Multipoint Volterra Series Interpolation and H2 Optimal Model Reduction of Bilinear Systems

Abstract

In this paper, we focus on model reduction of large-scale bilinear systems. The main contributions are threefold. First, we introduce a new framework for interpolatory model reduction of bilinear systems. In contrast to the existing methods where interpolation is forced on some of the leading subsystem transfer functions, the new framework shows how to enforce multipoint interpolation of the underlying Volterra series. Then, we show that the first-order conditions for optimal H2 model reduction of bilinear systems require multivariate Hermite interpolation in terms of the new Volterra series interpolation framework; and thus we extend the interpolation-based first-order necessary conditions for H2 optimality of LTI systems to the bilinear case. Finally, we show that multipoint interpolation on the truncated Volterra series representation of a bilinear system leads to an asymptotically optimal approach to H2 optimal model reduction, leading to an efficient model reduction algorithm. Several numerical examples illustrate the effectiveness of the proposed approach.

Cited by in corpus (8)

• A New Selection Operator for the Discrete Empirical Interpolation Method -- improved a priori error bound and extensions
• Krylov methods for low-rank commuting generalized Sylvester equations
• Structure-Preserving Interpolation of Bilinear Control Systems
• Structure-Preserving Interpolation for Model Reduction of Parametric Bilinear Systems
• A Unifying Framework for Interpolatory ${\mathcal{L}}_2$-optimal Reduced-order Modeling
• Time- and frequency-limited H2-optimal model order reduction of bilinear control systems
• Residual-based iterations for the generalized Lyapunov equation
• A unifying framework for tangential interpolation of structured bilinear control systems