基于原子稀疏分解的低频振荡模态参数辨识方法

摘要

为克服传统特征分析法不适合分析大规模高阶系统，而傅里叶算法和Prony算法等线性化方法又难以处理非平稳信号的缺点，本文将一种新的处理非线性、非平稳信号方法—原子稀疏分解法应用于电力系统低频振荡模态参数识别。该方法利用匹配追踪（MP）算法将初始信号从Gabor原子库中分解得到最佳匹配原子，并采用伪牛顿法对参变量进行优化，进而求出衰减正弦量原子的参变量，最终完成整个低频振荡模态参数的提取过程。仿真结果表明该方法的可行性和有效性，为电力系统稳定分析提供一种全新的途径和方法。%For the traditional eigenvalue analysis method is not suitable for analyzing the large scale high-order system and the Fourier method and Prony algorithm are incapable of dealing with nonstationary signals, a novel method, atomic sparse decomposition, is introduced in this paper which is invented to deal with nonlinear or nonstationary signals to the analysis of low-frequency oscillations. This method obtains the optimal atom from a signal by using matching pursuits (MP) algorithm with the Gabor dictionary, and makes a search for parameters using the Newton-like method and finally identified the parameters of damped sinusoids. The simulations and example analysis indicates the feasibility and validity of the new method proposed in this paper. It also provides a new way for power system stability analysis.