基于数据的化工过程无模型自适应即时学习控制方法研究

项目名称： 基于数据的化工过程无模型自适应即时学习控制方法研究

项目编号： No.21306234

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 化学工业

项目作者： 杨鑫

作者单位： 重庆理工大学

项目金额： 24万元

中文摘要： 复杂化工过程由于具有多变量、时滞和非线性等特性而难于对其建立机理模型并据此进行控制器设计；同时，化工生产过程中存在大量的过程数据，而这些数据包含的信息却很少得到利用。针对以上问题，本项目发展一种新的基于数据的无模型自适应即时学习控制器设计方法。基于化工过程的输入输出数据，抛弃过程建模步骤，直接用即时学习(JITL)方法设计控制器，重点研究控制器设计策略及基于即时学习技术的控制器设计方法。通过本项目的实施构建一套针对复杂化工过程的控制系统，并通过合理设计的数值算例验证控制系统的控制性能，最终形成一套稳定可靠的控制系统软件包。本项目拟发展的控制器设计方法，既保留了JITL方法的自适应特性，又摆脱了对化工过程数学模型的依赖，有望为复杂化工过程的无模型数据控制研究提供一种新的思路。

中文关键词： 基于数据的控制；无模型控制；自适应控制；即时学习；过程控制

英文摘要： It is not a trivial task to model most chemical processes since they are multivariable, time-delay, and nonlinear in nature. And the model-based control system is also hard to obtain.At the same time, many process data are routinely measured and automatically recorded in historical databases during chemical operation. Despite the large quantities of data available, the amount of online information on critical quality variables is usually rather low. To overcome the above difficulties, a new data-based model free and adaptive Just-in-Time Learning (JITL) process controller design method will be proposed in this research project. The JITL method will be directly used in adaptive controller design based on the process I/O data, without resorting to modeling of the process. The main focus of the present research includes controller design strategy study and controller design method study with JITL technique.A control system will be built and validated by reasonabley designed numerical examples on its control peformance.A control system software will be developed as the final goal.The controller design method not only inherits the adaptability of JITL but also is independent of the mathematicl model of chemical process and thus may provide a new way for the model free data-based control study.

英文关键词： Data-based Control；Model-free Control；Adaptive Control；Just-in-Time Learning (JITL)；Process Control