摘要
稀土混合萃取溶液中各元素组分含量的在线软测量是优化连续萃取生产过程、确保产品高纯化的前提.现有软测量方法可独立求解单个稀土元素组分含量,但忽略了多元素组分含量间或组分含量与其它相关因素(如浓度)间的共性.本文为探索多稀土元素组分含量间及组分含量与浓度间的共性,将多任务学习方法用于稀土元素组分含量软测量中.首先,构建多任务深度神经网络,提高模型的泛化能力和鲁棒性.其次,提出基于多目标优化算法的稀土多元素组分含量预测方法,通过搜索Pareto最优以提升各任务的预测精度.经多组对比实验表明,该方法在多元素组分含量或多元素组分含量与浓度同时训练时性能最佳,能满足稀土元素组分含量在线检测的精确性和实时性.
Online soft measurement of the component content of each element in a mixed rare earth extraction solution is a prerequisite for optimizing the continuous extraction production process and ensuring high purity of the product.Existing soft measurement methods can solve for individual rare earth element fractions independently,but ignore the commonality between multi-element fractions or between fractions and other relevant factors(e.g.concentration).A multi-task learning approach is used to explore the commonality between the component content of multiple rare earth elements and between the component content and concentration in soft measurements of rare earth elements.Firstly,a multi-task deep neural network is constructed to improve the generalization ability and robustness of the model.Secondly,a multi-objective optimization algorithm is proposed to improve the prediction accuracy of each task by searching the Pareto optimum.After several sets of comparison experimental results,it is shown that the method has the best performance when the multielement component content or multi-element component content and concentration are trained at the same time,which can meet the accuracy and real-time performance of online detection of rare earth elemental component content.
作者
张水平
张奇涵
王碧
ZHANG Shui-ping;ZHANG Qi-han;WANG Bi(School of Information Engineering,Jiangxi University of Science and Technology,Ganzhou Jiangxi 341000,China)
出处
《控制理论与应用》
EI
CAS
CSCD
北大核心
2024年第3期454-467,共14页
Control Theory & Applications
基金
江西理工大学博士科研启动基金项目(2022205200100595)
国家自然科学基金委员会项目(72261018)
江西省教育厅青年项目(GJJ2200868)资助.
关键词
稀土萃取
组分含量
多任务学习
多目标优化
机器学习
深度学习
帕累托
rare earth extraction
component content
multi-task learning
multi-objective optimization
machine learning
deep learning
Pareto
