共母线开绕组永磁同步牵引电机改进级联模型预测控制  

Common bus bar open winding permanent magnet synchronous traction motor improved cascade model predictive control

作　　者：高锋阳[1] 吴银波 徐昊 史志龙 岳文瀚 孙伟 王高强 GAO Fengyang;WU Yinbo;XU Hao;SHI Zhilong;YUE Wenhan;SUN Wei;WANG Gaoqiang(College of Automation and Electrical Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China;Qingshui Power Supply Company,State Grid Corporation of China,Qingshui 741000,China)

机构地区：[1]兰州交通大学自动化与电气工程学院,甘肃兰州730070 [2]国网清水供电公司,甘肃清水741000

出　　处：《铁道科学与工程学报》2025年第3期1254-1265,共12页Journal of Railway Science and Engineering

基　　金：国家自然科学基金资助项目(52267004)。

摘　　要：为降低共母线开绕组永磁同步牵引电机三矢量级联模型预测电流控制开关频率和控制系统对电机参数依赖性,提出一种基于变步长自适应线性神经网络(Adaline)可调参数改进级联模型预测电流控制策略。首先,针对共母线开绕组永磁同步牵引电机三矢量级联模型预测电流控制造成开关频率高的原因进行分析,剔除高开关频率和高共模电压的电压矢量,优化备选电压矢量范围,对剩余电压矢量根据其对q轴电流作用效果分组组合寻优和分配作用时间;基于变步长自适应线性神经网络改进PI控制器,使得改进PI控制器兼顾快速性与超调;然后,分析共母线开绕组永磁同步牵引电机模型预测控制参数变化特性,构建系统变步长自适应线性神经网络参数辨识模型,对电机参数分步辨识,形成参数可调节级联模型预测控制;最后,对所提策略和三矢量级联模型预测电流控制进行稳态和动态半实物测试对比。结果表明:所提策略对转矩脉动、零轴电流、总谐波畸变率、开关频率、调速超调都具有很好的抑制效果,避免了传统模型预测控制的多目标代价函数中权重系数整定和参数辨识模型构建欠秩问题,对系统的控制性能有明显的提升作用。研究结果为进一步将共母线开绕组永磁同步牵引电机传动系统应用于机车牵引提供参考。In order to reduce the switching frequency of the three-vector cascade model predictive current control and the dependence of the control system on the motor parameters for the common-busbar open-winding permanent magnet synchronous traction motor,a variable-step-size adaptive linear neural network(Adaline)adjustable-parameter-improved cascade model predictive current control strategy was proposed.Firstly,the reasons of high switching frequency caused by the three-vector cascade model predictive current control of permanent magnet synchronous traction motor with common busbar open winding were discussed.The voltage vectors with high switching frequency and high common-mode voltage were eliminated to optimize the range of alternative voltage vectors.The remaining voltage vectors were grouped to search for the optimal combination and allocation of the time of action according to the effect of their action on the q-axis current.Secondly,the PI controller was improved based on the variable step length adaptive linear neural network,so that the improved PI controller can be used to improve the control system’s dependence on motor parameters.The PI controller was improved based on the variable step-size adaptive linear neural network,which made the improved PI controller take into account the speed and overshoot.Then,the variation characteristics of the model predictive control parameters of the common-busbar open-winding permanent-magnet synchronous traction motor were analyzed.The parameter identification model of the system was constructed based on the variable step-size adaptive linear neural network,so that the parameters of the motor can be identified step by step,thus forming the parameteradjustable cascade model predictive control.Finally,the proposed strategy and three-vector cascade model predictive current control were tested.The proposed strategy was compared with the three-vector cascade model predictive current control in steady state and dynamic semi-physical tests.The results show that the proposed st

关 键 词：开绕组永磁同步牵引电机 变步长自适应线性神经网络 级联模型预测 转矩脉动 零轴电流 参数分步辨识 开关频率 

分 类 号：TM351[电气工程—电机]