基于改进增益规划法的导弹稳定控制回路设计  

A New Design Approach for Air Defense Missile Autopilot Based on Genetic Algorithms and Artificial Neural networks

作　　者：贝超[1] 周凤岐[1] 周军[1] 

机构地区：[1]西北工业大学,西安710072

出　　处：《航天控制》2002年第3期20-26,共7页Aerospace Control

摘　　要：定点设计和多点拟合是利用传统增益规划法进行控制回路设计的基本环节。针对传统增益规划方法中存在的问题 ,本文将基于遗传算法的特征点处控制系统参数设计方法和基于神经网络的稳定控制律拟合方法组合在一起 ,形成一种改进增益规划方法用于导弹稳定控制回路设计 ,以克服原有设计方法中人工试凑较多的弊端 ,降低人为因素对设计工作的影响 ,实现设计工作程序化、自动化和通用化。Following analysis of the traditional air defense missile autopilot design approach, a new gain scheduling method based on Genetic Algorithms (GA) and Artificial Neural networks (ANN) is presented, in which GA is used to automatically adjust the controller gains while ANN is used for the fitting of the control law over the missile flight envelop based on the prophase optimization done by GA Design methods and control schemes are tested by numerical simulations Both theoretical research and simulation results lead to the conclusion that the new approach is more designer-independent, automatic and general in use compared with the traditional ways

关 键 词：设计 导弹 控制系统 增益规划 遗传算法 神经网络 

分 类 号：TJ765[兵器科学与技术—武器系统与运用工程]