Research On Fusion Control Method And Its Application In UAV Flight Control

Precision flight control is one of the key problems for developing the technique of unmanned aerial vehicle (UAV) with integration of reconnaissance and attack. Because UAV is a multi-variable coupled system, the traditional single-loop design method is hard to satisfy performance reqirement of precision flight control. Modern control theory supplies advanced control methods for multi-variable systems. Optimal control is an important component of modern control, however, derivation process of the traditional solving methods for optimal control problem are complicate. Fusion estimation theory supplies a novel solving method. Therefore, researches on fusion control and its application in attitude and trajectory control system of an alternable thrust direction UAV are carried out.For the alternable thrust direction UAV, a total-varaible mathematical model, a longitudinal linear model and a lateral linear model of dynamics and kinematics are established. Comparing with the general UAV, it adds two control variables that are thrust deflection angles, and impacts of thrust alternable technique on mobility, agility and control ability of UAV are studied. For attitude stability and trajectory tracking problem, thrust alternable control strategies are obtained by single-loop design method, simulation results show that it can compensate the effect of pneumatic rudders, improve the attitude stability effect and decrease the trajectory tracking error.For attitude stability problem of the alternable thrust direction UAV under the gust disturbance, a fusion optimal control based attitude control method is presented. The essence of fusion control is using fusion estimation theory to solve the optimal control problem, thus multi-source information fusion estimation criterion is studied. For state regulator problem and tracker problem, centralized fusion methods and sequential fusion methods are proposed, equivalence among these methods and the traditional method is proved through theoretical analysis and simulations. Comparing with the traditional solving method, derivation of fusion solving method is simpler and it has explicit physical significance. Fusion optimal control method is applied to the alternable thrust direction UAV, simulations of a nonlinear total variable model described UAV are carried out, the results of which show that performance of fusion control method is better than that of single-loop design method.For the alternable thrust direction UAV with known flight path, a fusion preview control based trajectory control method is presented. Preview control is an optimal control method with receding horizon optimization. Traditional preview control method is based on error system that transforms tracker problem to regulator problem, which makes derivation process complicate and large calculation. Based on fusion estimation theory, error system and original system based fusion preview control methods are presented, respectively. Equavelence between the former and the traditiontal method when there is enough preview information is theoretically proved, and simulation results show that the former is better than the traditional method. The latter has smaller calculation, and simulation results show that the latter can reach a considerable level of the former when preview information is sufficient. It is proved that asymptotic properties of the fusion preview control method and the tradional method are same. Furthermore, fusion preview control is further applied to the linear system, multi-variable coupled system and nonlinear system, and approximate optimal control laws are obtained. For the UAV flight control problem, desired attitude information is transformed from known future trajectory information and is adjusted by current error information. And then fusion preview control based tracker is designed based on UAV linear models. Simulation results show that tracking performance of fusion preview control is better than that of single-loop design method. For improving adaptivity of the UAV flight control system and decreasing complexity and calculation, a brain emotional learning (BEL) algorithm based intelligent method for designing thrust alternable control law is presented, in order to keep safty, fusion preview control method is used in attitude control loop, simulation results show its effectiveness. For selection and tuning problem of the performance weight matrices of fusion controller, a modified particle swarm optimization based optimization method is proposed. Simulation results show its effectiveness. In order to improve reliability of the semi-physical simulation system for UAV flight control, turnable servo control problem and fault diagnosis problem are studied. BEL based direct and indirect intelligent control schemes are presented, simulation and experimental results show that these schems can overcome nonlinear factors such as friction and random noises. Ant colony optimization based intelligent diagnosis method is presented for precision servo turntable system, simulation results show that it can decrease redundant information of fault characteristic information, rduce diagnostic rules and condition terms, and improve diagnostic precision.Acomplishment of above researches will supply a novel method for solving variety optimal control problems, and supply an effective technique combined with modern control and intelligent control methods for flight control system, which will improve autonomous flight performance of UAV with integration of reconnaissance and attack.