Automated, robust controller design for active vibration damping

The controller design plays an important role in the design of an active vibration damping system. Only in the optimal interplay between structure and controller can the best possible damping be attained.


Matlab Toolbox  

Already during the design phase the interplay between structure optimization and controller design for the respective structure has to be taken into account. For example, changed actuator or sensor positions must be considered in the iterative optimization of the complete active structure. While tools like ANSYS or ADAMS are used for the structure modelling, the CAE tool Matlab is suitable for the controller design. Among other things, it provides a multitude of universally applicable functions for the widest range of control concepts. Suitable functions are selected for the respective control problem and the control loop description is adapted appropriately in each iteration.

By order of Volkswagen AG, a Matlab tool box was developed for automated controller design for active vibration damping considering non-linear actuator behavior. While the non-linearity is inverted with a precompensator, the actual active vibration damping is performed by a model-based linear controller which is optimized for the respective structure.

In order to limit the work for the developers to the essential work steps of structure optimization and the design of the linear controller, the system adaptations in Matlab required for the controller design are automatically executed in the background guided by a graphic user interface (GUI).

Calculation of a model-based controller

Initially, the system model will be exported from the modelling tool used by Volkswagen. Simulink models of the open and closed control loops are created from the data files and the parameters of a state space model of the linear physical subsystem are determined. Simulations of the original simulator, the Simulink model of the open control loop and the simplified state space model can be compared with each other for model validation.

Actuator and sensor positions as well as vibration quantities to be damped are selected and the resulting transfer functions are analyzed. In consideration of the desired frequency range characteristics for the control energy and the vibration damping behavior, which are adjustable by the user in the form of weighting matrices, a model-based controller is then calculated.

Time and frequency range analysis of the closed control loop of the linear subsystem as well as simulations of the controlled non-linear overall system, guided and visualized by the GUI, then provide essential information about the damping behavior of the designed active structure.