基于气动装置神经网络模型的anti-windup控制器设计

Anti-windup controller design based on neural network model of pneumatic actuator

为提高气动系统的控制效果,以Levenberg-Marquardt算法训练多层前馈神经网络,建立了一气动装置的神经网络模型并推导出ARX模型.基于气动装置的ARX模型,采用Ragazzini方法设计了anti-windup控制器.实时控制结果表明,所设计的控制器有效地克服了控制死区和阀的饱和效应,实现了对该气动装置快速和高精度的控制.

In order to improve control performance of pneumatic systems,utilizing multilayered feedforward neural network trained with the Levenberg-Marquard method,a neural network model of a pneumatic actuator is established,from which an ARX(auto-regressive with exogenous input) model is derived.Based on the built ARX model,an anti-windup controller is designed by the Ragazzini method for the pneumatic actuator.The real-time control result demonstrates that with this controller the dead-zone and valve saturation of...