稀土串级萃取分离过程元素组分含量的多模型软测量

针对稀土串级萃取分离过程中元素组分含量在线测量难的问题,提出了一种多模型软测量方法,用于在线预测元素的组分含量.首先,以物料平衡方程为基础,在多个工作点附近建立了描述萃取过程的局部线性模型.引入减法聚类算法对样本数据进行分类,用得到的分类数据对局部模型参数进行离线辨识.每一时刻根据积分性能指标选择最优模型,同时在线修正局部模型参数.利用某La,Ce,Pr,Nd 4组分串级萃取分离Ce/Pr生产线的实测数据进行了仿真研究,结果表明所提出的多模型方法有效、预测精度较高.

Soft-sensor of element component content based on multiple models for the rare earth cascade extraction process

To solve the on-line measurement problem for the rare earth cascade extraction process,a multiple models based on soft-sensor is proposed in this paper.Local linear models are obtained around multiple operating points,based on the material balance equations.By introducing the subtraction clustering algorithm,the sample data are classified and the local model parameters are identified off-line using the corresponding data set.At every instant the optimal model is chosen by a suitably defined performance index,and then the parameter of the local model is updated.The soft-sensor is conducted on a certain Ce/Pr extraction production line of La,Ce,Pr,Nd tetra-component system.By comparing with the measured data,the simulation results show the effectiveness and veracity of the soft-sensor.