Active Fault-tolerant Control Of Multi-agent Systems And Its Application In Unmanned Aerial Vehicle Formation

With the development of control science and the acceleration of industrialization,it is difficult to complete complex control tasks by relying on a single agent system.It is necessary to introduce a plurality of cooperative agent systems,namely multi-agent systems(MASs),in which various agents cooperate with each other to complete complex control tasks.Since MASs have better performance,more comprehensive functions and stronger adaptability,the application of MASs in unmanned aerial vehicles(UAVs)formation has received considerable attention.Close formation flight of multiple UAVs is a special tight formation pattern.Due to vortex effects in close formation flight of multi-UAV,UAVs in proper vortex-effect position can obtain smaller induced drag and greater lift,thus saving fuel con-sumption of close formation systems(CFSs)and increasing endurance mileage of multi-UAV as well as achieving better economic benefits.It cannot be ignored that MASs are vulnerable to faults and faults are easy to spread in MASs.In order to operate safely and reliably,MASs and multi-UAV CFSs need to establish an active fault-tolerant control(AFTC)methodology to guarantee the stability,reliability and safety even when MASs suffer from faults.In this dissertation,the AFTC methodologies of MASs and its applications in UAV formation are investigated.The main work in this dissertation is establishing a theoretical framework of AFTC methodologies of homogeneous MASs,heterogeneous MASs and multi-UAV CFSs,and provides a powerful solution for improving the reliability and safety of MASs and CFSs.The mechanism analysis of faults is firstly conducted for MASs,including the description and modeling of faults in homogeneous MASs and heterogeneous MASs.Then,the modeling of multi-UAV CFSs is conducted,including the analysis and modeling of vortex effects,multi-UAV CFSs and Lead-Wing CFSs in lines.The proposed modeling provides a model basis for the following chapters on AFTC schemes of MASs and its application in UAV formation.Three types of AFTC methodologies are proposed in MASs,namely,the distributed adaptive compensation-based AFTC scheme of homogeneous MASs,the decentralized fault estimation-based AFTC scheme of homogeneous MASs,and the decentralized sliding-mode control-based AFTC scheme of heterogeneous MASs.The main research contents are as follows:A distributed adaptive compensation-based AFTC methodology is proposed for homogeneous MASs in the presence of system nonlinearities,uncertainties and actuator faults.The fault detection hierarchy,the healthy control hierarchy and the AFTC hierarchy constitute the hierarchical structure of homogeneous MASs.The fault-tolerant consensus control protocol is switched from the healthy control hierarchy to the AFTC hierarchy once the actuator fault is detected in the fault detection hierarchy.A distributed adaptive compensation-based active fault-tolerant consensus control algorithm is developed in the AFTC hierarchy on the basis of the instantaneous and integral estimations to compensate simul-taneously for system nonlinearities,uncertainties and actuator faults and to guarantee the mean-square consensus of homogeneous MASs.A decentralized fault estimation-based AFTC methodology is proposed for homogeneous MASs with actuator and sensor faults.The states and faults of homogeneous MASs are estimated simul-taneously by merging the decentralized unknown input observers in fault estimation systems.The distributed active fault-tolerant controllers with asymptotically stable targets and with asymptotically tracking targets are proposed in fault-tolerant control systems.The constant gain-based and node-based AFTC algorithms are designed in the fault-tolerant controllers with asymptotically stable targets.The edge-based active fault-tolerant tracking control algorithm is added in the fault-tolerant controller with asymptotically tracking targets.The fault-tolerant control and fault-tolerant tracking control algorithms in the distributed active fault-tolerant controllers are developed to cope with bidirectional interactions and to guarantee the asymptotic stability and tracking property of homogeneous MASs.A decentralized sliding-mode control-based AFTC methodology is proposed for heterogeneous MASs in the presence of matched disturbances,unmatched nonlinear interaction and actuator faults.By designing an integral sliding-mode observer and merging an adaptive upper bounding law,the iteration and iteration-free algorithms in the decentralized output sliding-mode control-based AFTC schemes are developed to compensate the matched and unmatched components,and to guarantee the asymptotic stability of heterogeneous MASs and the ultimate boundedness of estimation errors.Three types of AFTC methodologies are proposed in multi-UAV CFSs,namely,the multiple-model adaptive state feedback control-based AFTC scheme and the_∞adaptive output feedback control-based AFTC scheme of Lead-Wing CFSs,and the integration of the decentralized fault estimation and distributed AFTC schemes of Lead-Wing CFSs in lines.The main research contents are as follows:The position-tracking and attitude-tracking control problem of Lead-Wing CFSs modeled with vor-tex effects in the presence of loss of effectiveness,lock-in-place/hard-over and unknown multiplicative actuator faults is investigated.On the one hand,a multiple-model adaptive state feedback control-based AFTC methodology is proposed on the basis of_∞control design to guarantee the asymptotic sta-bility of Lead-Wing CFSs,the boundedness of errors and the asymptotic tracking property of attitudes and positions.On the other hand,actuator fault parameters are identified by merging an unknown in-put observer and designing an adaptive law with projection functions.Then,an_∞adaptive output feedback control-based AFTC methodology is proposed to guarantee the asymptotic stability of Lead-Wing CFSs and the asymptotic tracking property of attitudes and positions while the Lead UAV is being maneuvered.The position-tracking and attitude-tracking control problem of Lead-Wing CFSs in lines modeled with vortex effects under simultaneous actuator and sensor faults is investigated.On the basis of the estimated state and fault information from unknown input observers in the fault estimation systems and the relative output information from neighboring heterogeneous UAVs,an integration of the decen-tralized fault estimation and distributed AFTC methodologies is developed to deal with bidirectional interactions between fault estimation systems and fault-tolerant control systems,and to guarantee the asymptotic stability and_∞performance of Lead-Wing CFSs in lines.In this dissertation,the designed AFTC algorithms are successfully applied to the multi-UAV CFSs.Simulation results are presented to validate the effectiveness of the proposed AFTC algorithms of MASs and CFSs.