摘要
针对自抗扰控制的跟踪-微分器模块中存在着待定参数的问题,借助人工神经网络便于建立非线性映射的思想,提出了基于神经网络的跟踪-微分器的设计方法。运用该方法可以对任意阶的跟踪-微分器进行参数设计,设计出的跟踪-微分器具有良好的鲁棒性。仿真实验表明,由该设计方法设计出的跟踪-微分器能够提供精确的微分信号,有一定的工程应用参考价值。
The main question existing in Tracking-Differentiator（TD） of Active Disturbance Rejection Control（ADRC） is to determinate some unknown parameters. Based upon the capacity of nonlinear approximation of multiplayer neural net- works, a design principle and algorithm for TD is presented. Using the method, the unknown parameters of any order TD are designed. With this work, the good robustness is guaranteed. Simulation results show that the accurate differential signal can be provided. The practical application of TD is possible.
出处
《计算机与数字工程》
2014年第3期378-381,432,共5页
Computer & Digital Engineering
基金
西安工业大学校长基金项目(编号:XAGDXJJ0617)资助
关键词
非线性映射
神经网络
跟踪-微分器
nonlinear mapping, neural network, tracking-differentiator
