Intelligent Reconfigurable Control for Systems with Multiple Effectors

Many modern systems rely on multiple innovative actuators and effectors to improve performance, versatility, and survivability. In systems where several actuators or actuator combinations are capable of affecting the behavior of a single controlled varaible, an opportunity exists to incorporate non-traditional features in the control law, including post-failure reconfiguration and optimization of high-level system functions (e.g., overall efficiency). Often, one requires an intelligent nonlinear control algorithm to achieve these benefits. This proposal describes such an algorithm based on neural network system modeling, robust on-line system identification, and receding-horizon optimal control. This algorithm has been evaluated favorably in piloted simulations and F-16 flight tests (including simulated impairments), and it is currently being applied to a tailless aircraft having over 14 independent effectors. The proposed Phase I research will (1) develop a prototype control system design toolkit based on the algorithm, (2) extend these algorithms with a control allocation module that provides desired closed-loop performance while achieving higher-level mission objectives such as reduced drag or low observability, and (3) demonstrate the algorithms and the prototype software tool on a complex system application of interest to the Navy.