Á¤º¸°úÇÐȸ ÄÄÇ»ÆÃÀÇ ½ÇÁ¦ ³í¹®Áö (KIISE Transactions on Computing Practices)
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
½º¸¶Æ® ÆÑÅ丮¸¦ À§ÇÑ ÇÏµÓ ¿¡ÄÚ ½Ã½ºÅÛ ¹× ¸Ó½Å·¯´× ±â¹ÝÀÇ °í¹« °øÁ¤ µ¥ÀÌÅÍ ºÐ¼® |
¿µ¹®Á¦¸ñ(English Title) |
Analysis of Data from a Rubber Manufacturing Process Based on Hadoop Ecosystem and Machine Learning for Smart Factor |
ÀúÀÚ(Author) |
»ç°ø¿î
À̽Âö
Àå¿ëÈÆ
¹ÚâÇö
Woon Sagong
SeungCheol Lee
Yonghun Jang
Changhyeon Park
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 26 NO. 12 PP. 0519 ~ 0527 (2020. 12) |
Çѱ۳»¿ë (Korean Abstract) |
3Â÷ »ê¾÷Çõ¸í ÀÌÈÄ ±Þ°ÝÇÏ°Ô Áõ°¡µÈ µ¥ÀÌÅÍ·Î ÀÎÇØ 4Â÷ »ê¾÷Çõ¸í ½Ã´ë¿¡¼´Â ºòµ¥ÀÌÅÍ¿¡ ´ëÇÑ Ã³¸®ÀÇ Çʿ伺ÀÌ ºÎ°¢µÇ°í ÀÖ´Ù. ¶ÇÇÑ ±¹³»¿Ü »ê¾÷ ÇöÀåÀº ºòµ¥ÀÌÅÍ Ã³¸®¸¦ ÅëÇÑ ½º¸¶Æ® ÆÑÅ丮¸¦ ±¸»ó ¹× ÁøÇà ÁßÀÌ´Ù. ÇÏÁö¸¸ ±¹³» »ê¾÷ ÇöÀåÀº ½º¸¶Æ® ÆÑÅ丮¸¦ ±¸ÃàÇϱâ À§ÇÑ ºòµ¥ÀÌÅÍ Ã³¸® ±â¼ú·Â°ú Àη ºÎÁ·À¸·Î ¾î·Á¿òÀ» °Þ°í ÀÖ´Ù. º» ³í¹®¿¡¼´Â ½º¸¶Æ® ÆÑÅ丮¸¦ ±¸ÃàÇϱâ À§ÇØ ºòµ¥ÀÌÅÍ¿Í ÇÏµÓ ¿¡ÄÚ ½Ã½º ÅÛÀ» ±â¹ÝÀ¸·Î ÇÑ °í¹« °øÁ¤ µ¥ÀÌÅ͸¦ ºÐ¼®ÇÑ´Ù. °í¹« »ý»ê °øÁ¤¿¡¼ ¼öÁýÇÑ ºòµ¥ÀÌÅ͸¦ È°¿ëÇϱâ À§ÇØ ÇÏµÓ ¿¡ÄÚ ½Ã½ºÅÛÀ» ÀÌ¿ëÇÏ¿© µ¥ÀÌÅ͸¦ ¼öÁýÇÏ¿´´Ù. ºÒ·®·ü°ú °üÇÑ ¿äÀÎ ºÐ¼®À» À§ÇØ µ¥ÀÌÅÍÀÇ Àü󸮸¦ ¼öÇàÇÏ¿´´Ù. Àüó¸® µÈ µ¥ÀÌÅ͸¦ Åë°è ºÐ¼®ÇÏ¿© ºÒ·®·ü°ú °ü·ÃÇÑ ¿äÀÎÀ» È®ÀÎÇÏ¿´´Ù. À̸¦ ÅëÇØ ¸Ó½Å·¯´× ±â¹ÝÀÇ °í¹« »ý»ê ºÒ·® ¿¹Ãø ¸ðµ¨¸µÀ» ¼öÇàÇÏ¿´´Ù. Á¦¾ÈÇÑ ¸ðµ¨ÀÇ Æò±Õ ¿¹Ãø ¼º´ÉÀº Macro F1 score 0.8554ÀÌ¸ç ¾çÇ°°ú ºÒ·®Ç°Àº °¢ 0.8912¿Í 0.8196À» ´Þ¼ºÇÏ¿´´Ù. |
¿µ¹®³»¿ë (English Abstract) |
There is an increasing need for big data processing in the era of the 4th industrial revolution, due to the rapid increases in the amounts of data following the 3rd industrial revolution. In addition, domestic and foreign industrial sites are conceiving of and proceeding with the development of smart factories through the use of big data processing. However, domestic industrial sites are experiencing difficulties due to a lack of big data processing technology and the manpower needed to build smart factories. In this paper, we analyze data from a rubber manufacturing process based on big data and Hadoop ecosystem to build a smart factory. Data were collected through the Hadoop eco system to utilize the big data collected during the rubber production manufacturing process. Data preprocessing was performed to analyze the factors related to the defective rate. Statistical analysis of the preprocessed data identified factors related to the defective rate. With this knowledge, we performed machine learning-based rubber production defect prediction modeling. The average predictive performance of the proposed model was a macro F1 score of 0.8554, and good and bad products achieved scores of 0.8912 and 0.8196, respectively. |
Å°¿öµå(Keyword) |
ºòµ¥ÀÌÅÍ
½º¸¶Æ® ÆÑÅ丮
µ¥ÀÌÅÍ ºÐ¼®
ÇϵÓ
¸Ó½Å·¯´×
big data
smart factory
data analysis
hadoop
machine learning
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|