H∞ robust control using LMI and unstructured uncertainty applied to a temperature process

In this work, an experimental methodology to design robust controllers using H
∞
optimization and linear matrix inequalities was applied to a two-input and two-output temperature system. This process is nonlinear and highly affected by the environment, so it is hard to obtain a good model for control. An identification procedure was outlined to estimate a set of possible models with uncertainties. The definition of frequency weights to account performance and uncertainty was discussed. A closed loop configuration for H
∞
robust control design is presented and a solution to the optimization problem using linear matrix inequalities is discussed. Finally, a controller who achieve nominal performance and robust stability was designed to the temperature process. To evaluate the controller performance, a closed loop experiment was performed.