| In recent years,as an important part of the coal industry,coal preparation plants have been accelerating the pace of intelligent construction and are developing towards high-end,intelligent and green directions.Slime flotation is an important process of coal separation.For a long time,there have been problems of high production cost and low production efficiency.It is imperative for the process to develop toward automation and intelligence.In the flotation production process,the ash content of the flotation tailings is an important production index,which is of great significance to realize the closed-loop optimization control of the flotation process.However,for a long time,there has been a lack of effective online ash detection technology,which has become an important reason for limiting the intelligent development of flotation processes.Based on this,this paper carried out the flotation tailings image acquisition experiment at the flotation industry site,focusing on the relationship between the color of the flotation tailings and its ash in the production process,and design and establish an on-line ash detection system based on flotation tailings images.This article first analyzes the problem of the difficulty of collecting images of flotation tailings in production.On the tailings tank of the flotation workshop of Liuwan Coal Preparation Plant,a real-time collection device for flotation tailings images is designed and built.The device adopts a bypass structure and uses a mud pump to extract flotation tailings samples for shooting.The main equipment includes industrial cameras,lenses and light sources.In order to optimize the relevant parameters of the image acquisition device,a comparative experiment was carried out.The results showed that when the light intensity is70%,the aperture size is F=5.6,and the exposure time is 8000,the image taken is the best.In order to study the relationship between the color of flotation tailings and its ash content,a flotation tailings image acquisition experiment was carried out.Through the flotation tailings image acquisition device,the ash content range was 30%—80% under different working conditions.There are 59 sets of images of flotation tailings,and the actual ash content corresponding to 59 sets of images was obtained through ash burning in the laboratory.After preprocessing 59 sets of flotation tailings images,the gray features,RGB color features and HSI color features of 59 sets of images are extracted and analyzed.The results show that the gray average value of the images and the R,G,B average value of the images has the strongest correlation with ash content,and the correlation coefficients are0.940,0.921,0.952 and 0.927,which show a linear upward trend with the increase of ash content.In order to study the soft measurement technology based on the image features of flotation tailings,BP neural network and support vector regression machine learning algorithms are used to establish and simulate the ash prediction model.The results show that:select gray + RGB + HSI features as input Variables,the ash prediction model established by the GSA-SVR algorithm has the best effect,the model fitting coefficient is 0.986,the mean square error is 1.828,and the average absolute error of prediction for 15 sets of test images is2.179.In order to further verify the feasibility of the on-line ash detection technology based on flotation tailings images,the hardware design and software design of the flotation tailings ash real-time detection system were carried out,and the trial operation was carried out on the industrial site.The results show that the system runs stably,has good real-time performance,is easy to use,safe and efficient,and the accuracy of the ash content detection is as high as96.54%,which can meet the basic requirements for ash content detection in the industrial field.The research in this paper provides a new idea for the online detection of ash content of flotation tailings and a new basis for the intelligent development of the flotation process. |
PDF Full Text Request