Research Stream 2 projects
Modelling and Optimisation
Chief Investigators
Professor Uwe Aickelin (Lead), Dr Ling Luo, Professor Bogdan Gabrys
Research Fellow
Dr Mohammad Golzarijalal
Research Assistant
Jennifer Duong
PhD Students
Daniel Pham, Mitra Heidari
This project will enable overall process optimisation by examining how relatively sparse high-quality data can be supplemented with lower quality data to enable ‘multi-fidelity’ optimisation for improved decision support for bioprocesses. It will consider approaches that allow searches outside the usual ‘search box’, dynamic optimisation and the development of criteria for the early abandonment of experiments that go wrong.
Machine Learning; Automated and Explainable Machine Learning Approaches for Digital Bioprocess Development and Monitoring
Chief Investigators
Professor Bogdan Gabrys (Lead), Professor Uwe Aickelin, Professor Sally Gras, Dr Abel Armas Cervantes
Research Fellow
Dr Thanh Tung Khuat
PhD Students
Johnny Peng, Dilshan Sonnadara
Automated and explainable machine learning (ML) solutions will be explored in this project to identify and predict product quality attributes, online features, and critical tasks, such as factor analysis, classification, pattern recognition, and outlier and anomaly detection either in data records or work practices. The research will particularly focus on building general online continuous learning and adaptation solutions for predictive models in bioprocesses and unit operations.
Modelling and Optimisation for Validation
Chief Investigators
Professor Michael Kirley (Lead), Associate Professor Artem Polvyvanyy, Professor Uwe Aickelin, Professor Sally Gras, Professor Bogdan Gabrys
Research Fellow
Dr Mansoureh Maadi
PhD Students
Paul Ou, Navid Akhavan Attar
Data-driven approaches can be used to enhance decision-making in biopharmaceutical manufacturing. This project specifically examines how multi-criteria decision-making methods can be used at stages such as clone selection, where the most effective high-performance cell lines or clones are selected from a group of genetically identical clones. The project will also consider fault detection, as well as predictive maintenance and advanced forecasting techniques.
Modelling and Optimisation of Unit Operations
Chief Investigators
Dr Ling Luo (Lead), Professor Uwe Aickelin, Professor Michael Kirley, Professor Sally Gras
Research Fellow
Dr Bastian Oetomo
PhD Student
Saumya Karunadhika
Unit operations within the downstream processing stages of bioprocessing are central to this project, which will construct machine learning models to enhance real-time monitoring and process control. ML and AI techniques will be designed to analyse large-scale operational data from the process control platform and optimise the experimental design of key unit operations. Real-time monitoring and control will be improved and predictive methods developed to assist process optimisation.
Digital Twin Approaches for Automated Monitoring, Controlling and Improvement of Bioprocesses
Chief Investigators
Professor Marcello La Rosa (Lead), Associate Professor Artem Polyvyanyy, Dr Abel Armas Cervantes, Professor Michael Kirley, Professor Bogdan Gabrys
Research Fellow
Dr Zahra Dasht Bozorgi
PhD Student
Deanna Coralage
In this project, digital twin technology will be designed, implemented, and evaluated for monitoring, controlling, and improving real-world bioprocesses. An end-to-end digital representation of bio-engineering processes will be built and calibrated using automatically discovered simulation parameters to reflect real-world bioprocesses. The digital twin will provide a mechanism for assessing different scenarios, including configurations that go beyond the limitations of current laboratory equipment. The model will forecast the performance of future bioprocesses, such as expected yield and quality, and aid to identify potential issues at an early stage when interventions are most effective.