Model reference adaptive control for nonminimum phase aerospace systems
Adaptive control techniques are often avoided in aerospace systems due to stringent plant structural requirements and validation difficulties. This dissertation seeks to broaden the range of aerospace engineering applications that can utilize an adaptive controller through the development of an extended model reference adaptive control (MRAC) design. First, a partitioned control framework is presented that permits the combined use of an adaptive control law and a nonadaptive control law. The partitioned framework is used to shift full control authority away from the adaptive portion of the system. Next, two MRAC variations that can accommodate the nonminimum phase zeros often seen in aerospace applications are discussed for use as the adaptive system. The parallel feedforward compensator approach proposes inclusion of a user--defied fictitious model in parallel with the plant that is designed to make the plant appear nonminimum phase. The surrogate tracking error approach modifies the typical MRAC structure to handle nonminimum phase plants by requiring knowledge of its nonminimum phase zeros. A tracking error convergence proof is provided for this continuous-time MRAC variant. The partitioned design using the surrogate tracking error approach is applied to the control tasks of an experimental, flexible wing aircraft. A simulation is used to demonstrate much improved flight path angle command tracking when compared to use of the aircraft's existing nonadaptive control law, even in the presence of large--scale modeling error. A second simulation is used to show the design applied to flexible motion control of the same aircraft model and exhibits similarly improved performance.