60 Application of Expert Group Neural Networks to Structural Health Diagnosis of Mau-Lo Creek Cable-Stayed Bridge

The Mau-Lo Creek cable-stayed bridge, located in central Taiwan, is mainly composed of a particular steel pylon with the parabolic arch shape and twin steel girders on the plane of the Clothoid curve. Since the bridge experiences a complicated structural characteristic, the structural health diagnosis is indeed necessary. This study proposes a structural health diagnosis method for cable-stayed bridge using Expert Group Neural Networks (EGNN) and field measurement data. The EGNN were used to analyze reversely the cor-responding axial forces of all the stayed cables using three sets of rotation measured of the pylon with easier and more efficient while solving this kind of inverse problem. Based on the cable force evaluated, the structural behavior including the deformation and stress state of the bridge can be traced successfully. The proposed optimization procedure is used to determine the appropriate axial force combination within the thirty-six stayed cables of Mau-Lo creek cable-stayed bridge finally.