Fermentation Parameter Prediction Using Neural Network Digital Twins
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Predicting the outputs of fermentation processes is challenging due to complex microbial interactions and the difficulty in determining the appropriate fermentation parameters to achieve desired product characteristics.
Innovation Solution
A system and method using neural networks and machine learning technologies to predict fermentation parameters and product attributes, minimizing experimental work by modeling microbial interactions and metabolic pathways, and providing insights through explainability methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physical experimentation is used to determine fermentation parameters, then product attribute prediction accuracy is improved, but time consumption and resource waste increase
Solution Approach 1:
The patent creates a virtual copy of the fermentation system through a digital twin model that replicates microbial interactions, metabolic pathways, and process parameters. This virtual model allows unlimited experimentation without physical constraints, enabling rapid prediction of product attributes for various fermentation scenarios without actual laboratory work.
Solution Approach 2:
The system performs preliminary modeling and simulation of fermentation processes before actual production. By pre-calculating optimal parameters and predicting outcomes through the digital twin, the system eliminates the need for extensive trial-and-error experimentation, significantly reducing time consumption while maintaining prediction accuracy.
2Reliability
If extensive physical experimentation is conducted, then reliable parameter determination is achieved, but development costs and waste increase
Solution Approach 1:
Instead of physically experimenting with actual microbial cultures and materials, the patent uses a virtual digital twin that copies all essential processes, interactions, and outcomes. This allows unlimited replication of experiments without consuming any physical resources, maintaining result reliability while eliminating waste entirely.
Solution Approach 2:
The patent replaces physical experimental systems with a computational model that uses algorithms and data processing to simulate fermentation dynamics. This substitution eliminates the need for physical materials, equipment, and laboratory infrastructure, reducing development costs and resource waste while maintaining scientific validity through rigorous mathematical modeling.
3Measurement precision
If complex microbial interactions are modeled, then product attribute prediction accuracy is improved, but model complexity increases
Solution Approach 1:
The patent divides the complex fermentation system into separate modular components, including individual microbial species models, metabolic pathways, and process parameters. Each component can be independently developed, validated, and adjusted, making the overall complex system manageable while maintaining high prediction accuracy through integrated simulation.
Data Source
AI summary
In variants, the method can include: determining a set of fermentation parameters; determining a set of features associated with the set of fermentation parameters; and determining a set of product attributes associated with the set of features. In examples, the method can optionally predict the attributes of a product manufactured using the set of fermentation parameters and/or predict the set of fermentation parameters that would create or replicate the attributes of a target product.