Computational simplification of ERA (Eigensystem Realization Algorithm)‎

Abstract EN

A computational simplification of the Ho-Kalman minimal realization procedure, recently referred to as the Eigensys،em Realization Algorithm (ERA), is proposed.

According to the simplified algorithm, instead of performing the singular value decomposition (SVD) of the Hankel matrix, which is of relatively high dimensions, and multiplying with matrices containing left and right singular vectors, a minimal system state space representation may be obtained by selecting appropriate rows or columns fiom the Hankel matrix.

The obtained state representation is in either P.seudo-Con،rollab؛e or Pseudo - Observable form.

An illustrative computational example is included.