An object-oriented toolbox for linear and nonlinear system identification

Nonlinear system identification techniques are often used to construct mathematical models of physiological systems. There is a wide variety of model structures, many of which require specialized techniques for their identification. Unfortunately, this variety may present a significant barrier to any non-specialists who wish to use these methods. This work describes a MATLAB™ based toolbox that addresses these problems by providing a common environment supporting a robust set of tools for the identification, simulation, and manipulation of important model structures. The design philosophy and its implementation are described. An illustrative example, constructing a parallel cascade model of the dynamic stiffness of the human ankle, is used to demonstrate many of the features of the toolbox.