Scholarlink

<p>Effective alkali management is vital for stable and efficient blast furnace operations. Accumulation of alkali elements like potassium&#xa0;(K) and sodium (Na) can cause refractory damage, scaffold formation, altered slag properties, increased energy consumption, negative effects on coke stability and reactivity, and reduced productivity. The Localized Parameter Influence Estimator (LPIE) is an explainable AI tool designed to analyze and quantify the impact of changeable process parameters on the slag alkali content.</p><p>LPIE integrates neural networks and Gaussian-weighted simulations to perform sensitivity analyses on key process variables. By isolating the influence of individual parameters, the tool provides reliable estimates and actionable insights for optimizing alkali management. Compared to traditional methods, such as Accumulated Local Effects (ALE), LPIE offers advanced visualization capabilities and focuses on operational optimization in highly dynamic industrial environments. Through simulations and historical data, LPIE evaluates parameter correlations, ranks their influence, and aids informed decision-making.</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

Localized Parameter Influence Estimator: An Explainable Artificial Intelligence Framework for Alkali Management in Blast Furnace Operations

  • 期刊论文

  • 发表日期：

    2025年5月2日

查看全文
  • Johann Wachlmayr,
  • Clemens Staudinger,
  • Christoph Feilmayr,
  • Christine Gruber

摘要

Effective alkali management is vital for stable and efficient blast furnace operations. Accumulation of alkali elements like potassium (K) and sodium (Na) can cause refractory damage, scaffold formation, altered slag properties, increased energy consumption, negative effects on coke stability and reactivity, and reduced productivity. The Localized Parameter Influence Estimator (LPIE) is an explainable AI tool designed to analyze and quantify the impact of changeable process parameters on the slag alkali content.

LPIE integrates neural networks and Gaussian-weighted simulations to perform sensitivity analyses on key process variables. By isolating the influence of individual parameters, the tool provides reliable estimates and actionable insights for optimizing alkali management. Compared to traditional methods, such as Accumulated Local Effects (ALE), LPIE offers advanced visualization capabilities and focuses on operational optimization in highly dynamic industrial environments. Through simulations and historical data, LPIE evaluates parameter correlations, ranks their influence, and aids informed decision-making.