Abstract:   The superheated steam system ultrasupercritical unit is a multiinput and multioutput system. In this paper, a datadriven subspace approach is introduced for the model identification of the superheated steam system. In the approach, the unit step response coefficients are drawn directly, then the loworder transfer function of the system is obtained by leastsquare regression. Finally, closed loop identification based on datadriven subspace approach is applied to the superheated steam system of ultrasupercritical coalfired power unit. Real application shows that proposed identification strategy has a small computational cost and can meet the accuracy requirements in industrial sites.

Key words:  , ultrasupercritical unit； superheated steam system； datadriven subspace approach； closed loop identification