Structural Damage Detection and Identification Using Learning Vector Quantization Neural Network

Abstract

Abstract This research deals with the structural damage detection by experimental measured modal parameters, such as the modal frequencies and the modal shapes. Changes of local structural parameters, induced by damage, will affect the local stiffness and cause the change of modal frequencies and modal shapes of structure. Use of these observable values to detect the damage of the structure is feasible and implement. Learning Vector Quantization (LVQ) Neural Network based on pattern classifier is used to detect the location of damage, and a method of releasing the dense of input vector to neural network is proposed to increase the accuracy of detection. Several numerical examples show the proposed method is effective to increase the rate of damage detection. Finally, a practical application example of damage detection for a turbine blade is used to demonstrate and verify the approach developed.