【作者】 薄翠梅；

【导师】 张湜； 蔡宁；

【作者基本信息】 南京工业大学 ， 控制理论与控制工程， 2003， 硕士

【摘要】 本文从工程应用角度，针对扬子石化丁二烯工业精馏装置的非线性、时变性和强耦合等特点，研究了软测量技术和先进控制策略，设计了基于软测量模型的先进控制系统，从而解决了与产品质量密切相关的重要过程变量无法在线实时测量和实时优化控制的问题，提高了产品收率、降低了能耗。全文共六章，可分为两大部分：第一部分研究了软测量技术及其工程设计；第二部分研究了先进控制系统理论及其在精馏装置上的工程应用。每一部分均从理论分析、仿真研究和现场应用三个层次展开论述。论文的主要内容如下： 1．概述了软测量技术和先进过程控制的发展。 2．深入研究了软测量技术的核心理论，即利用多种软测量建模方法(机理分析、多变量回归分析和神经网络)建立工业对象精确可靠的软测量模型，并提出了软测量工程设计步骤。 3．利用机理分析和仿真实验选择了软测量模型的辅助变量；设计了相应的数据采集系统；通过对采集的海量数据进行技术预处理，确立了软测量模型样本集。 4．利用机理方法、回归分析和神经网络建立了多个软测量模型，并通过仿真研究，对各模型的精度和外推能力进行了比较，确定了4个工业软测量模型。 5．在Delta V DCS上通过CL语言编程实施4个工业软测量模型，实现了软测量仪表的在线估计和实时显示。 6．设计了软测量校正模块，并进行了仿真研究。通过校正模块在线滚动优化，模型获得了可靠的在线估计精度，为先进控制系统的实施奠定了基础。 7．研究了推断控制、解耦控制、专家控制、节能控制等先进过程控制的相关理论。 8．针对丁二烯精馏装置实际存在的控制问题，设计了一套包括四个子系统的先进控制方案。重点讨论了其中两个子系统的设计，包括被控对象的选择、先进控制策略分析与设计、仿真实验、投运和调优。实现了成分的闭环控制，并取得可观的经济效益。 9．在先进控制系统设计中，通过增设逻辑控制模块、专家控制模块和模糊控制模块，最终构成了基于推断控制的复合控制策略。该策略既提高了控制系统的鲁棒性，又增强了推断控制系统的功能和工程实用性。更多还原

【Abstract】 From the engineering application aspect, aiming at the nonlinear, time-varying and strong coupling characteristic of butadiene distillation column, in this paper, the soft-sensor and the advanced process control (in short, APC) theory are studied. An advanced inferential control system based on soft-sensor are designed, which has solved the problem of the important process variables measured and controlled. The paper includes six chapters, which is classed to two parts. In the one part, soft sensor technology and engineering design are studied. In the other part, APC theory and application to the distillation operation are researched. The every part is discussed from the theory study to simulation analysis and engineering application. The main content of the paper is as follows:1. Summarize Soft sensor technology and the development of APC.2. Study the theory of soft sensor. Build the reliable soft sensor model of the industry object by many techniques, such as mechanism analysis, multivariate regress and neural network. At the same time, the soft-sensor engineering designed processes are given.3. The assistant variables of soft sensor model are selected with mechanism analysis and simulation; data acquisition system is designed correctly; the model sample sets are built.4. Build the soft sensor models by mechanism analysis, multivariate regress and neural network, and compare the smoothness and extrapolation of the models through simulation, and confirm four industrial models.5. Realize four industrial models on the DCS through CL programming; get the on-line estimation and real-time display of soft instruments.6. The correcting module is designed and simulated. After the model is corrected on-line, the models get the reliable estimation precision.7. The correlative theories of APC are studied, such as inferential control, decoupling control, expert control, saving energy control.8. An advanced control scheme including four subsystems is designed, aiming at the real control problems of butadiene distillation column. The two important subsystems are mainly discussed here, which included control object selecting, strategy analyzing, simulating and running. It realizes the component close-loop control, and gains the economic benefit.9. Logic control module and expert control module are added in the scheme. Which formed the hybrid control based on the inferential control. Thescheme strengthens the robustness of the control system, and improves the engineering application of the inferential control system.更多还原