STATISTICAL ANALYSIS OF THE KARHUNEN-LOEVE RANDOM PROCESS MODEL

摘要

Practical structural dynamic phenomena involve excitations and responses that are random processes. Nonstationary random processes are most frequently encountered, and an accurate and efficient model for them is the Karhunen-Loeve (KL) expansion. The KL expansion can be obtained using experimentally measured random process realizations, but if it is, there is some level of statistical error associated with the identified parameters of the model. This paper shows how bootstrap techniques can be used to perform confidence analysis on the parameters of a KL model. Laboratory-measured data are used to demonstrate use of the model and statistical analysis of the model parameters.