基于改进多元回归和MLP神经网络的转炉炉口微差压预测模型分析  

Analysis of Micro Differential Pressure Prediction Model for Converter Mouth Based on Improved Multiple Regression and MLP Neural Network

作　　者：赵夕东 Zhao Xidong(School of Electrical Engineering,Maanshan College,Maanshan Anhui 243100,China)

机构地区：[1]马鞍山学院电气工程学院,安徽马鞍山243100

出　　处：《山西冶金》2025年第2期147-150,共4页Shanxi Metallurgy

摘　　要：在转炉炼钢过程中,微差压的精确控制对于优化炉内气氛和温度分布、提升转炉煤气热值回收及钢水质量具有显著影响。针对传统微差压控制存在的精度不足和响应迟缓问题,通过深入分析转炉压力平衡机制,确定了影响炉口压力的关键因素,提出了基于改进多元回归和多层感知器(MLP)神经网络的转炉炉口压力预测模型,并利用转炉的实际生产数据,对模型进行了训练和验证结果实验。In the process of converter steelmaking,the precise control of micro differential pressure has a significant impact on the optimisation of furnace atmosphere and temperature distribution,the enhancement of calorific value recovery of converter gas and the quality of molten steel.Aiming at the problems of lack of precision and delayed response of traditional differential pressure control,this study proposes a converter hearth pressure prediction model based on improved multiple regression and multilayer perceptron(MLP)neural network by deeply analysing the pressure balance mechanism of the converter and identifying the key factors affecting the hearth pressure.And using the actual production data of the converter,the model was trained and validated with resultant experiments.

关 键 词：微压差控制 改进多元回归 MLP神经网络 压力平衡机制 

分 类 号：TF341.1[冶金工程—冶金机械及自动化]