一种鲁棒神经网络结构在温度控制系统中的应用

Application of Robust Neural Network Structure on Temperature Control System

在神经网络直接逆控制的基础上,加入闭环增益成形控制算法,构成闭环的控制系统,以提高系统的鲁棒性。通过对简单的温度控制系统仿真结果分析可知,当系统没加入死区模型摄动时,具有闭环增益成形控制算法的神经网络控制系统与神经网络直接逆控制的控制性能相近,当加入死区时,前者明显提高了系统的鲁棒性。

Closed loop gain shaping control algorithm is applie d on a direct inverse neural network so that a closed-loop control system is cons t iituted to improve the robust performance of the control system. Through analyzi ng the result of simple temperature control system simulation it follows the con trol performance of neural network control with closed loop gain shaping control algorithm is the same as that of the direct inverse neural network when a dead zone model perturbation is not introduced into the system. But the robust perfor mance is improved obviously by neural network control with closed loop gain shap ing control algorithm when a dead zone model perturbation is introduced into the system.