基于粒子滤波与EEMD的低频振荡模式识别方法研究

Pattern recognition research on low frequency oscillation based on particle filter and EEMD

为克服传统方法对非线性非高斯系统信号中噪声处理的缺点,提出一种基于粒子滤波算法与改进的EMD分解—EEMD分解法相结合的新方法。所提方法首先利用粒子滤波将非线性非高斯系统的初始信号的噪声去除,减少了噪声对后续操作的影响,再采用EEMD分解对去噪后的信号进行分解得到此征模态分量IMF,进而对此征模态分量IMF计算出瞬时频率,从而得出低频振荡的模式。通过算例仿真分析表明文中方法的可行性及有效性,并通过与Prony分析算法得到的结果进行了对比,验证了文中方法的正确性。为电力系统低频振荡处理非线性非高斯系统信号提供了一种新的途径和方法。

In order to overcome the shortcomings of traditional methods for noise processing in non-linear non-Gaussi an systems,this paper proposes a new method based on particle filter and improved EMD-EEMD decomposition.In this paper,the noise of the non-linear non-Gaussian system is removed by the particle filter,and the influence noise on the subsequent operation is reduced.Then,the EEMD decomposition is used to decompose the de-noise signal.And then,the instantaneous frequency is calculated for the intrinsic mode component IMF,and the mode of low frequency oscillation is obtained.The feasibility and effectiveness of the proposed method are demonstrated by simulation analysis.The results are compared with the results obtained by the Prony analysis algorithm,which verifies the correctness of the method,which provides a new way and method for the no-linear non-Gaussian system signal processing in power system low-frequency oscillation.