BP神经网络预测耐火浇注料调凝剂加入量  

Predicting the Amount of Adjusting Setting Agent in Cement Bonded Refractory Castables with BP Neural Network

作　　者：孙华云[1] 李明晶[1] 邵长波 SUN Huayun;LI Mingjing;SHAO Changbo(Shandong Vocational College of Industry,Zibo 256414,China;Shandong Sino-Refraetory Co.,Ltd,Zibo 256220,China)

机构地区：[1]山东工业职业学院,山东淄博256414 [2]山东中耐高温材料有限公司,山东淄博256220

出　　处：《工业炉》2018年第5期68-71,共4页Industrial Furnace

摘　　要：调凝剂(促凝剂或缓凝剂)是水泥结合耐火浇注料中重要的外加剂,其加入量受多种工艺因素影响,主要包括施工环境温度以及浇注料中水泥、微粉、水的用量。建立基于拟牛顿算法的BP神经网络预测模型,并利用铝酸钙水泥结合刚玉尖晶石质浇注料实际生产中的经验数据对模型进行训练和预测。结果表明,该模型能比较准确地预测调凝剂加入量,可以满足实际生产要求。Adjusting setting agent（accelerator or retarder） is an important additive in cement bonded refractory castables. Its addition amount is affected by many technological factors, including construction environment temperature and the amount of cement, micro-powder and water in the castables. A prediction model of BP neural networks based on quasi-newton algorithm was established. The model is trained and tested by the experienced data from the practical production of calcium aluminate cement bonded corundum-spinel refractory castables. The results show that the model can predict the amount of adjusting setting agent and meet the requirement for production.

关 键 词：调凝剂 耐火浇注料 BP神经网络 预测模型 

分 类 号：TQ175[化学工程—硅酸盐工业] TP183[自动化与计算机技术—控制理论与控制工程]