基于NNG-LSSVM的铜精矿品位软测量方法研究  

Research on Soft Sensor Method of Copper Concentrate Grade Based on NNG-LSSVM

作　　者：吴浩 潘冰清 杨会琳 孙凯 WU Hao;PAN Bingqing;YANG Huilin;SUN Kai(School of Electrical Engineering and Automation,Qilu University of Technology(Shandong Academy of Sciexces),Ji'nan 250353,China;State Key Laboratory of Automatic Control Technology for Mining and Metallurgy Process,Beijing 102628,China;Beijing Key Laboratory of Automatic Control Technology for Mining and Metallurgy Process,Beijing 102628,China)

机构地区：[1]齐鲁工业大学(山东省科学院)电气工程与自动化学院,济南250353 [2]矿冶过程自动控制技术国家重点实验室,北京102628 [3]矿冶过程自动控制技术北京市重点实验室,北京102628

出　　处：《有色金属（选矿部分）》2021年第3期88-92,共5页Nonferrous Metals（Mineral Processing Section）

基　　金：矿冶过程自动控制技术国家(北京市)重点实验室开放研究基金资助项目(BGRIMM-KZSKL-2018-01);国家自然科学基金资助项目(61603203)。

摘　　要：浮选技术是当今铜矿选矿最主要的方法并得到了广泛应用。浮选流程中铜精矿品位决定了最终产品的质量,是整个过程的关键变量。然而在实际生产中,该参数的测量耗时较长,难以实时在线测量。提出了一种基于非负绞杀(Nonnegative garrote)与最小二乘支持向量机(Least Squares Support Vector Machine)的软测量方法,并利用DCS系统提供的实际生产数据对该变量进行预测建模。仿真结果表明,所研究的软测量方法能够准确预测铜精矿品位的变化,能很好地实现精矿品位的实时预测及估计,并且在模型精度上明显优于其他软测量方法。Flotation technology is the most primary method of beneficiation of copper mines and has been widely used.In the flotation process,the grade of copper concentrate determines the quality of the final product,therefore it’s a key variable in the entire process.Nevertheless,in the actual production,the measurement of this parameter takes a long time,making it difficult to measure in real time online.The paper proposes a soft-sensing method based on non-negative garrote and least squares support vector machine,and uses the actual production data provided by the DCS system to predict and model this variable.The simulation results show that the researched soft-sensing method can accurately predict the change of copper concentrate grade,and can well realize the real-time prediction and estimation of concentrate grade,and the accuracy of model obviously superiors to other soft-sensing methods.

关 键 词：LSSVM NNG 铜矿浮选 软测量 铜精矿品位 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]