Study On The Intelligent Processing Technologies Of Testing Acoustics Signal

In this dissertation, for the purpose of meeting the needs of ultrasonic nondestructive testing (UNDT) and ultrasonic nondestructive evaluation (UNDE), several key technologies such as increasing the signal to noise ratio (SNR), improving the time resolution and identifying the defect type are studied to increase the accuracy and reliability of UNDT 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 nondestructive testing (NDT) techniques in modern industry is given and the review of their current research status is summarized. Then the key technologies for improving the accuracy and reliability of UNDT are discussed and the research direction of this dissertation is pointed out, too.Chapter 2, the model of ultrasonic testing signal transmission is established, meanwhile, some theories including digital signal processing and intelligent information processing which are related nearly to increase the accuracy and reliability of UNDT are introduced. So work foundation of this dissertation is supplied.Chapter 3, some adverse influence of various noises on UNDT are analyzed, the formation mechanism of structural noises is discussed, and the limitations of classical split spectrum processing (SSP) which is often used to suppress this kind of structural noises are indicated. Then, a new approach to eliminate structural noises based on support vector machine (SVM) pattern recognition theory and wavelet transform is presented. Afterwards, experiments with real situation are used to investigate this approach.Chapter 4, to improve the time resolution of ultrasonic echo signals in UNDT, a new approach of deconvolution technique based on wavelet transform and particle swarm optimization (PSO) algorithm is proposed after the shortcomings of conventional deconvolution techniques are analyzed. Simultaneously, the parameter selection guidelines of PSO algorithm is given, computer simulation and practical experiment results indicate that the new method owns a good many advantages such as more effective and more reliable.Chapter 5, a new approach of defect identification based on complex wavelet transform and SVM pattern recognition theory is presented to establish technical foundation for further QNDE. What's more, its validity and feasibility are proved by experiment.Chapter 6, as the last chapter, the important results and main conclusions of this dissertation are summarized and the prospect for further research in increasing the accuracy and reliability of UNDT are put forward.