Research on neural network model for sintering process based on iron ore characteristics

详细信息    查看全文 | 下载全文 | 推荐本文 |

• 作者：Dauter OliveiraWU ShengliDAI YumingZHANG GuoliangAlex Castro
• 会议时间：2013-06-01
• 关键词：neural network model ; sintering ; iron ore ; characteristics
• 作者单位：Dauter Oliveira(VALE,Singapore)WU Shengli,ZHANG Guoliang(School of Metallurgical and Ecological Engineering,University of Science and Technology Beijing,Beijing 100083,China)DAI Yuming,Alex Castro(VALE,China)
• 母体文献：第五届宝钢学术年会论文集
• 会议名称：第五届宝钢学术年会
• 会议地点：上海
• 主办单位：中国金属学会
• 语种：chi

摘要

A Neural Network Model ( NNM) for sintering process was investigated in this paper.According to raw material characteristics and its blending proportion in sintering plant,this intelligent model can effectively predict sintering performance indexes ( including productivity,fuel consumption,TI,particle size,,RDI and RI).Before building this Neural Network Model,it is important to understand the relationship between sintering performance indexes and iron ore characteristics.First,iron ore characteristics ( including chemical composition,physical quality,microstructure,and high temperature properties) of 41 kinds of Chinese and overseas ores were experimented.Second,70 different schemes of sinter pot tests were conducted.Third,basing on the result of the lst and 2nd steps,a software named TIBERIUS was used to build the model correlating sintering performance with iron ore characteristics.Finally,the effectiveness of the model was tested by using actual iron ore mix characteristics of a Chinese and a Brazilian steel mill.The model output predicted sintering performance indexes very close to the real production results,showing good efficiency and applicability of the model.