Online Detection System For Classification Effect Of Air Classifier Based On Machine Vision

The emergence of air classifier solves the frequent problem that sieve holes of the traditional screening equipments are easily blocked during the process of coal separation.The tradition method of detecting the classification effect of air classifier that sieving and weighing the coal sampled from the coarse-grained and the fine-grained products manually,then calculating undersize fraction and oversize fraction separately is time-consuming and laborious,the detection result is not representative enough and can not be fed back in time.Based on the phenomenon that experienced workers can observe the area of coarse and fine particles on the surface of the product by the naked eye to judge the classification effect of air classifier roughly,this paper replace the human eye with machine vision,design an online detection system for classification effect of air classifier based on machine vision,to detect the effect rapidly and continuously.According to the situation of the coarse-grained and the fine-grained products when transported on the conveyor belt,it was decided to select the fine-grained products as the object of detection.The core content of this article consists of two parts,the one part is determining the method of image processing.The steps of image processing in this research are: Step 1,transform RGB image into gray image;Step 2,use the Contrast Limited Adaptive Histogram Equalization(CLAHE)to adjust the gray distribution of the gray image;Step 3,retain the edge of particles while weaken the noise by using a bilateral filter;Step 4,use the binaryzation algorithm based on grayscale change rate to convert the gray image into binary image,the algorithm aimed at the features of the image in this study which is written by author;Step 5,after filling the small holes in the image,use the algorithms which is also written by author to segment adhesive particles.The image needs to be processed three times,those particles with particle size less than 6mm in the image need to be removed after each segmentation.Finally,the coarse particles in the image are preserved,and the ratio of the total area of coarse particles to the image area can be obtained by calculating.The other part is exploring the relationship between the mass fraction of coarse particles in coal samples and the ratio of the total area of coarse particles on the surface to the area of coal samples surface image by experiments when the thickness of coal samples is 30 mm,and the particle size range of coal is 0-10 mm,and determining the interval for judging the grading effect of air classifier.Considering the situation that coal is always wet in actual production,experiments about wet coal samples with different water content have been done to explore the influence of different water content on the judgement interval.At last,the design of online detection system for classification effect of air classifier based on machine vision has been finished by using MATLAB.This system can automatically obtain and process images of the fine-grained products transported on the conveyor belt at a set time intervals.The calculation results are displayed on the coordinate axis in the form of a line chart,with the judgement interval determined by experiments,the purpose of detecting classification effect of air classifier can be achieved.