THE JOURNAL OF INDUSTRIAL TECHNOLOGY

Volume 18, No. 02, Month MAY, Year 2022, Pages 203 - 216

Classification of carbon steels by automated spark test technique using feature extraction based on machine learning image processing

Teerawat Benjawilaikul , Thossaporn Kaewwichit

Abstract Download PDF

There are several methods of steel classification testing. The test through an emission spectrometer is one of the test methods for classifying steel. However, there are limitations involved in terms of time-consumption and cost-effectiveness in this method. The spark test is another method to classify steel, but this method still requires the knowledge and proficiency skills of the tester. Hence, this method is not very popular. The advantage of this analysis is that the testing process is not complicated and easy to do. However, the spark test requires high proficiency skills in classifying metals. This research applied the principle of steel classification to analyze the spark characteristics by categorizing metal groups using the machine learning characteristics according to JIS G 0566 standard. The results showed that low carbon steel was classified with an accuracy of 100%, medium carbon steel was classified with an accuracy of 95%, and high carbon steel was classified with an accuracy of 90%.

Keywords

Spark Testing, JIS G 0566, Image Processing, Machine Learning

---