Font Size: a A A

Research On Molten Steel Level And Covering Flux Thickness Measurement Based On Computer Vision

Posted on:2016-04-20Degree:MasterType:Thesis
Country:ChinaCandidate:X C ZhangFull Text:PDF
GTID:2371330542489454Subject:Detection Technology and Automation
Abstract/Summary:PDF Full Text Request
Modern industry is more and more strict in the quality of steel.Continuous casting is an important part of the steel production,and has a significant impact on the quality of steel.Tundish plays an important role in continuous casting machine,and its molten steel level and covering flux thickness directly influence the production efficiency,casting blank quality and molten steel yielding rate.Therefore,it is significant to do research on the measurement of the molten steel level and the covering flux thickness in the tundish.The method of computer vision was adopted by this thesis to measure the molten steel level and the covering flux thickness in the tundish.In order to solve the problems of the current measurement method,a thermal field model was established to verify the theory of the measurement method and analyze the influences of various interference factors on the measurement.A target extraction method based on the image characteristics and the Hough transform and a method of interface locating with multiple criterion s were proposed to carry out the measurement of the molten steel level and the covering agent thickness.This thesis made the following research:(1)Establishing the steady and transient thermal field model of the measuring barWith the measured data,this thesis established a thermal field model of the measuring bar.This thesis verified the measurement method by analyzing the simulation results,and studied the influences of the drop in temperature,the fluctuation of molten steel level and the adhesion of covering flux in liquid form on the measurement with the thermal field model.(2)Target area identification of the measuring barAs traditional image segmentation methods cannot effectively identify the target area of the measuring bar,a target extraction method based on the image characteristics and the Hough transform was proposed,which could separate the measuring bar area from the background image effectively.And the identification accuracy rate of the method is over 98%.(3)Research on the criterion of interface locating of the molten steel and covering agentInfluenced by various factors,the temperature distribution of the thermal image is varied.With the experiment measurement results,the characteristics of the thermal image,the temperature curve and the temperature gradient curve,this thesis categorized the thermal images,and proposed interface criterions for each type of image.(4)Factor analysis of interface locating and the solutionMulti images sorting and quadratic average method were put forward by this thesis to reduce the influences of the drop in temperature and the adhesion of covering flux,and they could restore the temperature information effectively.Compared with the results of manual measurement,the reliability of the results measured with the method proposed in this thesis is over 90.9%,and the error of the results is less than 10mm,which meets the industrial requirements on the measurement of molten steel level and covering flux thickness.
Keywords/Search Tags:Molten steel level, Covering flux thickness, Temperature gradient, Computer vision, Image segmentation
PDF Full Text Request
Related items