|Cite this article as:|
|Runhao Zhang, Jian Yang, Han Sun, and Wenkui Yang, Prediction of lime utilization ratio of dephosphorization in BOF steelmaking based on online sequential extreme learning machine with forgetting mechanism, Int. J. Miner. Metall. Mater.,(2023). https://doi.org/10.1007/s12613-023-2732-4|
The machine learning models of multiple linear regression (MLR), support vector regression (SVR), extreme learning machine (ELM), the proposed ELM model of online sequential ELM (OS-ELM) and OS-ELM with forgetting mechanism (FOS-ELM) are applied in predicting the lime utilization ratio of dephosphorization in BOF steelmaking process. Among the three basic models of MLR, SVR and ELM, ELM performs the best. OS-ELM and FOS-ELM are applied for the sequential learning and model updating. The optimal number of samples in validity term of the FOS-ELM model is determined to be 1500 with the smallest population MARE value of 0.058226. According to the variable importance analysis, the lime weight, initial P content and hot metal weight are the three most important variables for the lime utilization ratio. The lime utilization ratio will increase with the decrease of lime weight and the increases of initial P content and hot metal weight. A prediction system based on FOS-ELM is applied in the actual industrial production for one month. The hit ratios of the predicted lime utilization ratio in the error ranges of ±1%, ±3% and ±5% are 61.16%, 90.63% and 94.11%, respectively. The values of R2, MARE and RMSE are 0.8670, 0.06823 and 1.4265, respectively. The performance of the system is pretty good for the application in the actual industrial production.