基于数据驱动的浮选指标预测建模研究进展  

Research progress of flotation index prediction modeling based on data driven

作　　者：聂善煜 何桂春 石岩[1,2] 赵红宇 吴为波 NIE Shan-yu;HE Gui-chun;SHI Yan;ZHAO Hong-yu;WU Wei-bo(Jiangxi Province Key Laboratory of Mining Engineering,Ganzhou 341000,China;School of Resources and Environmental Engineering,Jiangxi University of Science and Technology,Ganzhou 341000,China)

机构地区：[1]江西省矿业工程重点实验室,赣州341000 [2]江西理工大学资源与环境工程学院,赣州341000

出　　处：《中国有色金属学报》2023年第7期2330-2338,共9页The Chinese Journal of Nonferrous Metals

基　　金：国家自然科学基金资助项目(52174249)。

摘　　要：针对浮选过程工业大数据具有高维非线性、动态时序性以及参数强耦合性等复杂特性,若使用单一模型进行浮选指标预测,则可能由于随机性的影响而导致模型的可信度低、泛化性差,预测效果不理想。为了突破单模型的瓶颈,本文对30多年来的基于数据驱动的浮选指标预测建模方法的研究文献进行了系统的评述,认为采用不同学习算法对浮选过程的多工况参数、泡沫图像的潜在内部特征进行深度挖掘,以这些工况参数特征和泡沫图像的特征作为输入,对浮选指标进行集成预测建模,可以面向选矿过程生产指标的智能预测和工况参数的智能协同优化,形成控制方法与策略,为解决给矿性质复杂、工况参数多变的浮选过程指标预测和工况识别的难题提供理论指导建议。In view of the complex characteristics of industrial big data in flotation process,such as high-dimensional nonlinearity,dynamic time series and strong parameter coupling,when a single model is used to predict flotation indexes,it may lead to low credibility and poor generalization of the model due to the influence of randomness,and the prediction effect is not ideal.In order to break through the bottleneck of single model,this paper systematically reviewed the research literature on data-driven flotation index prediction modeling methods for more than 30 years.It is considered that different learning algorithms should be used to deeply mine the multi-mode parameters of the flotation process and the potential internal characteristics of the foam image,and the characteristics of these working condition parameters and the characteristics of the foam images are taken as inputs to carry out integrated prediction and modeling of the flotation index.It can be oriented to the intelligent prediction of production indexes and the intelligent cooperation of working conditions parameters in mineral processing,and form an optimal control method and strategy,which provides theoretical guidance and suggestions for solving the problems of index prediction and working conditions identification in flotation process with complex feeding properties and changeable working conditions parameters.

关 键 词：浮选 浮选泡沫 集成学习 浮选指标预测 工况参数协同优化 

分 类 号：TD923[矿业工程—选矿]