Damage Detection Of Spatial Truss Structure Based On Wavelet Analysis

Large construction structure would be damaged because of corrosion, collision as well as environmental loads during service period. With the increase of structure service period, the damage is accumulated gradually, which will result in the attenuation of structure resistance and the reduction of security, serviceability and durability of structures. Furthermore, it can cause the failure of the structure when the damage accumulates to a certain degree. Therefore, it is valuable in practice to develop the structure damage identification technology, especially to diagnose the damage state of structure and to appraise the structure security without affecting the use of the structure. Based on this, the application of wavelet analysis in the damage detection of truss structure is studied in this paper with regard to the state-of-the-art of the damage detection techniques. The main contents are as following:(1) The primary application for the structure health monitoring is implemented based on wavelet analysis. The spatial truss structure is often simplified as a spatial bar structure in research because the components of spatial truss structure are considered to bear the axial force. When the damage appears in the structure, the structure stiffness will change and consequently the structure dynamical response signal obtained from online monitoring will be discontinued. Traditional Fourier transformation is only able to determine the entire property of function irregularity, but it fails to determine the spatial position and the distribution of the singular point. As a consequence,the vibrating signal is decomposed based on wavelet analysis,and the variance of its details is calculated,which is used to analyze the irregularity of signal and detect the exact time when the structure damage appears. The numerical example of roof spatial truss structure of Shenzhen Citizen Center shows that the proposed method is effective to detect structure damage.(2) The research on the wavelet packet in the damage identification of spatial truss structure is applied. In view of the requirement of the structure damage diagnosis, firstly the vibrating signal is decomposed to each frequency band based on wave packet analysis, then the wavelet packet (WP) signal component node energy of various frequency bands will be acted as characteristic vector to detect damage. The numerical result of a two-layer plate spatial truss indicates that wavelet packet analysis has the strong anti-noise jamming ability and can identify the structure damage effectively.(3) The two-step structure damage identification method based on wavelet analysis and support vector machines is proposed. Firstly the seismic response signal of the damaged structure is decomposed based on wavelet analysis, the variance of its detail is calculated, the irregularity of signal is analyzed and the exact time when the structure damage occurs is estimated. Secondly, the damage signal is decomposed based on wavelet packet analysis, extracting the WP signal component node energy containing structure damage information, which is selected as the sample of SVM network, the proper nuclear function is chosen to establish support vector machines (SVM) model to carry on damage localization of the structure. The example of a two layer plate truss structure shows that this method can distinguish the structural damage effectively.(4) The research on finite element (FE) model updating of spatial truss structure based on wavelet packet analysis is developed. Firstly, the element stiffness matrix expressed by the semi-rigid node coefficient with fixity factor is established. Secondly, the sensors' positions are optimized based on the effective independent method proposed by Kammer, the structure acceleration response signal is computed and decomposed by wavelet packet and the support vector machines SVM model after extracting the WP signal component node energy is established to act as the training sample of support vector machines. Finally the measured signal is decomposed and extracted by the WP signal component node energy acted as the test sample, which will carry on the support vector machines SVM model to predict and distinguish fixity factor of the structure node. Consequently, the total stiffness matrix is obtained by the coordinate transformation of the revised element stiffness matrix, that is, the FE model of the structure is updated. The example result has confirmed the validity of this method.(5) The application study for damage identification based on wavelet packet analysis and statistical process control is implemented. The method is developed to monitor the manufacturing process so as to reduce variation and build quality into the product at first. Then the vibrating signal is decomposed into each frequency band based on wavelet packet analysis and the WP coefficient node energy acts as characteristic vector of damage identification to establish damage indicator. Finally, structure damage critical point is established for damage pre-warning of the structure. The results of numerical example of spatial truss structure indicate that this method can be used to identify the damage effectively.(6) In order to confirm the application of the above method based on wavelet packet and the statistics process control in the actual structure, the experimental study is conducted to a simplified transmission tower model, the experimental results indicate that this method can be used to detect the structure damage effectively.This dissertation researches the damage identification of spatial truss structure based on wavelet theory by using the newest theory and draws some theoretical and technological conclusions for the online structure damage identification.