Abstract:
This thesis proposes the design of H∞ controllers for unstructured uncertain linear parameter varying (LPV) system. The LPV systems are linear systems that depend on time-varying, measurable parameters. In this work, both parameters and uncertainties are represented in linear fractional transformation. The H∞ controllers are also in linear fractional representation with the same parameters as those of the plants. We will convert the synthesis problem to a linear matrix inequality problem by the method of changing variables. In addition, we compare the results of the controller designed using the system uncertainty consideration with those designed neglecting the system uncertainty. Design examples of an inverted pendulum on a cart and a ball-beam experimental set demonstrate some limitations of this type of controller. In the case that the effective time period of the uncertainty is short or the value of the uncertainty is small, there is no difference whether or not the uncertainty is taken into consideration. On the other hand, if the time period is long or the value of the uncertainty is large, the controller with uncertainty consideration reveals a considerably better performance.