文献类型：期刊文献

中文题名：锑浮选过程加药量自适应迭代学习控制

英文题名：Adaptive iterative learning reagents control for antimony flotation process

作者：李中美[1];黄梦哲[2];桂卫华[3]

机构：[1]华东理工大学信息科学与工程学院,上海200237;[2]纽约大学工学院,美国纽约11201;[3]中南大学自动化学院,湖南长沙410083

年份：2020

卷号：37

期号：10

起止页码：2123

中文期刊名：控制理论与应用

外文期刊名：Control Theory & Applications

收录：CSTPCD;;EI(收录号：20204709517127);Scopus;北大核心:【北大核心2017】；CSCD:【CSCD2019_2020】；

基金：国家重点研发计划项目(2018YFB1701103);国家自然科学基金项目(61890930-3);国家杰出青年科学基金项目(61925305);上海市青年科技英才扬帆计划(20YF1411000);上海市自然科学基金项目(17ZR1406800)资助.

语种：中文

中文关键词：浮选过程;自适应控制;最优控制;抑制扰动

外文关键词：flotation process;adaptive control;optimal control;disturbance rejection

摘要：针对现有的加药量控制方法需要浮选过程动态模型或是鲁棒性不足的问题,提出一种数据驱动的浮选过程加药量自适应迭代学习控制方法.首先,将药剂量控制问题转化为两级优化问题(问题1和问题2).其中,基于前馈控制原理求解问题1得到前馈补偿分量以抑制外界扰动.然后,采用基于值迭代的ADP算法,仅通过工业现场工业运行数据求解问题2以得到最优反馈增益,从而设计一个最优的加药量控制策略使最终的生产指标(精矿品位和尾矿品位)跟踪给定值,且药剂量消耗最少.最后,通过工业生产数据进行仿真验证,证明所提方法的收敛性和稳定性.
In order to solve the problem that the existing control methods require dynamic model of flotation process or lack of robustness,an adaptive iterative learning reagents control scheme for flotation processes is proposed.First,the flotation reagents control problem is formulated as a two-stage optimization problem(problem 1 and problem 2).Specifically,the feed-forward compensation component is obtained by solving the problem 1 based on feed-forward control principle,which can be used in disturbance rejection.After that,a value-iteration based ADP algorithm is applied to deal with the problem 2 in order to derive the optimal feedback gain matrix by employing the online production data.Thus,an optimal reagents control strategy is designed to force the flotation indexes(concentrate grade and tailing grade)to track the desired values,and keep the reagents consumption to a minimum.In the end,the convergence and stability of the proposed data-driven method are proved by the simulation with industrial data.