No abstract

A mixed-integer linear programming approach to mixing continuous and pulsed control effectors is proposed. The method is aimed at applications involving reentry vehicles that are transitioning from exoatmospheric flight to endoatmospheric flight. In this flight phase, aerodynamic surfaces are weak and easily saturated, and vehicles typically rely on pulsed reaction control jets for attitude control. Control laws for these jets have historically been designed using single-axis phase-plane analysis, which has proven to be sufficient for many applications where multiaxis coupling is insignificant and when failures have not been encountered. Here, we propose using a mixed-integer linear programming technique to blend continuous control effectors and pulsed jets to generate moments commanded by linear or nonlinear control laws. When coupled with fault detection and isolation logic, the control effectors can be reconfigured to minimize the impact of control effector failures or damage. When the continuous effectors can provide the desired moments, standard linear programming methods can be used to mix the effectors; however, when the pulsed effectors must be used to augment the aerodynamic surfaces, mixed-integer linear programming techniques are used to determine the optimal combination of jets to fire. The reaction jet control allocator acts as a nonuniform quantizer that applies a moment vector to the vehicle, which approximates the desired moment generated by a continuous control law. Lyapunov theory is applied to develop a method for determining the region of attraction associated with a quantized vehicle attitude control system.

No abstract