一种改进型DRNN神经网络自学习PID控制  

Self-learning PID Control of Modified DRNN Neural Network

作　　者：刘甘霖[1] 沈玲[1] 左是[1] 

机构地区：[1]湖北工业职业技术学院电子工程系,湖北十堰442000

出　　处：《湖北工业职业技术学院学报》2016年第6期104-107,共4页Journal of Hubei Industrial Polytechnic

摘　　要：本文针对多变量耦合系统,采用DRNN神经网络对PID控制器参数进行自学习,提出了将学习因子在学习过程中进行动态调整,与传统DRNN神经网络自学习PID控制整定结果进行比较,使用matlab进行仿真,仿真结果表明,学习因子动态调整后的参数结果在超调量、调节时间、稳态性能上明显优于传统DRNN算法。Self learning is made with Muhivariable coupling system by DRNN neural network on PID controller parameters. This paper makes the comparison between dynamic adjustment of learning factor in the learning process and traditional DRNN neural network self-learning PID control setting the result. Mat-lab simulation is made for this comparison and the resuh shows that parameters adjusted dynamic of learning factnr are superior to traditional DRNN algorithm in overshoot,adjustmcnt time and steady state performance.

关 键 词：DRNN 收敛 学习速率 MATLAB 仿真 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]