某火箭武器抗干扰位置随动系统控制研究

发布时间：2018-03-26 20:03

本文选题：位置随动系统　切入点：负载扰动　出处：《南京理工大学》2017年硕士论文

【摘要】：随着国防科技的发展,对火箭炮武器系统的自动化水平和跟踪精度的要求也越来越高,位置随动系统是组成火箭炮整个系统的必要环节,为了得到良好的控制性能,我们必须对位置随动系统提出更高要求。火箭炮在发射时,系统的质心位置、刚度、阻尼和转动惯量均发生很大变化,系统参数具有不确定性,且火箭炮在发射状态时受连续燃气流冲击导致定向器产生振动,使得后续发射在此发射环境下命中精度降低。因此,如何克服这些扰动及系统不确定性对其影响,提高火箭炮位置随动系统的跟踪精度和抗干扰能力,是现阶段需要研究的问题。本文以某型多管火箭炮的位置随动系统为研究对象,介绍了多管火箭炮的机械结构以及随动系统的组成及工作原理,建立交流伺服电机的数学模型以及多管火箭炮的交流伺服系统的仿真数学模型,分析了位置随动系统的负载扰动,为辨识以及控制策略研究奠定了基础。采取了离线训练与在线调整的辨识策略。首先采用径向基函数(Radial Basis Function,RBF)神经网络对系统进行离线辨识,针对传统RBF神经网络参数确定问题,利用改进粒子群算法优化RBF神经网络的中心,宽度及权值,离线训练得出的参数作为在线辨识器参数初始值,避免了振荡现象,加快了神经网络的收敛速度。为抑制位置随动系统的负载扰动,减少对位置随动系统的干扰。本文设计了基于RBF神经网络的单神经元自抗扰控制器,利用自抗扰控制器中的改进fal函数的扩张观测器将系统发射时的燃气流冲击等扰动归于扩张状态。并考虑到传统自抗扰控制器的参数多,难确定的问题,本文利用单神经元控制器(Single Neuron Controller,SNC)来代替非线性状态误差反馈器(Nonlinear state error feedback control,NLSEF),其权值利用 RBF 神经网络在线辨识器的辨识信息来在线自动调整。结合国家973项目进行了实验,将所设计的控制策略在实验上进行验证,实验结果表明:该控制策略能够有效抑制负载扰动,具有较强的抗干扰能力。
[Abstract]:With the development of national defense science and technology, the requirement of automatic level and tracking precision of rocket launcher weapon system is more and more high. The position servo system is the necessary link to make up the whole system of rocket launcher, in order to obtain good control performance, We must put forward higher requirements for the position servo system. When launching the rocket launcher, the position of the center of mass, stiffness, damping and moment of inertia of the system all change greatly, and the system parameters are uncertain. Furthermore, the impact of continuous gas flow on the rocket launcher causes the vibration of the directional device, which reduces the accuracy of the subsequent launch in this environment. Therefore, how to overcome these disturbances and the influence of the system uncertainty on it, To improve the tracking accuracy and anti-jamming ability of the position tracking system of rocket launcher is a problem that needs to be studied at present. This paper takes the position servo system of a certain type of multi-barrel rocket launcher as the research object. This paper introduces the mechanical structure of multi-barrel rocket launcher, the composition and working principle of the servo system, establishes the mathematical model of AC servo motor and the simulation mathematical model of AC servo system of multi-barrel rocket launcher. The load disturbance of the position servo system is analyzed, which lays a foundation for the research of the identification and control strategy. The off-line training and on-line adjustment strategy are adopted. Firstly, the radial basis function (RBF) neural network is used to identify the system off-line. Aiming at the problem of parameter determination of traditional RBF neural network, the center, width and weight of RBF neural network are optimized by using improved particle swarm optimization algorithm. The parameters obtained by off-line training are taken as the initial values of parameters of on-line identifiers, and the oscillation phenomenon is avoided. In order to suppress the load disturbance of the position servo system and reduce the interference to the position follow-up system, a single neuron ADRC controller based on the RBF neural network is designed in this paper. The extended observer of the improved fal function in the ADRC is used to return the disturbance such as the gas flow shock to the extended state when the system is launched, and considering the problem that the traditional ADRC has many parameters and is difficult to determine. In this paper, a single Neuron controller (single Neuron controller) is used to replace nonlinear state error feedback control (NLSEFN) of nonlinear state error feedback control (NLSEFN). The weights are automatically adjusted on line by using the identification information of RBF neural network on-line identifiers. The experimental results show that the proposed control strategy can effectively suppress the load disturbance and has strong anti-interference ability.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TJ393;TP183

【参考文献】