基于双层结构的预测控制系统分析与应用
文献类型：学位论文
| 作者 | 潘昊
|
| 学位类别 | 博士 |
| 答辩日期 | 2015-12-06 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 授予地点 | 中国科学院沈阳自动化研究所 |
| 导师 | 于海斌 ; 邹涛 |
| 关键词 | 模型预测控制 双层结构预测控制 快速预测控制 分壁精馏塔 定制化设计 病态模型 |
| 其他题名 | Analysis and Application of Model Predictive Control Based on Two-layer Structure |
| 学位专业 | 机械电子工程 |
| 中文摘要 | 本文基于预测控制的双层结构原理从多变量系统的输入输出稳态值出发，针对一类多变量相关且开环稳定的线性MPC系统，通过系统的设定值可以直接一步计算出系统控制输入的稳态增量，避免了传统预测控制需要花费大量时间运算矩阵和求解大规模数学规划问题。进一步地，将双层结构预测控制中的稳态优化求解算法分解为离线计算和在线查表两部分，离线制表存储和在线查表更新结合进行，数值寻优的时间开销非常小，可以胜任高速对象控制的实时性要求。热力学计算显示分壁精馏塔比常规精馏塔具有降低三分之一能耗的潜力，但目前还无法成为化学工业的标准设计，内在的传质传热动力学行为认识不清晰、缺乏可靠的控制和操作技术支持等，都是限制分壁精馏塔大规模应用的主要因素。本文关于分壁精馏塔的多变量MPC系统设计方案、集成的控制和优化操作（MPC＋RTO）策略、基于快速预测控制算法实现不同工艺目标的平稳切换等研究工作，都是支持这类新型高效分离设备实现最佳操作目标的最新成果。伴随工艺设备的不断完善，对推进分壁精馏塔的大规模工业应用有促进作用。许多工业控制对象的模型可以用一阶或二阶时滞＋纯滞后的传递函数来表达，通过模型中的三类参数即增益、时间常数、纯滞后的不同组合，可以表达的模型范围非常广泛。现场工程师依据这些模型参数值，就大致可推断出闭环控制可实现的最佳响应特性。特别是当给定了预期的闭环响应特性后，可以直接得到预测控制作用计算的解析表达式，在线计算无需迭代极为快捷。依据这样的快速预测控制算法，再借助快速发展的嵌入式微控制器软硬件技术，为具体被控对象量身定制最佳控制规律成为可能。这样的定制化控制设计在可靠性、运行维护等方面优势显著，代表了未来先进控制应用技术发展的重要方向。本文开发了一套支持设计、仿真验证、实施的一体化系统平台，在试验装置上成功实现了定制化控制目标。 |
| 英文摘要 | Based on the steady-state inputs and outputs of multivariable two-layered MPC system, if for a class of multivariable coupled plants that is assumed the open-loop stable the economic objective function can be expressed as a linear relationship, the optimal incremental of control inputs can be directly obtained through the one-step calculation by our approach, therefore a lot of time to solve large-scale matrix computing and mathematical programming in traditional MPC are saved. Also by decomposing the optimal algorithm into two parts, that is offline calculation and online update, and further combining the offline tabulate storage and online look-up update, which making on-time numerical optimization time cost very small, so it can meet real-time requirements of high-speed object control. The dividing wall distillation column (DWC) has potential to reduce one third energy consumption compared to the conventional distillation columns in terms of thermodynamic calculations, however, DWC is still unable to become the standard design of chemical industry since the inherent behavior understanding of heat and mass transfer kinetics is not clear and the technical support of reliable control and operation is not enough, which large-scale applications of DWC are limited by above main factors. For multivariable MPC system design of DWC, the strategy integrated control and optimal operation based on fast MPC, is suggested to achieve smooth switching in different operating target. Our contributions will support these new high-efficient separation equipment to realize the desired optimal operation. With the continuous improvement of process equipment, DWC will be promoted to large-scale industrial applications. Many industrial plant models can be expressed as the transfer function of a first order lag plus delay or second order lag plus delay, which is called process model. In fact, many models can be identified by different combinations of gain and time constants and delay. Optimal response characteristics of closed loop control based on these model parameters can generally inferred by field engineers. Especially when the expected closed-loop response is given, the analytical expression of predictive control can be obtained directly, thus online calculation without iterative is extremely fast. Based on such a fast predictive control algorithm and with the rapid development of embedded microcontroller hardware and software technologies, the implementation of optimal control law based on custom-made for the controlled plant is possible. Such customized control design has significant advantages in reliability, operation and maintenance. It is an important direction for future advanced control technology development and applications. An integrated platform supporting the design, simulation, implement of fast MPC systems is developed. Such our customized control system is successfully applied to control the level of a pilot-scale two tank. |
| 语种 | 中文 |
| 公开日期 | 2015-12-27 |
| 产权排序 | 1 |
| 页码 | 118页 |
| 源URL | [http://ir.sia.ac.cn/handle/173321/17529]  |
| 专题 | 沈阳自动化研究所_信息服务与智能控制技术研究室 |
| 推荐引用方式 GB/T 7714 | 潘昊. 基于双层结构的预测控制系统分析与应用[D]. 中国科学院沈阳自动化研究所. 中国科学院沈阳自动化研究所. 2015. |
入库方式： OAI收割
来源：沈阳自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。