IDENTIFICATION, STABILITY, AND OPTIMAL CONTROL OF CHEMICAL ENGINEERING SYSTEMS

This thesis is concerned with the determination of the best configurations and controller parameters for feedback control of chemical processes. Control systems can be thought of as consisting of the combination of the process to be controlled and the controller itself. Four chapters are presented here, two dealing with the identification and behavior of a process and two with the type of controller to be used.; Chapter 1 is concerned with the fitting of a model to the actual dynamics of a process using time-domain analysis. This involves not only determining the form of the model, but also the numerical values of its parameters. Past methods have generally reduced to non-linear regression problems which can be solved approximately by the method of least squares. This method can be slow in converging and may lead to inaccurate estimates of model parameters. Moreover, convergence is sensitive to the initial estimates of these parameters. The method used in this study is based on the theory of normed linear space. For the cases studied here, this method was found to be substantially better than least squares in both convergence speed and accuracy. Because of this, this method should be attractive for self-adaptive control systems.; The second chapter focuses on the difficult problem of calculating the controllability and observability and matrices of a time-varying linear system. The theory and procedure are new in that block-pulse functions are used. The recursive algorithms for programming the state transition, controllability, and observability matrices are obtained and only involve the algebraic operations of matrices. The method makes it possible to examine the controllability and observability in a sequence of sampling times using a digital computer.; In Chapter 3, the problem of controller tuning is addressed. Weber and Bhalodia studied such processes in terms of the Continuous Cycling Method of controller tuning originally proposed by Ziegler and Nichols. In this thesis, that study has been further extended. Some modifications of the Ziegler-Nichols parameters for controller tuning are suggested.; The last chapter addresses the controller tuning problem from the standpoint of optimal state feedback control of a linear system with a quadratic functional index. A time-invariant linear system is considered under either optimal integral control or under proportional-integral control. Stability and offset under these controller modes are determined. (Abstract shortened with permission of author.)...