基于OSELM的无刷直流电机无位置传感器控制  

Sensorless control for brushless DC motors based on OSELM

作　　者：王欣[1,2] 梁辉 秦斌 WANG Xin;LIANG Hui;QIN Bin(School of Electrical and Information Engineering,Hunan University of Technology,Zhuzhou 412007,China;Key Laboratory for Electric Drive Control and Intelligent Equipment of Hunan Province,Zhuzhou 412007,China)

机构地区：[1]湖南工业大学电气与信息工程学院,湖南株洲412007 [2]电传动控制与智能装备湖南省重点实验室,湖南株洲412007

出　　处：《电机与控制学报》2018年第11期82-88,共7页Electric Machines and Control

基　　金：国家自然科学基金(61673166);湖南省自然科学基金(2017JJ4022;2018JJ4070);湖南省教育厅科研重点项目(15A050;17A053)

摘　　要：针对无刷直流电机转子位置检测问题,提出基于在线贯序极限学习机(OSELM)的无刷直流电机无位置传感器控制方法。该方法构建了一个单隐层前馈神经网络,经分析将定子端电压和电流作为OSELM网络的输入信号,逆变电路的逻辑换相信号作为OSELM网络的输出,将电流速度双闭环控制得到的PWM波形与OSELM网络输出进行逻辑处理得到功率开关管的控制信号,由此实现无刷直流电机的无位置传感器控制。网络参数通过离线训练得到,将训练好的网络模型应用到电机中进行在线测试。将该方法与传统反向传播(BP)神经网络方法进行比较,实验结果表明该方法避免了BP神经网络参数难以选取的问题,且在保证高精度的前提下,比传统的控制算法速度更快,验证了该方法的可行性和优越性。In order to solve the rotor position detection problem for brushless DC motors,a sensorless control method based on online sequential extreme learning machine(OSELM)control was proposed.A single hidden layer feedforward neural network was designed,in which the stator voltage and current were taken as the input signals,and the logic signal of the inverter circuit was used as the output.The power switching tube control signal was obtained by logic processing of PWM wave from current speed double-loop control and ELM network output,so as to realize sensorless control for brushless DC motors.The network parameters were trained offline,and the network model was applied to a motor for online testing.Compared with the traditional back-propagation(BP)neural network method,the experimental results show that the proposed method avoids the parameter selection difficulty and under the presupposition of ensuring high control accuracy,the speed is faster,which verifies the feasibility and superiority of the proposed method.

关 键 词：无刷直流电机 转子位置检测 极限学习机 前馈神经网络 PWM波形 

分 类 号：TM301.2[电气工程—电机]