基于GA-Newton法的异步电机改进模型参数辨识  

Parameter Identification of Improved Model of Asynchronous Motor Based on GA-Newton Method

作　　者：孟庆硕 许鸣珠 MENG Qing-shuo;XU Ming-zhu(School of Mechanical Engineering, Shijiazhuang Railway University, Shijiazhuang 050043, Chin)

机构地区：[1]石家庄铁道大学机械工程学院,石家庄050043

出　　处：《组合机床与自动化加工技术》2018年第5期47-50,共4页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：国家自然科学基金面上项目(11372198);河北省教育厅科学技术重点项目(ZD20131098)

摘　　要：针对传统的异步电机参数辨识存在磁链输入项无法直接获得、观测器设计复杂的缺点,提出了一种将电流定向坐标系下的磁链模型和两相静止坐标系下的电流一阶导数模型结合的改进参数辨识模型。该模型的所有信号均为可直接检测的状态变量,减少了其他干扰对电机参数辨识的影响,提高了参数辨识的准确性。为了解决传统遗传算法收敛速度慢,易局部收敛的缺点,将遗传算法与牛顿法结合,提高了遗传算法的收敛速度和搜索能力。实验结果表明,利用GA-Newton(遗传-牛顿)法进行参数辨识鲁棒性强、收敛性好。在额定转速下,待辨识参数能够在较短的时间内收敛,具有较高的精度,同时也克服了一般遗传算法对辨识参数初始值要求高的缺点。In view of the shortcomings of the flux linkage can not get directly and the flux linkage observer design complex,existing in traditional induction motor parameter identification.Proposed an improved parameter identification model,which combined the flux model under the current directional coordinates with the current first derivative model under the two-phase stationary coordinates.Signals are used by the model can directly detect,Thereby reducing the other interference effects on motor parameter identification,improved the precision of parameter identification.In order to solve the traditional genetic algorithm have the shortcomings of slow convergence speed and easy local convergence,combines genetic algorithm and Newton＇s method,improved the convergence speed and search ability of genetic algorithm.The experimental results shows that using GA-Newton method for parameter identification have strong robustness and good convergence.Under the rated speed,the parameters can be identified in a relatively short time and have highly precision,meanwhile overcome the disadvantage of the general genetic algorithm have highly request for the initial value.

关 键 词：遗传算法 牛顿法 异步电机 参数辨识 

分 类 号：TH139[机械工程—机械制造及自动化] TG506[金属学及工艺—金属切削加工及机床]