Scholarlink

<p>Accurately predicting the gas utilization ratio (GUR), a crucial metric that reflects the operating status and energy consumption of a blast furnace (BF), is extremely important to the production process. In this study, we propose a prediction model for GUR using a combination of empirical mode decomposition (MED) and a least squares support vector machine (LSSVM). Grey relational analysis and the Pearson correlation coefficient are used to select the influential parameters. The results show that GUR can be decomposed into eight intrinsic mode functions (IMFs), each corresponding to different components operating on different time scales. The correlation coefficients between the parameters and these different components range between 0.001 and 0.4, showing their different effects on GUR over different time scales. The model exhibits a high level of accuracy in predicting GUR, with a mean absolute deviation (MAD) of 0.0495 and a mean square error (MSE) of 0.0593, both of which are smaller than those of the LSSVM-based model.</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

Combination of Empirical Mode Decomposition and Least Squares Support Vector Machine for Gas Utilization Ratio Prediction of Blast Furnace

  • 期刊论文

  • 发表日期：

    2025年2月18日

查看全文
  • Xiaoming Li,
  • Zhiheng Yu,
  • Baorong Wang,
  • Xiangdong Xing

摘要

Accurately predicting the gas utilization ratio (GUR), a crucial metric that reflects the operating status and energy consumption of a blast furnace (BF), is extremely important to the production process. In this study, we propose a prediction model for GUR using a combination of empirical mode decomposition (MED) and a least squares support vector machine (LSSVM). Grey relational analysis and the Pearson correlation coefficient are used to select the influential parameters. The results show that GUR can be decomposed into eight intrinsic mode functions (IMFs), each corresponding to different components operating on different time scales. The correlation coefficients between the parameters and these different components range between 0.001 and 0.4, showing their different effects on GUR over different time scales. The model exhibits a high level of accuracy in predicting GUR, with a mean absolute deviation (MAD) of 0.0495 and a mean square error (MSE) of 0.0593, both of which are smaller than those of the LSSVM-based model.