Combined Application of Theoretical Modeling and Neural Networks in Vulcametry

06/2009 Juni

Results of vulcametry are related to mechanical properties of vulcanized rubber. Despite standing on firm theoretical grounds, these relations often prove to be inaccurate in practice because they are influenced by various other factors, hardly manageable by the theory. That makes predictions less reliable, thus demanding ample additional laboratory testing to be performed. The same goes for possibilities of theoretical extrapolations or simulations of extensive tests that would facilitate laboratory work. It is shown in this work how a sensible combination of both the theoretical models and neural networks can boost the performance in prediction of mechanical properties. Thus it can effectively be used to rationalize laboratory work in the field of rubber vulcametry and property predictions, without losing essentials, but rather gaining them.