Structural damage localization and quantification in cantilever beam structure using modal strain parameters and artificial neural network

Abstract. In this study, we proposed a damage identification method in two stages. In the first stage, Strain modal analysis is carried out using the finite element method on the cantilever beam model, and the first four-strain mode shapes are obtained for damaged and undamaged states of the cantilever beam along the length of its middle line. Then, the absolute difference of strain mode shapes and modal strain energy-based damage index are used to determine the damage position along the length of the cantilever beam. For this purpose, several damages are inserted in the finite element model of the beam at different positions to evaluate the performance of this damage indicators. The results obtained showed that the proposed indicators could determine the damage position at any location along the length of the beam. In the next stage, the values of the damage indicators at damaged sites of the beam are used as input parameters to train the artificial neural networks incorporated to predict the damage severity. The results of this study illustrated the accuracy and efficiency of the proposed two-stage damage detection methods for predicting the location and magnitude of damage in a cantilever beam.