Based On Ultrasonic Guided Wave To Inversion Of Elastic Constants Of Anisotropic Plates

Anisotropy, also known as "heterogeneous".Mean all or part of the physical,chemical and other properties of objects with the direction of the different andshow some of the differences of characteristics. Anisotropic materials with many uniqueadvantages, such as the strength and the stiffness ratio of high, good fatigue resistanceand vibration resistance and design of structures, which are more and more widely inmany fields of application. In turn, increasingly high demands on performanceimprovement, this requires more advanced detection methods to detect thecharacteristics of the material.Advanced detection methods of the mechanical properties of composite materialsusually use the structural dynamic response(such as displacement response, vibrationfrequency and wave velocity, etc)and the complex relationship between materialproperties When through a mathematical model to represent this relationship, called theforward problem. This forward problem can be resolved by analytical or numericalmethods. Then, if has a series of precise measurement of structural response data,combined with a large number of forward calculation, then the composite materialproperties can be achieved by properly build inversion model identification. Far morepromising intelligent inversion algorithms with neural network (neural network: NN),genetic algorithms (genetic algorithm: GA) and fuzzy logic (fuzzy logic: FL)algorithm,etc.This paper uses the Legendre polynomial series method was used to study guidedwave characteristics of anisotropic structure, obtained the structure of the dispersioncurve, solves the problem to calculate before. Then based on the structure of multimodal dispersion of guided wave group velocity, respectively, by using the neuralnetwork algorithm and genetic algorithm for two kinds of algorithm inversion model isestablished, through numerical simulation and inversion of the orthotropic plate and ananisotropic plate elastic constants, and summarizes the advantages and disadvantages ofthe two methods.The neural network inversion model area,the first ultrasonic signal to thetheoretical calculation pretreatment extract the signal feature information (this paper extracted at different frequencies of several low-level guided-wave mode the groupvelocity),to obtain a neural network training samples, and then design suitable ofnetwork model, the training sample into network for training and learning, to determinethe network parameters, finally the simulation waveforms as input, into the alreadytrained network, inverse material elastic constants. For the guide wave and thecombination of the genetic algorithms, the extracted remains the same in differentfrequency points prior to the theoretical calculation of the first few lower-order guidedwave group velocity, and then to design appropriate procedures, in elastic constants ofanisotropic plates within a range of possible according to the certain rule graduallysearch, optimization of a certain error function, until the error function to obtain theoptimal solution, then when the error function to obtain the optimal solutioncorresponding to the elastic constants for the anisotropic plate with elastic constants.