UC Merced

The increased attention towards sustainability, and the rapid advancement in technology in recent years, has introduced an unprecedented need towards more optimal process control. On the other hand, the increase in computing power per chip following the Moore’s law, and the decrease in computing power cost following the price-performance Moore’s law, enables implementation of more computationally demanding control algorithms that their application was deemed impractical in the recent past. This work presents a framework called self optimizing control (SOC) that attempts to address the new requirements for process control by pursuing the attainment and sustainability of optimal performance. Then, several control strategies that enable the realization of the SOC framework are introduced and implemented to solve control problems associated to thermal processing and radio frequency (RF) impedance matching. An iterative learning control (ILC) approach is used to control an experimental carbon activation plant. A plug and play model predictive control (MPC) toolbox is developed that allows for straightforward implementation of embedded MPC for process control; this toolbox is then used to solve control problems for several experimental and simulation-based examples associated to thermal processes. Finally, a double sensor configuration is proposed to solve RF impedance matching in the context of semiconductor manufacturing as opposed to the conventional single sensor configuration to achieve robust optimal matching performance. A model-reference adaptive control along with a gradient based reflection minimization approach are developed to control an L-type matching network based on the double sensor configuration.