面向超洁净流控的基于磁性薄膜标记柔性隔膜变形场重构方法研究  

Reconstruction Method of Deformation Field of Flexible Diaphragm Based on Magnetic Film Labeling for Ultra-clean Flow Control

作　　者：刘龙胤 潘慎逸 陈垚岗 申慧敏[1] LIU Longyin;PAN Shenyi;CHEN Yaogang;SHEN Huimin(School of Mechanical Engineering,University of Shanghai for Science and Technology,Shanghai 200093)

机构地区：[1]上海理工大学机械工程学院,上海200093

出　　处：《液压与气动》2024年第11期94-100,共7页Chinese Hydraulics & Pneumatics

基　　金：国家自然科学基金面上项目(52175055)。

摘　　要：面向半导体等领域的超洁净流控需求,提出了一种基于磁性薄膜标记的柔性隔膜变形场重构技术。利用磁性薄膜标记柔性隔膜形变,建立磁性薄膜激励空间磁场与柔性隔膜变形场之间的映射关系,基于长短期记忆神经网络构建磁性薄膜磁场和变形场的预测模型,实现了磁性薄膜在不同形变状态下的变形场重构,间接监测超洁净流控柔性部件隔膜的运行工况。通过COMSOL Multiphysics仿真平台获得训练样本数据,并根据训练完成的神经网络模型对预测变形场数据进行了分析,验证了方案的可行性。In order to meet the needs of ultra-clean flow control in semiconductor and other fields,a flexible diaphragm deformation field reconstruction technology based on magnetic film labeling is proposed.The magnetic film is used to track the deformation of the flexible diaphragm,then the mapping relationship between the magnetic film excitation space magnetic field and the deformation field of the flexible diaphragm can be established.Based on a long short-term memory neural network,the prediction model of the magnetic film magnetic field and deformation field is constructed,which achieves the reconstruction of the deformation field of magnetic film in different deformation states,thereby indirectly monitoring the operating conditions of the diaphragm.Additionally,training data samples are acquired through the COMSOL Multiphysics simulation platform,and the predicted deformation field data are analyzed by using the trained neural network model to validate the feasibility of the proposed approach.

关 键 词：超洁净 磁性薄膜 柔性隔膜 神经网络 变形场重构 

分 类 号：TH137[机械工程—机械制造及自动化]