Input Design for Structured Nonlinear System Identification

Structured nonlinear system identification is a block oriented identification method for systems that can be described by the interconnection of known linear dynamic systems and unknown static nonlinearities. An identification algorithm for this class of systems was previously presented in Hsu et al. [2008]. In this paper, we solve the related experiment design problem. Specifically, an expression for the expected estimate variance is derived, and a method for manufacturing an input sequence that minimizes an upper bound on this variance is developed, which depends on the solution to a convex optimization problem. Features of the solution include parameterization of the expected estimate variance by the input distribution, and a graph-based method for input generation.