神经元最优状态反馈控制及其在登月问题中的应用研究

以非线性系统自学习最优控制算法为基础，提出了一种基于人工神经元网络的非线性状态反馈最优控制策略，使非线性不确定性的被控系统能通过自学习实现某种最优的非线性反馈控制，本文对该控制策略在登月软着陆问题中应用进行仿真研究，验证了该控制策略的可行性和有效性。

Studies on Neuro-Optimal State-Feedback Control and Its Applications to Lunar Soft Landing

An optimal nonlinear state-feedback control law is presented which is based on an artificial neural network and a self-learning optimal control algorithm for nonlinear systems. The neural network acts as a nonlinear state-feedback controller. The learning algorithm is used to get a training sample set of the optimal control law,and its convergence has been proved. The control law is applied to the control problem of a lunar soft landing vehicle to make the vehicle descend as desired,and its feasibility is studied and verified in a simulation experiment. The relationship between training and robustness of the neure-controller is analysed and tested.