Discrete-time hypersonic flight control based on extreme learning machine
Discrete-time hypersonic flight control based on extreme learning machine
复制标题
DOI：
10.1016/j.neucom.2013.02.049
复制
发表时间：
2014-03
期刊：
影响因子：
6
通讯作者：
B. Xu;Yongping Pan;Danwei W. Wang;F. Sun
中科院分区：
文献类型：
--
作者：
B. Xu;Yongping Pan;Danwei W. Wang;F. Sun
本文描述了一种通用高超声速飞行器（HFV）纵向动力学的神经网络控制器设计。动力学转化为严格反馈形式。考虑到不确定性，神经控制器的单隐层前馈网络（SLFN）的基础上构建。隐节点参数的修改使用极端学习机（ELM）通过分配随机值。该方法不使用在线序贯学习算法，而是基于李雅普诺夫综合法更新输出权值，以保证闭环系统的稳定性。通过估计输出权向量的界，提出了一种新的反推设计方法，该方法需要在线调整的参数较少。仿真研究表明所提出的控制方法的有效性。
This paper describes the neural controller design for the longitudinal dynamics of a generic hypersonic flight vehicle (HFV). The dynamics are transformed into the strict-feedback form. Considering the uncertainty, the neural controller is constructed based on the single-hidden layer feedforward network(SLFN). The hidden node parameters are modified using extreme learning machine (ELM) by assigning random values. Instead of using online sequential learning algorithm (OSLA), the output weight is updated based on the Lyapunov synthesis approach to guarantee the stability of closed-loop system. By estimating the bound of output weight vector, a novel back-stepping design is presented where less online parameters are required to be tuned. The simulation study is presented to show the effectiveness of the proposed control approach.