摘要
针对传统预设性能控制(Prescribed Performance Control,PPC)方法处理输入受限问题时极易诱发控制奇异的缺陷,为输入受限乘波体飞行器(Waverider Vehicle,WV)提出了一种基于神经逼近的新型非脆弱PPC方法,设计了补偿系统分别处理速度控制输入与高度控制输入的饱和问题。进一步,利用补偿系统的状态,构造了新型自适应调整项,并对传统PPC的约束包络进行改进。引入神经网络对WV归一化的未知项进行在线逼近,保证了控制鲁棒性。所提方法的优越性在于弥补了传统PPC方法的脆弱性缺陷,并显著降低了控制复杂度与在线学习量。最后,通过数值仿真验证了所提方法的有效性与优越性。
Aiming at the defect that traditional PPC usually leads to control singularity when dealing with the input constrained problem,a new non-fragile prescribed performance control(PPC) methodology is proposed based on neural approximation for waverider vehicle(WV).To deal with the saturation of velocity control input and altitude control input,compensated systems are devised.Furthermore,the states of compensated systems are used to construct adaptive re-adjusting terms that are further applied to improve the constraint envelopes of traditional PPC.Besides,neural networks are introduced to approximate the WV normalized unknown terms,which guarantees the robust performance.The advantage of the proposed method is that it overcomes the fragile defect of traditional PPC,and also reduces the control complexity and online computational load.Finally,the effectiveness and superiority of the exploited method are validated via numerical simulation.
作者
卜祥伟
姜宝续
Bu Xiangwei;Jiang Baoxu(Air and Missile Defense College,Air Force Engineering University,Xi’an 710051,China;College of Graduate,Air Force Engineering University,Xi’an 710051,China)
出处
《航空兵器》
CSCD
北大核心
2022年第6期7-14,共8页
Aero Weaponry
基金
国家自然科学基金项目(61873278)。
关键词
乘波体飞行器
预设性能控制
脆弱性
神经网络
输入受限
控制系统
武器
waverider vehicle
prescribed performance control
fragility
neural network
input constraints
control system
weapon
