Using an Iterative and Affine Closed-Loop Identification and Controller Design Scheme for Local Linear Models*

In this paper an iterative, control-oriented identification scheme is used for generating nonlinear controllers for local linear models derived from observed input/output data of the system. For this, a locally affine identification and controller design method is used. An integral part of the iterative scheme is identification in closed-loop operation. It is shown that this procedure is capable of providing better controllers than the open loop identification. Characteristics of this closed-loop identification are discussed and the performance of generated Takagi-Sugeno fuzzy controllers for nonlinear systems examined. The results are shown on a longitudinal vehicle dynamics system with a standard vehicle testing cycle.