Research On Structural Damage Identification Method By The Wavelet-Neural Network Of Modal Pramaters

In the process of using, structures will be damaged differently for various reasons. The structural stiffness and bearing capacity will decline when these damages are accumulated to a certain degree. This will influence the usability and durability of the whole structure, and in severe cases may lead to catastrophic accidents, causing the significant loss of life and property. Therefore, how to quickly and efficiently identify the location and the degree of damages has become an important research subject in the research field of engineering structural damage diagnoses.As a time-frequency signal processing method, wavelet analysis can represent partial characteristics of signals in time and frequency domain. The neural network algorithm possesses both a highly nonlinear mapping ability and the capacity of self-organizing, self-learning and self-adaptation in processing signals. Combining with the advantages of both, in this paper, based on the combination of wavelet analysis and neural network theory, with the help of the wavelet coefficients figure, the position of damages can be detected. based on the wavelet modulus maximum from wavelet analysis, using neural network to identify the structure damage degree.Therefore, By combination of the two methods about wavelet analysis and neural network, we can effective identify the position and degree of structure’s damage.The thesis regards simply supported beams with damages as the object of research and establishes the identification method based on vibratory modes, rotation modes and curvature modes. Through the method, damage locations of breams with one place or multiple places of damage can be effectively detected. And the thesis also analyzes the results of damage identification under the three modes. Modal parameters of beams generate the coefficient figure of wavelets with the usage of wavelet transformation. The damage degree of structures can be identified by the non-linear relationship between wavelet modulus maximums and the damage degree which is simulated with neural networks. According to the numerical simulation, it shows that the damage location and degree of structures can be effectively identified with the help of wavelet analyses and neural networks.The thesis regards continuous beams with damages as the object of research and establishes the finite element model of beams with one place, two place and multiple places of damage. Based on vibratory modes, rotation modes and curvature modes, the work situation of all damages is dealt with wavelet transformation. The damage location of structures can be detected by the coefficient figure of wavelets. Then the non-linear relationship between wavelet modulus maximums and the damage degree is simulated with neural networks of which results can identify the degree of damages. The analysis shows that the damage location and degree of structures can be effectively identified with the help of wavelet analyses and neural networks. Therefore, the method in this thesis has a profound guiding significance for structural damage diagnoses.