Application Of Wavelet Analysis In Damage Identification For Beam/Arch Structures

Structures may damage due to various loading, fatigue and various accidents in their lives. The accumulated damage may give rise to the failure of structures and lead to the loss of property and life. To guarantee the safety of structure and to minimize the maintenance cost, it is desired to find an effective and efficient method to identify the damage occurrence, location and extent of damage.Defects in the structure will cause some changes in the dynamic character, such as frequency, mode shape and damp, which can be used to indicate the state of the structure. So a lot of researchers devote their attention to detect the damage in the structure and develop many methods based on the dynamic character of structure. Wavelet analysis, as a new tool of signal proceeding, is capable of presenting the local character of the signal and enables to identify the damage in the structure.In this paper, the wavelet analysis combined with structural response characteristic based vibration is used to detect damage. The question of how to use wavelet analysis to extract damage signature and how to minimize the noise are considered.The research works in this dissertation mainly include:First, a general view of the latest development in the fields of bridge, civil engineering, rotary machine and aero instrument is given. Also the new crack detection techniques based on dynamic characteristics are summarized.Secondly, frequency and mode shape are used as inspection parameters to detect cracks. When crack occurs in the beam and arch, frequency and mode shape are studied by numerical simulation, to predict the existence of a crack, the location of the crack and even the damage extent. Frequency can be measured easily and with high precision. The breakage of beam in the structure can be detected by frequency change. But frequency can not be utilized to predict an early crack because the frequency shift is too low.Parameters related to mode shape, such as mode vector, mode curvature and mode are much sensitive to early cracks. Mode curvature is the most sensitive one, but mode itself is not sensitive to early crack at all. Criteria based on mode shapes can be employed as local crack detection method to diagnose the damage and localize it.Thirdly, a new method to analyze the signal by wavelet transform is presented. This method can efficiently extract the feature of damage and can precisely predict crack. Mode signals can be analyzed by continuous wavelet transformed to find a crack at an extent of 5% in an arch, and also the crack can be located. Multi-resolution analysis (MRA) is illuminated and applied to mode signal in damage detection.Fourthly, the multi-damages identification in the structure is discussed. The damaged location in the mode node can not be identified, and so it should be integrated identified from the different order modes; the multiplication of wavelet coefficients in different scales can eliminate the effect of noise, and further detect the damage.