基于改进SVM的智能电网调控系统实时风险评估与预警技术  

Real-time risk-assessment and early-warning technology of smart grid regulation system based on improved SVM

作　　者：王宁[1,2] 田家英 董宁 韩盟[3] 陈艳霞[2] WANG Ning;TIAN Jia-ying;DONG Ning;HAN Meng;CHEN Yan-xia(School of Information and Electrical Engineering,China Agricultural University,Beijing 100083,China;Electric Power Research Institute,State Grid Beijing Electric Power Company,Beijing 100031,China;Power Dispatching Control Center,State Grid Beijing Electric Power Company,Beijing 100031,China)

机构地区：[1]中国农业大学信息与电气工程学院,北京100083 [2]国网北京市电力公司电力科学研究院,北京100031 [3]国网北京市电力公司电力调度控制中心,北京100031

出　　处：《沈阳工业大学学报》2022年第1期7-13,共7页Journal of Shenyang University of Technology

基　　金：北京市自然科学基金资助项目(4164101);国网北京市电力公司科技项目(52022319003R).

摘　　要：针对智能电网调控系统通信和数据安全难以保障的问题,提出了一种基于改进支持向量机(SVM)的智能电网调控系统实时风险评估与预警技术.采用卷积神经网络(CNN)改进SVM模型得到CNN-SVM分类模型,用以处理实时风险评估体系中的数据信息.通过将CNN输出的数据特征输入SVM分类器进行风险等级分类,完成对数据中可能出现的风险进行识别、评估定级及预警.仿真结果表明,所提技术能够对调控系统实时风险进行准确、可靠的评估与预警,且其分类准确率、召回率、F1分数的均值分别为92%、86%和90%,均优于对比方法并具有更优的可靠性.Aiming at the problem that it is difficult to ensure the communication and data security of smart grid regulation system,a real-time risk-assessment and early-warning technology of smart grid regulation system based on improved SVM was proposed.Convolutional neural network(CNN)was used to improve the support vector machine(SVM)model to obtain the CNN-SVM classification model,which was used to process the data information in the real-time risk-assessment system.By inputting the data features output by CNN into the SVM classifier for the classification of risk levels,the identification,evaluation,grading and early-warning of the risks that might appear in the data were completed.The simulation results show that the as-proposed technology can accurately and reliably evaluate and caution the real-time risk of regulation system.The mean values of classification accuracy,recall and F1 score are 93%,87%and 91%,respectively,which are better than those obtained by the reference method and facilitate better reliability.

关 键 词：卷积神经网络 支持向量机 CNN-SVM模型 智能电网 调控系统 数据处理 风险评估预警 实时风险评估体系 

分 类 号：TM734[电气工程—电力系统及自动化]