End‐Point Dynamic Control Model for 260 Tons BOF Steelmaking Based on PWWTSVR and LWOA

Abstract

The control of end‐point carbon content and temperature is very important for basic oxygen furnace (BOF) steelmaking. Herein, to improve the control precision of the BOF end‐point, a new dynamic control model of the BOF end‐point is proposed. First, based on the samples of low‐carbon steel collected from the steel plant, a prediction model of BOF end point is established by using projection wavelet weighted twin support vector regression (PWWTSVR). It is indicated in the simulation results that the error bound of carbon content and temperature are 0.005 wt% and 10 °C, respectively, and the hit rates of the prediction models are 92% and 90%. Moreover, a double hit rate of 84% is higher than the other five prediction models. Based on the PWWTSVR prediction model, a Lévy‐flying whale optimization algorithm (LWOA) is used to optimize the objective function to establish the control model. The control model shows that the mean absolute error of the end‐blow oxygen volume calculated by the model is 75.0201 Nm3, which is smaller by comparing the other two control models. The proposed control model can provide efficient guidance for on‐site smelting personnel.