Methodology for selecting measurement points that optimize information gain for model updating

Abstract Information collected through sensor measurements has the potential to improve knowledge of complex-system behavior, leading better decisions related system management. In this situation, and particularly when using digital twins, quality data determines improvement that sensors have on decision-making. The choice monitoring system, including types their configuration, is typically made engineering judgement alone. As price devices usually low, large networks been implemented. are often used monitor at high frequencies over long periods, very sets collected. However, model predictions behavior influenced by only a few parameters. Informative thus difficult extract as they hidden amid redundant other irrelevant updating key parameter values. This study presents methodology for selecting informative within given model-updating task. By smallest set maximizes information gain, can be significantly refined, increased data-interpretation efficiency. Results an excavation case show gains with refined measurement much smaller than entire prior refinement same probability identification, while computational time reduced. supports engineers significant filtering performance.