电弧炉温度预报模型的设计与实现

Design and Implementation of the Temperature Prediction Model for EAF Steelmaking

针对某厂150 t超高功率电弧炉的生产设备、冶炼工艺和终点预报问题进行研究,通过混合编程技术,建立基于BP神经网络的电弧炉温度预报模型,并基于现场生产数据进行优化。优化后当温度控制精度为±20℃时,模型预报命中率达到82.5%,可以满足实际生产的要求。而且模型与现场数据库建立连接,实现模型的自动运行和温度的实时预报。实践表明该预报模型对现场生产具有积极的指导意义。

Through the systematic investigations of the production devices,smelting process and the end-point prediction,the temperature prediction model for EAF steelmaking is constructed based on BP neural network,which was programmed with the mixed programming technique.Based on the field data,the structure of this model is optimized.After optimization,the hit ratio is 82.5% when the precision of temperature is controlled within ±20 ℃,which satisfies the demand of production.Moreover,the model is connected with the field database,which makes the model automatic operation.The practice shows that the application of this model to EAF process has lead to the good economic benefit.