Data science-based modeling of the lysine fermentation process.

Area of Science:

• Biotechnology and Biochemical Engineering
• Data Science and Machine Learning Applications

Background:

• Traditional mathematical and knowledge-based models struggle with unknown parameters in large-scale fermentation.
• Commercial-scale fermentors require advanced methods for predicting and controlling target production effectively.

Purpose of the Study:

• To develop an ensemble learning model for predicting amino acid fermentation behavior.
• To enable precise control and stabilization of lysine production in commercial-scale biotechnological processes.

Main Methods:

• Utilized an ensemble learning approach to predict fermentation process dynamics.
• Integrated observation values from fermentation tanks and future control inputs for model training.
• Applied dynamic fermentation controls to achieve high-order stability for lysine production trajectories.

Main Results:

• The ensemble model accurately predicted fermentation process behavior at a commercial plant scale.
• Identified the specific influence of control inputs on lysine production throughout the culturing period.
• Achieved stable, high-level lysine production through dynamic control strategies.

Conclusions:

• Ensemble learning offers a robust, data-science-based solution for complex biotechnological processes.
• This approach facilitates novel Industry 4.0 control systems for industrial fermentation.
• The developed model provides a pathway for enhanced production and control performance in biomanufacturing.