基于声波操纵的微小物体运动建模与控制

Modeling and control of micro object motion based on acoustic manipulation

中文关键词  声波操纵  微小物体控制  克拉尼平板  Faster R-CNN  局部加权回归

英文关键词  sonic manipulation  tiny object control  Chladni plate  Faster R-CNN  LOESS

中文摘要

      采用声波操纵克拉尼平板上的微小物体, 在精准医学、液滴和颗粒的工业控制等方面有着广阔的应用前景. 传统声波操纵认为在运动过程中, 声波对微粒的影响是无序的, 而近年来, 研究得出声振动是有序的, 但缺乏精确、有效的建模手段, 限制其应用. 针对该问题, 本文提出了结合Faster R-CNN算法与局部加权回归(LOESS)算法的方法对声波场进行建模. 采用图像识别技术辨识微粒在克拉尼平板上的位置, 计算在某一固定声波频率下微小粒子在克拉尼平板不同位置上的位移. 在积累大量位移数据后, 基于LOESS算法, 建立平板上的完整声波位移模型,分析声波力场如何在平面上对微小物体的运动施加影响, 并建立仿真模型, 进行克拉尼平板上微粒操纵的仿真实验. 最后, 本文将基于所搭建的声波操纵平台, 对易碎的速溶咖啡颗粒进行控制, 通过与仿真实验相对比, 验证了建模与控制方法的可行性, 证明该模型可实现微小粒子的运动控制.

英文摘要

      t: Manipulation of the tiny objects by acoustic waves shows broad application prospects in precision medicine, industrial control of droplets and particles. Traditional acoustic manipulation methods assumed that the influence of acoustic waves on particles is disordered in the motion process. Recently, it has been found that the acoustic vibration has some certain rules. However, the lack of accurate and effective modeling methods limits its application. Thus, to solve this problem, a method combining the Faster R-CNN algorithm and local weighted regression (LOESS) algorithm is designed to build the model the acoustic field. First, the image recognition technology is applied to identify the position of particles on the Chladni plate. Then, the displacement of small particles will be calculated under different positions on the Chladni plate with a fixed acoustic frequency. After collecting large sets of displacement data, based on the LOESS algorithm, a complete acoustic displacement model on the plate is built to analyze the influence of the acoustic force field on the motion of small objects. The simulation experiment of particle manipulation on the Chladni plate is also carried out with the proposed model. Finally, the motion control of the fragile instant coffee particles is verified based on the proposed acoustic manipulation platform. Compared with the simulation experiment, the feasibility of the modeling and control method is proved, and the motion control of micro particles is successfully realized.