Nondestructive Evaluation of Prestressed Concrete Beams using an Artificial Neural Network (ANN) Approach

An artificial neural network (ANN) based approach for the assessment of damage in prestressed concrete (PSC) beams using its present stiffness and natural frequency as the test inputs to the ANN has been proposed. The details of the extensive experimental programme designed and executed in this study to induce the known extent of damage in the PSC beams by a method that resembles natural damage processing techniques and to generate the training and test data for the ANN used to model damage levels have been presented. It has been demonstrated that it is possible to assess the damage with reasonable accuracy by the ANN learning by a back propagation algorithm and stiffness and natural frequency as test inputs. The efficiency of this damage assessment algorithm has been studied by testing this ANN with the test data available in the literature. The results indicate that this approach can be used as a cost effective and simple structural health monitoring tool for PSC beams since this procedure needs only limited nondestructive static and dynamic measurements on the structure under study.