Structure and Parameter Identification of a Batch Fermentation Process using Non-Linear Modelling

Recently powerful non-linear predictive error estimation algorithms have been developed based on the orthogonal forward regression procedure by Billings, Chen and coworkers. These techniques involve Radial Basis Functions (RBF) and linear-in-the-parameters non-linear polynomials both describing the system to be estimated. This paper examines both estimation techniques for their contribution as a process-predictor for biotechnology. Both techniques will be used to predict two outputs (specific growth rate and biomass concentration) based on two inputs (outlet gas-measurements of oxygen and carbondioxide) from a lab-scale batch fermentation process. Results show that both techniques can be applied to model the outputs.