DAMAGE DETECTION IN ITSMART BEAM THROUGH ITS VIBRATORY RESPONSE I

Abstract

The life and reliability of structures can be affected by structural ageing, environmental circumstances and reuse. This may be due to structural damage such as cracks. Cracks present a serious threat to the designed performance of structures. In recent years, vibrational investigation of damaged structures has proved to be a feasible approach for damage detection in Non Destructive Evaluation Category. This dissertation is aimed to present analytical technique for detection of damage and its severity. Here the approaches used are fracture mechanics based energy approach and artificial neural network (ANN) approach. The approach for finding the existence of crack in a beam, utilizes the changes in the value of first three natural frequencies. Using free vibrational analysis and applying Finite Element Method, some natural frequencies and mode shapes of the uncracked and cracked beam has been evaluated. These frequencies are compared with experimentally obtained values. In this dissertation, the artificial neural network approach has been applied to find the location and size of crack. The analytically computed modal frequency parameters for various crack locations and depths using a fracture mechanics based crack model, are used as input in the artificial neural network. These inputs are used to train a neural network to identify both the crack location and depth. A comparative study is made using the modular_ neural network architecture with two widely used neural networks, namely the multi-layer perception network and the single layer network. It is concluded that modular neural network architecture can be used, as a non-destructive procedure for health monitoring of structures