Seite - 123 - in Joint Austrian Computer Vision and Robotics Workshop 2020

Figure 2. Variance of the difference image: high red in- tensity=highvariance =missingpowder orhighporositystandout fromthe imagein theform of high grey values. The variance of the grey val- ues insmall, rectangular sectors isusedasan indica- tor of quality. The observation in sectors is needed to locate the defects. A low variance indicates that fewdefectsoccur in thepowderbed, the layer iswell compacted. High values of the variance can indicate defectsorhigh porosity. 3.1.Largedefects A difference image with powder missing over a large area can be seen in Figure 2. Areas with miss- ingpowder(ortoomuch,albeitanunlikelycase)pro- duce large shadows, which appear as high grey val- ues in the difference image. If the defects are larger than the sectors of the detection, the mean value of the grey values have to be considered in addition to the variance. A weighted average of the two factors isused asa criterion for theevaluation. 3.2.Porosity Another factor that is evaluated is the porosity of thepowder layers. Thestrengthcharacteristicsof the printedcomponentsaredirectly related to thedegree of compaction of the powder. If the porosity is too high during the printing process, lower strength values are expected. Porous powder layers create shadows in the area of the grain size of the powder used,whichappearasnoise in thedifferential image. The variance of the grey values in the sectors is sufficient for detection. It should be noted that areas close to the light source tend to be overexposed, which makesevaluationmoredifficult. 4.ExperimentalValidation Totestwhether thestrengthofcomponentscanbe estimated during the manufacturing process, several seriesof test specimenswereprintedand then tested. It was confirmed that layers with few defects and Layer number Series C Series D Mean value series C Mean value series D Figure3.Varianceof thegreyvaluesof thepowder layers of two layers of testpieces withdifferentporosities a low porosity achieve higher strength values. The evaluation of the test is shown in Figure 3. Series C achievesSC = 10.8MPawith a compression ratio of 10%, while series D achieves SC = 11.4MPa with a compression ratio of 15%. Test specimens withhigherdensityandbetterstrengthcharacteristics showlowerporosity in theprintingprocess,which is reflected in the differential images as a lower vari- anceof thegreyvalues. 5.Conclusion Our tests have shown that a simple system con- sisting of one camera and two light sources is well suited for process control of powder bed based ad- ditive manufacturing processes. Both coarse defects in the powder bed and different porosities can be de- tected, which avoids production downtimes and en- ablesqualitycontrol alreadywhileprinting. References [1] T. Craeghs, S. Clijsters, E. Yasa, and J.-P. Kruth. Online quality control of selective laser melting. 22nd Annual International Solid Freeform Fabrica- tionSymposium-AnAdditiveManufacturingConfer- ence,SFF2011, 01 2011. [2] M. Erler, A. Streek, C. Schulze, and H. Exner. Novel machineandmeasurementconcept formicromachin- ing by selective laser sintering. InProceedingsof the InternationalSolidFreeformFabricationSymposium, Austin,TX,USA, pages4–6,2014. [3] I. Gibson, D. Rosen, and B. Stucker. AdditiveManu- facturingTechnologies–RapidPrototyping toDirect DigitalManufacturing, volume5. 012010. [4] Z.Li,X.Liu,S.Wen,P.He,K.Zhong,Q.Wei,Y.Shi, and S. Liu. In situ 3d monitoring of geometric sig- natures in the powder-bed-fusion additive manufac- turing process via vision sensing methods. Sensors, 18(4):1180,2018. 123