Research On Image Detection And Modeling Of Ash Content And Particle Size Characteristics Of Coal Flotation Tailings

With the continuous improvement of the overall level of industrial process automation,the automatic control and intelligent control of the coal flotation process has also received increasing attention.One of the key aspects of the intelligent control of the flotation process is the on-line inspection of product quality,including flotation of clean coal and tailings.However,the development of intelligent flotation is severely limited by the lack of relevant detection technologies and sensors.The quality inspection research of flotation products has been more focused on the flotation of clean coal,while flotation tailings has been neglected.The quality of flotation tailings coal as an important feedback information plays a crucial role in achieving closed-loop optimal control of the flotation process.In the practical production process of coal flotation,on-site operators rely mainly on the naked eye to observe the color of tailings to judge ash content and the hand touch to judge whether the flotation tailings has"coarse particle loss"problem or not.Based on this,adjustments are made to the variables such as the amount of agent addition,the thickness of the froth layer,and the amount of aeration to ensure the quality of the flotation product.This article takes this as an entry point and focuses on the use of machine vision methods to realize the prediction of the ash content and coarse particle content of flotation tailings.Firstly,a set of image acquisition system to obtain the surface image of flotation tailings slurry and the particle size image is constructed in this paper.It mainly includes:#1 image acquisition system composed of color CCD camera,fixed focal lens and annular light is used for the surface image acquisition of floating-tail coal slurry.#2 image acquisition system composed of black-and-white CCD camera,double telecentric lens and parallel light is used for the acquisition of particle images of the coal slurry.A reasonable sample container was designed to create conditions for the simultaneous realization of the above image acquisition.In order to study the relationship between flotation tailings ash and coal slurry surface images,flotation tailings,flotation clean coal and gangue were used as the basic samples to prepare coal slurry samples with different ash content with concentration of 40g/L.Under the same conditions,the surface iamge of the coal slurry with different ash content was captured.The color image features were analyzed and extracted in the three color spaces of RGB,YUV and HSI.The results show that the color feature information of the coal slurry image mainly included in the first-order color moments;the first-order color moments of the U and V components converted from the R,G,and B linear conversions are the most sensitive and best correlation to changes in the ash of the coal slurry.At the same time,the U component and V component histograms have the lowest skewness and best symmetry,and the distributions are closer to a normal distribution.The univariate fitting models of R,G,and B components and the fitting models of Y,U,and V components are established respectively.The results show that the model fitting accuracy established by introducing the U,V components obtained by subtraction of the R,G,and B components is significantly increased.On this basis,the chromatic aberration variables with different forms were constructed using methods of linear combination and second-order polynomial combination respectively,which independently established a coal-slurry ash prediction models.The model fitting effects show that the model precision based on the chromatic aberration variable C_rgb of the R,G,B three-component linear combination is better than that of the two-component linear combination C_rb,C_rgandC_gb.The accuracy of the ash model established by the chromatic aberration variables constructed by the second-order polynomial combination of the R,G,and B components was further improved.Considering the effect of the concentration on the image characteristics,the coal slurry ash content prediction model was established using the chromatic aberration variables based on different combinations at each concentration which were prepared.Compared the fitting accuracy of the difference chromatic aberration variables under the condition of different concentrations and finally model prediction accuracy,global universality and model complexity,the chromatic aberration variables C_rgb is selected as the independent variable to establish soft-sensing model of flotation tailings ash content.At the same time,the law of C_rgb gradually decreases with the increase of concentration and finally becomes stable is found.The concentration correction function was constructed to modify the chromatic aberration variable C_rgb.Finally,a soft sensing model of ash content of flotation tailings based on the concentration and color image features was established.Based on lambert-beer law,the function between the shading effect K′of coal slurry flotation tailings coal particles and mass concentration M_s is defined,the variation rule of coal flotation tailings coal slurry transmission of access to the image grayscale characteristics along with the change of the concentration which based on#2 image acquisition system is researched,mean gray value of coal slurry transmission image increased with the increase of the concentration decreases is found under the same grain size.Based on the coal slurry transmission image gray level characteristics of coal flotation tailings,concentration soft-sensing model is established.The differences in particle images of different size coal slurry were studied.It was found that the image of particle morphology can be clearly captured using the#2 image acquisition system for+0.125mm particle size coal.For the-0.125mm particle size coal,it is difficult to obtain a clear image of the particle morphology owing to the phenomenon of serious particle adhesion.For different concentrations of coal slurry,it was found that when the concentration of the coal slurry is more than 30 g/L,it is difficult to obtain a clear particle image because the particle stacking and the adhesion are too severe.The final experimental coal slurry concentration is 20g/L.In order to realize the detection of coarse particles in flotation tailings,coal slurry samples with a concentration of 20 g/L of coarse particles of 5%,10%,15%,and 20%were prepared.The image was preprocessed using guided filtering,adaptive histogram equalization and maximum variance between classes.The method of morphological processing was used to eliminates particles below 200 um.For the problem of particle adhesion and overlap,a adhesion particle segmentation method of tailings which combines distance transform,marker control watershed and expansion algorithm is studied,achieving good segmentation of adherent particles in flotation tailings coal transmission image,which used Sobel operator to extract edges.Finally,the number and shape parameters of the coarse particles in tailings were obtained,and a prediction model for coarse particles content in flotation tailings（coarse particle loss）based on image detection was established.Finally,flotation process optimization control strategy based on product indexes is preliminarily studied and a coal flotation process optimization control framework is proposed,and a soft-sensing method of clean coal ash content was studied and simulated in this paper,the results show that when the flotation tailings ash content was added into input variables of clean coal ash soft-sensing model,the accuracy of model prediction is improved obviously.It also shows the importance of the detection of flotation tailings coal ash to some extent,and provides a useful idea for the soft-sensing model of flotation cleaned coal ash.This paper proposes a detection method for ash content,concentration and particle size of coal flotation tailings based on image processing.The detection device has the advantages of no contact with coal slurry,no radiation,and intuitive and accurate results.It is helpful to solve the technical problem that the quality of the flotation tailings is difficult to achieve online detection.It is of great significance to realize the intelligentization and improve the separation efficiency of coal flotation process.