Speaker
Description
Artificial intelligence (AI) is revolutionising measurement systems in electrical and electronic engineering, enabling advanced data analysis, real-time decision-making, and automation of complex measurement tasks. However, the deployment of AI-based models in measurement contexts introduces critical challenges, particularly regarding the reliability and traceability of their outputs and the sustainability of the measurement processes themselves.
This work addresses these challenges by developing rigorous methodologies to quantify the uncertainty associated with AI models used in measurement tasks. Traditional measurement systems rely on well-established physical models with defined uncertainty budgets and metrological traceability. In contrast, AI models are often treated as black boxes, lacking explicit evaluation of their measurement uncertainty, thereby limiting their acceptance in safety-critical or regulated contexts. Our research proposes the integration of metrological uncertainty quantification methods with AI model validation, enhancing the credibility and interpretability of AI-based measurements. The developed approach combines sensitivity analysis, probabilistic modelling, and performance metrics to provide comprehensive uncertainty budgets for AI-assisted measurements, thus enabling their integration into industrial, healthcare, and scientific applications that demand high confidence levels.
Furthermore, this work explores the evaluation and improvement of the environmental sustainability of measurement processes. Measurement activities, while typically considered low-impact, involve instrumentation, power consumption, maintenance, and data processing infrastructure that contribute to environmental footprints, particularly in large-scale or continuous monitoring applications. We introduce a systematic framework to assess the energy consumption and environmental impact of measurement systems, identify key contributors to unsustainable practices, and propose mitigation strategies, such as optimising measurement protocols, enhancing equipment efficiency, and adopting eco-design principles in instrumentation development.
By combining uncertainty quantification and sustainability assessment, this research contributes to building trustworthy, resource-efficient, and socially responsible measurement systems, thereby enabling safer and more sustainable adoption of AI technologies in critical sectors and aligning with global goals for industrial innovation and environmental protection.