Automated, Adaptive and Uncertainty-aware Smart Measurements using Machine Learning
Smart measurements are key enabling technologies to achieve and realise European strategic policies devoted to sustainability and digitalisation within the framework of Industry 4.0 and European Green Deal. These smart multidimensional measurements are often challenged by long acquisition times or limited resolution. This project will develop methods to significantly improve the efficiency of such measurements.
Smart measurements using machine learning
Multidimensional measurements that perform measurements sequentially are often time-consuming and costly. Performing these measurements "smart", using mathematical methods and tools from machine learning enables faster measurements, while maintaining high-accuracy.
Validated Uncertainty Quantification
Employing machine learning tools and deep learning techniques to perform a "smart" measurement requires validated uncertainty evaluation procedures to ensure thrustworthyness. Generative models that form prior knowledge and interpretable regularization techniques are explored.
Adaptive Measurement Protocols
Smart measurement procedures are automatically and adaptively adjusted during the measurement process. Incorporating physical properties and limitations in the development of methods is required to ensure practical implementation in experimental prototypes.
Generalizing guidance and software
Application to several industrial case studies will result in guidance documents and software solutions to generalize the developed approaches to other use-cases and facilitate the applicability.