FKFP (Research Projects in Higher Education)

Main evaluation window of the present (SISO) graphical identification tool

Identification of MIMO Systems in the Frequency Domain

FKFP 0098/2001

(Sponsored by the Hungarian Ministry of Education)

The goal of this research to extend the present frequency domain system identification methods from single input, single output (SISO) systems to multiple input, multiple output (MIMO) systems, and the development of a sample implementation. This implementation will be capable to demonstrate the working of the methods, teach their use, and to process MIMO data.

Frequency domain system identification methods are well elaborated for the SISO case [1], and a commercially available implementation also exists [2]. However, many of the possible applications contain MIMO models. A characteristic example is the investigation of vehicles: excitation hits the vehicle at the 2-4 or more wheels, and the number of outputs depends on the number of sensors (from a dozen up to several hundred). The routes of the propagation of the excitation signals are strongly interdependent, therefore MIMO data processing methods yield more precise results than channel-by-channel (SISO) ones. While the details of data representation and processing are well elaborated for SISO signals, the MIMO case needs further research. This research will contribute to these.

[1] Schoukens, J. and R. Pintelon, “Identification of Linear Systems – A Practical Guideline for Accurate Modeling”. Pergamon Press, Oxford, 1991.

[2] I. Kollár et al, “Frequency Domain System Identification Toolbox for Matlab”. http://elecwww.vub.ac.be/fdident/