Object‐oriented creation of input signals for system identification

Abstract

This study describes how a very large number of deterministic input signals for system identification may be created by object-oriented methods. The concepts of aggregation and inheritance, combined with the properties of m-sequences, are utilised to develop two new methods for the creation of pseudorandom perturbation signals with ideal spectral properties, two, three, five or seven levels and a very wide range of periods. The available signal levels ensure that the signals are suitable for both linear and non-linear system identification and the availability of a large number of signal periods ensures that the signals are suitable for both single and multi-input system identification. The methods are described in detail and illustrated by examples, together with their implementation in the open environment of an efficient, user-friendly and freely available Matlab program which provides third-party software to complement existing Matlab Identification Toolboxes.