Scholarlink

Abstract <p>Producing liquid and gaseous fuels from coal is a critical economic and social objective for our country. In this context, extensive research on the thermal processing of coal (coal pyrolysis, high-temperature dissolution, and hydrogenation) from Mongolia’s major deposits has been conducted over a considerable period. In this study, we utilized the results of pyrolysis experiments on coal from 15 Mongolian deposits (lignite, subbituminous, and bituminous coals). To model the tar yield from pyrolysis as a function of coal technical properties and pyrolysis temperature, we employed the Random Forest (RF) regression classification method for machine learning. For training the RF regression model, 80.0% of a total of 91 data points were used for model training, while the remaining 20.0% were allocated for model validation. The correlations between the variables used in the model were analyzed. Six key factors were selected to predict the yield of liquid products derived from coal. The random decision regression method was applied, and the calculations were performed using Python programming. Our model achieved a confidence level of 86%.</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

The Relationship between Coal Properties and the Parameters of the Pyrolysis Process

  • 期刊论文

  • 发表日期：

    2026年2月1日

查看全文
  • S. Batbileg,
  • Sh. Iderbayar,
  • J. Namkhainorov,
  • G. Shiirav,
  • B. Purevsuren

摘要

Abstract

Producing liquid and gaseous fuels from coal is a critical economic and social objective for our country. In this context, extensive research on the thermal processing of coal (coal pyrolysis, high-temperature dissolution, and hydrogenation) from Mongolia’s major deposits has been conducted over a considerable period. In this study, we utilized the results of pyrolysis experiments on coal from 15 Mongolian deposits (lignite, subbituminous, and bituminous coals). To model the tar yield from pyrolysis as a function of coal technical properties and pyrolysis temperature, we employed the Random Forest (RF) regression classification method for machine learning. For training the RF regression model, 80.0% of a total of 91 data points were used for model training, while the remaining 20.0% were allocated for model validation. The correlations between the variables used in the model were analyzed. Six key factors were selected to predict the yield of liquid products derived from coal. The random decision regression method was applied, and the calculations were performed using Python programming. Our model achieved a confidence level of 86%.