Condition Monitoring of Structures by Using Ambient Dynamic Responses

Abstract

The thesis presents the general procedure for structural condition monitoring of civil structures by using ambient dynamic responses. The proposed monitoring system is composed of three components: the data acquisition system, the damage identification algorithm, and the damage verification process. The thesis explores the use of distributed fiber optic sensors in sensing the ambient responses of the structures. The new damage detection algorithm is developed. The algorithm is composed of three components: the genetic algorithm (GA), the structural analysis toolbox, and the wavelet de-noising toolbox. The Dynamic Neighborhood Method (DNM) in genetic algorithm is developed to solve for the multiple solutions that arises from the spatial incompleteness in measurements and the redundancy of the structures. DNM increases the ability of GA in searching for the global optima. An approximate dynamic analysis with pseudo-force is used to reduce the computational time in fitness evaluation process. Noise in measurements is effectively handled by treating the de-noising parameters as the additional unknowns. The algorithm not only identifies the current conditions of the structures but also estimates the level of noise in the measurements. The proposed structural condition monitoring system is applied to detect the existence and extent of small fatigue cracks in a simulated railway bridge.