【作者】 段镇；

【导师】 陈娟； 贾宏光；

【作者基本信息】 中国科学院研究生院(长春光学精密机械与物理研究所) ， 机械电子工程， 2014， 博士

【摘要】 无人机在整个航空领域扮演着越来越重要的角色，飞控系统是无人机系统的核心组成部分，其性能直接影响着无人机的飞行性能和飞行品质，直接关系到无人机的飞行安全。本文在总结相关领域研究成果与发展现状基础上，针对某固定翼无人机，从空中飞行与地面滑跑动力学建模、模型简化、控制律设计及实现、干扰观测器设计、控制律评估与确认、半物理仿真与飞行试验等方面展开了相应研究。首先，在考虑发动机扭转力矩、推力偏心及初始停机角影响的情况下，建立了无人机空中飞行和地面滑跑的六自由度非线性模型，将无人机沿跑道中心线作定常直线无侧滑运动作为地面滑跑基准运动，实际运动是在基准运动基础上的小扰动运动，利用小扰动原理显式地给出了无人机空中飞行、地面三轮及两轮滑跑的线性化模型表达式，为控制系统设计与进一步的评估验证打下基础。研究了鲁棒伺服LQR控制方法，采用鲁棒伺服LQR控制及经典控制相结合的方法设计了无人机空中飞行纵向运动及横侧向运动的控制律，通过仿真证明上述方法比常规PID控制器响应更加平缓，俯仰角响应超调量减小50%，滚转角响应超调量减小70%，且很好地抑制了响应初期舵指令突变而导致无人机瞬间产生较大角速率的现象，大大降低了对机体可用过载的要求；对航迹跟踪回路采用偏航角速率、偏航角、侧偏速度及偏航距反馈的控制方式，仿真证明该方法能使无人机很好地跟踪预设航迹。为消除无人机飞行过程中铰链力矩干扰对控制精度的影响，将干扰观测器应用于飞行控制律的设计中，分析了干扰观测器与控制器设计的独立性以及干扰观测器的鲁棒稳定性；以盘旋飞行模态为例设计了基于干扰观测器的飞行控制律，通过仿真验证了带干扰观测器的控制律可消除不带干扰观测器时产生的3.2°滚转角稳态误差，滚转角控制精度得到大幅提高；采用分块离散方法和单模态瞬变抑制法分别解决了控制律工程实现中的非线性控制律离散化及控制律切换时舵面跳变的问题。针对自主滑跑模型的零极点分布范围较大且滑跑模型中实际参数不易准确获取的情况，对三轮滑跑的纵向控制采用升降舵保证前后轮压力比为定值的控制方式，对横侧向控制的偏航角速率、偏航角和侧偏距回路分别设计了基于参考模型的自适应控制律、定参数控制律及增益调节控制律；对两轮滑跑阶段横侧向控制的偏航角及侧偏距回路分别设计了基于PD增益自适应调节控制律和定参数控制律；通过仿真验证了当无人机存在2.5°初始偏航角和0.2m初始侧偏距的情况下，设计的控制律能迅速消除偏差，使无人机沿跑道中心线滑跑。为减少试飞次数，提高飞行试验成功率，在无人机空中飞行存在气动、惯性等不确定因素时，采用基于多项式的方法基于不稳定特征值判断准则对控制律进行了评估和确认，得到标称状态下无人机在整个飞行包线内的可用迎角为[-2°,18°]，单个不确定参数摄动时的可用迎角范围为[0°,15°]，多个不确定参数同时摄动时可用迎角范围减小为[0°,11°]。最后，通过半物理仿真试验、滑跑试验与飞行试验对上述研究工作的实际应用效果进行了验证，半物理仿真试验与数字仿真吻合较好；无人机外场飞行试验俯仰角控制的实际稳态误差为0.2°，滚转角控制的实际稳态误差为0.4°，高度控制的实际稳态误差为1m，满足国军标对飞行控制系统的指标要求；对自主滑跑控制律完成了外场自主滑跑试验，结果表明，整个滑跑过程中的最大偏航角为2°，最大侧偏距为1.5m，且对侧风干扰有较强的抑制作用。更多还原

【Abstract】 UAV plays an increasingly important role throughout the whole aviation ministry.Flight control system is one of the core components of the UAV, whose performancedirectly affects the flight performances, flying qualities and flight safety. On the basisof summarizing the research results and development status of related fields, thisthesis has a central research on a fixed-wing UAV, including dynamics modeling offlying and taxiing, model simplifying, control law design and implementation,disturbance observer design, clearance of control law, hardware in loop simulationand flight test.Firstly, considering the engine torque, thrust eccentric and ground angle, the sixdegree of freedom nonlinear equations of flying and taxiing is established.Taking theUAV moving along the centerline of the runway constantly without sliding as thereference movement, the actual movement is small perturbation motion based on thereference movement. The linear state space expression of flying, taxiing with threewheels and two wheels is derived using small perturbation principle, which lays afoundation for design and clearance of the control system.The robust servo LQR control method is studied, the longitudinal and lateralflight control law is designed combining the robust servo LQR and the classicalcontrol. The simulation results shows that the response of the method is more gentlethan conventional PID controller, with the overshoot of pitch angle responsedecreasing by50%, and the overshoot of roll angle response decreasing by50%. Thephenomenon of a larger angle rate because of servo command mutation is suppressedin initial response, the requirements of available overload is greatly reduced. The control structure of flight path tracking loop uses the feedback of yaw rate, yaw angle,yaw speed and yaw distance. The simulation results show that this control method cantrack the default flight path accurately.Disturbance observer is applied to the flight control law design to eliminate theeffects of hinge moment to control accuracy during flight, the independence ofdisturbance observer design and controller design is proved, the robust stability ofdisturbance observer is analyzed. Taking the circle flight mode for example, thesimulation results show that the control law with a disturbance observer can eliminatethe3.2°roll angle steady-state error generated by the control law without disturbanceobserver, the control accuracy of the roll angle is improved. The nonlinear control lawdiscretion issues in engineering implementation is solved using the block discretemethod and the servos transition issues when the control law switching is solvedusing single-mode transient suppression method.For the situation of extensive pole-zero distribution of the taxiing model and theactual parameters are hard to obtained accurately, longitudinal control of three wheelstaxiing is designed to ensure that the front and rear wheels pressure ratio is constantthroughout the deflection of elevator. The model reference adaptive control law, givenparameter control law and gain scheduling control law are designed respectively forthe yaw rate, yaw angle and yaw distance loop of lateral control. The PD gainadaptive control law and given parameter control law are designed respectively for theyaw angle and yaw distance loop of two wheels taxiing lateral control. The simulationresults show that the control law can eliminate the deviation quickly when there existsinitial yaw angle2.5°and initial yaw distance0.2m, so as to the UAV taxis alongcenterline of the runway.To increase the success rate of flight test, the control law is cleared and verifiedbased on unstable eigenvalues criterion using polynomial approach when there existsthe inertia, pneumatic and other uncertainties during flight. The clearance results showthat the available angle of attack throughout the whole flight envelope in nominalstate is [-2°,18°], the available angle of attack range with single uncertain parameterperturbation is [0°,15°]，the available angle of attack range with multiple uncertainparameters simultaneously perturbation is reduced to [0°,11°].Finally, the research works mentioned above are verified through hardware inloop simulation for taxiing test and flight test. The simulation results agree well withthe digital simulation, the actual pitch angle control steady-state error is0.2°in outfield flight test, the actual roll angle control steady-state error is0.4°, the actualheight control steady-state error is1m. All the above meet the military standardrequirements. The outfield autonomous taxiing test is completed to verify the adaptivecontrol law. The results show that the maximum yaw angle is2°, the maximum yawdistance is1.5m, the crosswind disturbance is well inhibited.更多还原