An algebraic theory for design of controllers for linear multivariable systems--Part I: Structure matrices and feedforward design

Abstract

In this two part paper a theory is presented in which several control problems can be solved. The theory is applicable to a general class of linear multivariable systems where the measured output does not have to be equal to the controlled output or the state. The system may be affected by nonmeasurable disturbances. Only controllers which stabilize the system and have proper transfer functions are allowed, i.e., the controllers have to be physically realizable. It is shown that the solutions to control problems of "servo type," e.g., problems of model matching, decoupling, and invertibility, are special cases of the solution to a more general problem. Analogously, the solutions to problems of "regulator type," e.g., disturbance decoupling, output regulation, and pole placement, are also speclal cases of the solution to a more general problem. It is shown that problems of "servo type" and "regulator type" can be solved independently of each other. In Part I generalized polynomials are introduced as a mathematical framework. Structue matrices, which describe how well a system can be controlled, are defined. Finally, problems of "servo type" are solved. Part II mainly deals with problems of "regulator type."