Application of a statistical procedure for the control of yeast production.

In many cases of fermentation processes, it is difficult to measure all of the state variables needed to indicate the culture state and to predict state changes. We propose the use of time-series data of measurable state variables instead of unmeasurable to describe the fermenatation process. We also employ a discrimination method of independent variables of regression analysis using statistical measures, PSS and AIC. These statistical procedure were applied to fed batch culture for yeast production under aerobic conditions, and the effectiveness was ascertained experimentally.