Principal Components Analysis for Processing Multiparameter Acoustic Signals of the Mirror-Shadow Technique for Bar Stock Control

The paper provides reasoning for using principal component analysis to assess the generalized characteristics of defects in processing of multiparameter acoustic signals of the multiple mirror-shadow technique for bar stock control. This technique made it possible to reduce the number of signal parameters when forming rejection criteria, develop a methodology for assessing the generalized defect characteristics, and formulate a complex rejection criterion based on the unacceptable value of the generalized defect characteristics for objects made of any grade of steel and any diameter. The results of testing the proposed approach for estimating the generalized characteristics of natural defects using the obtained regression dependence are in satisfactory agreement with the results of metallographic studies