基于无偏灰色GM(1,1)的铁精矿品位预测模型  

Prediction Model for Iron Concentrate Grade Based on Unbiased Grey GM(1,1)

作　　者：任传成[1] 夏文成[2] 王文博[1] 郑文艳[1] REN Chuancheng;XIA Wencheng;WANG Wenbo;ZHENG Wenyan(College of Computer and Information Engineering,Dezhou University,Dezhou 253023,Shandong,China;School of Chemical Engineering and Technology,China University of Mining and Technology,Xuzhou 221116,Jiangsu,Ch)

机构地区：[1]德州学院计算机与信息学院,山东德州253023 [2]中国矿业大学化工学院,江苏徐州221116

出　　处：《有色金属（选矿部分）》2023年第1期41-45,56,共6页Nonferrous Metals（Mineral Processing Section）

基　　金：国家自然科学基金青年基金资助项目(51604272)。

摘　　要：铁精矿品位的准确预测对铁矿选矿厂的生产和管理具有重要意义。为解决选矿厂生产过程中具有随机波动性的铁精矿品位预测问题,提出一种基于线性变换法的无偏灰色GM(1,1)铁精矿品位预测模型。通过采用一种线性变换方法降低铁精矿品位数据序列的波动干扰,将随机波动数据序列转换为单调增长的数据序列,然后将变换后铁精矿品位数据序列代入无偏灰色GM(1,1)模型以实现铁精矿品位预测模型的建模,最后将该预测模型用于两组铁精矿品位数据序列进行了验证。结果表明,基于线性变换的无偏灰色GM(1,1)铁精矿品位预测模型在预测精度和预测性能上优于两个改进的GM(1,1)预测模型,其预测精度均为一级,预测的最小相对误差为0.2%,平均绝对误差均小于1%,模型具有较好的应用性和有效性,为短期预测铁精矿品位提供了一种新途径。It is important to predict accurately iron concentrate grade for production and management of ore dressing plant in iron mine.In order to solve the prediction problem of iron concentrate grade with random and fluctuation in the ore dressing plant,a prediction model for iron concentrate grade based on the linear transformation method and the unbiased grey GM(1,1)method was proposed.The linear transformation method is firstly introduced for data processing,which reduces fluctuation interference of data sequence of iron concentrate grade and directly transforms it into monotone increasing data sequence.In order to establish the prediction model of iron concentrate grade,then the transformed data are put into the unbiased grey GM(1,1)method.Finally,the proposed model validation is carried out on the two sets of iron concentrate grade historical data sequence.The results show that the prediction accuracy and prediction performance of the prediction model for iron concentrate grade based on the linear transformation method and the unbiased grey GM(1,1)method are better than those of two improved GM(1,1)models,and both of prediction accuracies of the proposed model are evaluated as the 1st level,the minimum prediction relative error is only 0.2%,the mean absolute percentage error is below 1%.Some useful conclusions are drawn that the applicability and validity of the proposed model is superior to that of other two comparison prediction models,and it provides a new method for short-term prediction of iron concentrate grade.

关 键 词：铁精矿品位 无偏灰色GM(1 1) 线性变换 波动数据序列 预测模型 

分 类 号：TD926[矿业工程—选矿]