摘要
当系统发生严重级联故障导致失步运行时,需要快速准确地识别出系统中的同调机群,为下一步的自主解列控制提供基础。针对WAMS测量到的发电机动态轨迹信息往往具有非线性和非平稳性等特点,提出了一种在线识别同调机群的新方法,能充分考虑各种故障场景的动态特性和非线性系统的时变特征。首先根据WAMS量测可得到故障后发电机组的实时响应功角轨迹信息,利用基于类别信息和核空间的改进拉普拉斯特征映射算法提取特征信息,进而识别出各发电机的运行特性;再利用k-way余弦相似度因子分群算法对发电机组进行自主识别分群。最后通过新英格兰39节点系统仿真,验证了所提方法的有效性,并且适用于系统不同运行方式,能在线准确识别同调机群。
When power system suffers from cascading failures that would lead to system asynchronous operation,it is necessary to identify coherent generator groups accurately,which supplies basis for the study on controlled islanding strategy.Because dynamic trajectory of generator measured by WAMS tends to have nonlinear and non-stationary characteristic,a novel approach on identification of coherent generator groups is presented.This method can fully consider dynamic behavior of different fault modes and time varying characteristic of nonlinear system.Firstly,the real-time rotor angle trajectory information can be measured by WAMS directly.Based on categorization information and improved Laplacian Eigenmap algorithm in the kernel space,it extracts the feature information and identifies generator operation characteristic.Then with k-way Cosine similarity factor,it can provide automation for identifying coherent generator.Finally,the simulations on New England 39-bus system validate the effectiveness of such the novel approach proposed in this paper,which is suitable for different operation conditions.It can identify coherent generators on-line accurately.
出处
《电网技术》
EI
CSCD
北大核心
2013年第8期2157-2164,共8页
Power System Technology
基金
高等学校博士学科点专项科研基金资助项目(20100009110025)
中央高校基本科研业务费专项资金(2011YJS274)~~
关键词
广域量测系统
拉普拉斯特征映射
同调识别
特征提取
wide area measurement system（WAMS）
Laplacian Eigenmap algorithm
coherency identification
feature extraction
