Modeling of traffic excitation for system identification of bridge structures

In long-term health monitoring of bridge structures, system identification is often performed based only on the system output (bridge vibration responses) because the system input (traffic excitation) is difficult to measure. To facilitate the identification of the bridge properties, traffic excitation is commonly modeled as spatially uncorrelated white noise. A physical model of a stationary stream of vehicles (moving loads) arriving in accordance with a Poisson process, traversing an elastic beam, shows that the traffic excitation is spatially correlated. Employing the dynamic nodal loading approach, this spatial correlation results in a frequency-dependent excitation spectrum density matrix, and shifts the response spectra obtained from those excited by spatially uncorrelated white noise. It is shown that the application of system identification techniques based on the conventional excitation model may result in misleading structural properties. Hence, this study further proposes an output-only gray-box identification technique for bridge structures, in which knowledge about the nature of the traffic excitation, such as its spatial correlation, is implanted into an autoregressive-moving-average (ARMA) model. The identifiability of the ARMA model so constructed is assured and the feasibility of the proposed identification technique is demonstrated by a numerical example.