Research On The Distributed Cooperation Method Of UAV Swarm Based On Swarm Intelligence

UAV swarm is developing towards miniaturization,intelligence and swarming.It has been widely studied,demonstrated and applied in various mission scenarios,especially in combat scenarios.However,with the increasing number of UAV swarms,the continuous expansion of combat space,the gradual complexity and diversity of tasks and the limited information conditions,how to deal with the distributed cooperation of UAV swarms under complex constraints has become the key problem faced by UAV swarm system.Swarm intelligence is becoming a new collaborative model.Its basic idea is to solve the problem of distributed collaboration of UAV swarms by imitating the high selforganization of biological groups.The basic method is to make full use of the local information interaction of unmanned platform,and introduce group feedback and adaptability in the process of task division and execution,so that group collaboration is not limited by quantity,space and task.In this mode,how to realize the emergence of swarm intelligence and improve the task processing efficiency of UAV swarm through simple interaction rules is the core problem.Traditional research mainly focuses on urban application scenarios.It is assumed that the unmanned platform can obtain global information from the system or the communication range of the unmanned platform can cover the whole swarm space,but this assumption is not applicable to application scenarios in harsh environments such as battlefield.In addition,most of the existing collaborative research is limited to a simple task collaboration scenario,which fails to cover the collaborative tasks involved in the whole process of UAV swarm operation.Based on this,guided by the concept of UAV swarm operation,this paper studies the UAV swarm operation process and its key activity nodes,and focuses on analyzing and solving the distributed coordination problems of UAV swarm in tasks such as cooperative navigation,cooperative round-up attack and cooperative decentralized attack,including UAV swarm autonomous navigation,cooperative round-up Simple multi task collaboration and complex multi task collaboration.The main contents and innovations of this paper include the following aspects:Firstly,the concept of UAV swarm operation is summarized,and the combat tasks,combat scenes and combat characteristics of UAV swarm are analyzed.Then,taking the uncertain environment as the entry point,taking the existing conventional UAV combat system and the small UAV swarm that may be equipped in the future as the object,and taking the maritime penetration and attack operation as the background,this paper focuses on the impact of the uncertain environment on the UAV swarm combat effectiveness,and expounds the UAV swarm combat process in detail,Using the fuzzy UML formal modeling analysis method,the UAV swarm operation process is modeled and analyzed,and its fuzzy activity diagram is mapped to fuzzy Petri net.Combined with examples and expert evaluation data,various performance indexes of the model are analyzed.It is pointed out that the cooperative execution of operation tasks by UAV swarm is the key node affecting the operation efficiency of UAV swarm.Finally,the key problems of UAV swarm distributed cooperation are analyzed.Secondly,a UAV swarm cooperative navigation method based on improved flocking model is proposed.On the one hand,a novel flocking model(c-flocking)is designed,which comprehensively considers five motion control rules: short-range exclusion,medium-range maintenance,long-range attraction,obstacle avoidance and target area navigation; On the other hand,based on the c-flocking model,an improved flocking model(o-flocking)is obtained through the GOF optimization framework combining genetic algorithm and flocking model.Through simulation experiments,the reliability,adaptability,scalability and superiority of the improved flocking model in completing the swarm autonomous navigation task are verified.In v-rep simulation platform,the feasibility of this method is verified by UAV swarm autonomous navigation simulation experiment.In addition,we also provide a simple way of thinking for researchers or users to solve problems.As long as a simple model is established for a specific task and environment,and the velocity formula of UAV is abstracted,a solution with superior performance can be obtained quickly.This greatly reduces the workload of manually adjusting parameters and improves the efficiency of task completion.Thirdly,a UAV swarm cooperative Roundup method based on improved GRN model is proposed.The traditional round-up model requires UAVs to form groups and exchange information such as round-up location allocation and path planning in real time.This paper proposes a UAV swarm cooperative round-up model based on improved GRN.The cooperative round-up behavior can be completed only by relying on the interaction of concentration information and UAV independent decision-making,which greatly reduces the difficulty and complexity of UAV cooperative control.In the v-rep simulation platform,the feasibility,robustness and scalability of the method are verified by UAV swarm simulation experiments.The feasibility of the method proposed in this paper is verified in the real round-up experiment of multiple UAVs against ground UAVs.In addition,in the kilobots robot swarm experimental platform with limited communication,perception,positioning and motion capabilities,100 robots have realized the hunting behavior of more than 10 targets.Experimental results show that the proposed algorithm can well implement the UAV swarm cooperative round-up task,and the method has good generalization.Fourthly,a multi task cooperative decision-making method of UAV swarm based on threshold behavior tree is proposed.An improved adaptive task allocation method(pinao)is designed based on the background of UAV swarm cooperative decentralized attack on different types of targets.This method enables each UAV in the swarm to obtain and analyze the task demand information and task status information within the sensing range through infrared sensors,visual sensors and local communication,so as to infer the expected value of the global task allocation proportion.Then the task type and motion direction are determined independently,and finally the UAV group reaches the expected task allocation proportion.The feasibility of pinao is experimentally verified under different conditions such as fixed target proportion,dynamic target proportion,number of dynamic UAVs and complex target distribution.By comparing with the method of estimating the task transfer rate,the adaptive method and a variant of piano,it is verified that piano performs best in realizing the expected task allocation proportion of UAV group.In addition,in the v-rep simulation platform,the feasibility of this method is verified by the multi task collaborative simulation experiment of UAV swarm.Finally,a multi task cooperation method of UAV swarm based on multi-level variable gene regulation network is proposed.Based on the multi task scenarios such as UAV swarm search,swarm siege and swarm transportation,a new structural model mlv-grn is proposed,which enables the UAV swarm to complete multiple tasks in a limited environment.Mlv-grn consists of three main modules and corresponding methods.In the decision module,we use the threshold based group behavior tree method tsbt to determine which form should be generated according to the perceived environmental information.In the morphological generator,we generate a specific morphology called rage based on the multi gene regulatory network method.In the behavior driving module,we use the self-organizing motion method csgo based on protein concentration to control the action,so as to drive the UAV to the appropriate state according to the shape.The simulation results show that mlv-grn enables the UAV group to complete multiple tasks in a limited environment.In addition,the experimental results show that the feasibility,adaptability and scalability of mlv-grn are better than Che GRN,he GRN and ntmlv-grn,especially in the success rate,intensity and comprehensive performance.