Research On The Detection Method Of Pulverized Coal Injection Monitoring System For The Blast Furnace Based On Video Images

With the increasing rate of pulverized coal injection, the security and stability of the pulverized coal injection system has become more and more important. Monitoring the status of branch pipes of pulverized coal injection into blast furnace is indispensable to ensure smooth and stable operation in blast furnace and a high coal rate injection, which can also reduce the smelting costs.By installing tuyere monitoring camera into the blast furnace, the operators can observed the state of the pulverized coal injection directly in the main control room. At present, the status information of the blast tuyere monitoring camera is just estimated by artificial observation. Since there are so many tuyeres in the blast furnace (eg5500m3blast furnace in shougang jingtang united iron&steel co.,ltd. includes42tuyeres), omission or delay of the abnormal state often appeared by human judgment. What’s more, the consistency and uniformity of pulverized coal injection in each tuyere is difficult to determine accurately by human judgment.In response to these problems, this paper establishes the relationship between the flow of pulverized coal injection and the features of the tuyere images with digital image processing technology based on analyzing the video images of the blast furnace monitoring cameras, and then builds an automatic monitoring system to monitoring the status of branch pipes of pulverized coal injection into blast furnace. The main work of this paper is as follows:First, according to the characteristics of the coal injection images, an image preprocessing based on the combination of image enhancement algorithm and image filtering is presented. For the tuyere monitoring system working at high temperature and dust environments, the image degradation and distortion is inevitable. Therefore, an image enhancement algorithm based on the combination of window gray stretching method and power transform is proposed. An optimized compounding algorithm for image denoising is designed, which combined the improved median filtering and mathematical morphology, for the tuyere images has the characteristics like fuzzy edge and background noise. The experiment based on the images obtained from the steel mill live real-time shows that the image preprocessing algorithm proposed in this paper could enhance the image contrast and suppress the noise effectively. In addition, the edge details could be protected well.Secondly, the algorithm to find the background mask of raceway and injection lance is proposed. The tuyere accurate positioning method based on improved Hough circle fitting is proposed, which could fit the deficiency of image edge in pulverized coal injection images. In addition, the establishment algorithm of pulverized coal injection image real background is proposed, which combined with airbrush area detection. Using background subtraction method, the coal group image segmentation and feature extraction have been done. The experiment based on the images obtained live using the method adopted in this paper can segment out the pulverized coal accurately for the different tuyere.Finally, the consistency and uniformity detection algorithms of each tube pulverized coal injection are proposed. In order to establish the correspondence between the coal group region and branch pipe flow, coal group feature extraction algorithm based on area statistics is proposed through analysising the characteristics of the coal group. So as to determine the spatial information of coal group, weighted area method, the approximate ellipse method and the method of spindle rotation are researched. The experiment based on the images obtained from live shows that this four kinds of feature extraction algorithm could reflect the flow changes of the blast furnace pulverized coal injection accurately.Pulverized coal injection monitoring system for the blast furnace based on video images is built based on the Matlab_GUI graphical user interface design platform. The experiments are carried out by using the captured video of1#blast furnace and2#blast furnace in shougang jingtang united iron&steel co.,ltd.. Experimental results show that the pulverized coal feature extraction algorithm based on video image could monitor the status of coal power injection in a blast furnace effectively. The clogging and unimpeded transporting can be detected accurately and the consistency and uniformity state of the blast furnace coal injection tuyere can also be get in time.