基于控制器动态线性化的数据驱动控制方法及在精馏过程的应用

项目名称： 基于控制器动态线性化的数据驱动控制方法及在精馏过程的应用

项目编号： No.61503138

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 自动化技术、计算机技术

项目作者： 朱远明

作者单位： 华东理工大学

项目金额： 19万元

中文摘要： 精馏过程是石油化工生产的重要环节，该过程具有多变量、大时滞特点，且不同控制回路间相互关联、耦合严重，难以建立精确系统模型，给控制系统设计带来困难。数据驱动控制方法由于不依赖于被控对象精确数学模型，在模型不可用时，能通过利用系统输入输出数据直接进行控制器设计，有助于解决复杂化工过程中的控制问题。.本研究针对精馏过程控制问题中存在的影响因素多、耦合严重、建模困难等特点，研究基于控制器等价动态线性化的数据驱动自适应控制方案，包括直接利用系统测量数据进行控制器结构优化配置、控制器参数整定以及控制系统稳定性分析方法。同时，针对复杂系统四种基本联接结构，即串联、并联、反馈及互联联接，研究互联影响的估计方法，并设计相应的数据驱动控制方案，以提升控制效果。最后，初步建立面向复杂化工精馏过程的数据驱动控制框架，将该理论结果应用到实际化工生产环节中，验证并进一步改进该方法。

中文关键词： 无模型自适应控制；数据驱动控制；动态线性化；精馏控制

英文摘要： Distillation process is one of the important parts in petrochemical production. The main characteristics of the distillation process include multiple variables, large time delay, and strong coupling interaction, which would lead to difficulty in building accurate plant model and further result in controller design trouble. To avoid relying on the accurate plant model, data-driven control method has been developed, which could design the controller by I/O data directly. This kind of control method serves the purpose of solving aforementioned control problems in complex chemical process..In this research, a controller dynamic linearisation based data-driven control method will be developed for distillation process, whose primary features include multiple influence factors, strong coupling, and difficult modeling problem. The research efforts mainly compose of optimal configuration of controller structure, controller parameter self-tuning, and system stability analysis, which are all achieved by only using system measurement data. In addition, notice that for most of complex systems, they could be decomposed into several sub-loops, which are connected via four basic connection modes, including series connection, parallel connection, feedback connection, and interconnected connection. The relationships among sub-loops will be analyzed and utilized here for data-driven controller design to improve the control performance. At last, the proposed controller will be improved and enriched to build a data-driven control framework for complex distillation process, which will be further verified in both experimental platform and practical production process.

英文关键词： model free adaptive control;data-driven control;dynamic linearization;distillation control