Research On Detection Method Of Plate Defects In Plate Heat Exchanger

Plate heat exchanger is an ideal equipment for heat exchange with compact corrugated plate.The quality of the plate directly determines the efficiency of heat exchange,and if the defective plate is put into use,it will have a significant impact on personal and environmental safety.At present,there are manual sampling inspection and penetration flaw detection methods for plate heat exchanger plate,but this kind of methods have high requirements for the detection environment,low detection efficiency,and are limited by the subjective factors of quality inspection personnel.In view of the deficiency of manual detection,the plate defect detection method of plate heat exchanger based on machine vision is designed to realize the automatic detection of plate surface defects.Firstly,the ROI(region of interest)boundary is located based on the understanding of the defect area.The ROI area in this thesis refers to the dark area of the corrugated plate.Positioning the ROI area can greatly reduce unnecessary detection work and increase the detection efficiency and accuracy.In this thesis,two ROI region boundary positioning algorithms are designed:(1)dark area boundary detection based on one-dimensional directional gradient.According to the gradient operator from the global gray scale change trend,calculate the gradient threshold value through the maximum interclass variance method(OTSU),and further extract the ROI region using the boundary.This method solves the problem of locating the dark area of the corrugated plate,or when the detection plate type is unchanged.(2)Border positioning of the dark area based on the transition line segment of the gray-scale distribution curve.This method explores a idea of using the convolution window to obtain image gradient,analyze the gray distribution curve of convex line segment,search the extreme point of image gray curve by the data statistics,give feedback on the data,the fundamental view of the boundary transition segment real height difference,the height difference is called the amplitude,and the image,locate the dark area boundary,and extract the ROI area.This method solves the positioning problem of corrugated plate ROI area in the case of many corrugated plate types or uncertain corrugated plate type,and realizes the automatic detection of area positioning.Then,the amplitude features of the defects are extracted through the concave and convex line segment in the located ROI region,and the amplitude features of the defects and the background are divided by adaptive threshold,and the defects are labeled.Finally,for reflective misdetection,the thesis proposes two algorithms based on the section width of convex line and distinguishing the concave and convex surface,and proposes the method based on the connected domain area to suppress the misdetection caused by rough position,so as to reduce the defect misdetection rate.In the field of machine vision,this research provides a new solution for plate detection,which makes up for the deficiency of manual detection to some extent,and has certain practical value.