Robust Nonlinear Control Design with Proportional-Integral-Observer Technique
In this thesis, the motivation is explained in the introduction part based on a review of the development of observer technique especially the Proportional-Integral-Observer (PI-Observer) technique and an analysis of robust control for nonlinear systems. After that, the goals of the work are set as: improvement of the high gain PI-Observer design in the way that the high observer gains of PI-Observer can be automatically chosen and fit the current situation; and the application of PI-Observer in robust nonlinear control for nonlinear systems with unknown effects. To the first point, the high-gain PI-Observer design is systematically discussed and the online adjustment of the PI-Observer gains proposed as an advanced PI-Observer (API-Observer) is detailed. At the same time, the implementation of the addressed adjustment algorithm is included showing the adaption of the PI-Observer gains to the current situation of system dynamics and disturbances on a practical system with simulation results and real measurements respectively. To the other aspect, a high gain PI-Observer-based nonlinear control method is proposed as a combination of the PI-Observer and the classical exact feedback linearization method (EFL-PIO) and directly after it the applicability on mechanical systems is surveyed. Numerical examples of mechanical systems are given to illustrate the application of the EFL-PIO approach. At last, the implementation of the proposed PI-Observer-based nonlinear control method is illustrated in detail on a hydraulic cylinder system for its position control. Besides that, the position control for the cylinder system without direct position measurement is realized with the PI-Observer applied as virtual sensor. Here a discrete-time control algorithm is also designed and programmed to satisfy the normal industrial requirement. At last, a broader application of the robust EFL-PIO approach, a more efficient and compact numerical realization of the adaption algorithm for the API-Observer, and an API-Observer-based robust control or fault detection are suggested as future work.