Design of suboptimal robust controllers under unknown weights of perturbations

In the classical formulations of the problems of design of the robust optimal controllers, the equations of the nominal plant and the weights of the permissible perturbations are assumed to be known. In the present paper, consideration was given to a nonclassical formulation of the design problem where the weights of perturbations are assumed to be unknown and subject to estimation from the measurement data. The multivariable discrete-time controlled plant was described by a given transfer matrix with perturbations in the irreducible factors and a bounded external perturbation. The problem of determination with a predefined accuracy of the perturbation weights that are best coordinated with the measurements and of their corresponding suboptimal controller was solved.