基于动力学系统的两阶段低频振荡模态识别算法

Two-stage Low-frequency Oscillation Mode Recognition Algorithm Based on Dynamical System

低频振荡已经成为目前电力系统运行所面临的严重问题,在大量测噪声条件下,如何通过WAMS(wide ar?ea measurement system)量测值进行精准有效的振荡模态识别至关重要.本文提出了一种新的模态识别方法,通过构造商梯度系统,并追踪与最优参数估计值对应的退化稳定平衡流形进行求解.根据实际系统运行提出了两阶段算法:第一阶段通过改进随机子空间方法获得初始参数辨识值,若不满足残差精度要求,则第二阶段以其为初值通过基于商梯度系统的方法求解.通过在高比例噪声条件下与Prony算法对比,并对IEEE 39节点系统仿真数据分析,表明该算法在抗噪性能、辨识精度和适用性上具有优良特性.

Low-frequency oscillation has become a serious problem in the existing power system operation.Under the condition of a large number of noise measurements,it is of great importance to accurately and effectively identify the os⁃cillation modes by means of wide area measurement system(WAMS)measurements.In this paper,a novel modal recog⁃nition method is proposed,which is based on constructing a quotient gradient system(QGS)and tracing the degenerate stable equilibrium manifold corresponding to the optimal parameter estimation.According to the actual system opera⁃tion,a two-stage algorithm is put forward.In the first stage,the initial parameter identification value is obtained by im⁃proving the random subspace method.If the accuracy of residual error is not satisfied,then the obtained value will be used as an initial value in the second stage and the identification value will be solved using the method based on QGS.From the comparison with the Prony algorithm under the condition of high proportion noise and the analysis of simula⁃tion data from an IEEE 39-bus system,it is shown that the proposed algorithm has excellent characteristics in anti-noise performance,identification accuracy and applicability.