A globally optimized state-space model identification method

State space models are greatly favored by scientific researchers on account of particular superiority. Aiming at the minimized error criterion and using the matrix differential theory, the paper represented a global optimal state space model-identifying algorithm for stochastic state space model. This is a new identifying algorithm that integrates system parameters identification, structural identification and state estimation. In this method, hypothesis state space model is disturbed by the measuring noise and process noise. First, state space vector x is identified according to error criterion J(E_s ^TE_s). Then, system parameters matrix A, B, C and D is identified by criterion function J(E_i ^TE_i). Results of mathematical simulation proved that this identifying method is characterized as simple calculation and higher identifying precision.