Schmidberger, E.E.SchmidbergerMuessigmann, U.U.MuessigmannAbele, A.A.AbeleSchmutz, M.M.Schmutz2022-03-082022-03-081995https://publica.fraunhofer.de/handle/publica/303290Process for the monitoring of a manufacturing process, whereby (a) process codes (Pk1...Pkn) are determined from the signal signatures at several points of the process during a significant process section for the manufacture of a product; (b) the process codes occurring during the manufacture of products are allocated to a neural network (NN1) as its input parameters; (c) the values of quality features (Qk1...Qkm) are allocated to the nerual network (NN1) in an evaluation phase; (d) the neural network (NN1) is trained; (e) quality features values or classes are derived from the process codes by means of the trained neural network (NN1) in a monitoring phase, said values being displayed and/or documented and/or used to sort the manufactured products (OK and NOK) and/or to control the process.de608670patent1994-44185321995-95157370