Master thesis and internship[BR]- Master's thesis : Identification of tensions in cross-cables network[BR]- Integration Internship

Abstract

Cable tension monitoring is essential to ensure the continuous safety of cable structures throughout their entire lifetime. The present paper aims at developing a frequency-based identification procedure of the axial forces for a network of two cross-cables with non-negligible bending stiffness anchored to possible flexible supports. Physical models relying on the in-plane and out-of-plane dynamic behavior are presented through a Finite Element Method. The previous results are then compared to the outcomes of an analytical approach for validation purposes. To acquire a first set of observed eigenfrequencies, experimental measurements performed on a reduced scale model of the structural network are carried out. The latter shows during the inverse approach a certain limitation, namely that the bending stiffness of the cable is too small to be accurately identified. However, these measurements allow the validation of the inverse identification approach. To perform these identifications, the use of a non-linear Bayesian regression turned out to be a powerful tool to get overall ranges of estimated parameter values with their own probability to be observed. As second method, the function \texttt{fminsearch} from the optimization toolbox of \texttt{Matlab} proves to be effective in the identification strategy to obtain a single optimal value of the unknown parameters. The use of the Laplace approximation allows to complete this last method by providing a Gaussian PDF around the optimal value. Finally, the validation of the identification method is achieved through the experimental natural frequencies acquired on one selected cable network on the Milsaucy bridge in Liège, leading among other things to the consideration of flexible anchor supports.