Extensions of a Simplified Continuous-Time Multivariable Adaptive Control Algorithm

A simplified adaptive control algorithm was recently shown to guarantee robustness with parasitic dynamics and disturbances. The algorithm uses parallel feedforward in order to satisfy necessary positive realness conditions. On the other hand, the order of the plant and the pole excess may be very large and unknown, while the model reference may be of very low order. This paper presents an attempt to reduce the prior knowledge needed for implementation of the adaptive algorithm. It is only assumed that a necessary feedforward of known order exists and the eigenvalues of this configuration are calculated adaptively. Although the stability analysis still arises some difficulties, the proposed algorithm seems to perform very well in difficult environments, including nonminimum-phase plants with rapidly changing parameters.