Summary: | 碩士 === 中原大學 === 機械工程研究所 === 86 === Research for bubbling in a fluidized beds has been going onfor many years inacademy and industry. Traditionally the academia obtained the data relatedto bubbles directly from the pressure perturbation, but this method isintrusive. Several optical instruments such as x-ray or γ-ray have alsobeen used in the study on 3-D fluidized bed, but the apparatuses are muchmore expensive and bubbles in a 3-D fluidized bed may overlap each other.Such phenomenon will make study of the behavior of individual bubbles moredifficult.The purpose of this study is to establish a measurement tool for2-D column based on image processing technology . Using CCD and image grabbercard to record the bubble data,therefor the process is non-intrusive. Inthe segmentation of the image, thresholding method is used. SOM (self-organizationfeature map) of artificial neural network is used to find out the threshold value. To reduce noise of the image, spatial low-pass filter isapplied on each frame. A bubble auto judgment rule is developedin this study. Several bubble parameter will be counted such as : area, velocity, shape ratio, aspect ratio, ect.
|