Simulation of Structure Control and Controller Design Within a Finite Element Code

Abstract

This paper investigates some aspects of control by using large-order finite element models and low-order controllers for suppression of vibration (active damping), in terms of both controller simulation and design. It is fundamental for understanding interaction in an actively controlled structure to have an accurate model. The basis structure as well as active elements can be modeled with appropriate elements within a finite element (FE) code. It leads to rather complex FE models in particular if 3-D modeling is applied. Optimal sensor and actuator location and collocation are not topics of the current paper. However complex the FE model might be, it always lacks in reality. Moreover, model reduction techniques are necessary to make a controller design feasible. To close the gap between the FE model and reality, experimental modal identification is used to correct the FE model with an updating procedure before putting to use the controller design and simulation. To integrate controller design methods and the simulation of control action into an FE package, appropriate interfaces between structure and controller have to be designed. From the control system designer's point of view, the definition of these interfaces must absolutely take into consideration the nature of the plant and the possible control design methods. The control of mechanical structures will in most cases be a multiple input—multiple output (MIMO) problem. To do all that preliminary work before testing and final tuning on real structure within the FE system is a challenging undertaking, but it can reduce the number of experiments necessary for final design of the controller. And, on the other hand, influences on the controller of change in the structure can be examined. The paper concludes with four examples of different complexity to illustrate methods and tools that can be very helpful in design and testing of structural controllers.