Three-dimensional operational modal analysis based on self-iteration principal component extraction and direct matrix assembly

Abstract

This paper proposes a self-iteration principal component extraction (SIPCE) and direct matrix assembly method for three-dimensional structures. Different from calculating principal components (PCs) by matrix decomposition in traditional principal component analysis (PCA), SIPCE extracts PCs one by one through self-iteration, so SIPCE has lower space-time complexity. Besides that, it avoids singular-value and ill-posed problems of matrix decomposition. The previous method of solving three-dimensional structures is using modal coordinate response back general reversion of least square algorithm, while the new matrix assembly method calculates three-dimensional modal shapes at one time. So, the new matrix assembly method has less calculation error. The numerical simulation results in a cylindrical shell demonstrate that this method can be practically and effectively applied in operational modal analysis (OMA) of three-dimensional structures. The new method is also robust to noise, and has higher identification accuracy and lower space-time consumption than previous method.