A theoretical and algorithmic framework is proposed for optimal identification of rational transfer function parameters of discrete-time linear systems from input-output (IO) data. The nonlinear optimization criterion is theoretically decoupled into a purely linear problem for estimating the optimal numerator and nonlinear problem for the optimal denominator. An iterative computational algorithm is given for estimation of the denominator. Once the denominator is estimated, the numerator is obtained with only a single step of linear estimation. The effectiveness of the proposed algorithm is verified with simulation examples.
展开▼