Damage detection in structural elements with the help of vibration data using artificial neural networks and genetic algorithm optimization

In this study, the effectiveness and potential of Artificial Neural Networks (ANN) and Genetic Algorithm (GA) optimization techniques in damage detection applications based on vibration data were evaluated specifically for beam structures. Damage in beams was defined by reducing the stiffness (modulus of elasticity) in the relevant elements. In the first part of the study, a cantilever beam model was created with the help of ANSYS Workbench to evaluate the effectiveness of ANN in damage detection; 1000 random damage cases were simulated with the help of Design Experiments tool. This obtained data was used in ANN training. Performance evaluation was performed on 10 different damage scenarios using the trained network. In the second part of the study, damage detection was performed with GA optimization. For this purpose, a stiffness loss set that minimizes an objective function considering the frequency and mode shape changes of the two states of the structure was investigated. To compare the limited data obtained with the one obtained from the finite element model, a model reduction process was applied to the numerical model with the Guyan method. The performance of ANN and GA was evaluated and compared with each other.