小波时间序列分析在铁矿品位波动预测中的应用  

Application of Wavelet Analysis of Time Series in Prediction of Iron Ore Grade Variation

作　　者：应海松[1] 王于鹤 陈颖娜[1] 李斐真[3] 余春晖[1] 李雪莲[1] 

机构地区：[1]北仑出入境检验检疫局,浙江宁波315800 [2]中国社会科学院经济研究所,北京100836 [3]宁波大学,浙江宁波315211

出　　处：《矿冶工程》2017年第1期96-99,102,共5页Mining and Metallurgical Engineering

基　　金：国家质检总局计划科研项目(2012IK045)

摘　　要：利用时间序列预测,针对某进口铁矿石卸载时的在线粒度检测结果,应用MATLAB语言,分别采用自回归移动预测模型、特征点聚类法、双窗口贝叶斯后验检验法3种模型,利用小波分解的高通软阈值和相关特征参数预测其品位波动情况,然后根据预测的品位波动情况按相关标准规定制订取样方案。该方法简单易用,可以节省大量的人力财力。Based on the on-stream particle-size analysis results of some imported iron ore during unloading,variation in the grade of iron ore was predicted by using time series prediction technofogy and MATLAB through three models （autoregressive integrated moving average model, feature points clustering method and the double window Bayesian posterior estimation ）,respectively, based on the high-pass soft-threshofd and related parameters of wavelet decomposition. This prediction was then taken as a basis for developing a sampling scheme. This method,being simple and easy to operate,can save both manpower and material resources.

关 键 词：小波分析 时间序列 铁矿 品位波动 预测 

分 类 号：TD926.3[矿业工程—选矿]