In the scientific fields, more and more data are to be analysed. Those data may take the form of Time Series, which are a temporal data class, containing a record of chronological values. Those values are considered as a whole instead of a list of individual and independent data. Moreover, Time Series generally contain a large number of values and can be stored in a database in quite large quantity. In this article, we propose a storage tool for Time Series, by using their abstract model (differential equation), leading to the new notion of Models Warehouse aiming at giving a new representation of Time Series and an alternative storage solution. Unfortunately, this method triggers a significant cost in terms of computation time. Our proposal is implemented and validated using real dataset issued from automatic.