ARMAV Techniques for Traffic Excited Bridges

Abstract

In this paper ARMAV (Auto Regressive Moving Average Vector) models are used for system identification and modal analysis purposes. This time domain technique allows to estimate a discrete time system response function without performing any domain change (i.e. it doesn’t use FFT and IFFT to evaluate the model parameters) and without applying any time window (also when sampled data are non periodic): this leads to well-estimated system parameters, also for short data records. These models are useful to perform system identification for multiple input-output cases also when the excitation is just statistically known. The present analysis is dedicated to a scaled bridge, designed according to the theory of models, whose static and dynamic characteristics are compatible to those of real bridges. The aim of the tests is to collect a series of supervised measurements in a controlled environment, with statistically defined traffic conditions; the comparison of the model results with those acquired on the real bridge is the compulsory step towards a correct modelling of bridges for their identification and monitoring. The paper reports encouraging results obtained with experimental simulations on the model.