BRIDGE-MODEL-DAMAGE DETECTION BASED ON THE ENERGY METHOD AND NEURO-FUZZY INFERENCE SYSTEM

Abstract

This paper presents a bridge-model-damage detection method based on neuro-fuzzy networks and characteristics of displacement-distributing state of vibrating beam, which is used for the bridge model. The structure of neuro-fuzzy inference system named ANFIS of [1] is used. A two-step method is proposed in order to detect of damage. In the first step, damaged locations are identified based on the energy-based damage index of [7]. In the second step, the magnitude of damage is determined based on using the ANFIS. Series of experiments are executed, in which the stiffness of the structure is reduced in one location or more, and then, the structure is roused to be vibration to calculate damage indices. The damage indices are used to detect, locate, and quantify the structure damage. The results show that efficiency of the proposed approach is quite independent of vibration frequencies of the beam. In addition, this method is quite effective in determination the locations and magnitude of damage in the case of the measurement errors are small enough.