Model reduction studies in LQG optimal control design for high-speed tilting railway carriages

The paper studies the utilisation of model reduction techniques, both physical-based and mathematicalbased, in designing simplified LQG optimal tilt controllers to improve the curving performance of railway coaches at increased running speed. The schemes make exclusive use of local practical signal measurements, i.e. sensors mounted on the current passenger coach. The fundamental problem related with straightforward classical nulling-feedback control is presented, while the commercially-used command-driven with precedence scheme is introduced. A combination of simulation results and, a recently proposed, tilt control system assessment method are employed for assessing the performance of the designed LQG controller.