- Building models or digital twins from limited data
- Bayesian methods and probabilistic modelling
- Uncertainty management and strategies for model learning and validation
- Management and optioneering for high integrity assets
- Machine learning, artificial intelligence and decision support
I have a background in material science and asset management for complex and safety critical systems. More generally, I’m interested in understanding how high integrity systems perform in service and to build models which describe this behaviour.
At Frazer-Nash, I lead a multidisciplinary team developing methods which are helping our clients understand the current and future performance of their assets. We build probabilistic models using a wide range of data analytic and machine learning approaches backed up with scientific or expert understanding. By understanding the future uncertainty, we can use models to explore options. This helps our clients to make operational decisions which have the best chance of ensuring that their assets meet their required safety, operational and financial performance.
In addition to building and training models, we have also developed methods for testing and validating them. The process of building trust in any model used for decision making is important, but can be essential when they are used autonomously in an AI approach.
I have specific experience of developing and applying these methods in a broad range of sectors including: the cores of both new and existing graphite moderated nuclear reactors; warships; land based gas turbines and offshore wind installations.