Furuta pendulum

My masterthesis project was about the implementaton of a novel off-line controller tuning algorithm that uses linear time varying identification to reduce the number of experiments necessary for one tuning iteration. My professor (Jonas Sjöberg) devised this method and showed in simulations that it works well for different types of controllers and plants. My job was to show that this novel tuning algorithm also works well on a non-linear physical system. I implemented the novel tuning algorithm on a Furuta pendulum, which is a highly non-linear system making it a good candidate to test the algorithm. I implemented the tuning algorithm using the real-time Matlab toolbox and simulink. I was able to show that the physical results were close to the simulated ones, which made my thesis a big succes.

My thesis is available for download here: Masterthesis.pdf