MULTI VARIABLE CONTROL USING FEEDBACK AND FECDFBRWARD BLOCKS - A COMPARATIVE STUDY

Abstract

The specific objective of this dissertation is to develop linear multivariable controllers that would achieve stability and requlation, in the presence of parameter variations and disturbance inputs, for a general system. Basically two approache's have been discussed in this thesis and a comparative study has been made in the end. In the first approach [6] a simple method for the design of PID controllers for linear multivariable systems, where the controllers use only the available system outputs, is discussed. The resulting closed loop system has a specified set of poles and in the steady state the outputs follow step commands, and reject disturbances of any form with constant final values. For an m input I output system of order n, the PID controller can place 3m poles of the (n+l)th order closed loop system at any desired locations. The controller 01 matrices are assumed to have unity rank, and are obtained from linear equations. In the second approach 171, a command matching controller, or a feedforward controller (inverse dynamics) has been introduced along with the PID controller (pole placement). In contrast to the PID controller, the multivariable command matching controller, to be proposed in this approach, ensures that the output variable track their corresponding commands both during the transient and steady state. In other words, the output must comply with the input during the entire transition from the (iii) initial to the final conditions. This clearly implies a "tighter" control over the output, than a steady state regulator such as a PID. Also it has been shown that..the feedforward controller and the feedback controller can be designed separately. This greatly simplifies the control system design.