基于连续回转电液伺服马达模糊RBF神经网络控制研究  

Fuzzy RBF Neural Network Control of Continuous Rotary Electro-hydraulic Servo Motor

作　　者：范文静 王晓晶[1] 孙培元 何超群 刘亚楠 李岩 FAN Wen-jing;WANG Xiao-jing;SUN Pei-yuan;HE Chao-qun;LIU Ya-nan;LI Yan(School of Mechanical Power Engineering,Harbin University of Science and Technology,Harbin,Heilongjiang 150080)

机构地区：[1]哈尔滨理工大学机械动力工程学院,黑龙江哈尔滨150080

出　　处：《液压与气动》2020年第12期89-95,共7页Chinese Hydraulics & Pneumatics

基　　金：国家级大学生创新创业训练项目(201910214031)。

摘　　要：针对仿真转台用连续回转电液伺服马达,由系统的非线性和摩擦、泄漏等外界因素导致的不确定性,严重影响了连续回转电液伺服马达的控制精度,提出了一种模糊RBF神经网络控制策略。将RBF神经网络的学习能力引入模糊机制中,利用神经网络高效的非线性拟合能力以及基于专家经验的模糊规则,以避免RBF神经网络的权值更新陷入最优解,同时选择遗传算法优化模糊RBF神经网络的中心宽度、阈值和权值的初始值,以提高控制算法的收敛速度以及收敛精度;最后通过仿真对比说明,该控制算法较PID更能有效提高系统的低速稳定性,拓展系统的频响,实现伺服系统的精确控制。Aiming at the uncertainties caused by high nonlinearity and external factors such as friction and leakage of the continuous rotary electro-hydraulic servo motor for simulation turntable,this paper proposes a fuzzy RBF neural network control strategy.The learning ability of RBF neural network is introduced into the fuzzy mechanism.The reasoning ability of the fuzzy mechanism is improved by the efficient parallel processing and self-learning ability of the neural network.At the same time,the genetic algorithm is selected to optimize the central width,threshold and weight of the fuzzy RBF neural network.The control accuracy of the network is improved by gradient descent method.Finally,the simulation comparison shows that the control algorithm can effectively improve the low-speed stability of the system,expand the frequency response of the system and realize the precise control of the servo system.

关 键 词：连续回转电液伺服马达 模糊RBF神经网络 遗传算法 梯度下降法 

分 类 号：TH137.5[机械工程—机械制造及自动化]