Metallurgy, Vol. 63 No. 2, 2024.
Original scientific paper
Prediction of oxygen consumption in steelmaking based on LAOA-TSVR
Z. C. Ma
; School of Electrical and Automation Engineering, Liaoning Institute of Science and Technology, Benxi, China
L. Zhang
; School of Electrical and Automation Engineering, Liaoning Institute of Science and Technology, Benxi, China
C. Y. Shi
; School of Electrical and Automation Engineering, Liaoning Institute of Science and Technology, Benxi, China
*
X. Wang
; Shandong Iron & Steel Group Rizhao Co. Ltd., Rizhao, China
Y. K. Wang
; School of Electrical and Automation Engineering, Liaoning Institute of Science and Technology, Benxi, China
P. L. Tao
; School of Electrical and Automation Engineering, Liaoning Institute of Science and Technology, Benxi, China
P. Sun
; School of Electrical and Automation Engineering, Liaoning Institute of Science and Technology, Benxi, China
* Corresponding author.
Abstract
To solve the issue of oxygen consumption forecasting, the researchers suggested a twin support vector machine for regression (LAOA-TSVR) prediction model based on an improved arithmetic optimization algorithm. The model has beneficial generalization, high prediction accuracy, and the ability to jump out of the local optimum and other characteristics. The group used the method of mechanism analysis to determine the main influencing factors of oxygen consumption. To confirm the model’s prediction effect, it is compared to the Back Propagation, Radial Basis Function, and Twin Support Vector Regression prediction models. The LAOA-TSVR oxygen consumption forecasting prediction model was then tested on actual steel mill production. The test phase consisted of 200 production cycles, and the results revealed that the LAOA-TSVR model had an 85,1 % hit rate for oxygen consumption within 5 m3/t. The model can suit the actual needs of predicting oxygen consumption in steel.
Keywords
steelmaking oxygen consumption forecasting hit rate industrial test; LAOA-TSVR
Hrčak ID:
312247
URI
Publication date:
1.4.2024.
Visits: 287 *