The spectral kurtosis (SK) is a useful tool for locating the non-stationary component in the single source signal. It has been verified by the mechanical fault diagnosis. The fast kurtogram (FK) is an improvement method based on the SK. However, for the signals with lower signal noise ratio (SNR), FK is not able to locate the chatter. In this paper, an on-line chatter monitoring using the FK and the frequency band power (FBP) is proposed to resolve this problem. Through the application of the FK, the frequency band with the largest SK can be found. In order to enhance the SK, a bandpass filter is designed. However, for the unstable cutting process, SK is not increased at all. Hence, FBP is adopted to monitor the chatter according to the energy change. The proposed method is verified by the milling experiments and satisfactory detection results are obtained.

谱峰度 (SK) 是用于定位单源信号中的非平稳分量的有用工具。经机械故障诊断验证。快速峰图（FK）是一种基于SK的改进方法。但是，对于信噪比（SNR）较低的信号，FK 无法定位抖动。在本文中，提出了一种使用 FK 和频带功率 (FBP) 的在线颤动监测来解决这个问题。通过FK的应用，可以找到SK最大的频段。为了增强SK，设计了带通滤波器。但是，对于不稳定的切削过程，SK 根本没有增加。因此，采用 FBP 根据能量变化来监测颤振。