Batch to Batch Improving Control of Yeast Fermentation

Publisher Summary This chapter discusses an extended data-driven modeling framework for iterative learning control of batch processes by means of simple linear models. The framework is extended to include a batch-to-batch dynamic description of initial conditions and their effect on batch evolution. The framework is extended with measurement noise. Based on desired model properties, regularization can be applied to obtain models applicable for control from sparse and noisy data. The potential of the proposed framework is demonstrated on simulated fed-batch yeast fermentations. The model development and the identification scheme are given, and an iterative learning model predictive control algorithm is presented. The tool is demonstrated with simulation results. The extended modeling and control framework as well as the identification scheme are validated through the simulated trajectory tracking of fed-batch fermentations with highly satisfactory results. Hence, the presented extended framework for modeling and control of batch processes exhibits significant potential for industrial implementation.