Research On Dynamic Control Allocation With Constraints For Near Space Vehicles

The technology of near space vehicle(NSV) has attracted much attention by the world countries in recent years. As a novel aircraft, due to its hypersonic velocity, high flight height, the NSV is hard to be attacked by the existing ant-aircrafts. Moreover, it can be reused, and the cost is lower than the satellites. Thus, NSV has a unique advantage in high altitude reconnaissance,prompt strike, and so on. To improve the aircraft performace, such as reliability, security and controllability, the NSV usually adopts the mulit-effectors aerodynamic configuration. Since the design method of the traditional aircraft flight control system couldn’t apply to NSV, it is important to distribute the virtual control command to the effectors for the NSV. Aiming at the distribution problem of control commands in the design of NSV’s control system, this thesis studies the constrained control allocation with the physical constraints of actuators. The main works are given as follows:Firstly, the control effect caused by different control moment of actuators is analyzed among the range of physical constraints. And, the control system model of NSV with the control allocation module is established. The dynamic of actuators is taken into account, and the dynamic responses of two typical kinds of actuator models of the control command are analyzed.Then, aiming at the constraints of control surfaces, the fish swarm based hybrid optimation algorithm is used to implement the control allocation of NSV. The genetic algorithm(GA) is introduced to solve the problem of local convergence in the late process of fish swarm algorithm(FSA). The simulation results show that GA based hybrid optimization fish swarm algorithm(HOFSA) can effectively assign control command to each actuator and the tacking control performance is good.Next, for the case with control allocation error due to the external disturbance and the uncertainty of configuration matrix, an anti-disturbance finite-time convergence terminal sliding mode controller is proposed. And, a robust least squares control allocation(RLSCA) is combined with it to achieve the goal of finite-time control allocation under the polyhedron uncertainty of configuration matrix. Simulation results are presented to demonstrate the effectiveness of the proposed scheme.Following, the nonlinear control allocation problem is studied, which includes obtaining the segmentation sections and the optimization of corresponding section coefficients. The segmentations are obtained by employing GA to search the best combination cells of piecewise linear controleffectiveness. The optimal section coefficients are given through pigeon inspired optimization algorithm(PIOA). Control surface deflection command is calculated by the optimal results of GA and PIOA. Simulation results show that the developed control allocation method is effective and the control requirement can be achieved.Finally, the dynamic control allocation combined with radial basis function(RBF) neural network and differential evolution algorithm(DEA) is studied by considering the dynamic of actuators. The ability of approximate nonlinear function both online and offline of RBF neural network is used to calucate the individual’s control effect using the iterative process of DEA with the consideration of actuator dynamic. The best commands of control surfaces are obtained by the DEA with calucated value. Simulation results are given to demonstrate the proposed method can improve the effectiveness of control allocation.