核极限学习机在浮选回收率中的研究与应用  

Research and application of kernel extreme learning machine in flotation recovery rate

作　　者：王欢[1] 徐鑫[1] 鲁鹏云 张军[1] 彭文娟[2,3] 

机构地区：[1]鞍钢集团矿业公司,辽宁鞍山114001 [2]北京科技大学计算机与通信工程学院,北京100083 [3]材料领域知识工程北京市重点实验室,北京100083

出　　处：《中国矿业》2016年第7期118-124,共7页China Mining Magazine

摘　　要：浮选回收率是浮选过程中重要的生产指标。需要通过人工检测得到的浮选回收率,可知性具有较大的时间延迟,使工人不能及时有效地对生产做出相应控制调整。由于浮选过程相当复杂,变量维数高、关联性强、噪声大、检测信号不完备等因素,难以建立较精确的回收率预测模型。然而,人工智能与机器学习技术能在机理不清楚、信息不完备的情况下,对复杂系统建立基于数据驱动的经验模型。因此,本文为提高回收率检测的及时性、有效性,在分析浮选过程相关因素影响的基础上,提出基于核极限学习机建立浮选回收率的预测模型。仿真实验结果表明,该建模方法可有效辨识浮选过程中,输入数据与回收率测量值之间的非线性关系,且具有更高的预测精度与训练性能。The flotation recovery rate is an important index in the process of flotation.The flotation recovery rate is obtained by manual detection,which has a large time delay,so that workers can not effectively control the production to make the corresponding adjustment.Due to the complexity of the flotation process,the high variable dimension,strong correlation,large noise and incomplete detection signal,it is difficult to establish a more accurate prediction model of recovery rate.However,artificial intelligence and machine learning technology can establish based on data driven model of complex system in the case of unknown mechanism and incomplete information.Therefore,in order to improve the efficiency and effectiveness of the detection of the recovery rate,this paper proposes a prediction model based on the establishment of the flotation recovery rate based on the analysis of the factors affecting the flotation process.The simulation results show that the proposed method can effectively identify the nonlinear relationship between the input data and the recovery rate,and has higher prediction accuracy and training performance.

关 键 词：核极限学习机 浮选回收率 人工智能 预测模型 

分 类 号：TD951[矿业工程—选矿]