Scholarlink

<p>This article presents key findings from the PhD thesis of this article’s first author on optimizing industrial vibrating screen operation using insights derived from vibration data. Controlled laboratory experiments were conducted to establish a&#xa0;basic understanding of screen behavior, revealing how vibration patterns shift under varying loads. Building on these insights, two industrial case studies were carried out using vibration sensors and machine learning models to analyze the condition states and process-related performance over a&#xa0;monitoring period of more than one year.</p><p>In Case Study&#xa0;I, a&#xa0;linear motion screen used for dewatering was analyzed using an MLP model that classified operating states with 88% accuracy, identifying manual cleaning as a&#xa0;key factor in reducing efficiency. A&#xa0;threshold-based expert system also detected typical failures such as loose screen decks and bearing damage. In Case Study&#xa0;II, a&#xa0;circular motion screen in an aggregate plant was monitored, and models were trained to predict feed flowrates, achieving an R<sup>2</sup> score of 90%.</p><p>Overall, the investigations conducted as part of the dissertation and the resulting insights demonstrate that the combination of vibration monitoring, manual analysis, and machine learning enables an effective condition classification, fault detection, and real-time process prediction. Implementing vibration sensors for continuous monitoring proves to be a&#xa0;suitable and promising tool for significantly increasing the efficiency and reliability of industrial screening processes.</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

Insights Gained from the Digital Monitoring of Vibrating Screens Using Smart Acceleration Sensors of Type SES by IFE Aufbereitungstechnik GmbH

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年5月23日

查看全文
  • Philip Krukenfellner,
  • Helmut Tröbinger,
  • Helmut Flachberger

摘要

This article presents key findings from the PhD thesis of this article’s first author on optimizing industrial vibrating screen operation using insights derived from vibration data. Controlled laboratory experiments were conducted to establish a basic understanding of screen behavior, revealing how vibration patterns shift under varying loads. Building on these insights, two industrial case studies were carried out using vibration sensors and machine learning models to analyze the condition states and process-related performance over a monitoring period of more than one year.

In Case Study I, a linear motion screen used for dewatering was analyzed using an MLP model that classified operating states with 88% accuracy, identifying manual cleaning as a key factor in reducing efficiency. A threshold-based expert system also detected typical failures such as loose screen decks and bearing damage. In Case Study II, a circular motion screen in an aggregate plant was monitored, and models were trained to predict feed flowrates, achieving an R2 score of 90%.

Overall, the investigations conducted as part of the dissertation and the resulting insights demonstrate that the combination of vibration monitoring, manual analysis, and machine learning enables an effective condition classification, fault detection, and real-time process prediction. Implementing vibration sensors for continuous monitoring proves to be a suitable and promising tool for significantly increasing the efficiency and reliability of industrial screening processes.