基于模糊神经网络的导弹直接力/气动力复合控制系统设计  

Improving Maneuverability and Guidance Performance of Air-to-Air Missile with Compound Control System

作　　者：陈旿[1] 贾晓洪[2] 李友年[2] 刘忠[2] 

机构地区：[1]西北工业大学自动化学院,陕西西安710072 [2]中国空空导弹研究院,洛阳471009

出　　处：《西北工业大学学报》2006年第4期423-426,共4页Journal of Northwestern Polytechnical University

基　　金：航空科学基金(03D12004)资助

摘　　要：为了提高导弹的机动性、敏捷性和导引精度,新型导弹大多采用直接力/气动力复合控制方案。由于神经网络对于系统非线性变化具有较强的适应能力,因而在解决直接力/气动力复合控制中的时变非线性问题时有较明显的优点。在建立导弹非线性模型的基础上,采用模糊小脑模型神经网络(FCMAC)与动态逆相结合的方法设计导弹控制器,该方法结构简单,收敛速度快,易于硬件实现。数字仿真结果表明该方法对导弹系统参数的非线性变化具有很强的适应性。Aim. Ever advancing air war technology is constantly pushing air-to-air missile to have better and better maneuverability and performance. We seek to go one step further in this endless game by designing a new compound control system, which combines traditional aerodynamic control with reaction jets. In the full paper, we explain in detail the design of this new compound control system; in this abstract, we just add some pertinent remarks to listing the three topics of explanation= （A） FCMAC （Fuzzy Cerebellar Model Articulation Controller） neural network ; CMAC was first proposed by J. S. Albus in 1975; Fig. 1 in the full paper gives the architecture of FCMAC neural network; （B） nonlinear pitch model of missile with aerodynamic-control/reaction-iet compound system; under topic （B）, we derive eqs. （1） through （4） in the full paper; （C） design of controller; under topic （C） we derive eqs. （5） through （12） in the full paper; we like to mention in particular that eq. （11） is used by designers to perform what we call feedback linearization or dynamic inversion. The new compound controller has simple architecture, quick learning convergence, and is easy to be implemented by hardware. Numerical simulation results, given in Figs. 2 and 3 in the full paper, illustrate that the presented neural network control method possesses strong adaptability to system nonlinear variations.

关 键 词：动态逆 神经网络FCMAC 直接力控制 非线性 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]