An application of expert hierarchical control to piecewise linear systems

A hierarchical control structure is presented for the output feedback control of nonlinear systems approximated by piecewise linear models. At the lowest level, traditional output feedback control techniques are used to design robust local controllers for each operating segment. The low-level control is then enhanced by the use of a supervisory level, which combines clustering and binary tree classification techniques from artificial intelligence to identify the transitions between the models. The controller is trained on typical operating scenarios. The structure is applied to the case of a nonlinear DC servomotor with a piecewise linear model of the thyristor converter dynamics