设为首页 联系我们 加入收藏

当前位置: 网站首页 期刊分类目录 2018第4期 正文

数据驱动的自寻阶最小二乘法及其在电机中的应用

作者:时间:2018-07-10点击数:

PDF全文下载:  201804017.pdf

 

文章编号: 16726987201804011207 DOI 10.16351/j.16726987.2018.04.017

 

瞿叶奇, 高德欣* 张佳伟

(青岛科技大学 自动化与电子工程学院,山东 青岛 266061)

 

摘要: 为解决电机性能测试中某些性能曲线阶次难以确定,无法进行在线最佳阶次拟合的问题,本研究证明了转速转矩曲线的泰勒多项式逼近,提出了一种基于数据驱动的自寻阶最小二乘算法,并应用于Visual Studio平台开发的电机测试系统中。该算法基于电机测试中采集的电参数数据,利用随机梯度算法,对要拟合的电机性能多项式曲线的参数进行逐次逼近,通过阶次评价函数选择出最佳阶次,从而实现在线最优拟合。经现场验证表明:该算法拟合速度快,能够给出最佳逼近模型,达到在线最佳阶次拟合的要求。

关键词: 数据驱动; 三相异步电动机; 自寻阶最小二乘法; 最佳阶次拟合

中图分类号: O 224文献标志码: A

引用格式: 瞿叶奇, 高德欣, 张佳伟. 数据驱动的自寻阶最小二乘法及其在电机中的应用\[J\]. 青岛科技大学学报(自然科学版), 2018 394): 112118.

QU Yeqi, GAO Dexin, ZHANG Jiawei. Data driven selfseeking least squares method and its application in electrical machines\[J\]. Journal of Qingdao University of Science and TechnologyNatural Science Edition), 2018, 39(4) 112118.

 

Data Driven SelfSeeking Least Squares Method and

Its Application in Electrical Machines

 

QU Yeqi, GAO Dexin, ZHANG Jiawei

(College of Automation and Electronic Engineering, Qingdao University of Science and Technology, Qingdao 266061China)

 

Abstract: As it is difficult to determine the order of some performance curves in motor performance test, Taylor polynomial approximation of the speedtorque curve proves to solve online bestorder fitting problemand a datadriven selfseeking least squares algorithm is proposed to the motor test system developed by Visual Studio platform. The algorithm is based on the electrical parameters collected in the motor test, using stochastic gradient algorithm. In order to fit polynomial curve parameters of motor performance, the optimal order is selected by the order evaluation function, then it can achieve online optimal fitting. The optimal order is selected by the order evaluation function, to achieve online optimal fit. Rotational speedtorque test is difficult to determine model orderand the system can give the best approximation model for two groups of threephase asynchronous motor data with different sources collected in the fieldwhich meets the best order fitting demand.

Key words: data driven; threephase asynchronous motor; self seeking least square method; optimal order fitting

 

收稿日期:  20170925

基金项目: 山东省自然科学基金项目(ZR2017LF009,ZR2018LF008);山东省高等学校科学技术项目(J18KA323.

作者简介: 瞿叶奇(1990—),男,硕士研究生.*通信联系人.

 

Copyright © 2011-2017 青岛科技大学学报 (自然科学版)