论文部分内容阅读
针对积分时滞系统控制中存在的稳态误差和建模误差问题,文章使用BP神经网络设计了一种新型的神经网络改进下的内模控制方案。李雅普诺夫直接法被用于证明适当的选择学习速率可以保证神经网络内模控制器的收敛率。仿真结果表明,新的设计可以成功消除在积分时滞系统控制中由于扰动带来的稳态误差,同时在应对模型误差时表现出较强的鲁棒性。
Aiming at the problem of steady-state error and modeling error in the system control of integral time-delay, a new neural network-based internal model control scheme with BP neural network is designed. Lyapunov direct method is used to prove that the appropriate choice of learning rate can guarantee the convergence rate of neural network model controller. The simulation results show that the new design can successfully eliminate the steady-state error caused by disturbance in the system control of integral time-delay and show strong robustness in response to model error.