Rotorcraft Flight Envelope Protection by Model Predictive Control

摘要

A novel flight envelope protection method is presented. The approach is based on a model predictive receding horizon formulation, which computes at each instant in time the future extremal control inputs that would lead the vehicle to ride the flight envelope boundary without ever exceeding it. The extremal inputs are then used for pilot cueing. The calculation of the extremal inputs is based on the constrained optimization of a quadratic figure of merit for a reduced-order linear parameter varying model of the vehicle. The model accounts for the cross-couplings among the inputs that characterize the flight mechanics of rotorcraft vehicles and spans the entire flight envelope of interest by piecewise linear interpolation of given trim points. The approach leads to a convex optimization problem, which can be computed very efficiently in real time using a deterministic number of operations. A heuristic modification to the limits of the critical parameters, driven by a reduced-order nonlinear model of the vehicle, is used for online solution adaptation against possible model mismatch. The new approach is demonstrated by numerical simulation of multiple complex pilot-in-the-loop aggressive maneuvers.