Ventilation Prediction for an Industrial Cement Raw Ball Mill by BNN—A 'Conscious Lab' Approach
| Autor: | Esmaiel Hadavandi, Samaneh Yazdani, Saeed Chehreh Chelgani, Hossein Siavoshi, Rasoul Fatahi, Rasoul Khosravi |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2021 |
| Předmět: |
cement
Technology Mean squared error Computer science 02 engineering and technology Article law.invention 020401 chemical engineering law 0202 electrical engineering electronic engineering information engineering Mill General Materials Science 0204 chemical engineering support vector regression Process engineering Microscopy QC120-168.85 Artificial neural network business.industry Multivariable calculus QH201-278.5 conscious laboratory Engineering (General). Civil engineering (General) ball mill TK1-9971 Random forest Support vector machine Descriptive and experimental mechanics Ventilation (architecture) 020201 artificial intelligence & image processing Electrical engineering. Electronics. Nuclear engineering Comminution TA1-2040 business random forest |
| Zdroj: | Materials Volume 14 Issue 12 Materials, Vol 14, Iss 3220, p 3220 (2021) |
| ISSN: | 1996-1944 |
| DOI: | 10.3390/ma14123220 |
| Popis: | In cement mills, ventilation is a critical key for maintaining temperature and material transportation. However, relationships between operational variables and ventilation factors for an industrial cement ball mill were not addressed until today. This investigation is going to fill this gap based on a newly developed concept named “conscious laboratory (CL)”. For constructing the CL, a boosted neural network (BNN), as a recently developed comprehensive artificial intelligence model, was applied through over 35 different variables, with more than 2000 records monitored for an industrial cement ball mill. BNN could assess multivariable nonlinear relationships among this vast dataset, and indicated mill outlet pressure and the ampere of the separator fan had the highest rank for the ventilation prediction. BNN could accurately model ventilation factors based on the operational variables with a root mean square error (RMSE) of 0.6. BNN showed a lower error than other traditional machine learning models (RMSE: random forest 0.71, support vector regression: 0.76). Since improving the milling efficiency has an essential role in machine development and energy utilization, these results can open a new window to the optimal designing of comminution units for the material technologies. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
| Nepřihlášeným uživatelům se plný text nezobrazuje | K zobrazení výsledku je třeba se přihlásit. |