The Application And Research Of Ultrasonic Non-destructive Testing Technology

This paper first presents the current research situation and development tendency of ultrasonicnondestructive testing technique at home and abroad, furthermore, expound the fundamental theoryof ultrasonic nondestructive testing in details. Then on the basis of A-mode pulse Ultrasonicdetecting theory, the digital ultrasonic detection system hardware solution is constructed centeringon DSP and CPLD. Then, in accordance with various modules functional requirements, the relevanthardware circuits such as ultrasonic transmit, programmable gain amplifier, analog band-pass filter,envelope detector, high speed data acquisition, processing and man-machine interaction aredesigned and system software is programmed. At last, the debug of system hardware and software iscompleted. The test results show that this system design is correct and feasible.The flaw signals detected by this Ultrasonic detection system are often mixed up with a largequantity of noise, affecting the correct judgment. While the conventional digital filters cannot filterout the noise effectively. For this reason, this paper analyzes the multi-wavelet de-noising;furthermore, improve matrix preprocessing method for GHM multi-wavelet and soft-thresholdfunction. The experimental results indicate that multi-wavelet de-noising can obtain a good effect.At last, according to the designing ideas of virtual instruments, an analysis system of ultrasonicecho signal is designed. It has realized the functions of Ultrasonic detection signal de-noising, thewaveforms displaying, de-noised data saving and spectrum analysis etc.As the system adopted A-mode pulse reflecting type ultrasonic testing method cannot give aqualitative recognition to the defects, this paper discusses the artificial neural network in theapplication of ultrasonic flaw qualitative, according to fine time-frequency analysis ability ofwavelet packet and good classification ability of BP neural network, putting forward defectqualitative identification method based on the wavelet packet energy spectrum characteristics andBP neural network, verifying the effectiveness of the method through the experimental results torealize the automatic qualitative identification to the defects.