Digital Control of Nonlinear Processes

摘要-克拉瓦里斯-9403432化学工业被表现出非线性的过程所主导。例如，聚合反应塔、高纯精馏塔、生物过程等都是高度非线性的过程。工业中基于数字计算机的非线性控制律的实现推动了非线性、离散时间过程控制方法的发展。所有离散时间控制算法的基石是连续时间、非线性动态系统不可避免的时间离散化，这是一个开放的问题。基于在许多工业控制问题中，某些测量是在低采样率下执行的事实，针对非线性、连续时间系统，特别是当采样周期变大时，开发可靠的离散化过程。例如，化学成分测量通常是在比温度测量低得多的采样率下进行的(例如，气相色谱)。这项研究的总体目标是开发方法来分析受慢采样影响的非线性化学过程的行为，并合成满足设计规范的稳定、健壮的数字控制系统。微分几何的概念框架将被用来：1.发展一种非线性、连续时间系统的时间离散化过程，特别是在采样周期变大的情况下。2.研究了采样和特殊离散化过程对非线性离散过程模型非最小相位特性的影响。3.开发了非线性多速率控制算法，该算法有效地利用了不频繁和频繁的测量，以便为所考虑的过程诱导出理想的闭环系统行为。4.设计离散时间的非线性状态估值器，用于重构不可测状态变量。5.通过在聚合反应器中的应用，对这些控制策略进行了实验验证。

Abstract - Kravaris - 9403432 The chemical industry is dominated by processes that exhibit nonlinearities. For example, polymerization reactors, high-purity distillation columns, bioprocesses, etc. are highly nonlinear processes. The digital computer-based implementation of nonlinear control laws in industry has motivated the development of nonlinear, discrete-time process control methods. The cornerstone of all the discrete-time control algorithms is the inevitable time-discretization of a continuous-time, nonlinear dynamic system, which is an open problem. The development of a reliable discretization procedure for nonlinear, continuous-time systems, especially when the sampling period becomes large, based on the fact that in many industrial control problems, certain measurements are performed at low sampling rates. For example, chemical composition measurements are usually performed under a much lower sampling rate (e.g. gas chromatography) compared to temperature measurements. The overall objective of this research is the development of methods for analyzing the behavior of nonlinear chemical processes subject to slow sampling and synthesizing stable and robust, digital control systems that meet design specifications. The conceptual framework of differential geometry will be use to: 1. Develop a time-discretization procedure for nonlinear, continuous-time systems, especially where the sampling period becomes large. 2. Study the effect of sampling and the particular discretization procedure upon the nonminimum-phase characteristics of a nonlinear, discrete-time process model. 3. Develop nonlinear multirate control algorithms which use the infrequent and frequent measurements effectively in order to induce a desirable closed-loop behavior for the process under consideration. 4. Design discrete-time, nonlinear state estimators which will be used to reconstruct the unmeasurable state variables. 5. Experimentally verify these control strategies by application to polymerizat ion reactors.