Estimation of Cable Tension with Unknown Parameters Using Artificial Neural Networks

Abstract

For existing stay cables, bending stiffness and boundary conditions at cable ends often differ from those of the design owing to the degradation of cable materials and complexity at the cable supports. This paper presents a method for vibration-based cable tension estimation using artificial neural networks (ANNs) regardless of the uncertainties of cable boundary conditions and unknown cable bending stiffness. Finite difference formulation of a discretized cable with rotational restraint ends is developed to generate datasets for training, validation, and testing in ANNs. The proposed method was applied to identify tensions in cables of an existing bridge as a case study. Results showed that the suggested that suggested methodology is highly capable of identifying cable tension with unknown cable bending stiffness and uncertain boundary conditions.KeywordsANNsCable tensionFinite difference modelRotational restraintsUnknown parameters