一种新的阻尼正弦原子分解算法辨识SSO模态参数

针对大多数线性化方法难以实现对次同步振荡（subsynchronous oscillation，SSO）模态参数的有效辨识，提出了基于改进入侵杂草优化（invasive weed optimization，IWO）算法优化的阻尼正弦原子分解算法。该方法根据次同步振荡信号特点构造过完备阻尼正弦原子库，引入混沌序列、选择机制、小生境分类策略以及矢量跟踪思想对IWO算法进行改进，利用改进后的IWO算法对传统的匹配追踪算法（matching pursuit，MP）进行优化，通过原子分解得到最佳阻尼正弦原子，将最佳阻尼正弦原子转换为次同步振荡信号的模态参数，即可实现对次同步振荡模态参数的有效辨识。算例结果表明，该算法具有良好的时频特性，辨识精度高，适用于扰动源定位、故障诊断等领域。

Modal Parameter Identification of SSO Based on Damping Sine Atomic Decomposition Optimized by Improved IWO

Since the existing linearization method can’t effectively identify subsynchronous oscillation modal, the damping sine atomic decomposition based on improved Invasive Weed Optimization (IWO) algorithm is proposed. The complete damping sine atomic library that represents subsynchronous oscillation signal is constructed. The chaotic sequence, selection mechanism, Niche classification strategy, and vector tracing ideas are introduced into the improved IWO algorithm to optimize the traditional matching pursuit(MP) algorithm. The optimized MP algorithm is used for damping sine atomic decomposition of subsynchronous oscillation signal. And then the parameters of the obtained optimal damping sine atomic are converted into subsynchronous oscillation modal parameters. The identified results indicate that the damping sine atomic decomposition optimized by improved IWO has advantages of good time-frequency features and high identification accuracy. And it is applicable to the disturbance source localization, fault diagnosis and other fields.