Abstract
A design method of globally bounded robust adaptive controllers is proposed in the present paper. On the assumption that the overall system is minimum phase and that the relative degree of it is known exactly, it is shown that 1) the resulting adaptive control system is globally bounded in the presence of arbitrarily large unmodeled parts of the process, and for arbitrary initial conditions, 2) even if unmodeled parts of arbitrary magnitudes exist, the output error converges to a residual region whose amplitude can be made arbitrarily small, and 3) when unmodeled parts of the process are absent in the ideal case, the zero residual tracking error is attained. This can be done by introducing two types of adaptive controllers into the control schemes, one of which is composed of linear compensators and adjusted for the modeled (nominal) part of the process, and the other is of nonlinear high gain feedback-based compensators and tuned for the unmodeled part (unknown parasitics and disturbances) of the process.
