Abstract
In the case of probe rotation type eddy current defect detection of round steel bars in hot rolling, the most serious problem is noise caused by vibration of round steel bars.
This paper reports the result of calculation and spectrum analysis of this vibration noise and improvement of defect detectability by characteristics extraction method using mini-computer.
By the calculation and spectrum analysis of vibration noise, it became clear that practical SN ratio of defect signal and the vibration noise is not obtained by conventional frequency discrimination method using band pass filter when the vibration amplitude is large.
To improve defect detectability, a signal processing to extract the characteristics of defect signal which are obtained by differential coil type probe were contrived and adopted to defect detection signals of round steel bars in hot rolling.
Consequently, SN ratio and defect detection rate have been improved about 40% and 50% respectively as compared with conventional frequency discrimination method.
