Artificial neural networks to model formulation-property correlations in the process of inline-compounding on an injection moulding machine

摘要

Today the global market poses great challenges for industrial product development. Complexity, diversity of variants, flexibility and individuality are just some of the features that products have to offer today. In addition, the product series have shorter lifetimes. Because of their high capacity for adaption, polymers are increasingly able to displace traditional materials such as wood, glass and metals from various fields of application. Polymers can only be used to substitute other materials, however, if they are optimally suited to the applications in question. Hence, product-specific material development is becoming increasingly important. Integrating the compounding step in the injection moulding process permits a more efficient and faster development process for a new polymer formulation, making it possible to create new product-specific materials. This process is called inline-compounding on an injection moulding machine. The entire process sequence is supported by software from Bayer Technology called Product Design Workbench (PDWB), which provides assistance in all the individual steps from data management, via analysis and model compilation, right through to the optimization of the formulation and the design of experiments. The software is based on artificial neural networks and can model the formulation-property correlations and thus enable different formulations to be optimized. In the study presented, the workflow and the modelling with the software are presented.