Schlüsselwort: Riccati-Regler

The subject of the thesis at hand is the controller design for a vehicle with active suspension to achieve an increase in ride comfort and safety. This is achieved by taking the current conditions of the road unevenness into account. The unevenness of the road can be modeled as a stochastic signal with additional single bumps. As these differ in their characteristic it would seem advisable to face them with two different parallel control concepts. In relation to the stochastic part a new approximative unevenness model is presented to be used for the design of a quadratic optimal controller. To estimate the vehicle and unevenness states a Kalman filter is designed using the plant and unevenness model. The design of both the control and the Kalman filter are in need of the unknown and time-variant parameters of the unevenness model. Therefore a recursive least squares algorithm is utilized to identify the parameters online. In doing so, the vehicle suspension becomes adaptive to the actual road unevenness conditions. To face single bumps in the road profile, an unevenness adaptive bump detection is presented. It becomes possible to separate the bumps and the stochastic part of the unevenness profile. A modelbased feedforward control for the bumps is shown to reduce the negative effects in respect to comfort and safety. New criteria are proposed. A two-degree-of-freedom control structure is used to make the vertical system response relating to both the excitation by the horizontal movement and the excitation by the unevenness adjustable independently of each other. A feedforward control to handle horizontal movement excitation is introduced. In respect to model uncertainties the Kalman filter is extended to increase robustness. Furthermore a feedforward structure for the filter is shown to deal with nonlinearities of a passive damper and thus making it possible to stick with the linear filter design. To validate the concept a quarter-car test rig is used. In addition a vehicle trailer is presented to put the concept of unevenness adaptive Kalman filtering with parameter identification and bump detection into action on real roads. After calibration on a test rig, results of road measurements are given. A statistical evaluation of the identified parameter values in respect to probability and correlation shows the necessity of online parameter identification and allows the reader to choose realistic and matching parameter values in future simulation.