摘要
提出了基于最小二乘支持向量机回归模型的氧化铝浓度软测量方法,分析了分布式槽电压、阳极导杆电流与氧化铝浓度之间的关系;建立了面向氧化铝浓度预测的最小二乘支持向量机回归模型,提出了基于网格搜索和交叉验证的惩罚因子和核函数宽度的寻优方法;最后,基于100组现场采集的数据,验证了所提方法的有效性,预测的整体相对误差达到了0.051 2,标准差为0.132 1,与已有文献中的ELM算法相比,相对误差降低了约22%。
A soft sensor based on least square support vector machine regression model was proposed.The relationship between distributed tank voltage,anode lead current and alumina concentration was analyzed A least squares support vector machine regression model for alumina concentration prediction was established.An optimization method of penalty factor and kernel function width was proposed based on grid search and cross-validation.Finally,the validity of the proposed method was verified using the data.The overall relative error of the prediction reaches 0.051 2 and the standard deviation is 0.132 1.Compared with the ELM algorithm in the existing literature,the relative error is reduced by about 22%.
作者
崔家瑞
张政伟
李擎
崔家山
CUI Jiarui;ZHANG Zhengwei;LI Qing;CUI Jiashan(University of Science and Technology Beijing,Beijing 100083,China)
出处
《兵器装备工程学报》
CAS
北大核心
2018年第12期243-247,共5页
Journal of Ordnance Equipment Engineering
基金
国家高技术研究发展计划项目(2013AA040705)
国家自然科学基金项目(61603034)
北京市自然科学基金项目(3182027)
中央高校基本科研业务费赞助项目(FRF-GF-17-B44
FRF-OT-18-016SY)
关键词
铝电解
氧化铝浓度
最小二乘支持向量机
交叉验证
回归预测
aluminum electrolysis
alumina concentration
least squares support vector machine
cross validation
regression prediction
