摘要
以时域、频域和小波分析 3种处理方法提取摇摆曲线的特征 ,对有限时段内稳定的多机摇摆曲线进行同调群识别 ,分离出所有潜在的失稳群。将摇摆曲线经过数字低通滤波器后重新采样 ,再求取各曲线的时域特征 ;用快速傅里叶变换得到其频域特征 ;用小波变换得到曲线在特定频段内的特征。然后用规则库分析这些特征 ,对摇摆曲线进行分群 ,其结果与 EEAC算法对后继时段分析所得到的失稳模式基本吻合。
It is a big challenge for a numerical integration method to assess multimachine system stability beyond the actually performed integration period,and reliable criterion for early stopping of the integration is very much needed.EEAC Extended Equal Area Criterion integrates the whole dynamic equations in and then divides the resultant n- machine trajectory into a complementary- subset in all possible ways.For each pair of the complementary clusters,the trajectories are mapped into a plane with a stability preserving transformation,and the stability of each image can be quantitatively assessed. If and only if all images are stable,the global system is stable.This not only makes quantitative analysis of multimachine system stability possible,but also opens a way to early stop the integration.If there are several machines which loss stability after a few swings,their trajectories mustbe well coherent.The present paper proposes a new approach to solve the problem of early stopping.It sorts stable trajectories of an n- machine system into coherent groups and analyzes the dynamic behaviors in the images defined by EEAC.The individual machine trajectories within a short time period are characterized by time domain analysis,frequency domain analysis and wavelet analysis successively.After passing a low pass digital filter,the data of swing curves are re- sampled to eliminate the local peaks and valleys.Then the time domain features,such as the average period and the positions of peaks and valleys,are extracted,and most of the coherent groups can be figured out.The remaining curves,which have non- predominant elements with either low frequency or high amplitude,are then processed in frequency domain.The Blackman Window is applied for fast Fourier transformation,and the curves are characterized with frequency elements,phases and relative amplitude.If there are still some curve remaining,wavelet analysis with a special spectrum of1 Hz～ 2 Hz is applied.An expertsystem is used to coordinate the classification procedures.Tests are performed to cases those are marginally stable within 1 0 seconds,as well as to cases those are marginally unstable after 1 0 seconds.The sorting procedure uses only the data within the first3seconds of the dynamics,and good performances are demonstrated in this paper.
出处
《电力系统自动化》
EI
CSCD
北大核心
2000年第1期13-16,共4页
Automation of Electric Power Systems
基金
国家重点基础研究专项经费资助项目! (G1 9980 2 0 3)
国家电力公司科技项目! (1 999SPKJ0 1 0 - 2 0 )
关键词
同调群
失稳模式
EEAC
小波变换
分群
电网
coherent group
unstable mode
recognition
EEAC
fast Fourier transform
wavelet transform
