AN APPROACH TO MODAL PARAMETER-IDENTIFICATION USING DATA-COMPRESSION

Data compression techniques have been introduced as an additional step in the modal analysis process, with the possible benefits of data volume reduction and time savings. A feasibility study was performed to determine if this addition would adversely aff ect the identification of natural frequencies and damping ratios. Frequency response functions for both single- and multiple-degree-of-freedom (SDOF and MDOF, respectively) systems were simulated for noise-free and 10%, 20%, and 50% rms noise levels. This data was analyzed in its uncompressed form as well as for several levels of compression, using both an SDOF Nyquist circle-fit algorithm as well as the Eigensystem Realization Algorithm (ERA). Overall results indicate that a significant reduction in volume is possible without any deleterious effects to the parameter identification.