低频振荡模式辨识的改进衰减正弦原子分解法

为克服传统衰减正弦原子分解法的估计偏差,提出一种改进的衰减正弦原子分解法用于低频振荡模式辨识。该方法在匹配追踪中引入松弛算法的思想,通过迭代的策略分离各个振荡模式,并采用改进萤火虫算法和伪牛顿法相结合的优化方法,在衰减正弦原子库中进行参数寻优。通过数值信号仿真、IEEE4机2区系统仿真和相量测量单元PMU(phasor measurement unit)实测信号仿真,验证了本文所提方法可获得振荡参数的无偏估计,提高了辨识精度,且具有良好的抗噪能力,能够满足低频振荡模式辨识的要求。

Low-frequency Oscillation Mode Identification Using the Improved Damped Sinusoidal Atomic Decomposition Method

To overcome the estimation bias due to the traditional damped sinusoidal atomic decomposition method,an improved method for low-frequency oscillation mode identification is proposed in this paper.This method introduces the idea of relax algorithm to the matching pursuit and separates different oscillation modes in an iterative manner.In addi tion,it uses a combination method of the improved firefly algorithm and pseudo-Newton method for parameter optimiza tion in the damped sinusoidal atom dictionary.Simulation results of numerical signals,an IEEE 4-machine 2-area sys tem,and phasor measurement unit(PMU)actual signals show that the proposed method is able to obtain unbiased esti mations for mode parameters,thus improving the identification accuracy.Moreover,it has a strong anti-noise capability and can meet the requirements of low-frequency oscillation mode identification.