Colloquia

Samenvatting

Gravitational waves are ripples in space-time that carry valuable information about interactions between large interstellar objects, such as neutron stars, black holes or supernovae. The information that gravitational waves carry can reveal new insights about physics and our understanding of the universe. These signals are measured by gravitational wave observatories, which employ laser interferometry to measure a phase difference between two perpendicularly reflected lasers. These observatories are extremely sensitive to seismic activities from the earth's surface. Therefore, an elaborate vibration isolation suspension guards the mirrors, that alter the path of the lasers, from the seismic noise. Without these suspensions, a gravitational wave measurement would vanish in noise. Usually, the control systems for gravitational wave applications are designed via classical frequency domain design methods, such as loop-shaping, which is generally time consuming, labour intensive and often requires much expertise and knowledge about the suspension design in order to successfully design suitable controllers. A more modern approach for the design of controllers are optimal controller synthesis methods, which automate the controller design process. This allows to quickly evaluate the achievable performance for many suspension configurations in the available design space. This makes optimal controller synthesis especially usefull for future gravitational wave observatories that are still in the design phase. This research investigates the usefulness of a non-smooth optimal controller design strategy, specifically for the controller design for the mirror suspensions that are present in gravitational wave observatories. Additionally, optimization of the actuation distribution and suspension mechanics is integrated with the controller design for a more holistic design approach. Part of the research is dedicated to the shaping of the requirements into a relevant optimization problem. Besides the optimization of the controller and actuation distribution, some relevant extensions are investigated to showcase the versatility and flexibility of the non-smooth optimization algorithm. These extensions involve simultaneous optimization of both the controller and the dynamics of the plant. The results show how the optimal controller synthesis approach is able to shape controllers and the distribution of the actuation over the stages of the suspension according to requirements.