利索-全球专利检索

  1. Login
  2. Sign Up

Search Results

279 records (took 0.013 seconds)

    OPTICAL DETERMINATION OF A CONTROL SIGNAL FOR SLABSTOCK POLYURETHANE FOAM PRODUCTION
    5.
    发明公开
    OPTICAL DETERMINATION OF A CONTROL SIGNAL FOR SLABSTOCK POLYURETHANE FOAM PRODUCTION 审中-公开

    公开(公告)号：EP4257327A1

    公开(公告)日：2023-10-11

    申请号：EP22167358.5

    申请日：2022-04-08

    Applicant: Evonik Operations GmbH

    Inventor： KIWITT, Jörn ,  LANDERS, Rüdiger ,  TOMUSCHAT, Philipp

    IPC: B29C44/60 ,  B29C44/28 ,  B29C44/56 ,  B29C44/46

    Abstract: Disclosed herein is a method of producing a polyurethane foam block (137) using a foam production machine (100) configured for producing polyurethane foam as a slabstock (137) using a continuous slabstock process defined by at least one process parameter, such as a raw material output factor or a conveyor speed, the foam production machine including a mixing head (102) configured to mix precursor reagents for forming a reactive mixture, and a conveyor (106) configured to receive the reactive mixture. The method comprises: mixing (800) the precursor reagents in the mixing head to provide the reactive mixture; dispensing (802) the reactive mixture to initiate foaming of the reactive mixture; receiving (804) the reactive mixture by the conveyor; cutting (806) the polyurethane foam block from an end portion (720) of the slabstock to produce a cut surface (724), wherein the polyurethane foam comprises foam cells, wherein the intersection of the cut surface with the foam cells forms a foam cell structure (906); acquiring (808) a digital image (742) of at least a portion of the cut surface using an imaging system (706), wherein the digital image depicts the foam cell structure; inputting (810) the digital image into a trained machine learning module (762, 1000, 1200), the trained machine learning module being trained to produce a two-dimensional segmentation (1012) of the foam cell structure at the cut surface; and receiving (812) a control signal (744) for changing the least one process parameter of the continuous slabstock process in response to inputting the digital image into the trained machine learning module.

Patent Agency Ranking