模拟复合正交神经网络在轴向磁轴承中的控制研究

作者：徐永灿1,叶友红2 日期：2007-08-31/span> 浏览：3359 查看PDF文档

                                   模拟复合正交神经网络在轴向磁轴承中的控制研究
                                                      徐永灿1,叶友红2
                               (1.绍兴制水有限公司，浙江 绍兴 312035;
                                   2.浙江浙大中控信息技术有限公司，浙江 杭州 310035）

摘 要：由于磁轴承的动态性能主要取决于所采用的控制规律，控制器是磁轴承系统的关键。在数字复合正交神经网络（NN）的基础上，提出了一种模拟复合正交神经网络，并用于轴向磁轴承的控制中。控制器采用模拟复合正交神经网络与PID的并行控制方法，对带有负载干扰的轴向磁轴承控制系统作了PID控制与NN+PID控制的仿真实验。仿真结果表明，相对于常规PID控制器，该并行控制法具有较高的抗干扰与自适应能力，控制效果理想。
关键词：轴向磁轴承;模拟复合正交神经网络;连续学习算法;并行控制

                          Control study of analog compound orthogonal neural network in axial magnetic bearing
                                                        XU Yongcan1, YE Youhong2
                                (1.Shaoxin Water Treating Co.,Ltd., Shaoxing 312035, China;
                             2.Zhejiang Supcon Information Technique Co., Ltd., Hangzhou 310035, China)

Abstract: The dynamic properties are mostly decided by the controller, so the control method is key of active magnetic bearing. An analog compound orthogonal neural network (NN) was presented on the basis of the digital compound orthogonal neural network and was applied in the control of the axial magnetic bearing. The controller was a parallel control method with an analog compound orthogonal neural network and PID. Simulation experiments were given for control system of the axial magnetic bearing with load disturbance by using PID controller and NN+PID controller. The simulation results prove that the parallel control method has excellent antidisturbance and speed response ability by comparison with PID controller, and obtains satisfactory control effect.
Key words: axial magnetic bearing; analog compound orthogonal neural network; continuously learning algorithm; parallel control

参考文献(Reference)：
［1］HAN B CH, LI Y F, JIA H G, et al. Summarization on control system of active magnetic bearing［J］. Machine Tool & Hydraulics,2004(4):1-3,88.
［2］YE J, ZHANG X H. High speed learning algorithm for a multilayer feedforward neural network and its application［J］. Control and Decision,2002,17(z1):158-159.
［3］叶 军.伺服系统在摩擦条件下的模拟复合正交神经网络控制［J］.中国电机工程学报,2005,25(17):127-130.
［4］叶 军.基于模拟正交神经网络的电热干燥器温度控制［J］.农业机械学报,2005,21(10):105-108.
［5］叶 军.基于拉盖尔模拟神经网络的过热汽温直接自适应控制系统［J］.中国电机工程学报,2006,26(23):67-71.
［6］ZHU H Q. Parameter design and control system study for active magnetic bearing in the axial direction of bearing less motors［J］. Transaetions of China Electrotechnical Society,2002,17(3):12-16.