Study On Maneuverable Decision Of The Pursuit-Evasion Problem And Coordination Algorithm For The Multiple Unmanned Aerial Vehicle Task

Maneuvering decision and cooperative work for some air missions forthe UAVs are the key technologies in modern intelligent air-combat system.In this paper two problems of pursuit-evasion for one vs. one of the UAVair confrontation and multi-fighter cooperation in the UAV intelligentconfrontational system are studied. And there are three major workcompleted in this thesis and listed as followings,1) A three-Degree Of Freedom kinetic model of the UAV isestablished and a basic maneuvering library is setup by using the throttle,normal acceleration and roll-angle as control variables via momentumtheorem to solve maneuvering decision.2) A nonlinear dynamic model of pursuit-evasion for the UAV isestablished, and a maneuverable decision algorithm based ondifferential-game for the one vs. one UAV air confrontation in intelligentcombat system is proposed. According to the maximum and minimumvalue principle, the above maneuverable decision problem can betransformed into a two-point bound value problem and a feasiblenumerical solution based on allocation method is achieved. Furthermore,considering of the constraint of dynamic pressure for the UAV, a modifiedmaneuverable decision algorithm based on differential-game is proposedand this constraint can be converted into normal load constraint and thenachieved the solution by using the allocation method. The results show thatalgorithms of maneuvering decision not only meet the dynamicpressure-normal load constraint, but also be suitable for the practicalengineering. 3) Target assignment problem for the Multi-UAV cooperation isstudied. An autonomous priority function is established, and a targetassignment algorithm based on Back Propagation neural network isproposed for the Multi-UAV cooperative work, and a simulation of airconfrontation situation between two UAVs vs. two UAVs is verified theproposed target assignment algorithm based on BP NNs feasible.In general, novel maneuvering decision algorithms based ondifferential game are useful for the UAV air confrontation with or withoutconsideration of constraints of dynamic pressure and normal loads. Andthe BP neural network based target assignment algorithm is adaptive forvarious air confrontation situations. All of these works will improve thetheory and engineering application of maneuvering decision andcooperative work in UAV air confrontation in China.