Study On Online Detection Methods Of Particle Size Based On Digital Image Processing Technology

In the mineral processing,the ore particle size distribution is an important parameter toevaluate the crushing effect and realize the automation of mineral processing.The minepre-mill is an efficient broken grinding equipment which includes crushing and grinding.Torealize online detection of particle size distribution and based on the statistical results,theworking parameters of pre-mill will be effectively controlled and the capacity will beimproved into the highest level that the efficiency of equipment will achieve the best.So,thereal time and accuracy of size measurement is very important.At present,sieving granulationis the most commonly used to determine the particle size distribution,the efficiency is low andcan not be used for online detection, so,its can not meet the growing requirements of onlinereal-time detection of particle size in industrial production.Therefore,there has high researchvalue and a broad perspective in application to develop a new particle detection method forgrinding requirements.The mainly aims of this thesis is using image processing technology by MATLAB todesign a set of reliable online particle detection method according the productionenvironment.The main research achievements:(1)By comparing the different methods of image segmentation,the gray,median filter,gradient performed on the particle image and segmentation the image by marked a watershedsegmentation.the overlap edge of particles has be well segmented.(2)Calibration the pixel of each grade of ore particles,and creating area detector operatorcorresponding through the results of calibration,then,statistics the number of each gradethrough the HMT(hit/miss transformation).Compared the particle size distribution ofstatistical results with the screening results,it shows that the statistical results is in accordancewith the true size distribution and verify the accuracy of the detection operator.(3)Established the BP neural network model by he particle size measurement results ofsurface layer as input, the overall size distribution as output.The size detection experiment inthe stationary state shows that the model has certain accuracy.(4)Established the particle size measurement device and design the operation interface bycombining the image acquisition toolbox, image processing toolbox, neural network toolboxand graphical user interface development environment (GUIDE) of MATLAB.Through thedynamic particle detection experiments,the results show that the detection method has better practicability that it’s can detect the particle size of ore that in compact accumulationaccurately.Writing each test results into EXCEL form through the communication between theMATLAB and EXCEL,so that record test data in real time.This research innovates in the online particle size measurement in the crushing andgrinding processing, the product quality of crushing and grinding processes will be effectivelycontrolled by combined the detection method with current control system of crushing andgrinding equipment.So,the detection method has broad application prospects.