Autonomous Safety Control Strategy Research And Verification Of Unmanned Aerial Vehicle

Unmanned aerial vehicle has the considerable potential of being applied in both military and civilian areas,where safety control is one of the critical techniques to guarantee its safe operation.With the development of the unmanned aerial vehicle,the control object,mission environment,and flight status tend to be complex increasingly.Meanwhile,the requirements for mission operation quality and safety performance are getting stricter.However,the practical application of the unmanned aerial vehicle suffers from multiple factors,such as static and dynamic obstacles in the complex environment,complex faults of system components,multiple disturbances,and system capability limitations.Firstly,as the environment change and flight space expansion,the unmanned aerial vehicle may encounter static and moving obstacles during the operation of a specific mission.A collision accident may take place due to aerodynamic constraints,the convergence rate of path planning,and the moving uncertainty of obstacles.Secondly,the physical attack and heavy working burden may take place during mission operation,which induces surface damage and actuator degradation.Thus,both the aerodynamic features and control input effectiveness are affected.Lastly,except for the modeling error including component installation error and aerodynamic coupling,load vibration and fuel consumption during the flight will further change the model parameters of the unmanned aerial vehicle,thus deteriorating the tracking accuracy.Therefore,how to further improve the collision avoidance,fault tolerance,and disturbance rejection capability in the presence of complex mission environments is a vital issue to prompt the safety control performance of the unmanned aerial vehicle.To solve these problems,the high safety control methods of the unmanned aerial vehicle are investigated in the thesis.The main innovative works are summarized as follows:1.The collision avoidance problem of the unmanned aerial vehicle is considered when confronting static and dynamic obstacles in complex environments.By incorporating the factors including inherent kinematics constraints,moving uncertainty of obstacles,and online convergence rate,a varying cell strategy-based fast path planning method is proposed.To be more specific,a collision prediction model with moving uncertainty of obstacles is established,where the kinematic constraints of the unmanned aerial vehicle are analyzed.These factors are incorporated into the reinforcement learning framework.The kinematic constraints of the unmanned aerial vehicle are integrated into the basic avoidance action design,where a varying cellbased strategy is developed to enable more flexible avoidance maneuvers.Furthermore,because of the low convergence rate of the current fixed-cell-based method in largescale space,a path planning method based on the forward search is developed.In comparison of the conventional methods,the proposed path planning method not only satisfies kinematic constraints but also improves the online convergence and collision avoidance rates highly.2.The degradation of mission quality and system capability of the unmanned aerial vehicle is considered under actuator fault.In order to achieve the quantitative estimation of system fault adaptability and system controllability,a fault-tolerant control method is developed with trajectory replanning,differing from path planning with healthy status(the first work).To be more specific,the trajectory tracking process is backpropagated to evaluate the mission object reachability based on the system dynamic model.The system capability analysis method is given thereby.Furthermore,a flight trajectory with performance degradation is replanned based on the fault information and evaluation index of system capability.According to the fault,position and attitude models,an integrated control-oriented model is established,where a refined fault-tolerant control method is proposed based on dynamic surface control.As compared to PID control and backstepping control methods,the proposed method not only enables to estimate system fault adaptability and mission accomplishment quantitatively,but also highly improves the trajectory tracking accuracy.3.The complex environment,external wind disturbance,and actuator fault are considered,which affect the flight safety of the unmanned aerial vehicle.When comparing to the second work with actuator faults solely,the differential flatness technique and dual-loop nonlinear disturbance observers are combined,such that an anti-disturbance and fault-tolerant control method based on trajectory dynamic adjustment is proposed.First,a differential flatness tool is implemented such that the velocity and acceleration can be formulated as flat variables.By making a trade-off between the trajectory parameter and the control inputs,a collision-free and dynamic adjustment trajectory planning method is developed online,where the obstacles in the environment are considered.Secondly,in the event of actuator faults and wind disturbance,the disturbance observers are exploited to identify and attenuate the effect of fault and disturbance.Moreover,by quantitatively evaluating system capability,the trajectory is adjusted dynamically or even emergency landing is activated,thus avoiding loss of flight control due to breaching system capability.The estimation result of fault and disturbance is fully used.Therefore,the safety control performance and mission success capability of the unmanned aerial vehicle are clearly ameliorated in working conditions with obstacles,faults,and disturbances.4.The coupling of model uncertainty,complex faults,and load vibration is considered,which affect the unmanned aerial vehicle during the mission operation.In order to reduce the conservativeness as done in the third work,due to regarding faults and disturbances as “lumped disturbances”,a composite hierarchical control method based on fault and disturbance separation is developed.Both finite time observer and disturbance observer are combined.Specifically,the transmission mechanism,action channel,and time-varying characteristic of faults and disturbances are investigated,which starts from the physical mechanism of malfunctions and inner and external disturbances.As a consequence,the separable condition for multiple disturbances and faults is proposed.By virtue of the frequency of load vibration disturbance,a disturbance model and disturbance observer are designed to achieve accurate estimation and compensation of periodic disturbance.Meanwhile,a finite time observer is developed to online compensate for the faults and disturbances with slow time-varying and derivation-bounded properties.In comparison of the third work,the complex faults and multiple disturbances of unmanned aerial vehicles are separated and compensated,which highly enhance the tracking accuracy,fault-tolerant and antidisturbance capability of the trajectory tracking process.Lastly,both numerical simulations and real-world experiments are carried out for the sake of verifying the effectiveness of the developed safety control schemes.