The Application Of Convex Optimization Based Model Predictive Control In Guidance Of Reentry Vehicles

The hypersonic vehicle has great value in military and civilian application, and hasthe ability of fast global strike. Reentry process is very complicated, for the guidancecontrol system with the character of strong nonlinearity and uncertainty, and also theprocess is subject to heat flow, dynamic pressure and overload which are needed to besatisfied strictly. This dissertation studied the reentry guidance control problem forCommon Aero Vehicle based on the method of model predictive control (MPC) usingconvex optimization.First, the3DOF reentry equations of motion are formulated by analyzing the forceson the hypersonic vehicle. With analysis of the constraints action and introducing zerooblique balance gliding condition, reentry corridor is builded in the height-velocitysurface and we can clearly get reentry border of the hypersonic vehicle.Second, the use of the convex tool CVXGEN is studied. CVXGEN is used to solvethe model predictive control problem. The optimization problem of MPC is transformedto a convex problem. The transformation steps are given and an example is analyzed fortypical MPC problems.Then MPC reenty guidance control algorithm is proposed for hypersonic vehiclefast and correctly tracking given trajectory. The real-time character is guaranteed byconverting the each step optimization problem of MPC into convex optimizationproblem. The tracking rusults is analyzed with the variation of initial point, optimizationhorizon and coefficients of performance index respectively. The simulation resultsshows that even with the existence of disturbance the proposed algorithm still tracks thegiven tractory correctly and meanwhile satisfies all the constaint conditions.At last, fast trajectory planning and real-time trajectory tracking problems areinvesgated for the hypersonic vehicle last phase. Trajectory planning belongs toterminal time free optimization problem, and we first finish the time-mappingtransformation, make the problem convex and discretized, propose multi-trajectoryoptimization algorithm. By using the algorithm we finally get the suitable trajectory,and then we use the prior MPC algorithm to track the optimized trajectory. Simulationresults shows that multi-trajectory optimization algorithm satisfies real-time andcontraint conditions.