چكيده به لاتين
Various damages in buildings, bridges, oil rigs and in general all kinds of structures during their lifetime are inevitable. So far, many examples of various types of damage have been recorded in various engineering structures, which have resulted in many human and financial losses. Early detection of damage and action to eliminate its defects increase the useful life of the structure. Therefore, in recent decades, a lot of research has been done to identify damage to structures. Since 1960, the design and construction of steel arch bridges by engineers has increased. Steel arch bridges are a good way to create communication paths that have received a lot of attention in recent decades due to their rapid construction (compared to multi-span straight bridges that also require a large number of columns), but a correct understanding of Their behavior is still needed. The presence of a horizontal arch in the bridge deck, due to the creation of centrifugal forces, will change the interaction of the structure and the moving load, which will greatly affect the vibrational response of the structure used in health monitoring methods. Was. Also, the presence of horizontal or vertical arches in bridges will have a strong effect on the conditions of its support, which can affect the response of the structure to the loads and, as a result, common methods in monitoriIn this study, the Nonlinear Auto-Regressive with eXogenous input is used. The NARX is a nonlinear model, with different types of nonlinearity functions such as wavelet, Artificial Neural Network, tree partition etc., able to predict time series. In a group of sensors which are set on a structure, each sensor as a main sensor and it’s neighbor is assumed as a cluster. An NARX model is created for each cluster. The accelerations of each cluster due to free vibration test is used as the inputs of each model. Each model predicts the acceleration of it’s main sensor. The Fit Ratio between predicted and measured acceleration is used as Damage Factor NARX methodology was tested on a 5 DOF numerical model. Furthermore, this approach was used for damage assessment on a concrete bridge deck. Different damage cases, single and multiple, was considered on the deck. The results show that the damages can be detected, located and quantified with reasonable accuracyng the health of bridges.
