基于气泡尺寸动态PDF的铜粗选药剂添加量评估应用研究  

Study on application of copper rougher flotation reagent-addition assessment based on dynamic bubble size PDF

作　　者：谭珍[1] 王伟[1] 唐朝晖[1] 刘金平[1] 

机构地区：[1]中南大学信息科学与工程学院,长沙410083

出　　处：《计算机应用研究》2015年第1期138-141,174,共5页Application Research of Computers

基　　金：国家自然科学基金重点资助项目(61134006);国家自然科学基金面上资助项目(61071176;61272337);国家创新研究群体科学基金项目(61321003)

摘　　要：在浮选生产中,泡沫表面的气泡尺寸分布特征与浮选工况紧密相关。针对浮选药剂添加量变化会动态改变气泡尺寸分布的特点,提出一种基于气泡尺寸动态概率密度函数统计特征的自动评估铜粗选药剂添加量的方法。该方法利用B样条基函数方法获得PDF权值向量以表征气泡尺寸的分布特征,然后采用层次聚类方法中的最远邻聚类和模糊C-均值聚类相结合的方法对各典型药剂添加量下气泡尺寸分布的权值向量进行特征学习,最后利用贝叶斯推理方法获得药剂添加量状态的评估结果。工业现场应用结果表明,该方法能够实时获得气泡尺寸概率密度函数,有效地实现药剂添加量状态的自动评估。In the flotation process, the surface bubble size distribution is closely related to the flotation operation conditions. As the variation of reagent-addition could dynamically change the bubble size distribution, this paper proposed an automatic evaluation method for the reagent-addition operation in the copper rougher flotation based on the dynamic probability density function of bubble sizes. Firstly, it utilized the B-spline basis function method to obtain the PDF （ probability density function） weight vectors representing the distribution features of bubble sizes. The clustering analysis method based on the combination of the farthest neighbor clustering and fuzzy C-means clustering was subsequently employed to learn the weight vector features of bubble size distribution under typical dosage-addition conditions. Finally, it inferred the reagent-addition operation status by Bayesian inference. The industrial application results demonstrate that this method can obtain real-time the probability density function of bubble sizes, and effectively achieve the automatic assessment of reagent-addition status.

关 键 词：泡沫浮选 B样条基函数 PDF权值向量 特征学习 药剂添加量评估 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]