In this paper, a new modal analysis method based on a novel multivariate data analysis technique called smooth orthogonal decomposition (SOD) is proposed. The development of the SOD and its main properties are described. The mathematical justification for the application in modal analysis is also provided. The proper orthogonal decomposition (POD) and its application in modal analysis are provided for comparison. Numerical simulations of discrete and continuous systems are used in this comparison. It is demonstrated that the SOD-based analysis overcomes main deficiencies of the POD. The SOD identifies linear normal modes and corresponding frequencies without requiring any a priori information about the distribution of mass in the system.

This content is only available via PDF.
You do not currently have access to this content.