摘要
烧结是高炉炼铁的重要环节之一,其中烧结矿FeO的含量对高炉的使用寿命,烧结的原料成本都有重要影响。为了提高烧结矿FeO预测的准确度,提出了一种基于灰狼算法支持向量机回归(GWO-SVR)的烧结矿FeO含量预测模型,利用原料使用数据对烧结矿中FeO含量进行预测,为烧结过程提供理论依据。将98个原料样本数据和烧结矿FeO含量数据进行预处理,选取79个样本作为训练集,19个样本作为测试集,分别建立烧结矿FeO支持向量回归模型(SVR),GA-SVR模型,PSO-SVR模型和GWO-SVR模型,采用决定系数(R2),均方误差(MSE)和绝对平均误差(MAPE)作为模型的对比标准。结果表明,GWO-SVR算法预测精度高、误差小,耗时较短,使用灰狼算法优化支持向量机回归对烧结矿FeO含量预测分析合理、高效。
Sintering is one of the important links in blast furnace ironmaking.The content of sinter FeO has an important influence on the service life of blast furnace and the raw material cost of sintering.In order to improve the accuracy of sinter FeO prediction,a prediction model of sinter FeO content based on gray wolf algorithm support vector machine regression(GWO-SVR)was proposed.The FeO content in sinter was predicted by raw material use data,which was sintered.The process provides a theoretical basis.98 raw material sample data and sinter FeO content data were preprocessed,79 samples were selected as training set,and 19 samples were used as test sets to establish sinter FeO support vector regression model(SVR),GA-SVR model,PSO-SVR model and GWO-SVR model,using coefficient of determination(R2),mean square error(MSE)and absolute mean error(MAPE)as the comparison criteria of the model.The results show that the GWO-SVR algorithm has high prediction accuracy,small error and short time consumption.It is reasonable and efficient to use the grey wolf algorithm to optimize the support vector machine regression to predict and analyze the FeO content of sintered ore.
作者
史振杰
董兆伟
孙立辉
武晓婧
SHI Zhen-jie;DONG Zhao-wei;SUN Li-hui;WU Xiao-jing(College of Information Technology,Hebei University of Economics and Business,Shijiazhuang Hebei 050061,China)
出处
《河北省科学院学报》
CAS
2019年第4期1-6,共6页
Journal of The Hebei Academy of Sciences
基金
河北省科技计划项目(17210310D)
