IDENTIFICATION OF MULTIVARIABLE STEP MODEL BASED ON MULTI-STEP TEST METHOD

A least-square-based multi-step test method for model identification of multivariable step response is presented. It overcomes the disadvantages of the traditional one-step test, and especially applicable to multivariable system. The notable characteristic of this method is that the multi-step test can be done under the normal operation, and that it has strong practicality on the real engineering. Simulation demonstrates its feasibility.