基于单网络ADP的一类未知非线性系统的近似最优控制

针对一类未知的连续非线性系统，提出一个基于单网络近似动态规划(ADP)的近似最优控制方案。该方案通过设计一个新型的递归神经网络(RNN)辨识器放松了系统模型需已知或部分已知的要求，并利用一个神经网络(NN)近似系统的性能指标函数消除了常规ADP方法中的控制网络。通过Lyapunov理论分析严格证明了闭环系统内所有信号一致最终有界，并且所获得的性能指标函数和控制输入分别收敛到最优性能指标函数和最优控制输入的小邻域内。仿真结果验证了所提出控制方案的有效性。

Near-optimal control of a class of unknown nonlinear systems based on single network ADP

The near-optimal control scheme is proposed for a class of unknown continuous-time nonlinear systems based
on single network approximate dynamic programming (ADP). The proposed scheme relaxes the requirement of the system
model being known or partly known by designing a novel recurrent neural network(RNN) identifier, and eliminates the action
network of ordinary ADP methods by employing a neural network(NN) to approximate the performance index function. By
Lyapunov theory, it is proved that all the signals in the closed-loop system are ultimately uniformly bounded and the obtained
optimal performance index function and control input lie in small neighborhoods of the optimal performance index function
and the optimal control input, respectively. Simulation results demonstrate the effectiveness of the proposed scheme.