Identification of a spatial linear model based on earthquake-induced data and genetic algorithm with parallel selection

Abstract

This paper deals with the parametric identification of a small-scale bridge model that was intended to approximately reproduce the transversal dynamic behaviour of its corresponding prototype. This is representative of a typical multi-span continuous-deck irregular bridge. A linear model with viscous damping is proposed to reproduce its dynamic response. The identification is carried out in the time domain using experimental earthquake-induced data and assuming the mass is known. An identification procedure including a genetic algorithm with parallel selection has been developed. In the studied case, the procedure has demonstrated to be robust. It is shown that the model could be improved by using a non linear approach for the dissipative forces.

Keywords

• model identification
• linear model
• bridge
• genetic algorithms
• parallel selection
• signal processing