Proposition of a single-degree-of-freedom experimental modal analysis method considering the effect of mode components close to the target mode

Abstract

Modal parameters are sometimes identified by using single-degree-of-freedom (SDOF) methods in the experimental modal analysis, where each resonance peak of the frequency response function (FRF) is considered as a SDOF system. The SDOF method is widely used in the actual application for identifying the modal parameters because it can be easily analyzed and introduced at low cost. It is noted that the conventional SDOF method can be only applied when the target mode is dominant. But in some cases, the effect of the non-target mode can't be ignored. In such cases, the multidegree-of-freedom methods are used, while it is more effective if the modal parameters can be easily identified by the SDOF method. In this study, we propose a new SDOF method for identifying the modal parameters even th ough the vibration modes are relatively close. The distinguished feature is that the FRF components of the non-target mode, that is the residual terms, are expressed by a polynomial with respect to the frequency, and then the coefficients of polynomial are identified by the linear fit method as well as the modal parameters. The appropriate order of the polynomial expressing the residual term was determined by theoretical approach, then we confirmed that it could be approximated by a cubic function. Furthermore, the validity and the applicability of proposed method were checked in the numerical examples, where the identification results were compared with the conventional linear fit method and the proposed method using with and without noise in the FRF data. The results showed that the identification accuracy is improved by using a cubic function as the residual term.