基于EMD的Prony算法在低频振荡模态参数辨识中的应用

Application of Prony algorithm based on EMD for identifying model parameters of low-frequency oscillations

Prony算法对分析数据的噪声非常敏感,对输入信号要求较高,鉴于此,提出将EMD和Prony算法有机结合的电力系统低频振荡模式的辨识方法。该方法以广域测量信号作为输入,首先利用EMD对非平稳、非线性信号的能力进行分解,通过能量权重比找出含有主导振荡模式的IMF;最后利用Prony算法对其进行分析后获得电力系统低频振荡模态参数,扩展了Prony法应用范围。通过对PSASP仿真轨迹的算例分析,验证了此方法提取非轴对称振荡信号主导模式的有效性,并通过与特征根分析进行比较,表明了此法能相对精确地进行振荡模式辨识,同时又有很好的复合模式分离能力和良好的抗噪能力。

Prony algorithm is sensitive to the noise of analysis data, and has a higher requirement to the input signal. Because of that, the paper proposes a method for identifying power system low-frequency oscillation, which combines EMD （Empirical Mode Decomposition） with Prony algorithm. This method considers the wide area measurement signals as the inputs of EMD. First of all, it takes advantage of EMD to decompose both non-stationary and nonlinear signal, find IMF （Intrinsic Mode Function） through energy weighting ratio, which has main-mode of low frequency oscillations, then model parameters of low-frequency oscillation in power systems are obtained by analyzing the IMF with Prony algorithm, which enlarge the scope of Prony. In the end, by analyzing the case studies on the curves simulated by PSASP, we found that the result verifies the validity of extracting non-axisymmetric oscillation signal. In addition, by comparing the Eigen-analysis, we can see that the proposed method has a good performance to separate complex mode with off-axis-symmetrical signal, and has a better ability of anti-noise.