Vector Triggering Random Decrement for High Identification Accuracy

Abstract

Using the Random Decrement (RD) technique to obtain free response estimates and combining this with time domain modal identification methods to obtain the poles and the mode shapes is acknowledged as a fast and accurate way of analysing measured responses of structures subject to ambient loads. When commonly accepted triggering conditions are used however, the user is restricted to use a combination of auto RD and cross RD functions with high noise contents on the cross RD functions. Use of the auto RD functions alone causes the loss of phase information and thus the possibility of estimating mode shapes. In this paper a new algorithm based on pure auto triggering is suggested. Equivalent auto RD functions are estimated for all channels to obtain functions with a minimum of noise, using a vector triggering condition that preserves phase information, and thus, allows for estimation of both poles and mode shapes. The proposed technique (VRD) is compared with the traditional RD technique by evaluating modal parameters extracted from the RD and the VRD functions using ITD identification technique on simulated and experimentally obtained data.