基于SVMD和能量转移SR-MLS反演识别技术的低频振荡信号特征辨识

为解决多通道低频振荡信号特征辨识在噪声背景下提取精度低的问题,提出采用基于带宽总和限定的变分模态分解算法(bandwidth sum limit variational modal decomposition algorithm, SVMD)和随机共振–移动最小二乘(stochastic resonance-moving least squares, SR-MLS)反演识别技术相结合的方法进行低频振荡信号特征提取。以广域测量系统(wide area measurement system,WAMS)检测数据作为原始输入信号,利用SVMD算法对信号进行自适应去趋势项主导模态分离;再利用SR-MLS反演识别技术以噪声能量转移的方式进行强噪背景下带参信号反演,进而获得辨识频率、阻尼比、振幅等特征信息。最后,通过自合成模拟信号、IEEE16机68节点系统仿真以及东北电网实测数据3个算例进行分析,仿真结果表明,所提方法相比于传统方法 Prony和希尔伯特-黄变换法(Hilbert-Huang transform,HHT)算法具有更高的识别精度和稳定性。

Identification of Low Frequency Oscillation Signals Based on SVMD and Energy Transfer SR-MLS Inversion Recognition Technique

In order to solve the problem of low precision of multi-channel low-frequency oscillatory signal feature identification in noise background,we put forward a combination of bandwidth sum limit variational mode decomposition algorithm(SVMD)and stochastic resonance-moving least-squares(SR-MLS)inversion recognition technique for low-frequency oscillatory signal feature extraction.In this paper,the wide area measurement system(WAMS)detection data is used as the original input signal.The SVMD algorithm is used to separate the signal from the leading mode of adaptive detrend term.Then the SR-MLS inversion recognition technique is used to retrieve the band parameter signal in the strong noise-energy transfer mode,and then the identification frequency,damping ratio,amplitude and other characteristic information are obtained.Finally,the self-synthetic analog signal,IEEE 16-machine 68-node system simulation and three examples of measured data from the northeast power grid are analyzed.The proposed method has higher recognition accuracy and stability compared to the traditional methods such as the Prony and Hilbert-Huang transform(HHT).