Research And Development Of Valve Rocker Arm Installation Position Detection System Based On Machine Vision

Automobile engine is a technology-intensive product, therefore, the structure and theaccuracy of rocker arm make an important effect on engine valve system, and the installationposition error which began on the cylinder block can cause a big impact on the whole engineperformance. We need to rely on a lot of artificial works to finish the valve rocker arminstallation position detection before, this will not only increase the cost of production, butalso it is difficult to realize the integration of the information, what’s more, the human eyesare particularly susceptible to fatigue and instable, artificial detection may increase the errorrate of the product. As limited by detection accuracy, velocity and automation, artificialdetection can’t meet the requirements of modern industrial production line speed. Whilemachine vision inspection technology has some advantages as high efficiency, high flexible,high automation,24-hours non-stop work, easy to realize the information integration, which isa good way to satisfy the detection requirement for rocker arm installation position detection.According to the specific size test requirements of rocker arm installation positiondetection systems, designing the system of hardware structure and detection software. Itshardware mainly includes: image acquisition equipment, motion control mechanism, andVision processing equipment. Image acquisition device is combined by industrial camera,image acquisition card, optical lens and the light source. Industrial camera is connected to byGigE interface whose role is to process the image data to computer, motion control organ ismade up by the motion control card,7250I/O card, Siemens PLC, RFID communicationmodule/Siemens ASM456and stepper motor, etc.System software design is mainly used to implement the test of the filter gasket androcker arm installation position:To achieve the test for filter gasket, we should solve the problem of valve rocker armimage segmentation, this topic uses the improved Artificial Bee Colony algorithm for imagesegmentation, and through comparing image histogram with separated area to realize thegasket’s presence of detection. Artificial Bee Colony algorithm mainly adjust the traditionalArtificial Bee Colony algorithm’s bee honey of choosing way and defects, and it found outthe best threshold according to the information sharing and collaboration between leadingbees、scouting bees and following bees.Rocker arm installation position detection should solve these two problems abouttemplate matching and blob feature extraction. Template matching uses image templatematching algorithm of extended integral image with mask, Mask is used to exclude areaswhich are susceptible to interference, the algorithm’s template matching speed is improved byreducing the traversal window and template to match the image pixels, and simplifying thecalculation process. After finding matching template is a Blob analysis algorithm design, Inorder to reduce algorithm’s sensitivity to noise level, this paper puts forward a rapid algorithmfor calculating blob analysis which is suitable for industrial application based on the analysisof the previous methods. The algorithm only deals with the image data of adjacent two linesevery time, And a single scan images can finish all the connected regions of the tag, Because abandoned equivalent table, the algorithm has avoided problems by establishing equivalenttable to solve the tag conflict. Then adding traceability marking algorithms scan algorithm todeal with conflict tag, Makes sure that it can mark all shapes connected area accurately.Run time testing of the system, show that the detection system meets the requirements ofpractical application, and the actual operation results also show that the precision of thesystem size detection can be controlled at99%, system testing time can be controlled within1.5s. All of which show the system can meet the demands of industrial detection.