A NONLINEAR PREDICTIVE CONTROL OF MICROSCOPIC PROCESSES USING A FUZZY SYSTEM IDENTIFICATION APPROACH

Open Access
Author:
Rahnamoun, Ali
Graduate Program:
Chemical Engineering
Degree:
Master of Engineering
Document Type:
Master Thesis
Date of Defense:
July 06, 2011
Committee Members:
  • Antonios Armaou, Thesis Advisor
Keywords:
  • nonlinear model predictive control
  • kinetic monte carlo
  • nonlinear fuzzy system identification
Abstract:
System identification and controller design for systems with atomic-scale dimensions is a challenging concept. In this work, the problem of model-based control of a microscopic process is investigated. The unavailability of closed-form models, as well as the ill-definition of variables to describe the process evolution, makes the controller design task challenging. We address this problem via a fuzzy system identification of the dominant process dynamics. The data required for the system identification of such processes is produced employing atomistic simulations. A methodology is developed in which fuzzy logic for nonlinear system identification is coupled with nonlinear model predictive control for regulation of microscopic processes. We illustrate the applicability of the proposed methodology on a Kinetic Monte Carlo (KMC) realization of a simplified surface reaction scheme that describes the dynamics of CO oxidation by O2 on a Pt catalytic surface. The second case study on which this procedure performance will be evaluated is thin film deposition of silicon that has wide applicability in photovoltaic industry. Thin film deposition is of the most important manufacturing processes in the category of microscopic simulation and controller design. The quality of thin film which can be used in several different useful fields is highly dependent on the manipulated variables of the process which can be intentionally changed to achieve the highest quality. Using a lattice model of thin film evolution, the important characteristics of thin film, like thickness, roughness and porosity, can be simulated. The stochastic realization of this lattice model evolution is captured with KMC Simulation. The nonlinear fuzzy model gives a reasonable approximation to the system even without using filter for the system and the proposed controller successfully forces the process from one stationary state to another state for the CO oxidation on Pt catalyst. In the thin film deposition process, two important parameters are porosity and roughness of the film, which affect the quality of the thin film devices.The modeling and controller design approach is used for controlling porosity and roughness of thin film deposition process.