基于对角回归神经网络的转台伺服系统逆控制

针对转台伺服系统中存在的不确定性和非线性因素,提出一种基于对角回归神经网络(diagonal re-current neural network,DRNN)的逆控制方法。逆控制器由对角回归辨识网络(DRNNI)和对角回归控制网络(DRNNC)组成,利用神经网络的逼近能力,在线辨识系统的逆模型,直接将辨识器的拷贝作为系统的控制器。该方法结合了神经网络和逆系统控制的优点,能够克服系统中的不确定性和非线性因素。仿真结果表明,有效提高了转台伺服系统的动态跟踪精度,并具有较好的鲁棒性能。控制器的运算量小,能够满足实时控制要求。

Reseach on the DRNN-based inverse control for a servo system

To resolve the influence of uncertainty and nonlinearity on the high-precision turntable servo system, an inverse control method based on diagonal recurrent neural network (DRNN) is presented. The inverse controller is composed of two DRNNs, the DRNNI is used to identify the system inverse model on line, the copy of the DRNNI is used as the controller (DRNNC). The proposed controller combines the advantages of neural network and inverse system control, and can compensate the influence of uncertainty and nonlinearity. The simulation result on the servo system indicates that the control method above improves effectively dynamic tracking performance of the servo system and has good robustness. The calculation time of the controller is reduced, thus, the requirement of real-time control is better satisfied.