Scholarlink

<p>Low-grade coal in India is impregnated with large amounts of ash and carbonaceous materials that restrict its industrial applicability, whereas blast furnace (BF) slag, which is one of the major byproducts of the steel-making process, is disposable despite its potential usefulness as a cementitious secondary raw material. The present paper examines a recently designed slope-controlled dry separation apparatus that has a 2&#xa0;mm mesh screen to efficiently handle +0–4&#xa0;mm feeds of coal and BF slag. Experiments within a frequency range of 6–13&#xa0;Hz were done with an occurrence of ±1° slope using an experiment under the characteristics of 4%, 6%, and 8% moisture content. At higher frequencies, separation efficiency between dry and moist particles declined with moisture owing to agglomeration of particles together with aperture plugging and loss of stratification strength. In optimal conditions (±1° upward slope and medium frequency), coal had maximum efficiencies of 85.96%, 75.64%, and 63.11% at 4%, 6%, and 8% of moisture, respectively, and 88.31%, 80.34%, and 71.92% at BF slag. The upward slope arrangement was better in terms of increasing residence time and stratification of the particles, and the downward slope arrangement was better in terms of enhancing the mobility of the particles in the condition of high moisture coal. In order to analyze the nonlinear frequency–efficiency relationship, classical (kernel ridge regression and support vector regression) and quantum-inspired (quantum kernel ridge regression and quantum support vector regression) schemes were modeled on the basis of grid-search with fivefold cross-validation. Coefficient of determination (<i>R</i><sup>2</sup>), root mean square error, and mean absolute error were used to estimate predictive performance, and diagnostics of residual effects, such as probability analysis and histogram, were used. Quantum Support Vector Regression (QSVR) showed the highest predictive accuracy at all the moisture levels, with the highest R<sup>2</sup> and lowest error measures as compared with both the classical model and quantum kernel ridge regression.</p> Graphical Abstract <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

Statistical and Residual Analysis of Slope-Controlled Dry Screening for Fine Coal and Blast Furnace Slag Using Classical and Quantum Machine Learning-Inspired Regression Models

  • 期刊论文

  • 发表日期：

    2026年7月30日

查看全文
  • Bharath Kumar Shanmugam,
  • G. Santhosh Kumar,
  • U. Pavan Kumar,
  • Kavyashree I. Pattan,
  • S. Gokulakrishnan,
  • L. Chandrakala,
  • Aishwarya Kavitha Kumar,
  • Harish Hanumanthappa,
  • Pramod Kumar Naik

摘要

Low-grade coal in India is impregnated with large amounts of ash and carbonaceous materials that restrict its industrial applicability, whereas blast furnace (BF) slag, which is one of the major byproducts of the steel-making process, is disposable despite its potential usefulness as a cementitious secondary raw material. The present paper examines a recently designed slope-controlled dry separation apparatus that has a 2 mm mesh screen to efficiently handle +0–4 mm feeds of coal and BF slag. Experiments within a frequency range of 6–13 Hz were done with an occurrence of ±1° slope using an experiment under the characteristics of 4%, 6%, and 8% moisture content. At higher frequencies, separation efficiency between dry and moist particles declined with moisture owing to agglomeration of particles together with aperture plugging and loss of stratification strength. In optimal conditions (±1° upward slope and medium frequency), coal had maximum efficiencies of 85.96%, 75.64%, and 63.11% at 4%, 6%, and 8% of moisture, respectively, and 88.31%, 80.34%, and 71.92% at BF slag. The upward slope arrangement was better in terms of increasing residence time and stratification of the particles, and the downward slope arrangement was better in terms of enhancing the mobility of the particles in the condition of high moisture coal. In order to analyze the nonlinear frequency–efficiency relationship, classical (kernel ridge regression and support vector regression) and quantum-inspired (quantum kernel ridge regression and quantum support vector regression) schemes were modeled on the basis of grid-search with fivefold cross-validation. Coefficient of determination (R2), root mean square error, and mean absolute error were used to estimate predictive performance, and diagnostics of residual effects, such as probability analysis and histogram, were used. Quantum Support Vector Regression (QSVR) showed the highest predictive accuracy at all the moisture levels, with the highest R2 and lowest error measures as compared with both the classical model and quantum kernel ridge regression.

Graphical Abstract