10th International IFAC Symposium on Computer Applications in Biotechnology KNOWLEDGE BASED DISCOVERY IN FED-BATCH BIOPROCESS Andrei Doncescu * Sebastien Regis ** * LAAS CNRS, Avenue du Colonel Roche,31007 Toulouse France ** GRIMAAG Group, French West Indies University, 97159 Pointe-`a-Pitre Guadeloupe, France Abstract: We present a data mining approach based on a clustering method to detect and characterize states in a fed-batch processes. This method is based on the detection of singularities in biochemical signals and on the correlation between these signals. A segmentation based on maxima of wavelets transform is used to make an adaptive and dynamical correlation of the signals. The segmentation enables the detection of the borders of states whereas the correlation enables to characterize the physiological states. The method is applied successfully on a fed- batch process and particular states (difficult to detect with classical methods of classification) are detected and characterized. Keywords: Bioreactor, Fed-batch fermentation, Classification, Wavelet, Correlation. 1. INTRODUCTION Yeasts are a very well-studied micro-organisms and today, such micro-organism like Saccha- romyces Cerevisiae are largely used in various sectors of the biomedical and biotechnology in- dustrial bioproces. So, this is a critical point to control such processes. Model-based methods are the most used tool for the bioprocesses because of the mathematical modelisation of the phenom- ena (see (Roels, 1983)), but these methods using simulation techniques can lead to wrong conclu- sions because of lack of description parameters or during an unexpected situation. Nowadays non- model-based methods have an increasing success in bioprocesses. The non-model-based methods are mainly based on the analysis of biochemical signals (also called biochemical parameters). Two directions have been explored: (1) the ”manual” on-line analysis : it does not allow to identify in an instantaneous manner and with certainty the physiological state of the yeast. (2) the ”manual” off-line analysis : it allows to soundly characterize the current state, but generally too late to take into account this information and to adjust the process on the fly by actions of regulators allowing to adjust some critical parameters such that pH, temperature (addition of basis, heat, cooling). To remedy these drawbacks, computer scientists in collaboration with micro-biologists develop tools for supervised control of the bioprocess. They use the totality of informations provided by the sensors during a set of sample processes to infer some general rules to which the biological process obeys (see for example in (Aguilar-Martin et al., 1999)). These rules (Steyer et al., 1991) can be used to control the next processes. Clas- sification, supervised methods, learning and more generally data mining are also used for these Preprints Vol.1, June 4-6, 2007, Cancún, Mexico 339