复杂工业过程运行指标闭环优化方法研究

Supported by the National Science Fund for Young Scholar, the structure and design approaches of closed-loop optimal decision-making for technical indices under uncertainty environment have been proposed, which include the predictive model of the production indices, multi-objective optimization of technical indices, priori- and posteriori-evaluation, and dynamic tuning. The platform for simulation research of the proposed approach has been developed. The validations have been carried out by simulation and industrial application, and the results show the effectiveness of the proposed approach. There are 15 papers have been published or accepted in the international journals and conferences, where 5 papers have been published or accepted by international journals indexed by SCI and 2 papers have been published by IEEE Transactions as Regular paper. Also, 12 papers are indexed by EI.. During the research, some new problems reserved to be studied in depth have come forth. Therefore, this project focuses on these problems and looks at: 1) The mathematical description of the statistical characteristics, such as the mean value, variance and probability distribution, etc., between the operational indices and the production indices. 2) Combining the Case-based reasoning and multi-group strategy, elitism reservation, prediction and mapping mechanism, propose the fast and efficient dynamic multi-objective optimization algorithm, at the same time, carry out the performance analysis. 3) The rule mining approach of dynamic tuning will be studied for the operational data with the characteristics of discord, deficiency, unlabelled and multiple sampling periods, etc. 4) Construct the closed-loop optimization system for operational indices and carry out the performance analysis for the closed-loop optimization system. Meanwhile, 5) The validations through the simulation and industrial application will be performed.

在青年基金项目的支持下，提出了动态环境下工艺指标闭环优化决策结构和设计方法，包括全流程生产指标预报模型建模、工艺指标多目标优化、指标前验/后验评估与动态校正方法等；研制了仿真实验平台，并结合选矿过程开展仿真验证和工业应用验证研究，结果表明了所提结构与方法有效性。在国际期刊和会议上发表或录用论文15篇，其中国际SCI期刊论文5篇，第一作者IEEE长文2篇，EI检索12篇。. 在上述研究过程中，凝练出新的需解决的关键科学问题。针对这些问题，本项目研究各个工序运行指标与全流程生产指标间的均值、方差和概率分布等统计特性的数学描述；将案例推理与多种群和多目标优化的精英保留策略、预测映射机制结合，研究动态多目标优化算法及其理论分析；研究针对不一致、不完备、标签缺失、指标采样周期与速率不同的过程运行数据的动态校正规则挖掘方法；构造运行指标闭环优化系统，进行系统性能分析，并进行仿真与工业应用验证研究。

复杂工业过程运行指标的优化对于实现企业全局优化具有重要的意义。本项目针对复杂工业生产全流程的多工序、多层次、多时间和空间尺度特征，提出了实现综合生产指标优化为目标的复杂工业过程的全流程集成优化策略。针对复杂工业过程运行指标的优化问题，在前期研究基础上，将多目标静态优化方法与综合生产指标预报、运行指标的前验/后验评估与动态校正相结合，提出完善了由运行指标初值优化、综合生产指标预报、基于综合生产指标预报值/实际值的指标前验/后验评估与动态校正组成的运行指标多目标动态闭环优化结构和算法。研发了基于智能移动终端和云计算的运行指标闭环优化系统实验验证与性能分析平台，并结合选矿生产全流程开展了实验验证和工业应用研究，取得显著的应用效果。.研究成果发表论文24篇，其中被SCI检索14篇，EI检索23篇。在国际著名期刊IEEE 汇刊和IFAC会刊发表8篇。1篇论文2014年获国际自动控制联合会IFAC会刊Control Engineering Practice期刊2011-2013年度最佳论文奖（国内单位和国内学者首次入选）。1篇论文2013年获国际会议ICAMechS最佳论文奖。在国际期刊组织两个专辑。申请国际发明专利PCT和美国专利1项并已获得卢森堡授权。授权国家发明专利10项，申请6项。研究成果获国家技术发明二等奖1项（第二完成人）。.形成了一支基础研究与应用基础研究密切结合的从事复杂工业过程运行优化控制研究的团队。1人获第十四届中国青年科技奖，获国家杰出青年科学基金资助、入选2016年度教育部长江学者特聘教授（已公示），入选IEEE Senior Member。培养研究生18人，其中毕业博士1人，毕业硕士8人。其中获东北大学优秀硕士论文1篇（正在申请辽宁省优秀硕士论文）。