Bubble Image Segmentation And Dynamic Feature Extraction Of Gas-liquid Two-phase Transient Flow

As an important link in the structure of gas-liquid two-phase flow equipment,the hydraulic pipeline water hammer process affects the efficiency of industrial production.The violent movement of bubbles in the pipeline can be regarded as a gas-liquid two-phase transient flow.With the increasing development of modern transmission tools,the requirements for the main pipeline transmission system are getting higher and higher,so the research on the pipeline system is more and more important.It is of great significance to grasp the change law of bubbles in the gas-liquid two-phase transient flow during the water hammer process to reduce the abrasion of the pipeline and maintain the normal operation of the pipeline system.The main research contents of this paper are:1)Through the built experimental platform,adjust the motor frequency,use a high-speed camera to take pictures of the bubble changes in the water hammer process under different initial flow rates of the hydraulic pipeline,analyze the characteristics of the bubble images,and compare the effects of various image preprocessing algorithms on these images.Perform corresponding image denoising and enhancement.2)According to the acquired image characteristics,the algorithm of moving object detection is selected to segment the bubbles,the effects of three moving object detection algorithms are compared,the Gaussian mixture model moving object detection algorithm is selected,and the watershed segmentation algorithm is used to separate the bubbles in the image.Then,according to the long and short axes of the bubbles,circular fitting and ellipse fitting are performed on the bubbles to obtain a complete bubble outline,and the area,perimeter,radius,volume and other parameters of each bubble are obtained through the MATLAB function.3)Analyze the law of bubble generation,collapse and movement position and track the bubble,determine the direction of the bubble movement through the frame-to-frame difference method,predict the position of the bubble at the next moment,and obtain the movement trajectory of the bubble.The bubble trajectories and corresponding velocities of the first bubble generation and death in the water hammer process were analyzed when the flow velocity of the hydraulic pipeline was 0.75m/s.4)Through the change of the position of the center of the steel ball in the adjacent images obtained by the Gaussian mixture model moving target detection in this paper,the speed of the steel ball before it collides with the valve seat can be obtained,and the initial velocity of the water flow in the pipeline can be obtained.And the change of gas content in the pipeline water hammer process under different initial velocities obtained by the algorithm in this paper is analyzed,and the change of air content is compared with the image obtained by simulation.