基于深度置信网络的电力系统扰动后频率曲线预测  

A Method of Frequency Curve Prediction Based on Deep Belief Network of Post-disturbance Power System

作　　者：仉怡超 闻达 王晓茹[1] 林进钿 ZHANG Yichao;WEN Da;WANG Xiaoru;LIN Jintian(School of Electrical Engineering,Southwest Jiaotong University,Chengdu 610031,Sichuan Province,China)

机构地区：[1]西南交通大学电气工程学院

出　　处：《中国电机工程学报》2019年第17期5095-5104,共10页Proceedings of the CSEE

基　　金：中国电力科学研究院有限公司项目(1808-00790)~~

摘　　要：为快速、准确地预测扰动后电力系统的动态频率,该文基于深度置信网络(deep belief networks,DBN)提出一种预测扰动后电力系统频率曲线的方法。该方法以发电机的电磁功率、机械功率、同步发电机最大出力限制、各发电机对动态频率的影响因子等在内的22维数据作为深度置信网络的输入特征值,输出为系统的动态频率。该文采用新英格兰10机39节点系统和美国南卡罗来纳州的90机500母线系统作为仿真研究算例,通过与PSS/E中的仿真结果相对比,证明使用深度置信网络可以快速准确地对扰动后系统的动态频率进行预测。该方法适用于频率的在线稳定分析,可为后续制定频率稳定控制措施提供依据,对防止系统频率崩溃具有重要意义。In order to quickly and accurately predict the dynamic frequency of the post-disturbance power system, this paper proposed a method for predicting the dynamic frequency of the power system based on deep belief network. The method considered the 22-dimensional data as the input features of the deep belief network, including the electromagnetic power and the mechanical power, the maximum output limit of each generator, and the influence factor of each generator on the dynamic frequency. The New England 39-bus system and the South Carolina 500-bus system were used as the simulation systems. Compared with the simulation results in PSS/E, it is proved that the deep belief network can quickly and accurately predict the dynamic frequency of the post-disturbance power system. The method is applicable to online stability analysis of frequency, and it can provide a basis for setting frequency stability control measures which is important to prevent frequency collapse.

关 键 词：电力系统扰动 频率动态预测 深度置信网络 机器学习 

分 类 号：TM712[电气工程—电力系统及自动化]