Scholarlink

<p>The efficiency of sugar mill is significantly impacted by the level of mineral trash entering with the cane. This type of trash adversely affects milling equipment, raises ash content in bagasse, and diminishes clarification and filtration effectiveness due to higher mud levels, impacting extraction, cogeneration, and sucrose recovery. These effects have intensified with the increased mechanical harvest, from 40% in 2013 to 75% in 2023. Combined with climate variability and high precipitation events, this has led to mineral trash levels occasionally exceeding 5%, when normal levels are between 0.8 to 1.0%. An exploratory study was carried out to estimate of mineral trash content by turbidity measurements in mixed juice, aiming to provide real-time data for timely interventions. At laboratory scale, mineral impact was simulated by doping clean juices from the stalks press with soil, which showed a strong correlation (R2 &gt; 0.90) between turbidity and mineral trash, above other variables evaluated such as juice deterioration and color variations due to varietal differences. Based on these findings, an analysis of historical data recorded during nearly two years by a sugar mill was conducted, evaluating variables such as turbidity and insoluble solids in mixed juice, ash in bagasse, and filter cake as a percentage of cane. Clustering techniques were used, and relationships were identified between turbidity and these variables of interest. Finally, a K-nearest neighbors (KNN) classification model was trained, achieving an accuracy above 75%, allowing the classification of mineral trash levels into low, medium, and high categories. This model could be implemented online with the installation of a turbidity sensor in mixed juice, enabling timely process interventions.</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

A Machine Learning Approach for Detecting Mineral Trash in Sugarcane Processing Through Turbidity Measurements

  • 期刊论文

  • 发表日期：

    2026年2月7日

查看全文
  • Julián David Montes,
  • Maria Alejandra Gomez,
  • Roosembert Polanco,
  • Ana Milena Escobar,
  • Nicolas Gil Zapata

摘要

The efficiency of sugar mill is significantly impacted by the level of mineral trash entering with the cane. This type of trash adversely affects milling equipment, raises ash content in bagasse, and diminishes clarification and filtration effectiveness due to higher mud levels, impacting extraction, cogeneration, and sucrose recovery. These effects have intensified with the increased mechanical harvest, from 40% in 2013 to 75% in 2023. Combined with climate variability and high precipitation events, this has led to mineral trash levels occasionally exceeding 5%, when normal levels are between 0.8 to 1.0%. An exploratory study was carried out to estimate of mineral trash content by turbidity measurements in mixed juice, aiming to provide real-time data for timely interventions. At laboratory scale, mineral impact was simulated by doping clean juices from the stalks press with soil, which showed a strong correlation (R2 > 0.90) between turbidity and mineral trash, above other variables evaluated such as juice deterioration and color variations due to varietal differences. Based on these findings, an analysis of historical data recorded during nearly two years by a sugar mill was conducted, evaluating variables such as turbidity and insoluble solids in mixed juice, ash in bagasse, and filter cake as a percentage of cane. Clustering techniques were used, and relationships were identified between turbidity and these variables of interest. Finally, a K-nearest neighbors (KNN) classification model was trained, achieving an accuracy above 75%, allowing the classification of mineral trash levels into low, medium, and high categories. This model could be implemented online with the installation of a turbidity sensor in mixed juice, enabling timely process interventions.