摘要
目前,钢厂大多采用人工测温方式,该方式成本高、精确度低、工作强度高。针对某钢厂的LF精炼炉,基于大量生产与物耗数据,经数据预处理和相关性分析,结合专家经验,确定了温度预测模型的输入输出。引入深度网络,深度网络结构更加复杂,提取特征更加细致,提出了基于深度信念网络的LF炉钢液温度预测模型,并与基于灰狼优化算法的BP神经网络模型进行对比研究。结果表明,深度信念网络模型预测准确率为93%,比浅层网络精度高,可为实际生产提供理论指导,也将为以温度预测值与实际值温差最小和耗电量最少为目标的精炼过程优化奠定基础。
At present,most steel mills use manual temperature measurement,which has high cost,low accuracy and strong work force.For the LF refining furnace of a steel plant,based on mass production and material consumption data,data preprocessing and correlation analysis,combined with expert experience,determine the input and output of the temperature prediction model.Introducing the deep network,the structure of the deep network is more complex,and the extracted features are more detailed,and a prediction model of the molten steel temperature of LF furnace based on the deep belief network is proposed.And compared with the improved BP neural network.The results show that the prediction accuracy of the deep belief network model is 93%,which is higher than that of the shallow network.It can provide theoretical guidance for actual production,and will also be refined with the minimum temperature difference between the predicted temperature value and the actual value and power consumption as the optimization goal.Lay the foundation for process optimization.
出处
《工业控制计算机》
2021年第2期68-70,共3页
Industrial Control Computer
关键词
LF精炼炉
温度预测
深度信念网络
LF refining furnace
temperature prediction
deep belief network