The ultrasonic NDT of adherent area between metal and nonmetal surfaces and the prediction of its destruction strength

A new approach so-called multi-reflection scanning method by transverse ultrasonic wave is described for the NDT of adherent between metal and nonmetal surfaces. According to the adherent state, the destruction strength can be predicted by Bayesian inference. A skew probe scanning on the metal side is used to transform the longitudinal wave, dispatched by a PZT transducer, after entering the metal layer into transversal. Then the latter propagates in a zigzag path along the adhesive layer. The reflected wave strength are detected and integrated with respect to time. The relation between the adherent area along the scanned path and the time integral of reflected wave strength is found. The prediction model by Bayesian inference has been studied. As the adhesive process is fixed, the relation between the destruction strength of a part versus the measured adherent area proportion and their distribution can be established. Practice shows that, even the rate of spoilage as high as 60%, the software can strictly reject the spoils and maintain the probability of false rejection not exceed 7%.<>