| This thesis presents a novel approach to designing the input signal for an identification experiment, in which the process model estimate obtained is used to formulate and solve a robust (in a worst case sense) optimal controller. The input signal is designed to contain the information that is relevant for the end use of the model, that is for process control purposes. As a result, the experiment time required to achieve a certain level of achievable controller performance can be minimized. The new approach uses sensitivity analysis to determine the input signal frequencies that are important with respect to a certain measure of achievable controller performance. Using the sensitivity analysis framework, an overall experiment design methodology (which deals with specific design issues such as when to stop the experiment, how much data should be acquired at one time, etc.) is developed. Preliminary results demonstrate the effectiveness of the proposed design methodology. |
