【作者】 张迎春；

【导师】 高长生；

【作者基本信息】 哈尔滨工业大学 ， 航天工程（专业学位）， 2019， 硕士

【摘要】 导弹是借助旋转体制可以减弱弹体不对称因素影响,省去一套舵控系统、制导天线等,因此制导兵器以及再入弹头多采用旋转体制,随着制导和硬件发展,旋转导弹双通道控制成为主流。新的攻防形式下旋转导弹又被赋予了新的使命,导弹通过旋转可以减弱激光反导武器的毁伤,也可以形成旋转弹道用于机动突防,对于高空拦截,旋转更有利于脉冲发动机下优化控制,因此旋转导弹仍有发展的活力。但是旋转也带了马格努斯效应、陀螺效应以及舵控延迟等通道间耦合因素,导致了自动驾驶仪设计参数稳定域减小,降低了导弹的锥形运动稳定性以及产生指令响应误差,因此要对施加必要的解耦措施,通道导弹飞行空间、速度包线增大,制导对机动性要求提高需要大攻角飞行,旋转下的气动就更加复杂,因此气动参数存在非线性、大范围快时变的的特点,因此控制设计还必须具有抗干扰和鲁棒性。所以本文开展了以下几个方面的研究工作:建立了准弹体坐标系下和弹体系下的旋转导弹非线性耦合动力学模型,仿真分析了旋转导弹单通道方波调宽和双通道旋正弦余弦控制从准弹体下到弹体系下的舵控实现,初步分析了旋转导弹的动力学和耦合特性,为后文奠定了基础。基于自抗扰控制技术设计了姿态和过载自抗扰控制器。为此首先介绍了控制方法并相应地变换了控制模型,模型中忽略的小量由控制的鲁棒性和抗干扰性保证,主要是通过状态实时观测和补偿把系统解耦成二阶串联积分型,考虑到鸭舵对过载的贡献非小量,利用指令信号减去舵控贡献量后滤波和微分信号反馈置零的办法有效克服了舵控震荡,最后给出了参数摄动、外加扰动下仿真,验证了控制器的无超调、抗扰动、鲁棒性强等良好品质。基于RBF神经网络设计了直接逆解耦控制器。为此介绍了逆系统理论证明了旋转导弹短周期角运动方程的可逆性,给出了RBF神经网络学习训练过程,它的引入克服了动态逆依赖于准确数学模型的缺点,将神经网络串联在系统之前构成了伪线性系统,在此解耦基础上使用非线性反馈作为控制器,同解析形式的逆系统相比验证了神经网络很好逼近了系统的逆,在抗扰动分析中发现以上神经网络逆控制缺乏抗扰动性,为此分析伪线性系统特点增加了简单反馈回路,极大的增强了抗扰动性。将导弹视线系加速度和准弹体系加速度近似等同,从而制导控制连接构成一体化的控制方程,并基于此采用自抗扰控制的方法进行解耦设计,并根据零化视线角速度定义了新的状态量作为制导状态量,完成了对制导单独解耦和制导控制一体化解耦设计,有效的避免了对此状态量指令跟踪式观测误差的影响,在非线性反馈中论述了精度因子对此制导中控制量的作用,并提出了分段采用不同精度因子达到良好的制导控制实现的想法。更多还原

【Abstract】 Spinning missiles can reduce the influence of asymmetry factors,and save a pair of antennas and rudders,thus guided weapons and reentry warheads are generally in rolling motion.With the development of guidance precision and hardware,dual-actuator control scheme is increasingly used in spinning missiles.In face of new situation of attack-defense counterwork,spinning missiles are expected to protect body from laser weapons and enhance penetration capability by spiral trajectory and conduce to the use of impulse thrust.Thus,spinning missiles are still developing actively.But spinning are inevitably accompanied by magnus effect,gyroscopic effect and control coupling,furthermore these coupling effects reduce the range of autopilot stability domain and impact the stabilization of cone motion,eventually leading to response error.In conclusion,decoupling method should be applied to controller.With consideration of flight environment,the speed and altitude of spinning missiles vary in a large range,and aerodynamic characteristics under high angles of attack become more complicated.In conclusion,aerodynamic is with characteristics of nonlinearity and sharp variation.So,the controller should be designed with characteristic of robustness and anti-interference.Therefore,the research work is listed as follows:The nonlinear and coupling dynamic model is established and analyzed in quasi-body as well as body coordinate system.Simulations are given that rudder signal is transformed from quasi-body coordinate to body coordinate,which demonstrates the method of control of single or double channel spinning missiles.An attitude controller and an acceleration tracking controller are designed based on active disturbance rejection control(ADRC)technique,then the model is decoupled as second order integrator system.Those parts ignored can be tolerated by the anti-interference controller when model is transformed to adapt to ADRC.Further than that,this thesis considers that canard wings will contribute to lift which cannot be ignored in this model,so this thesis uses command signal compensation and doesn’t use its differential as feedback,rudders are saved from vibration in this way.Finally,simulation results are given under condition of parameter perturbation and disturbance that prove the controller performs well.A decoupling controller is also designed based on RBF neural network and dynamic inversion.This thesis proves model of short-period motion is invertible,so RBF neural network are used to approximate its invertible model,then they are combined as pseudo-linear system,with this the decoupling work is done.On this basis,this thesis uses NLSFF as a controller.In order to enhance ability of anti-interference,this thesis add extra compensation for error from inversion to the controller,which is testified to be reasonable and effective.Acceleration in line of sight coordinate system and quasi-body coordinate system is proximate,thus,guidance equation and control equation can be connected as integrated guidance and control model.Then this thesis completes the decoupling design based on ADRC,this thesis redefines state variables in order to avoid influence of error from ESO and discusses the effect of precision coefficients of NLSFF on control variables,and the controller are thought to perform better with different coefficients in different sections of guidance.更多还原