基于粒子群神经网络的导弹自适应逆控制  

Adaptive Inversion Control of the Missile Based on Neural Network with Particle Swarm Optimization

作　　者：梁坤[1] 宋书中[1] 马建伟[1] 

机构地区：[1]河南科技大学电子信息工程学院,洛阳471003

出　　处：《导弹与航天运载技术》2013年第2期50-54,共5页Missiles and Space Vehicles

基　　金：航空科学基金资助项目(20110142002)

摘　　要：针对直/气复合控制导弹的模型和参数的不确定性和多操纵机构的耦合问题,采用动态逆作为基本控制律,利用神经网络对非线性函数的逼近特性,设计在线模型参考自适应神经网络对慢回路的逆误差进行补偿,并用粒子群算法优化神经网络的参数,避免了局部寻优并提高了学习效率。通过仿真证明该方法克服了模型和参数不确定性以及外来干扰对系统稳定性的影响,对逆误差进行了有效的补偿,提高了系统的鲁棒性。As the nonlinear effect and coupling character of the flight dynamics become a big problem to the blended aero and reaction jet flight control systems, the base control law is dynamic inversion that is sensitive to the accurateness. Considering fitting characteristics of neural network, we designed an adaptive Neural Network controller with Particle Swarm optimization （PSO） to account for the Dynamic Inverse error. The method avoided the neural network of local optimization and improve the learning efficiency .The simulation results proves that the new flight control system conquered the aerodynamic modeling inaccuracies and the external disturbances. The compensation of the inverse error is effective, and the robustness of the control system is improved greatly.

关 键 词：复合控制 动态逆 神经网络 粒子群算法 鲁棒性 

分 类 号：TJ765.2[兵器科学与技术—武器系统与运用工程]