Performance evaluation of data mining techniques in steel manufacturing industry

Thembinkosi Nkonyana, Yanxia Sun, Bhekisipho Twala, Eustace Dogo

Research output: Contribution to journalConference articlepeer-review

21 Citations (Scopus)


Industry 4.0 has evolved and created a huge interest in automation and data analytics in manufacturing technologies. Internet of Things (IoT) and Cyber Physical System (CPS) are some of the recent topics of interest in the manufacturing sector. Steel manufacturing process relies on monitoring strategies such as fault detection to reduce number of errors which can lead to huge losses. Proper fault diagnosis can assist in accurate decision-making. We use in this study predictive analysis to help solve the complex challenges faced in industrial data. Random Forest, Artificial Neural Networks and Support Vector Machines are used to train and test our industrial data. We evaluate how ensemble methods compare to classical machine learning algorithms. Finally we evaluate our models' performance and significance. Random Forest outperformed other ML methods in our study.

Original languageEnglish
Pages (from-to)623-628
Number of pages6
JournalProcedia Manufacturing
Publication statusPublished - 2019
Event2nd International Conference on Sustainable Materials Processing and Manufacturing, SMPM 2019 - Sun City, South Africa
Duration: 8 Mar 201910 Mar 2019


  • Fault Diagnostics
  • Machine Learning
  • Manufacturing

ASJC Scopus subject areas

  • Industrial and Manufacturing Engineering
  • Artificial Intelligence


Dive into the research topics of 'Performance evaluation of data mining techniques in steel manufacturing industry'. Together they form a unique fingerprint.

Cite this