基于改进的黑翅鸢优化算法-轻梯度提升机建立的转炉炼钢终点温度预测模型OA

In order to realize the accurate prediction of converter steelmaking endpoint temperature,the actual SPCC steel data collected in the field are selected.By using the 3σ principle,mean-filling method for cleaning data outliers and missing values and determining 10 input features based on gray correlation analysis and process theory,and then establishing the light gradient boosting machine(LightGBM)-based endpoint temperature prediction model for converter steelmaking,and adopting support vector machine(SVM)and extreme gradient boosting(XGBoost)model as a comparison,the adaptability of the LightGBM model in the converter data and the superiority of the prediction performance is verified.Aiming at the problem that LightGBM models require manual parameter tuning,making it difficult to improve prediction accuracy,a multi-strategy improved black-winged kite optimization algorithm(IBKA)is proposed to obtain the optimal combination of its important parameters.The results show that compared with the other five optimization models(JAYA-LightGBM,GWO-LightGBM,WOA-LightGBM,RBMO-LightGBM,BKA-LightGBM),the IBKA-LightGBM model performs optimally in terms of prediction accuracy and performance evaluation indices,and achieves better prediction results,with prediction error hit rates of 85.56％and 96.67％at±10 ℃ and±15 ℃,respectively,which can provide effective operational guidance for steelmaking production.