Marcus Haywood-Alexander

Research area

Physics-Enhanced Machine Learning for Monitoring Systems

Data-driven approaches offer a method to overcome the challenges faced by the complexity of modern structures and systems, provided enough data. However, such data-driven models suffer from being restricted to the domain of the instance in which the data was collected; i.e. they lack generalisability. The fusion of physics with machine learning technologies allows for further progression by; improving generalisability, reducing the amount of data required, and have potential for improving public trust. In the context of Structural Health Monitoring, informed machine learning models will have a significant impact by reducing the data required to adequately encompass the environmental and operational envelope, reducing costs and improving true-detection rates.