Online- Qualitätsüberwachung attributiver Merkmale beim Spritzgießprozess

FKM 1998
Booklet No. 231
Project No. 207

Increasing demands on technical injection molded parts are leading to tougher competition. In addition to the high quality of his products, the producer must prove the consistency of production. IKV has developed a method for predicting continuous characteristics. This method is based on the calculation of measurable quality characteristics, such as dimensions and weights, from measured process variables. However, quality is not only determined by continuous characteristics. Attributive quality characteristics, such as burner, sink marks and surface gloss, quickly lead to unacceptable molded parts. If continuous characteristics can be measured objectively using appropriate equipment, attributive characteristics must be assessed subjectively by employees in production or quality assurance. The aim of this research project was to investigate the use of logistic regression and neural networks for online monitoring of attributive quality characteristics. Both neural networks and logistic regressions have proven to be suitable for predicting attributive characteristics from measured process variables. The model qualities achieved with neural networks are higher than those of logistic regressions. In the analyses carried out, model accuracies of 100 % were achieved for the quality characteristics investigated. However, this requires a parameter data set that is optimally adapted to the process. The direct processing of measuring points from the curve progression of process variables developed in this project enables the use of online monitoring without the need for process knowledge, which must necessarily be available when forming the characteristic values. Investigations into disruptive influences have shown that the process models are stable against batch fluctuations and colorations. However, grade changes and the use of the same materials from different manufacturers lead to significantly reduced model grades. Scope of report:
79 pages Start of work:
01. 08. 1995 End of work:
31. 07. 1997 Funding body:
AiF-No. 10304 Research center:
Institute for Plastics Processing, RWTH Aachen University Head:
Prof. Dr.-lng. W. Michaeli Processors and authors:
C. Schnerr, E. Henze, F. Ehrig Head of the expert group:
J. Rabe, Sachsenring Automobiltechnik AG Chairman of the advisory board:
Prof. Dr.-lng. H. Kipphan, Heidelberger Druckmaschinen