LF精炼终点钢水温度灰箱预报模型  

Grey box model for predicting the LF end-point temperature of molten steel

作　　者：付国庆[1,2] 刘青[1,2] 汪宙[1,2] 常健 王彬[1,2] 谢飞鸣 卢新春[2,3] 居琪萍 

机构地区：[1]北京科技大学钢铁冶金新技术国家重点实验室,北京100083 [2]北京科技大学冶金与生态工程学院,北京100083 [3]方大特钢科技股份有限公司,南昌330012

出　　处：《北京科技大学学报》2013年第7期948-954,共7页Journal of University of Science and Technology Beijing

基　　金：国家自然科学基金资助项目(50874014);教育部博士学科点专项基金(20090006110024)

摘　　要：LF精炼工序在炼钢过程起着调节温度的关键作用,准确预报LF精炼终点钢水温度对实际生产有重要意义.传统的LF精炼预报模型包括机理模型与黑箱模型.机理预报模型能够体现各工艺因素对终点钢水温度的影响,但由于LF精炼传热机理研究尚不完善,依靠机理模型预报终点钢水温度,难以达到预期效果;黑箱预报模型能够准确预报终点钢水温度,但不能反映精炼过程各工艺因素对钢水温度的影响,尤其当生产工艺条件发生改变时,黑箱模型在应用上会受到限制.本文以方大特钢LF精炼炉为研究对象,建立一种机理预报模型与黑箱预报模型(BP神经网络预报模型)相结合的LF精炼终点钢水温度灰箱预报模型.该模型既能反映各工艺因素对终点钢水温度的影响,又能准确预测终点钢水温度,其终点钢水温度预测误差在±5℃以内的命中率可以达到95%以上.LF refining process plays an important role in the temperature adjustment of molten steel, and precisely predicting the LF end-point temperature of molten steel is of great importance to actual production. Generally speaking, the prediction models of LF end-point temperature include the mechanism model and the black box model. The mechanism model can reflect the influence of each factor on the end-point temperature of molten steel, but it is dimcult to obtain the expected prediction accuracy due to the limited comprehension of heat transfer in LF refining process. The black box model can usually achieve high prediction accuracy, whereas it does not reveal the effect of each factor. Moreover, the black box model has limited applications when process conditions are changed. Taking LF refining process in Fangda special steel plants as an object of study, this paper establishes a grey box model for predicting the LF end-point temperature of molten steel based on the mechanism model and the black box model. The grey box prediction model can not only indicate the impact of each factor, but also provide the precise prediction of LF end-point temperature. Verification results show that the hit rate of the grey box model is greater than 95% while the predictive error is within ＋5 ℃.

关 键 词：炼钢炉 钢包精炼 温度 预报 神经网络 灰箱模型 

分 类 号：TF769.2[冶金工程—钢铁冶金]