摘要
针对传统傅里叶变换(fast Fourier transformation,FFT)算法分辨率不高,不能准确提取固有频率的问题,提出将矩阵束(matrix pencil,MP)、自回归滑动平均(auto-regressive and moving average,ARMA)模型、稀疏时间需求(sparse time demain,STD)时域模态参数辨识方法应用于提取固有频率。为此,介绍了这几种算法的基本原理,并采用电力系统计算机辅助设计(power system computer aided design,PSCAD)联合MATLAB软件对相关算例进行大量仿真。仿真结果表明,MP、ARMA、STD方法能大大提高频率分辨率,准确提取固有频率值,较FFT方法有明显的优越性。
Aiming at low distinguishability of fast Fourier transformer （FFT） and failure in accurate extraction of natural frequency, the paper suggests applying matrix pencil （MP）, auto-regressive and moving average （ARMA） model and sparse time demand （STD） time domain modal parameter recognition method in extracting natural frequency. Therefore, the paper introduces principles of the algorithms and simulates in bulk on relevant examples by power system computer aided design （PSCAD） combining with MATLAB. The simulation result shows that MP, ARMA and STD methods are of superiority to FFT method in frequency recognition and natural frequency extraction.
出处
《广东电力》
2012年第8期91-96,共6页
Guangdong Electric Power
