Product Quality Sensors for ML-Guided Process Optimization and Control of Modularized Production Plants
Members: Iowa State University, Novonesis
Project dates: 2023 - 2026
Researchers at Iowa State University designed a bioreactor that can be controlled and optimized via a reinforcement learning (RL) agent. They designed a digital twin to demonstrate R-based control and a bioreactor array system to obtain the real-time variables – like enzyme activity level and cell concentration – necessary to train the RL agent.
Process monitoring is very important during the production process of enzyme production. Currently, parameters that can be monitored are limited to the available on-line sensor, such as temperature, pH value, and oxygen level. Some critical parameters, such as the produced enzyme activity, still need to be manually sampled and measured in the laboratory. This new tool allows biomanufacturers to automatically and continuously measure additional variables like cell density and enzyme activity. This more efficient design and process control tool will be critical to more rapid market entry and better product quality.
Through the course of this work, the team also created low-cost reactors that they discovered could be used to train students in bioprocessing, which are now being built out as educational kits. Going forward, researchers are partnering with a private sector company to build a highly autonomous bioreactor that includes sensors for real-time monitoring of enzyme activity for commercial use, allowing others in the industry to scale up and produce more biomanufactured products.
Funding source: Schmidt Sciences
Related news:
Learn how bioreactors – the engines that power bioindustrial manufacturing – are getting an upgrade through this and other projects. Thanks to BioMADE projects funded through Schmidt Sciences, advancements in AI-guided control systems, vortex-powered mixing chambers, wireless sensors, and modular architectures will supercharge the future of bioindustrial manufacturing.