Ultrasonic Detection And Pattern Recognition For Bonding Quality Of Multi-interface Adhesive Materials

With a more widespread use of adhesive materials, testing and evaluation of bonding quality is emerging as a hot topic in the field of nondestructive test. In recent years, ultrasonic detection has been used in bonding quality test of multi-interface adhesive material incrementally. But the use of this detection technique is restricted because feature extraction of echo signals in different bonding conditions is particularly difficult which is due to influence by the superposition of echo signals from different interfaces and noise jamming.This thesis took use of steel/rubber/phenolic resin adhesive structure as tested model object, to simulating the solid rocket motor vessel structure. Based on the theory of pulse reflection method, ultrasonic inspection system is designed for tested objects, which include drive pulse generate module with NE555 and 74LS221 as the core and data acquisition module with EP4CE6F17C8N FPGA as the core. The parameters of this system are optimized by experiments, in order to realize the detection of bonding quality of multi-interface adhesive composites efficiently and the acquisition of echo signals.In consideration of the weak echo signals from deep interface which because of easily overwhelmed by the strong signals reflected from the surface, the thesis utilized adaptive filter to realize interface separation of echo signals, and wavelet transform to enhance the features of characteristic signals and extract the features from original signals. Finally, defects recognition and classification of multi-interface adhesive materials were implemented through BP neural network.