Enhanced Ultrasonic Characterization of Assemblies,TLL{_}9

Thomas, G. ; Chinn, D.

Lawrence Livermore National Laboratory

关键词: Artificial Intelligence;    Mechanical Properties;    Classification;    42 Engineering;    Algorithms;   

DOI  :  10.2172/791734
RP-ID  :  UCRL-ID-131699-Rev-1
RP-ID  :  W-7405-Eng-48
RP-ID  :  791734

美国|英语

来源: UNT Digital Library

【 摘 要 】

The solid state bonded joint between two components; called an autoclave bond, is critical to the performance of a weapon system. A nondestructive method to assess the integrity of these joints is needed to certify the weapon for extended life. This project is developing ultrasonic technologies for bond quality assessment. Existing ultrasonic technology easily maps totally unbonded areas in a bond line. As an example, Figure 1 is an ultrasonic image of the bondline in a tensile specimen that was taken from a surrogate autoclave bond. We enhanced this technology to quantify the mechanical properties of a bond. There are situations when a bond interface appears intact by existing inspection methods, but fails under minimal loading. We developed an ultrasonic technique to eliminate this problem and assess the durability of the bond. Our approach is based on advanced signal processing and artificial intelligence techniques that extract information from the ultrasonic signal after it interacts with the bondline. We successfully demonstrated this technique on surrogate samples. We also designed and began assembly of an ultrasonic system to evaluate weapon components. Our next step is to acquire ultrasonic data on real parts and tailor the bond classification algorithm to detect and image defective bond regions.