基于密度聚类的磁悬浮平面电机模态参数估计  

Modal parameter estimates for a magnetic levitation planar motor based on density clustering

作　　者：孙浩博 杨开明[1] 朱煜[1] 鲁森[1] SUN Haobo;YANG Kaiming;ZHU Yu;LU Sen(Beijing Key Laboratory of Precision/Ultra-Precision Manufacturing Equipments and Control,Tsinghua University,Beijing 100084,China)

机构地区：[1]清华大学精密超精密制造装备及控制北京市重点实验室,北京100084

出　　处：《清华大学学报（自然科学版）》2023年第1期33-43,共11页Journal of Tsinghua University(Science and Technology)

基　　金：国家科技重大专项(2017ZX02102004)。

摘　　要：磁悬浮平面电机的高加减速运动需要轻量化设计,但轻量化设计也会导致磁悬浮平面电机出现不可接受的振动。对磁悬浮平面电机模态参数的准确估计是抑制振动的关键环节。该文提出了一种基于密度聚类的模态参数估计方法。首先利用两步迭代的系统辨识算法,得到系统的参数化频率响应函数;然后利用基于密度的有噪声空间聚类(DBSCAN)算法对系统进行模态分析,去除不稳定的数学模态点,并对物理模态点进行基于正态分布的离群点剔除,以获取最终的模态参数。仿真和试验表明,所提方法可以实现系统模态参数的准确估计。Lightweight designs are needed for high acceleration and deceleration rates of a magnetic levitation planar motor(MLPM),but lightweight designs also lead to unacceptable vibrations in the MLPM.Accurate estimates of the MLPM modal parameters are the key to suppressing the vibrations.This paper presents a modal parameter estimation method based on density clustering.The system parametric frequency response function is obtained using a two-step iterative identification algorithm.Then,the DBSCAN algorithm is used for the modal analysis to remove the unstable mathematical modes.The outliers of the physical modes are also removed based on a normal distribution to obtain the final modal parameters.Simulations and tests show that this method can accurately estimate the system modal parameters.

关 键 词：磁悬浮平面电机 密度聚类 模态参数估计 

分 类 号：TM351[电气工程—电机]