Performance investigation of SLICOT system identification toolbox

This paper summarizes the results obtained through a systematic and extensive investigation of the performance of the new system identification toolbox, SLIDENT, incorporated in the freely available Fortran 77 Subroutine Library in Control Theory (SLICOT). This toolbox provides drivers, computational routines, and Matlab or Scilab interfaces, which implement two algorithmic subspace-based approaches (MOESP and N4SID) and their combination, either standard or fast techniques for data compression, and offers abilities to process multiple data batches, connected or not. The input-output data sets considered in this investigation include those in the DAISY collection, freely available at the site www.esat.kuleuven.ac.be/sista/daisy, which contains several large data sets from various domains. Extensive comparisons with the available subspace techniques have been made. The results show that SLIDENT is reliable, efficient, and powerful enough to solve large identification problems.