Seed Data Analysis for Production Fermenter Performance Estimation

Abstract

The ability to supervise highly non-linear and time variant bioprocesses is of considerable importance to the bio-industries which are continually striving to obtain improved productivity and to reduce process variability. In this contribution the seed stage (a batch process) of an industrial antibiotic fermentation is considered to correlate the performance of the main production fermentations with the quality of the seed. Four pattern recognition techniques were applied to the seed data: Artificial Neural Networics (ANNs), Principal Component Analysis (PCA), Non-Linear Principal Component Analysis (NLPCA) and Kohonen Self Organising Feature Map (SOFM). It is shown that high and low performing fermentations can be distinguished using information from only the seed stage. Data from industrial penicillin G fermenters is used to demonstrate the procedures.