基于自适应动态规划的未知模型非线性系统H_2/H_∞控制

H_2/H_∞ Control of an Unknown Model Nonlinear System Based on Adaptive Dynamic Programming

提出了一种在线的自适应动态规划算法,近似求解耦合的哈密尔顿雅可比(Hamilton-Jacobi-Isaacs,HJI)方程,获得非线性系统混合H_2/H_∞控制的纳什均衡策略。通过在控制策略和干扰策略中加入已知噪声,从而不依赖系统的模型信息,得到一个求解混合H_2/H_∞控制问题的未知模型的近似动态规划算法。分别使用2个评价神经网络和2个执行神经网络,同步在线更新2个值函数、控制策略和干扰策略,神经网络未知参数通过最小二乘法进行估计。仿真结果验证了算法的可行性。

An online adaptive dynamic programming algorithm is proposed for getting the approximate solution of the coupled Hamihon-Jacobi-Isaacs Equations （HJIE）, and obtaining the Nash equilibrium strategy of mixed H2/H8 control of nonlinear system. By adding the detection signal to the control strategy and the interference strategy, an approximate dynamic programming algorithm is acquired for solving mixed HJH, control problems with unknown model without depending on model information of the system. Two critic neural networks and two executive neural networks are used to synchronously update two value functions, control strategies and interference strategies online. The unknown parameters of the neural network are estimated by generalized least squares. The simulation results verify the feasibility of the algorithm.