Study On The Key Technologies For Acoustic Impedance Reconstruction Of Material With High Accuracy

In this dissertation, for the purpose of meeting the needs of UNDT (Ultrasonic Nondestructive Testing)&UNDE(Ultrasonic Nondestructive Evaluation), several key technologies such as increasing the SNR(Signal-Noise Ratio), eliminating the multiple reflections and improving the resolution are studied to increase the accuracy and reliability of acoustic impedance reconstruction, by means of digital signal processing and intelligent information processing technologies. The details are presented as below:Chapter 1, a survey of important role of UNDT techniques is given and the review of their current research status is summarized. Then the significances of acoustic impedance reconstruction used in UNDE are summed up, at the same time, the key technologies for acoustic impedance reconstruction is discussed and the research direction of this dissertation is pointed out also.Chapter 2, the model of acoustic impedance distributing of multi-layer material is established, meanwhile, some theories including digital signal processing, intelligent information processing which are related nearly to the acoustic impedance reconstruction are introduced.Chapter 3, some adverse influences of various noises on the acoustic impedance reconstruction are analyzed, the formation mechanism of structural noises is discussed, and the limitations of SSP(Split Spectrum Processing) which is often employed to suppress this kind of structural noises are indicated. Then, a new approach to eliminate structural noises based on RBF(Radius Basis Function) artificial neural network and wavelet analysis is presented, afterwards, experimentation with real situation is used to investigate this approach.Chapter 4, the formation mechanism and the effects on acoustic impedance reconstruction of multiple reflections are illustrated, besides, the efficacy of PRF(Pulse Repetition Frequency) for eliminating multiple reflections is testified. And then, a new method based on wavelet analysis is introduced to improve the PRF technology for high accuracy and reliability, experiment results show the advantages of the presented method.Chapter 5, the performance of some deconvolution technologies in common use is analyzed and compared, the deficiencies of these technologies are listed also. Subsequently, an innovative algorithm based on CNN(Chaotic Neural Network) and GA(Genetic Algorithm) is developed to improve the resolution performance of deconvoluation technology when it's applied to acoustic impedance reconstruction. Ultimately, the parameter selection guidelines of this new algorithm is given, computer simulation and practical experiment results indicate that the new methodowns a good many advantages such as more effective and more reliable.Chapter 6, a embedded portable system of the acoustic impedance reconstruction is planned generally, and a design scheme is produced, which takes the ARM architecture chip-AT91M55800A as the main control unit, the DSP chip-TMS32F2812 as the signal processing unit, and the two units are combined with Dual-RAM in order to achieve high data Communication speed. The design scheme can ensure that the embedded portable system has high performance of real-time and reliability.Chapter 7, as the last chapter, the important results and main conclusions of this dissertation are summarized and the prospects for the application of digital signal processing and intelligent information processing in the acoustic impedance reconstruction are opened up.