SH-Wave Scattering By An Cylindrical Interface-Inclusion With An Inclination Crack Via The Center And The Inverse Problem

As for the theory of elastic wave, issues are generally focused on the discussion about physical model's stress and displacement responses by the given incidence waves or point source. With the help of time-domain Fourier integral, common responses could be obtained, and problems in some fields, such as explosion and shock, underwater acoustic engineering, and earthquake engineering, are also resoluble. As it is ill-posed, the so-called inverse problem, that is, conditionally analyzing parameters of physical model or load starting from known response, has been the focus of discussion for years with relatively less researching work. In areas of non-destructive inspection, objective identification, machinery monitoring, seismic survey, and so on, solutions to inverse problems of elastic wave have extraordinary broad application prospects.To demonstrate its strategy, this thesis presents the issue of SH-wave scattering by elastic cylindrical interface-inclusion that contains a center-via acclivitous radial crack. Solution of dynamic stress intensity factor (DSIF) of crack tip is discussed firstly. The interface with cylindrical inclusion and the center-via acclivitous radial crack are formatted step by step, using the thought of "corresponding". A group of integral equations that contain a series of unknown forces are obtained by expressing displacement continuity in forms of stress and Green function's integral. According to the definition of crack tip's DSIF in fracture mechanics, this thesis put out the transform expression of the unknown forces and dimensionless DSIF, and then it presents the group of integral equations that contain dimensionless DSIF. The group of integral equations is transited into a series of linear algebraic equations, discretization-based-on, to solve out the dimensionless DSIF numerically. Factors that effect veracity are analyzed while programming with the thought of modularization in Matlab. Error-controlling measures corresponding are discussed detailedly and as for numerical parameters, formulas that base on experience are formulated.On basis of Matlab's colorful GUI design and artificial neural network (ANN) functions, this thesis completes the design of Crack Parameters Detection on ANN (CPDA). After computing on a large scale, data that are relatively veracious are sought out to format configuration-reasonable and quantity-adequate sample data. Generalized regression neural network (GRNN) is trained with study samples, and then it is evaluated with examining samples, with the purpose to determine the initial parameter for the network. The jarless GRNN then works on inputs and presents simulation outputs: size and acclivitous angle of the crack. Practice indicates that both precision and consuming time is satisfying, means size and acclivitous angle being given out effectively. The strategy based on ANN to the so-called inverse problems is feasible.