Modelling and control of the fed batch fermentation process using statistical techniques

In order to achieve better control and hence improve the productivity of the fed batch fermentation process, important state variables such as product concentration and soluble nitrogen level have to be estimated at each moment of time. In industrial fed batch fermentation process these variables are determined by infrequent off-line laboratory analysis making this set of variables of limited use for control. In this paper statistical techniques such as Karhunen-Loeve (K-L), principal component regression (PCR) and partial least squares (PLS) are used to estimate the nitrogen utilization rate and from this, the soluble nitrogen level. The performance of these approaches for the online estimation of the unmeasurable state variables is discussed and illustrated. The aim of this work is to control the soluble nitrogen around a desired level based on estimating the nitrogen utilization rate and by controlling the amount of nitrogen fed to the fermenter