数据驱动的电力系统静态电压稳定态势评估  

Data-Driven Situation Assessment of Power System Static Voltage Stability

作　　者：梁修锐 刘道伟 杨红英 李卫星[1] 邵广惠[3] 徐兴伟[3] 王克非 李宗翰 赵高尚 LIANG Xiurui;LIU Daowei;YANG Hongying;LI Weixing;SHAO Guanghui;XU Xingwei;WANG Kefei;LI Zonghan;ZHAO Gaoshang(School of Electrical Engineering and Automation,Harbin Institute of Technology,Harbin 150001,China;China Electric Power Research Institute,Beijing 100192,China;Northeast Branch of State Grid Corporation of China,Shenyang 110000,China)

机构地区：[1]哈尔滨工业大学电气工程及自动化学院,哈尔滨市150001 [2]中国电力科学研究院有限公司,北京市100192 [3]国家电网公司东北分部,沈阳市110000

出　　处：《电力建设》2020年第1期126-132,共7页Electric Power Construction

基　　金：国家电网公司科技项目(XTB11201705943)~~

摘　　要：随着电力系统的日益复杂,传统静态电压稳定分析方法难以满足精度和速度要求。为此,提出一种数据驱动的静态电压稳定评估方法。首先,通过仿真工具沿时间轨迹捕捉不同时间断面上的系统状态数据,筛选出与系统电压稳定性关联度较高的特征属性,采用静态负荷有功功率裕度对数据进行标签分类。然后,应用集成多个决策树的随机森林分类器模型将原始状态标签数据转化为稳定信息和行为模型,从数据中感知和学习静态电压稳定行为。最后,通过查询随机森林的分类结果,实现对静态电压稳定性的判断。仿真结果表明,该模型可行有效,适用于静态电压稳定态势评估。With the increasing complexity of power systems,traditional static voltage stability analysis methods face challenges to meet the accuracy and speed of real-time static stability assessment. In this paper,a data-driven static voltage stability evaluation method is proposed. First,the system state data in different time sections are captured along the time trajectory by simulation tools,and the characteristic attributes with high correlation with voltage stability are screened out.The static load active power margin is used to classify the data. Second,a random forest classifier model integrating multiple decision trees is used to transform the original state label data into stable information and behavior models,and the static voltage stability behavior is sensed and learned from the data. Finally,by querying the classification results of the random forest,the judgment of the static voltage stability is achieved. Simulation results show that the model is feasible and suitable for static voltage stability situation assessment.

关 键 词：电力系统 静态电压稳定 数据驱动 态势感知 机器学习 随机森林 

分 类 号：TM73[电气工程—电力系统及自动化]