A closed-loop comparison of control allocation methods

This paper compares closed- and open-loop performance measures for sixteen control allocation approaches. The approaches are regular and discrete time versions of direct allocation, linear programming, quadratic programming, and several variations of the weighted pseudo-inverse approach including a cascaded generalized inverse. The control allocation methods are first evaluated using open-loop measures including the percent volume of the attainable moment subset for which the allocation method produces admissible solutions and the ability to attain commanded moments for a prescribed rolling maneuver. Then, the control allocation methods are compared closed-loop using a dynamic inversion control law to perform the same rolling maneuver. All maneuver results are completed using a simplified 5 degree-of-freedom nonlinear simulation with seven control effectors. The results show that the open-loop measures do not necessarily translate to closed-loop performance.