Scholarlink

<p>In traditional relining practices, aluminum smelter cells are shut down in a planned manner to avoid failure which carries risk to environment and humans. This limits the effective utilization of the cell lining and profitability of the industry. Thus, calling for modern strategies such as prognostic and health management systems which can utilize the abundant data generated by the industry to predict the time taken for the cell failure to occur. This study explores two of the popular statistical techniques, similarity and survival models in aluminum smelter cells and estimate the remaining useful life of the cell. It also integrates machine learning-based algorithms such as principal component analysis and Random Forest Regression for building the models. The models have been trained and tested using the historical data extracted from the plant database and further validated by predicting the life of cells which were running and planned to be shut down.</p> Graphical Abstract <p>The below picture shows the overall working of the remaining useful life model that was been developed in the study. This model determines the remaining life of cell to raise alarm for required relining.</p> <p></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

Prolonging Aluminum Reduction Cell Life Using Machine Learning

  • 期刊论文

  • 发表日期：

    2025年4月25日

查看全文
  • Shanmukh Rajgire,
  • Avinash Beena,
  • Rohit Chaudhari,
  • Amit Gupta

摘要

In traditional relining practices, aluminum smelter cells are shut down in a planned manner to avoid failure which carries risk to environment and humans. This limits the effective utilization of the cell lining and profitability of the industry. Thus, calling for modern strategies such as prognostic and health management systems which can utilize the abundant data generated by the industry to predict the time taken for the cell failure to occur. This study explores two of the popular statistical techniques, similarity and survival models in aluminum smelter cells and estimate the remaining useful life of the cell. It also integrates machine learning-based algorithms such as principal component analysis and Random Forest Regression for building the models. The models have been trained and tested using the historical data extracted from the plant database and further validated by predicting the life of cells which were running and planned to be shut down.

Graphical Abstract

The below picture shows the overall working of the remaining useful life model that was been developed in the study. This model determines the remaining life of cell to raise alarm for required relining.