Study And Implementation Of Detection And Removal Of Harmonic Modes For Structures In Operation

In the field of rotating machines, the periodic excitations introduce harmonic modes intostructural modes, which greatly affect the accuracy and reliability of modal analysis. Therefore, it isvery important for the modal analysis of structures in operation to detect harmonic modes and removetheir effects as much as possible.The techniques of detection and removal of harmonic modes for structures in operation aresystematically studied in this thesis.(1) Two traditional time-domain detection methods based onprobability density and kurtosis are improved.(2) The effects of the selection of frequency band andfilter order in the narrow-band filter are analyzed.(3) The concept of higher-order cumulant isintroduced to present a high-efficiency spectral kurtosis algorithm, by which all the harmonic modescan be simultaneously detected in a broad frequency band.(4) The influence of harmonic modes invarious operational modal parameter identification algorithms is anlyzed.(5) The effects of differentstop-band filters on the removal efficiency is studied, and a proper filter bandwidth selection methodis presented.(6) A modal parameter correction method on the half power spectral density matrix isproposed, which is of high efficiency and easy to implement.(7) Based on MATLAB GUIprogramming technique, a software with good human-computer interaction is developed for thedetection and removal of harmonic modes in the operational modal analysis.