首页> 外文会议>AIAA infotech@aerospace conference and exhibit >Adaptive Critic Optimization of Dynamic Inverse Control
【24h】

Adaptive Critic Optimization of Dynamic Inverse Control

机译:动态逆控制的自适应评论批评优化

获取原文

摘要

Dynamic Inverse (DI) control is combined with adaptive critic optimizati7on to create a hybrid adaptive controller. Because the adaptive critic is applied to the combination of the system and its dynamic inverse, the critic co-state network does not need to be pre-trained as it can be initialized with the solution to the Algebraic Ricatti Equation (ARE) applied to a "perfect inverse" identity system. Moreover, since the critic network need only represent deviations of the co-state from this initial perfect inverse solution, it can be implemented as a simple low order sigma-pi network. Thus, adaptive critic and SNAC methods bring adaptive optimal control to dynamic inversion control of nonlinear systems. Likewise, dynamic inversion brings the straightforward control design to adaptive critic and SNAC methods and replaces pretraining with a straight forward initialization. In the presence of system modeling error, damage, or failure, this hybrid system adapts the control to continue to minimize the cost function and achieve optimal control as defined by that cost function. Application to a simple nonlinear SISO system which has an exact optimal control solution is presented, as well as application to a general aviation aircraft longitudinal flight controller.
机译:动态逆(DI)控件与Adaptive Instemizati7on组合以创建混合自适应控制器。由于自适应批评者应用于系统的组合及其动态逆,因此批评者共同状态网络不需要预先接受预先训练,因为它可以通过应用于应用于A的代数Ricatti等式(A)的解决方案初始化“完美逆”身份系统。此外,由于批评网络只需要从该初始完美的逆解决方案代表共同状态的偏差,因此它可以实现为简单的低阶Sigma-PI网络。因此,自适应评论家和SNAC方法为非线性系统的动态反演控制带来了自适应最优控制。同样地,动态反演为自适应批评和SNAC方法带来了直接的控制设计,并用直线初始化替换预先预先预测。在系统建模误差,损坏或失败的存在下,这种混合系统适应了控制,继续最小化成本函数并实现由该成本函数定义的最佳控制。应用于简单的非线性SISO系统,其具有精确的最佳控制解决方案,以及应用于通用航空飞机纵向飞行控制器。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号