Localization of damage in beam-like structures by using support vector machine

Support vector machine (SVM) is a machine learning algorithm based on statistical learning theory, and it has recently been established as a powerful tool for classification and regression problems. This paper presents a novel SVM-based approach for damage location identification in beam-like structures. The curvature mode shapes are used as inputs of the SVM. The proposed approach involves two steps. The first step uses support vector classification to obtain the structural damage probability distribution. The second step uses support vector regression to identify the precise damage location after reconstructing the training set. Furthermore, a series of simulations in the cantilever beams involving different damage scenarios (at different location and different extent), have been conducted to verify this method. In order to check the robustness of the input used in the analysis and to simulate the experimental uncertainties, artificial random noise has been generated numerically and added to noise-free data during the training of the SVM. The results show that this approach is a promising method for damage diagnosis.