Research On Particle Manipulation Technology Based On Chladni Plate

Particle manipulation technology based on chladni plate fits the motion of particles outside the sound depression by statistical means to get the displacement field of particle motion.Particle manipulation based on this displacement field can get rid of the dependence on sound depression and complete more complex particle manipulation task with simple equipment.In this thesis,particle manipulation technology based on chladni plate was analyzed in detail,particle manipulation experimental system was built,corresponding visual recognition algorithm and particle manipulation algorithm were designed,and several realtime particle motion control tasks were completed.The main contents of this thesis are as follows:(1)Real-time detection algorithm of particle motion trajectory on chladni plate.Accurate visual recognition algorithm is the prerequisite for real-time particle manipulation.The visual recognition algorithm designed in this thesis performs chladni plate area extraction,image segmentation,Hough circle detection and other operations on the chladni plate images collected by industrial cameras in turn,and it can accurately identify the coordinates of particles on the chladni plate.Then,Hungarian algorithm was used to calculate the corresponding displacement relationship of several particles on two images,and precisely calculated the motion vector of particles,so as to track the particle motion trace in real time.(2)Mapping of particle displacement field.Displacement field is the basis for precise particle manipulation and modeling of particle manipulation and control system.The vibration frequencies of 59 chladni plates have been selected in this thesis,and the particles with a diameter of 500 microns were used as the operating objects.The motion vectors of a large number of particles randomly placed on the plate at selected frequencies were counted and fitted to obtain the particle motion displacement field.The test results showd that the displacement field can quickly and correctly predict the movement trend of particles on the chladni plate.(3)Research on particle manipulation algorithm based on displacement field.Based on the prediction of particle motion trajectory by displacement field,a particle motion control algorithm was designed in this thesis.The algorithm can select the optimum frequency from 59 frequencies and control the vibration of the chladni plate so that the particles on the control plate can move according to the set requirements.In this thesis,the influence of several decision-making algorithms on particle manipulation algorithm was compared and analyzed,and the task of single/multi-particle orientation,fixed trajectory motion control and large-scale particle patterning was realized.(4)Research on particle manipulation algorithm based on reinforcement learning.This thesis applied Q-learning algorithm to particle manipulation,designed Q-learning algorithm for particle manipulation,and built a simulation training environment for Q-learning algorithm.After 500 rounds of training,the algorithm can automatically complete particle motion control,and the dynamic greedy factor designed greatly improves the training efficiency of the algorithm.Finally,the real-time particle manipulation task was successfully performed on the experimental platform using the trained Q-table,which verified the effectiveness of the Q-learning algorithm.