Adaptive Fault-Tolerant Control of Multivariable Systems with Applications to NASA Generic Transport Model

For performance-critical systems, such as aircraft flight systems, structural damage, actuator failures, actuator nonlinearities, sensor uncertainties, and environmental disturbances may lead to severe accidents if not promptly and properly mitigated, since adverse conditions can cause large unknown variations of system dynamics, introduce undesired disturbance inputs, and limit the performance of feedback control.

In this research, we will develop novel adaptive fault-tolerant control and fault-detection schemes for multi-input and multi-output (MIMO) systems with structural damage and component failures (such as actuator failures, actuator nonlinearities, and sensor uncertainties) to guarantee desired and safe system performance.

To handle complexities and uncertainties of nonlinear system dynamics, we use linearization-based design methods, where control schemes are developed for linearized system models, with both continuous-time and discrete-time control designs being considered. To accommodate uncertain damage and failures for the MIMO systems, the multivariable model reference adaptive control (MRAC) design method is employed. A key design condition--system infinite zero structure is investigated for both continuous-time linearized models and discrete-time linearized models before and after the adverse conditions occur, and invariance of this essential condition can be concluded under realistic failure and damage conditions. With such an invariance property, some novel fault-tolerant state feedback for output tracking and output feedback for output tracking multivariable MRAC schemes, whose plant-model matching conditions are much less restrictive than that of the state feedback for state tracking design, are developed to ensure stability and asymptotic output tracking for systems in the presence of parametric and structural uncertainties caused by damage and component failures.

Equipped with the developed feedback adaptive fault-tolerant control to ensure system signal boundedness requirements, novel adaptive fault detectors are constructed based on system dynamic coupling features and different failure patterns to identify and isolate damage and failures in order to enhance situational awareness for control personnel.

The developed adaptive fault-tolerant control and fault detection designs have been evaluated on a high-fidelity aircraft Matlab/Simulink model--the nonlinear NASA generic transport model (GTM), which offers a realistic representation of the aircraft. Extensive simulation studies have been conducted and simulation results have demonstrated the desired performance of our developed designs.