基于小脑模型关节控制器的无人机集群快速终端滑模容错控制  

Fast terminal sliding mode fault tolerant control for UAV swarm based on cerebellar model articulation controller

作　　者：钱默抒 吴柱[1] 王村松 展凤江 QIAN Moshu;WU Zhu;WANG Cunsong;ZHAN Fengjiang(College of Electrical Engineering and Control Science,Nanjing Tech University,Nanjing 211816,China;Key Laboratory of Advanced Technology for Small and Medium-Sized Unmanned Aerial Vehicle,Ministry of Industry and Information Technology(NUAA),Nanjing 210016,China)

机构地区：[1]南京工业大学电气工程与控制科学学院,南京211816 [2]南京航空航天大学工信部中小型无人机先进技术重点实验室,南京210016

出　　处：《航空动力学报》2023年第10期2441-2449,共9页Journal of Aerospace Power

基　　金：国家重点研发计划(2022YFB3305300);国家自然科学基金重点项目(62333010)。

摘　　要：针对受执行器故障和外界干扰影响的集群无人机(UAV)协同编队控制问题,提出了一种自适应快速非奇异积分滑模(FNISM)容错控制(FTC)方法。为使集群无人机在执行任务时具有良好的协同跟踪性能,通过对无人机实际飞行情况的分析,考虑了无人机编队飞行时的执行器故障和尾涡扰动等对跟踪性能的不利影响。采用小脑模型关节控制器神经网络(CMANN)来估计并消除外部干扰的影响,同时运用CMANN逼近补偿执行器故障。研究表明:所提出的容错控制方案可以保证无人机编队闭环系统在故障情况下的最终一致有界稳定,并且可以通过减小滑模设计参数提高收敛速度,通过增大虚拟和实际控制器参数提高控制精度。4架无人机的集群编队在该方法、基于径向基神经网络(RBFNN)的鲁棒动态面容错控制、比例微分(PD)滑模容错控制3种方法下的对比仿真结果表明该方法在无人机集群编队出现故障时具有更优异的协同控制性能。An adaptive fast nonsingular integral sliding mode(FNISM)fault tolerant control(FTC)approach was developed for the cooperative formation control problem of unmanned aerial vehicle(UAV)swarm affected by actuator faults and external disturbances.To achieve a good cooperative tracking performance of UAV swarm during mission execution,the actual flight condition was analyzed for the adverse effect of actuator failures,vortex disturbances and so on.The cerebellar model articulation controller neural network(CMANN)was introduced to approximately compensate the external disturbances and the effect of actuator faults.Result showed that the closed-loop formation system of UAVs achieved ultimately uniformly bounded stability under the failures,meanwhile the convergency rate was increased by reducing the sliding mode parameter and the control precision was improved by increasing the virtual and actual controller parameters.The simulation experiments of four UAVs swarm were carried out respectively under three approaches,including the proposed approach,RBFNN based robust dynamic surface FTC and robust PD sliding mode FTC.The comparative results showed that the designed approach has superior control performance for the faulty formation control system of UAV swarm.

关 键 词：无人机集群 容错控制 快速非奇异积分滑模 协同控制 小脑模型神经网络 

分 类 号：V249.122.2[航空宇航科学与技术—飞行器设计]