Research On Multi-radar Coalition Game Theory And Methods For Intelligent Cooperative Target Tracking

Multi-radar systems can improve anti-jamming and target detection performance by cooperating in space,time,frequency,energy,and processing domains.However,with the development of swarming and intelligent targets and jammers,the adaptive ability of multi-radar system based on traditional cooperative methods to fast time-varying environments is insufficient and the control and utilization efficiency of resources is low,so that the anti-jamming and detection capabilities in complex environments cannot be guaranteed.Intelligent cooperative tracking technology is an effective means to achieve high-precision target tracking in complex environments.To realize this technology,it is necessary to solve the intelligent cooperative mechanism design and fast high-dimensional strategy optimization problems of multiple radars.For the above problems,this dissertation studies the mechanism model and theoretical method of multi-radar intelligent cooperative tracking on the basis of the coalition game theory.The main work and contributions are as follows:(1)For autonomous and unmanned multi-radar cooperative tracking application requirements,the interaction mechanism of multiple radars in task,resource,and physical motion spaces is analyzed,the closed-loop architecture of multi-radar cooperative tracking coalition game is designed as a natural bee colony,the swarm intelligent behavior mode of autonomous division of labor and cooperation for multiple radars is characterized,and the mathematical model of the multi-radar coalition game integrating coalition formation subgame and strategic subgame is developed,which provides the model foundation for the subsequent methods.(2)For the intelligent cooperative task assignment problem of multiple radars for multi-target tracking,a model-based multi-agent random Fourier feature reinforcement learning algorithm is proposed,which uses random Fourier features to approximate the state-action value function,thus significantly reducing the strategy optimization time for task assignment and the number of radars performing tracking tasks on the premise of ensuring tracking accuracy.(3)For the intelligent cooperative joint task assignment and path planning problem of multiple radar-enabled unmanned aerial vehicles(RUAVs)for multi-target tracking and pursuit,a distributed Nash equilibrium seeking multi-target pursuit algorithm is proposed,which enables RUAVs to assign target pursuit tasks autonomously and plan flight paths cooperatively,realizing the safe,effective,and fast pursuit and interception of enemy penetration targets by multiple RUAVs in offensive and defensive confrontation scenarios.(4)For the intelligent cooperative joint track-sleep assignment and power allocation problem of multiple radars for maneuvering target tracking under self-defensive suppressive jamming,an alternate coalition formation and projection gradient algorithm is proposed,which enables radars to assign tracking and sleeping tasks autonomously and allocate transmitting power cooperatively,reducing the number of radars performing tracking task and the total transmitting power of the multi-radar system under the condition of ensuring target tracking accuracy.(5)For the intelligent cooperative joint transmit-receive task assignment and path planning problems of multiple RUAVs for target tracking under stand-forward swarm suppressive jamming,an alternate coalition formation and particle swarm optimization algorithm is proposed,which enables RUAVs to assign transmitting and receiving tasks autonomously and plan flight paths cooperatively.By constructing a moving and variable multi-static radar system,the target tracking accuracy in jamming is improved.The application scenarios focused by above methods are from static radar platform to moving radar platform,from no jamming to the presence of jamming,from single jamming to multiple jammings,and from monostatic radar network to multi-static radar network.That is,the relationship of the considered scenarios is progressive.All above theories and methods are verified by numerical simulations.The results show that these methods can realize the autonomous division and cooperation of multiple radars in the multi-dimensional space of task,resources and kinematics,and significantly improve the adaptability to complex dynamic environments and the overall target tracking utility of multiple radars.The theories and methods proposed in this dissertation can support the mechanism,model,and method for the future development of swarmming and intelligent radar equipment,ensuring the perception and countermeasure performance of radar equipment in the future electromagnetic spectrum warfare,which has important research value for protecting the land and aerospace security.