Scholarlink

<p>We consider the main principles and approaches used for the development of a&#xa0;predictive model aimed at minimizing the downtimes of equipment used for the production of pipes in continuous rolling mills. We perform comparative analysis of the available modern failure-prediction methods, including machine-training algorithms, artificial neural networks, and mathematical models. An upgraded system of data collection and transmission guaranteeing the possibility of efficient integration of predictive algorithms in the process of production is presented. We study two methods that can be used for constructing the models, namely, gradient boosting and artificial neural networks. The application of artificial neural networks makes it possible to attain a&#xa0;predictive accuracy of 99.6%, which exceeds the characteristics of traditional methods and promotes a&#xa0;decrease in the downtimes of equipment. The obtained results can be used for the improvement of control systems of the manufacturing processes both in metallurgy and in other branches of industry.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Key principles and approaches to the development of a predictive model aimed at minimizing the downtimes of equipment in continuous pipe-rolling mills

  • 期刊论文

  • 发表日期：

    2025年9月29日

查看全文
  • E. A. Shkuratov,
  • N. V. Ulman,
  • D. V. Levshin,
  • K. P. Pyankov

摘要

We consider the main principles and approaches used for the development of a predictive model aimed at minimizing the downtimes of equipment used for the production of pipes in continuous rolling mills. We perform comparative analysis of the available modern failure-prediction methods, including machine-training algorithms, artificial neural networks, and mathematical models. An upgraded system of data collection and transmission guaranteeing the possibility of efficient integration of predictive algorithms in the process of production is presented. We study two methods that can be used for constructing the models, namely, gradient boosting and artificial neural networks. The application of artificial neural networks makes it possible to attain a predictive accuracy of 99.6%, which exceeds the characteristics of traditional methods and promotes a decrease in the downtimes of equipment. The obtained results can be used for the improvement of control systems of the manufacturing processes both in metallurgy and in other branches of industry.