Multivariable Self-Tuning Controller with Integral Action

摘要

A generalized minimum variance self-tuning control algorithm for multivariable systems described by a CARIMA model is presented. It is shown that the resulting control system exhibits effective integrating properties, thus ensuring the robust zero-error regulation for any constant set point and load disturbance. The major limitations of the technique are concerned with the knowledge a-priori required on the system delay structure. Simulations show that the algorithm is well suited to control minimum phase and/or asymptotically stable plants, while dealing with nonminimum phase and unstable systems calls for a careful tuning of the parameters entering the performance index. In order to improve the performance, modifications to the basic algorithm can be given, such as the direct feedforward compensation of measurable load disturbances, and the development of a continuous time equivalent.