<title>Performance-oriented optimization of structures with mixed passive and active control</title>

Abstract

ABSTRACT Smart structures include active elements, hence the integration of structure design with design isinevitable. This paper provides a step in this integration in order to achieve optimal performance. The approachpresented here solves a mixed passive and active problem with performances characterized by the so-called mixed H2/H performances. Keywords: passive control, active control, vibration control, mixed H2/H control, structural 1 . INTRODUCTION In traditional designs, structures are usually designed first, and then active elements are designed based on the designed structures. It is a well-known fact that the structure and its design problems are not independent.'Most often the two designs are competing witheach other, and the effort often combats certain dynamics ofthe structure. These dynamics could have beeff;designed easier to control , without degrading other performancerequirements. Hence, for modern smart structures, a theory for integrating structure design with controller designis required. An ad hoc approach for this integration is provided for the NASA LaRC CS! Evolutionary Model in.2Theoretically an ideal approach is to simultaneously optimize the structure and the controller. However this idealapproach is far from computationally tractable due to the complex and non-convex nature of the optimization.35Passive modifies structure parameters (i.e., structure redesign) without injecting external energy,while active uses external energy. In order to overcome the computational difficulty,6'7 propose acomputationally tractable iteration to perform the integrated design, which continues the work of,8'9 where theactive energy is minimized subject to performance requirements. The key to this approach is to add onemore step to connect the traditional separated design steps. This step performs simultaneous passive and activecontrol design while preserving some feature of the desired system behavior. We call this method the feature