论文部分内容阅读
现有的线性化方法难以有效辨识电力系统次同步振荡模态。提出一种处理非线性、非平稳信号的阻尼正弦原子分解方法。该方法在过完备阻尼正弦原子库基础上，采用匹配追踪(matching pursuit，MP)算法对次同步振荡信号进行原子分解，并通过改进粒子群算法(improved particle swarm optimization，IPSO)降低MP搜索过程的时间复杂度，得到表征次同步振荡信号的阻尼正弦原子参变量，完成整个次同步振荡模态参数辨识，并与改进Prony算法及快速傅里叶变换的辨识结果进行对比分析。结果表明，基于阻尼正弦原子分解的次同步振荡模态辨识方法能快速准确地辨识次同步振荡模态，且具有良好的时频特性。“,” It is hard to use the existing linearization method to effectively identify power system subsynchronous oscillation modal. In this paper, a treatment method for nonlinear or nonstationary signal-damping sine atomic decomposition was proposed. The method was based on complete damping sine atomic library, used the matching pursuit (MP) algorithm for atomic decomposition of subsynchronous oscillation signal, and adopted the improved PSO (IPSO) to reduce time complexity of MP search process. As the parameters of damping sine atomic that represent subsynchronous oscillation signal were obtained, the whole process of subsynchronous oscillation modal identification could be accomplished. Comparison with the improved Prony algorithm and fast Fourier transform (FFT) indicates the presented method can identify the modal faster and more accurately, and it has better time-frequency features.