ORDER ESTIMATION FOR SUBSPACE SYSTEM IDENTIFICATION METHODS
YANG Chun1, 2, OU Jin-ping1, 3
(1. School of Civil Engineering, Harbin Institute of Technology, Harbin 150090, China;2. Research and Development Center, Shanghai Municipal Engineering Design General Institute, Shanghai 200092, China;3. School of Civil and Hydraulic Engineering, Dalian University of Technology, Dalian 116024, China)
Order estimation is an important step for subspace system identification methods, and has not been solved well. Four different estimation criterions, which are the singular value sequences gap, the gradient gap of singular value sequences, NIC and singular value criterion (SVC), are reviewed and an improved criterion MSVC is proposed based on SVC. The performance of these five criterions is compared through Monte Carlo test. The results indicate that the effect of MSVC is superior to others. Some suggestions are presented for the choices of the block row and block column number of the data block Hankel matrix based on the test results.