| The utilization rate of potassium resource is directly related to the sustainable development of potassium fertilizer industry in China.With the development of intelligent plant,the demand of potash fertilizer production in Qinghai Salt lake is higher and higher.Through machine vision technology,industrial camera identification technology instead of manual identification method is used as the basis of potassium fertilizer production process optimization.The data collected by machine vision technology on the production process line is not only more accurate and faster,but also can greatly improve the efficiency of the production process.The field intelligence level of potash fertilizer production line is low,the feedback speed of problems in each stage of the production line is slow,and the whole production line has a large production error.In view of the characteristics of the process,it is found that the particle size of the raw ore containing sodium carnallite has a certain influence on the grade of potassium fertilizer produced,and the particle size directly affects the dosage and water amount of the crystallization process.Therefore,based on the process optimization of the cold crystallization process in the production process of potash fertilizer,this paper studied and designed a sodium-containing carnallite raw ore particle size measurement system based on machine vision,and completed the particle size measurement of the sodium-containing carnallite raw ore on the upper belt of the cold crystallization process.In this paper,algorithms of image preprocessing,feature extraction,edge detection and radius measurement of the minimum circumferent circle of edge contour are studied.The main contents of the research are as follows:1.according to the factory production environment to select suitable for the site of industrial cameras and light source equipment.In order to meet the requirements of field detection speed,industrial CMOS wire-sweep camera is used,equipped with professional wire-sweep lens,and special wire-sweep light is used to collect the picture information of raw ore containing sodium carnallite.2.Study the image analysis algorithm of the collected images.The traditional image analysis method is adopted,and the algorithm is written based on pythonOpen CV algorithm library.The processing results of the same image by different algorithms are compared and analyzed.Finally,the median filtering,flooding filling,image sharpening and Gaussian filtering are used to preprocess the image.3.According to the results of image pretreatment,the edge detection and contour extraction of raw ore particles containing sodium carnallite on the image are completed,and the minimum peripheral circle of the contour is drawn and the output result of radius is measured.The results obtained were screened.Due to the high humidity of raw ore,adhesion phenomenon would occur,so the results with a radius of 1 mm were selected as effective data,and the data beyond the range were the particle sizes of multiple particles adhered together and removed.4.Field practice and application.The mechanism model was established according to the theoretical formula,and the optimized process mathematical model was obtained by inputting the measured particle size results as new independent variables into the mechanism model of the crystallizer process section.Adjust dosage and water amount under different particle size.It was found that the accuracy of the crystallization process mathematical model was improved and the yield and yield of potash fertilizer were improved with the addition of particle size as independent variable. |
