Abstract
In earthquake engineering applications of feedback control, the inherent vulnerability of external power sources has stimulated significant interest in dissipative systems with controllable passive parameters. Semiactive dampers (i.e. viscous dampers with controllable coefficients) constitute the simplest example of such actuation. Regenerative force actuation (RFA) networks are an extension of semiactive technology, whereby mechanical energy is first converted to electrical energy, which is then dissipated in a controllable resistive network. They have advantages in that they can transmit energy between remote locations in a structure, and can store energy for reuse. Recent research has shown that they, like semiactive systems, can be viewed as variable‐coefficient damping systems. In this paper, a general approach to control synthesis for semiactive and regenerative systems is presented which is analytically guaranteed to outperform optimal linear viscous damping in stationary stochastic response, under the standard quadratic‐Gaussian performance measure. Furthermore, it is shown that control systems utilizing RFA networks can be designed to outperform optimal linear viscous damping by an analytically computable margin. In Part 1 of this paper, the analysis assumes a linear structural model with feedback access to the full, uncorrupted system state vector. Part 2 extends the method to the more practical case of noise‐corrupted and incomplete (i.e. output) feedback measurements. The approach is illustrated on a simple simulation example. Comparisons are made to clipped‐optimal controllers, which are designed by a similar process but which do not exhibit guaranteed performance margins. Copyright © 2007 John Wiley & Sons, Ltd.

This publication has 14 references indexed in Scilit: