An efficient model construction strategy to simulate microalgal lutein photo-production dynamic process

Area of Science:

• Biotechnology
• Bioprocess Engineering
• Microalgal Cultivation

Background:

• Lutein, a valuable bioproduct from Desmodesmus sp. microalgae, has significant industrial applications.
• Increasing global demand necessitates enhanced lutein production and accurate bioprocess simulation.

Purpose of the Study:

• To design and validate novel artificial neural networks (ANNs) for simulating complex microalgal bioprocess dynamics.
• To improve the accuracy and predictive power of models for lutein bioproduction.

Main Methods:

• Developed two novel ANNs that model the rate of change in dynamic systems.
• Employed hyper-parameter optimization, artificial data generation with noise, and input standardization for model accuracy.
• Validated models through experimental verification for real-time and offline bioprocess simulation.

Main Results:

• Demonstrated high accuracy and predictive power of the developed ANNs for long-term dynamic bioprocess simulation.
• Confirmed the models' effectiveness in both real-time and offline simulation frameworks.
• Established a robust model construction strategy applicable to other bioprocesses.

Conclusions:

• The novel ANNs provide a powerful tool for simulating and optimizing lutein bioproduction.
• This research facilitates future control and optimization strategies for microalgal bioproducts.
• The developed modeling approach has broad applicability across various bioprocesses.