2005 Volume 91 Issue 6 Pages 537-545
We developed the internal defects and inclusions evaluation method in continuous casting cleanliness slabs by using supersonic flaw detection technique.
The size and position of defect in a slab was detected by several echoes of supersonic by a neural network signal treatment. Internal defect and inclusion was distinguished according to the phase and amplitude of the echoes. The size of defects was calibrated by preliminary artificial defect testing. Those results was compared with the optical microscope observation of the defects which revealed by polishing at 0.1 mm interval.
The results were follows.
(1) Whole pinhole type defects detected by the present method agreed with that by the optical microscope observation. And the 90 percent of inclusion type defects by this method agreed with that by the optical microscope observation. This result suggested that this method was enough precision for practical use.
(2) On the other hand, some kinds of flaw such as Al2O3 cluster and pin hole could not be distinguished by this neural network treatment.
(3) The different shape of pinhole type defects and the cluster type inclusions could be classified with a slope α which was obtained by size distribution fitting of doubly exponential function.
