抄録
In is very important to understand the dynamic characteristics to control vibrations of machinery. Generally, the conventional identification methods for experimental modal analysis are employed to time-invariant systems. However, there are a lot of requirements to extract modal parameters of time-varying systems. It is hard to implement the vibration testing for time-varying systems, because of excitation difficulties, boundary conditions, etc. In this paper, the subspace algorithm is proposed to obtain natural frequencies of time-varying systems. Then, a combination procedure of SVD and QR decomposition is proposed for the identification algorithm to improve computational load. The effectiveness of the proposed time-varying identification algorithm is verified by numerical results. Finally, the procedure is applied to an experimental manipulator model.
