| Abstrakt: |
Bimetal components play a vital role in fulfilling diverse material requirements of lightweight structures, heat transfer, and functionally graded applications. In bimetallic castings, the presence of microgap defects significantly affects the bond strength between dissimilar materials. The dissimilar joining of Al-SS304, achieving a high-quality bond, relies on several factors, such as pouring temperature of molten aluminum, preheating temperature of SS insert, thickness of SS inserts, and the choice of interlayer coating materials. In the present work, the Zn, Cu, and Sn interlayers were used to produce Al-SS304 bimetallic castings since they showed good bond between aluminum and stainless steel. In addition to that, a machine learning algorithm was employed to predict the microgap at the interface between aluminum and stainless steel. Both destructive and nondestructive testing techniques were utilized to predict the microgap defects formed at the interface. Response surface methodology was employed to systematically perform the experiments, and a dataset for a machine learning algorithm was formed using the results from liquid penetrant testing. The results from this prediction model were compared with those obtained from the destructive shear punch test results. Notably, the samples without defects displayed a maximum load exceeding 26 kN, whereas the samples with microgap defects exhibited a maximum load below 26 kN. A confusion matrix was used to validate the machine learning classification model. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink