CC BY 4.0Gauweiler, PascalPascalGauweilerZheng, XiaorongXiaorongZhengCornehl, LucieLucieCornehlKicherer, AnnaAnnaKichererGruna, RobinRobinGruna2025-04-102025-04-102025https://doi.org/10.24406/publica-4521https://publica.fraunhofer.de/handle/publica/48645010.24406/publica-4521What distinguishes good wine from premium wine is the quality of the berry material. The sugar and acid content are the key parameters in this case. To measure these, a timeconsuming and costly labor analysis is usually required during delivery process to the cellar. Near-infrared (NIR) spectroscopy offers a faster alternative, but is also used in laboratories due to its size and integrability. The development of the miniaturization of NIR-spectrometer opens up new possibilities for integration. To explore this, such a spectrometer and a pump with particle filter were integrated into a harvester to measure the quality parameters online during harvesting. The filter size of the pump that delivers the grape must to the NIR-spectrometer for analysis is a critical parameter as suspended solids in the must affect accuracy. This study examined the filtration requirements to reduce this problem using 38 samples from four grape varieties, filtered in 9 steps from 1mm to 40μm and centrifuged as the final step considered as free of suspended solids resulting in 359 spectra. Chemometric analysis via PLSR showed that, for sugar prediction, a finer filter is needed, whereas a proper preprocessing effectively eliminates the influence of solids on acidsenNIR-SpectroscopyAgricultureFood QualityChemometricsVineGrape Mustconference paper