基于DRNN的某交流伺服系统控制研究

AC Servo System Control Based on Dynamic Recurrent Neural Network

某火箭炮交流位置伺服系统存在非线性、强耦合、时变性,传统控制难以取得很好的控制效果,为此提出了基于对角递归神经网络(DRNN)的模型参考自适应控制,设计了对角递归神经网络辨识器与控制器,并引入了动量因子和自适应学习因子,提高了自适应控制器的稳定性和响应速度,仿真结果表明该控制器具有精度高,响应快,鲁棒性好的特点。

AC position servo system should be characteristic of non-linear,strong coupling and variability by time,so it is difficult to get a good control result in the traditional control system. In consideration of this,a controller and an identifier based on Dynamic Recurrent Neural Network are designed. The momentum factors and adaptive learning factors are introduced to it to ensure the stability and fast convergence of the system. Simulation results show that the control system is characteristic of high precision,fast response and good robustness.