Prony算法的低频振荡主导模式识别

从Prony算法拟合阶数和数据预处理对主导振荡模式的有效识别进行了探讨.在分析归一化比值法和归一化奇异值法2个不同的拟合阶数判据的基础上,提出了应用基于数量级概念的归一化奇异值法阈值进行拟合阶数的确定.算例表明按相对变化大于3个数量级的阈值定阶能有效识别低频振荡的主导模式且计算量相对较小.应用快速且易实现的有限脉冲响应滤波方法进行高频噪声的滤波,能适用于Prony分析中的数据处理,有效识别主导振荡模式.用不同类型的样本数据对以上研究进行了仿真,仿真结果验证了其有效性.

Identification of control modes in low frequency oscillation analysis by Prony method

The valid identification of control oscillation modes is studied on fitting orders and preprocessing of Prony method.The threshold of unitary singular is put forward by the magnitude on the basis of analysis of different fitting orders criterions by unitary ratio and unitary singular methods.The threshold of relative change more than three orders of magnitude can effectively identify control oscillation modes and has less calculation through simulation of samples.The fast and tractable method of finite impulse respond filter is applied to filter high frequency noise,which avails preprocessing of Prony analysis and effectively identifies control oscillation modes.The proposed scheme is evaluated using simulation of various samples and results show its validity.