Structural Damage Identification Based on AR Model and PSO-SVM

In this paper, we put forward a structural damage identification method based on autoregressive (AR) model and particle swarm optimization support vector machine (PSO-SVM). Firstly, AR model is established for acceleration response signal, and the model coefficients are extracted. Secondly, the difference of AR model coefficients under damaged and undamaged conditions is selected as the damage feature. Finally, PSO-SVM is used to distinguish the damage location of a structure. The validity and noise resistance of the method are verified by numerical simulation and analysis results of a five-floor steel offshore platform.