摘要： 针对稀土萃取液中有颜色特征和无颜色特征的离子在共存工况下组分含量难以快速检测的问题，提出一种基于遗传算法（GA）-极限学习机（ELM）的多组分含量预测方法。确定稀土萃取槽体混合溶液图像特性和描述图像信息的H、S颜色特征分量，利用ELM速度快、泛化能力强的优点，建立基于颜色特征的多组分含量模型，鉴于传统ELM模型初始权值和阈值的随机性易影响模型性能，使用GA对初始值进行优化确定。基于CePr/Nd萃取溶液样本数据的实验结果表明，与ELM、BP、LSSVM以及GA-BP、PSO-ELM等算法相比，该方法具有较高的预测精度且稳定性较好，可为稀土萃取现场快速获取多组分含量值提供技术支撑。

关键词: 稀土萃取, 多组分含量, 颜色特征, 遗传算法, 极限学习机模型

Abstract: To solve the problem that it is difficult to detect the component content quickly under the coexistence conditions of colored and non-colored ions in rare earth extracts,this paper proposes a multi-component content prediction method based on Genetic Algorithm(GA)-Extreme Learning Machine(ELM).The image characteristics of the mixed solution of rare earth extraction tank and the H and S color characteristic components for image information description are determined.On this basis,a multi-component content model based on color characteristics is established by utilizing the high speed and strong generalization of ELM.In view of the randomness of the initial weight and threshold of the traditional ELM model,which easily affects the performance,GA is used to optimize the initial value.Experimental results based on CePr/Nd extraction solution sample data show that compared with ELM,BP,LSSVM,GA-BP,PSO-ELM and other algorithms,this method has higher prediction accuracy and stability.It enables rapid on-site acquisition of the values of multi-component content of the rare earth extraction.

Key words: rare earth extraction, multi-component content, color feature, Genetic Algorithm(GA), Extreme Learning Machine(ELM) model

中图分类号: 

• TP391