基于选择性集成算法的浸出率混合预测模型  

Hybrid prediction modeling of leaching rate based on selected ensemble algorithm

作　　者：胡广浩[1] 毛志忠[2] 杨菲[2] 

机构地区：[1]沈阳大学信息工程学院,沈阳110044 [2]东北大学信息科学与工程学院,沈阳110004

出　　处：《仪器仪表学报》2013年第5期1049-1053,共5页Chinese Journal of Scientific Instrument

摘　　要：一个精确的模型对浸出过程中浸出率预测是十分重要的。针对湿法冶金浸出生产过程中浸出率在线检测的难点,提出一种有效的混合建模方法,建立浸出率的预测模型。在对浸出过程进行分析后,建立了一个浸出过程机理模型。由于机理模型与实际之间存在着较大的误差,因此建立了混合模型来减少误差。随后,针对小样本建模问题,提出了基于二进制PSO算法的选择性bagging集成算法,并将该算法应用于混合模型的误差补偿中去。实验结果表明该混合模型的预测精度比其他模型的预测精度高。An accurate mathematical model of leaching process is important for the prediction of leaching rate. Aiming at the difficulties of on-line measuring leaching rate in hydrometallurgy leaching process, an effective hybrid modeling method is proposed to build the prediction model of leaching rate. After analyzing the leaching process, a mechanism model is proposed to describe and analyze the heat-stirring-acid leaching process. Because there are large modeling errors between the mechanism model and actual system, a hybrid model is established to reduce the error. Afterwards, a new selective bagging ensemble algorithm based on binary PSO algorithm is proposed aiming at the problem of small sample regression. This algorithm is applied to the error compensation of the proposed hybrid model. Experiment re- suhs indicate that the proposed hybrid model has better prediction performance than other models.

关 键 词：浸出过程 预测 混合模型 选择性bagging集成算法 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]