We are developing the techniques and platforms needed to support the new measurement modalities that are emerging in the digital age: large, high-dimensional data sets; inference techniques for optimisation and sampling; sparse/compressed sensing; sensor networks; AI and machine learning. We will ensure that these new technologies can be used with confidence in a metrological context by establishing the necessary uncertainty quantification methodologies and providing capability for assessing trust in algorithms and software.
As instruments and systems become fully digital, digital twins – or replicas – and models are needed to drive our understanding and supplement increasingly automated control systems. Currently, data are used to provide model inputs and for validation purposes, but the digital twin approach goes beyond this paradigm by using data gathered at every stage of manufacturing and usage to update the model throughout a product's lifetime. This approach means that lifetime predictions and maintenance schedules can be individually tailored for high-value assets. The approach depends on bringing together high-fidelity modelling, data of known quality, and a rigorous approach to uncertainty evaluation. Our work is developing tools for all three of these cornerstones.
We will develop the techniques, platforms and standards needed to underpin the digitisation of metrology and digital traceability chains: ontology, metadata standards, curation, data quality indicators; distributed ledger technology and data security. In addition, we will provide the science leadership for a number of NPL laboratory-of-the-future pilots as we take the first steps to ensure that NPL becomes an exemplar for digital and data-intensive metrology.