Bioautomation (Oct 2004)
Ultrasonic Measurements and its Evaluation for the Monitoring of Saccharomyces cerevisiae Cultivation
Abstract
The monitoring and supervision of batch Saccharomyces cerevisiae cultivations are presented by ultrasonic velocity measurements. The measurements are performed in a by-pass to reduce the influence of bubbles. Using these signals the typical phases of such cultivations can be identified. Applying a multi-linear regression model the ultrasonic velocity can be estimated by the biomass, the glucose and the ethanol concentration with a mean estimation error of 1.6 m/s. The multi-linear regression model has also been used to predict one of the three process variables by the other two and the ultrasonic velocity. Here the mean error of prediction is 0.6 g/L, 2.3 g/L and 1.5 g/L for biomass, glucose and ethanol concentration respectively. Using a Kalman filter theses variables have been estimated with mean errors of 0.6 g/L, 1.8 g/L and 1.6 g/L.
