Machine Vision-based Sealing Test Of Automotive Filter Element

Honeycomb air filter is a new type of structured air filter used in European and American heavy-duty trucks.During the manufacturing process,the gap between the holes caused by the uneven amount of hot melt adhesive and the bonding time interval are inaccurate.It will cause local failure to effectively bond,forming a bypass of the filter element,external air will directly enter the engine,and the filter will fail,exacerbating the wear of the piston and the cylinder wall.If the entered particles are too large,it will cause a "pulling cylinder" phenomenon.Honeycomb air filter is very necessary for quality inspection.At present,the industry’s defect detection of honeycomb air filters is through manual inspection.A light source is placed at the bottom of the filter.Workers hold a magnifying glass to scan the upper surface of the filter and judge whether the filter has defects based on whether or not "light leakage".The complex honeycomb shape inside the filter The structure makes its detection process take a lot of time and labor costs.The detection efficiency and accuracy depend on the proficiency of the inspector,the degree of mood fluctuations and the continuous working time.Long-term detection will cause great eye damage,and if the size of the filter element is too large,,Is a task that cannot be completed at all,and cannot meet the needs of large-scale production.In view of the above problems,combined with the structural characteristics and working principles of the filter element and the characteristics of the defects to be detected,this paper proposes a complete visual inspection system for the defects of the honeycomb air filter element,combined with the experience and defects of manual inspection,to develop an overall design plan for this topic.To realize the automatic loading and unloading of the filter element and the automatic identification of defects.First of all,one-time imaging technology for ordinary small and medium-sized objects cannot meet the detection accuracy requirements.Large-size imaging technology is introduced.The designed rectangular coordinate robot drives the camera to divide the filter detection area into several sub-areas.To avoid unsealed chambers,It was mis-segmented,and designed the format segmentation scheme and the overlapping size of adjacent sub-regions.In order to clearly shoot the unsealed cavity holes and reduce the impact of distortion,a telecentric lens was used for imaging;the filter-unsealed cavity defect recognition algorithm is based on The grayscale abruptness characteristics of the local area,the obtained image is sequentially subjected to median filtering noise reduction,Otsu threshold segmentation,and defect areas are marked based on 8 neighborhoods.Secondly,in order to improve the intelligence and detection efficiency of the filter element defect detection device,an automatic filter element loading and unloading device is designed.The incoming conveyor belt is designed to have variable speed and fixed-point stop functions to cooperate with the parallel robot to perform fixed-point grabbing of the filter element to be inspected.The gripper is designed with an end manipulator.Finally,the control system is analyzed and designed.The industrial computer is used as the upper computer to realize the automatic loading and unloading of the filter element,automatic detection,real-time display of the detection results,receiving the information sent by the PLC and operation information monitoring;PLC as the lower computer,directly control the moving parts.