| People pay more attention to the research of state space model recentlybecause of the development of modern control theory. It needs the highveracity for system model with the greatness, complication andsystematism of control system. The classical system identification methodshave many limits and shortage, especially for multivariable system. Theycan't achieve a good identification result and meet the control demand. Sowe have the necessity to study multivariable system identification methodsadapting to the development of modern industrial system. We introduce andstudy the subspace state space system identification method in the paper. First, we study deeply the theory and methods of multivariable systemidentification, and emphasize on the study of subspace state space systemidentification method. We check the validity and veracity through thesimulation experiments, and compare to the classical identification methodsby the simulation experiments. Second, the paper introduces the AIC criterion in order to acquiring thesystem model order exactly. The system order is decided by non-zerosingular values of the observation matrix in the subspace state space systemidentification method. But the identification result is not true when thenon-zero singular values are very small or the disturbance is great, whichwill lead to a bad deflection, sometimes we can't identify the system model.But we can get the system order exactly with the introduce of AIC criterionand improve the identification effect. -IV-北京化工大å¦ç¡•å£«å¦ä½è®ºæ–‡ Third, the paper studies deeply the closed-loop subspace state spacesystem identification method that will help to the application of the method.We also check its validity and veracity through the simulation experiments. Last, we develop an identification software of subspace state spaceidentification methods, and get a good identification result through somesimulation samples. The software will help to the study and simulation ofthe identification methods.
|
PDF Full Text Request