Using Continuous Time Stochastic Modelling and Nonparametric Statistics to Improve the Quality of First Principles Models

Abstract A methodology is presented that combines modelling based on first principles and data based modelling into a modelling cycle that facilitates fast decision-making based on statistical methods. A strong feature of this methodology is that given a first principles model along with process data, the corresponding modelling cycle can be used to easily, rapidly and in a statistically sound way produce a more reliable model of the given system for a given purpose. A computer-aided tool, which integrates the elements of the modelling cycle, is also presented, and an example is given of modelling a fed-batch bioreactor.