基于G1赋权法的新型配电网安全态势综合评估技术  

A New Comprehensive Assessment Technology of Distribution Network Security Situation Based on G1 Weighting Method

作　　者：高捷 孔世炜 方珺 刘涌 李远舸 GAO Jie;KONG Shi-wei;FANG Jun;LIU Yong;LI Yuan-ge(State Grid Zhejiang Shaoxing Power Supply Company,Shaoxing 312000 China;State Grid Zhejiang Xinchang Power Supply Company,Shaoxing 312000 China;Shanghai Proinvent Information Technology Co.,Ltd.,Shanghai 200240 China)

机构地区：[1]国网浙江省电力有限公司绍兴供电公司,浙江绍兴312000 [2]国网浙江省电力有限公司新昌县供电公司,浙江绍兴312000 [3]上海博英信息科技有限公司,上海200240

出　　处：《自动化技术与应用》2025年第3期34-37,99,共5页Techniques of Automation and Applications

基　　金：国网浙江省电力有限公司绍兴供电公司经济技术服务项目(B7SBSX200010)。

摘　　要：电力系统缺乏对于新型配电网运行状态的实时掌控,智能化水平与自动化水平均较低,设计基于G1赋权法的新型配电网安全态势综合评估技术,合理评估其安全态势。以新型配电网安全态势为评估目标,构建评估指标体系;设计Autoencoder深度学习自编码网络,其由数据输入层、最终输出层以及中间限制波兹曼机(RBM)层构成,对各评估指标数据实施降维处理;通过G1赋权法获取各指标的权重,利用效用理论与ER算法等实现新型配电网的安全态势综合评估。测试结果表明,设计技术的态势评估误差在后期趋于平稳,稳定在0.02以下,评估耗时较低。The power system lacks real-time control over the operation status of the new distribution network,and the level of intelligence and automation is low.A new comprehensive assessment technology for the security situation of distribution network based on G1 weighting method is designed to reasonably assess its security situation.Taking the new distribution network security situation as the evaluation target,the evaluation index system is constructed.Autoencoder deep learning self coding network is designed,which is composed of data input layer,final output layer and intermediate bound Boltzmann machine(RBM)layer,to reduce the dimension of each evaluation index data.The G1 weighting method is used to obtain the weight of each index,and the utility theory and ER algorithm are used to realize the comprehensive assessment of the security situation of the new distribution network.The test results show that the situation assessment error of the design technology tends to be stable in the later stage,which is less than 0.02,and the assessment time is low.

关 键 词：G1赋权法 新型配电网 网架结构 Autoencoder 卡方距离 安全态势评估 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]