Premium
An improved equivalent linearization procedure for non‐linear systems under bounded random excitations
Author(s) -
Cai G. Q.,
Lin Y. K.
Publication year - 2006
Publication title -
structural control and health monitoring
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.587
H-Index - 62
eISSN - 1545-2263
pISSN - 1545-2255
DOI - 10.1002/stc.123
Subject(s) - bounded function , linearization , linear system , gaussian , mathematics , gaussian process , nonlinear system , control theory (sociology) , mathematical analysis , statistical physics , computer science , physics , quantum mechanics , control (management) , artificial intelligence
Taking into consideration that a physically realistic random process must be bounded in variation, a type of non‐linear filters is employed to model bounded random processes, which are then used as excitation processes for linear and non‐linear dynamical systems. In the case of a linear system, the statistical moments of any order can be obtained exactly for the system response. It is shown that the response of a linear system may be strongly non‐Gaussian, especially if the bounded random excitation deviates far from being Gaussian distributed, and/or is narrowly banded in frequency. Based on this discovery, a procedure is proposed to improve the well‐known equivalent linearization procedure for the case of non‐linear systems under bounded random excitations. In the procedure, the exact statistical moments obtained from the replacing linear system are used to calculate the linearization coefficients; thus, the usual Gaussian assumption is not required. It is shown in numerical examples that the accuracy of the calculated system response can be improved substantially, especially when the system is strongly non‐linear and/or the excitation process is far from being Gaussian distributed. Copyright © 2005 John Wiley & Sons, Ltd.