在旋转机械领域,周期激励的存在使运行模态分析中含有谐波模态。根据结构模态和谐波模态的谱峭度值的不同,该文提出了基于谱峭度的谐波模态辨识方法。首先,给出了谱峭度的定义及其无偏估计算法;其次,从理论上证明应用此方法不需要滤波处理,并提出了简单且易于实现的优化方法;然后,应用频域空间域分解算法分离出谐波模态去除其对结构模态的影响;最后,通过仿真及实验成功检测出了谐波模态并消除其影响,验证了方法的有效性和可靠性。
In the operational modal analysis of rotating machines, the periodic excitations introduce harmonic modes into structural modes. Based on the different spectral kurtosis values between a structural mode and a harmonic mode, an approach for detecting harmonic modes is proposed. The definition of spectral kurtosis is introduced and an optimized fast algorithm of its estimation is presented. The frequency spatial domain decomposition modal identification method is applied to separate harmonic modes from structural modes. Finally, a simulation example and a real experiment are used to validate the approach, in which the harmonic modes are successfully distinguished and removed.