We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Machine Learning Modeling of Gas Utilization Rate in Blast Furnace.
- Authors
Jiang, Dewen; Wang, Zhenyang; Zhang, Jianliang; Jiang, Dejun; Li, Kejiang; Liu, Fulong
- Abstract
In the field of ironmaking, it is crucial to predict the gas utilization rate (GUR), which is strongly related to the product quality and the smooth operation of a blast furnace (BF). The present work proposes a model based on the multi-layer perceptron (MLP) algorithm and data-driven to predict GUR after 1 h, 2 h, and 3 h, respectively. First, the collected data are preprocessed using the 3σ criterion. A maximal information coefficient (MIC) is then used for feature selection. Meanwhile, a grid search is used to select best hyperparameters for the MLP and an extreme learning machine (ELM) algorithm. Based on the above steps, the MLP and ELM models are built separately. Finally, the prediction performance of the two models is compared in several dimensions. The results show that the prediction result of both models is excellent when the chosen output parameter is the GUR after 1 h. In addition, when different output parameters are selected, the prediction accuracy of the MLP model is always higher than that of the ELM model.
- Subjects
BLAST furnaces; FEATURE selection; PRODUCT quality; GASES
- Publication
JOM: The Journal of The Minerals, Metals & Materials Society (TMS), 2022, Vol 74, Issue 4, p1633
- ISSN
1047-4838
- Publication type
Article
- DOI
10.1007/s11837-022-05166-7