取的锂离子电池健康状态评估及影响因素分析［J］. 电工技术学报， 2023 ， 38 （ 19 ）： 5330 - 5342 .

Gu Juping ， Jiang Ling ， Zhang Xinsong ， et al . Estimation and influencing factor analysis of lithium-ion batteries state of health based on features extraction ［J］. Transactions of China Electrotechnical Society ， 2023 ， 38 （ 19 ）： 5330 - 5342 .

祖文静，李勇，谭益，等 . 高渗透率可再生能源配电网测试系统建模与有效性分析［J］. 电力自动化设备， 2019 ， 39 （ 7 ）： 45 - 50 .

Zu Wenjing ， Li Yong ， Tan Yi ， et al . Modeling and effectiveness analysis on test system for distribution networks with high penetration of renewable resource ［J］. Electric Power Automation Equipment ， 2019 ， 39 （ 7 ）： 45 - 50 .

侯慧，朱韶华，俞菊芳，等 . 基于高效数据降维的配电网风灾停电用户数量预测模型［J］. 电力系统自动化， 2022 ， 46 （ 7 ）： 69 - 76 .

Hou Hui ， Zhu Shaohua ， Yu Jufang ， et al . Prediction model for user number in power outage caused by wind disaster for distribution networks based on high-efficient data dimensionality reduction ［J］. Automation of Electric Power Systems ， 2022 ， 46 （ 7 ）： 69 - 76 .

Faustine A ， Nunes N J ， Pereira L . Efficiency through simplicity： MLP-based approach for net-load forecasting with uncertainty estimates in low-voltage distribution networks ［J］. IEEE Transactions on Power Systems ， 2025 ， 40 （ 1 ）： 46 - 56 .

Zhu L P ， Hill D J . Networked time series shapelet learning for power system transient stability assessment ［J］. IEEE Transactions on Power Systems ， 2022 ， 37 （ 1 ）： 416 - 428 .

Hokmabad H N ， Husev O ， Belikov J . Day-ahead solar power forecasting using LightGBM and self-attention based encoder-decoder networks ［J］. IEEE Transactions on Sustainable Energy ， 2025 ， 16 （ 2 ）： 866 - 879 .

Arzamasov V . Electrical grid stability simulated data ［PP/OL］. ［ 2025-10-20 ］. https：//doi.org/10.24432/C5PG66 https://doi.org/10.24432/C5PG66 .

Arzamasov V ， BöHm K ， Jochem P . Predicting smart grid stability with deep learning ［EB/OL］. ［ 2025-10-20 ］. https：//www.kaggle.com/code/pcbreviglieri/predicting-smart-grid-stability-with-deep-learning https://www.kaggle.com/code/pcbreviglieri/predicting-smart-grid-stability-with-deep-learning .

ZIYA 07 . Smart grid monitoring dataset［DS/OL］ . ［ 2025-10-20 ］. https：//www.kaggle.com/datasets/ziya07/smart-grid-monitoring-dataset https://www.kaggle.com/datasets/ziya07/smart-grid-monitoring-dataset .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Risk Assessment and Reasoning Method for Power System Operation Based on Large Language Models

Short-term Load Forecasting Method for Electric Vehicle Charging Stations Based on Improved Spatio-Temporal Graph Convolutional Network

Large Language Model Based Decision-making for Distribution Emergency Maintenance and Its Derived Risk Reasoning

Mechanism and Clearing Model of Regulatory Capacity Market for New Power System

Suppression Strategy of High-frequency Oscillation for DFIG Grid-connected System Based on Stator Voltage Feedforward Compensation

Related Author

WANG Runzhu

WANG Hao

SHAO Changzheng

XIE Kaigui

HU Bo

HUANG Wenqi

ZHAO Bo

LIU Yihang

Related Institution

College of Electrical Engineering, Chongqing University

China Electric Power Research Institute

Beijing Huairou Laboratory

Southern Power Grid New Energy System (Beijing) Research Institute Co., Ltd.

Shanghai Artificial Intelligence Laboratory

⁰