摘要
FeO含量是球团质量的重要指标之一。为了更加准确地计算球团化学成分指标,提出了FeO氧化系数的概念。FeO氧化系数按照定义需要球团成分的离线计算,无法进行直接检测,通过相关因素的分析实现了FeO氧化系数的软测量。在分析球团生产过程中FeO影响因素的基础上,利用神经网络建立了FeO氧化系数的软测量模型,通过灰色关联分析方法确定了神经网络软测量模型的输入。使用实际的生产数据对模型的参数进行训练和验证,结果表明,提出的球团FeO氧化系数软测量模型能够获得满意的精度。
The FeO content is one of the most important indexes of pellet quality.In order to calculate the chemical component indexes of pellet accurately,a novel concept named FeO oxidation factor is put forward to.However,the FeO oxidation factor can not be measured directly,and need to calculate pellet component by off-line.And the soft-sensing of the FeO oxidation factor is got by analyzing the related factors.On the basis of analyzing the influencing factor of FeO in pellet production process,the soft-sensing model of FeO oxidation factor is set up by artificial neural network.Furthermore,the inputs of neural network soft-sensing model were? acquired by gray correlation analysis method.The results trained and tested by practical data show that the proposed FeO oxidation factor soft-sensing model has the satisfactory precision.
出处
《控制工程》
CSCD
北大核心
2012年第S1期142-144,148,共4页
Control Engineering of China
基金
国家自然科学基金资助项目(61074098)