双层结构预测控制系统全流程协调与经济性能优化研究

Two-Layer predictive control systems are widely used in industrial processes and are very important for the steady operation and cost reduction of process industries. Due to the hierarchical strucutre of predictive control systems on different process units, the cooperation and coordination between them are rare which results the restriction of the implementation of plant-wide optimization objectives of process industry. Combining the concept of distributed control, in three levels, this project studies different coordination strategies for the improvement of global economic performance of process industry: With repect to the ordinary connection structure of process sub-systems, this project exploits the reduced and robust coordination strategy under model uncertainties. By establishing the variation restraining model, the influence of dynamic-control-layer coordination over the backoff strategy and performance of steady-optimization-layer will be analyzed. Moreover, stimutaneous coordination of two layers of different predictve control systems are studied. The coupling mechanisim of optimization objectives of different sub-systems are studied and the reconstruction method of control objectives is proposed. The research findings will be tested and validated on aromatic hydrocarbon extraction process. This project will deliver plant-wide control system coordination and performance assessment algorithms and software for process industries, which help the improvement of application and maintainance of industrial advanced control systems and have great significance of theory and application.

双层结构预测控制系统在工业过程中应用广泛，是流程工业装置级平稳运行、节能降耗的重要手段。但由于流程工业不同子装置预测控制系统普遍采用递阶结构，彼此间缺乏合作与协调，非合作博弈的结果制约了预测控制系统在全流程层面实现优化运行目标。本项目结合分布式控制思想，分三层逐次研究不同协调策略，实现对全流程经济性能的改进：针对流程工业常见子系统拓扑结构，探索预测控制系统稳态层简约协调技术，研究模型不确定性下稳态层协调方法；通过建立子系统回路波动约束模型，采用随机动态规划，分析动态控制层协调对稳态层退避机制及性能的影响关系，研究双层同步协调机制；研究合作博弈下子系统预测控制优化目标耦合机理，给出面向全流程优化的双层预测控制目标重构机制。最后通过芳烃联合装置应用和试点，研究出全流程预测控制协调与经济性能优化算法与软件，提高流程工业先进控制系统应用维护水平，保障全流程优化运行，具有重要理论意义与应用价值。

双层结构预测控制系统在工业过程中应用广泛，是流程工业装置级平稳运行、节能降耗的重要手段。但由于流程工业不同子装置预测控制系统普遍采用递阶结构，彼此间缺乏合作与协调，非合作博弈的结果制约了预测控制系统在全流程层面实现优化运行目标。本项目结合分布式控制思想，逐次研究不同协调策略，实现对全流程经济性能的改进。本项目实施期间，项目组共出版中文专著1本；发表、录用论文14篇，其中录用SCI期刊论文12篇，包括TOP期刊Automatica, Journal of Process Control等杂志；提交专利申请6项，软件著作权授权1项；参加国际会议3人次，国内会议4人次；培养硕士生2人，博士生4人。