Grey-box stochastic modelling of industrial fed-batch cultivation

This paper presents application of a grey-box stochastic modelling framework for developing continuous time models for dynamic systems based on experimental data. The framework is used to develop predictive models suited for control purposes. The main idea behind the framework is to combine stochastic modelling with data to obtain information on parameter values and model (in-) validity. In case the proposed model is falsified the method can point out specific functional deficiencies which facilitate further model development. The developed model can be used for monitoring purposes as well as serve as a basis for advanced multivariablc control to reject both intrabatch and interbatch disturbances. The industrial fermentation investigated in this case is production of an amylase enzyme by a filamentous fungus.