Abstract:
|
The determination of shelf life is fundamental to ensure quality and safety of products. To this purpose kinetic models have been proposed in the literature. In particular, Bracket method allows us to estimate the degradation rate and the shelf life of a product under specific temperature conditions. This approach, however, does not take into account a potentially relevant factor: humidity. To overcome this limitation, we introduce a modified Bracket method which applies the humidity-corrected Arrhenius equation to consider the effect of both temperature and humidity. On the other hand, we also propose the adoption of machine learning models, to avoid specific assumptions about the functional form of the relationship between variables. The performances of these proposals are evaluated through a toy example, which allows us to show the importance of adopting appropriate modelling techniques considering humidity for shelf life estimation.
|