Robustness and parameter selection in Stochastic Subspace Identification

• Main Authors: Chun-Yu Tsai, 蔡淳宇
• Other Authors: Yum Ji Chan
• Format: Others
• Language: en_US
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/33m3zd

碩士 === 國立中興大學 === 機械工程學系所 === 106 === Stochastic subspace identification operational modal analysis (SSI-OMA) is a recent method in modal testing. In this method, determining the size of the Hankel matrix, demoted I and N, is an important step. Different sizes of the Hankel matrix may lead to variations in identified modal parameters. In this thesis, the projection procedure is analyzed using probability and digital signal processing theory, and the variation of the projection result under Hankel matrix sizes is sought. It is found that the projection can be seen as a predictive filter and the projection quality is influenced by the location of passband if I=(3/4+1/2N_SL) T_Toep, the location of passband in Bode diagram would be located at the signal frequency. Also, the optimal value of I increases with sampling frequency. Therefore, under limited computer memory, it is beneficial to use low sampling frequency and a reduced value of I. In addition, it is found when the value of N is increased, the predictive filter would converge gradually The identified modal parameters are also influenced by variation of excitation force. Using Monte Carlo simulation, it is found that identified damping ratios are sensitive to the variation of excitation force and negative values may emerge on a system with positive damping loss factors. Finally, the SSI is applied to a multi-degrees-of-freedom test rig and the trends shown in simulations are validated.