Knoth, Sven

Accelerometers play a crucial role in measurement systems utilized for structural health monitoring (SHM) in the field of civil engineering. Structural vibrations are analyzed to gather data that helps evaluate the integrity, performance, and safety of infrastructure. Their accurate measurements allow for early identification of abnormalities, support proactive maintenance, and guarantee the resilience of civil constructions. Yet, practical conditions encountered in the field produce variations in signal response that impact the precision and performance of sensor responses. This highlights the necessity for a quality assurance approach for SHM measurements. This report systematically investigates the sources of variation contributing to measurement uncertainty in sensor responses. To study the factors causing these variations, data were collected from two accelerometers using a shaker table apparatus and a factorial design of experiments. Amplitude, a fundamental signal feature, was evaluated in eight different ways to characterize measurement uncertainty, study static characteristics such as repeatability and linearity, and perform in situ calibration. The characteristics were compared over both short and long time‐windows as well as between two sensors. The findings reveal that amplitude, depending on how one calculates it, is a simple feature that effectively demonstrates both linearity and proportionality when correlated with shaker acceleration. This makes them exceptionally suitable for fast, efficient quality assurance (QA) tasks. Such analyses help evaluate uncertainties caused by environmental, operational, instrumentation, and human factors affecting sensing systems. The presented methodologies for evaluating measurement uncertainty contribute to the essential set of tools needed for QA of sensing systems.

The list-level proportion congruency effect (PCE) and the context-specific PC (CSPC) effect are typical findings in experimental conflict protocols, which competing explanations attribute to different mechanisms. Of these mechanisms, stimulus-unspecific conflict-induced selectivity adjustments have attracted the most interest, from various disciplines. Recent methodological advances have yielded an experimental procedure for entirely ruling out all stimulus-specific alternatives. However, there is a stimulus-unspecific alternative–temporal learning–which cannot even be ruled out as the sole cause of either effect with any established experimental procedure. That is because it is very difficult to create a scenario in which selectivity adjustments and temporal learning make different predictions–with traditional approaches, it is arguably impossible. Here, we take a step towards solving this problem, and experimentally dissociating the two mechanisms. First, we present our novel approach which is a combination of abstract experimental conditions and theoretical assumptions. As we illustrate with two computational models, given this particular combination, the two mechanisms predict opposite modulations of an as yet unexplored hybrid form of the list-level PCE and the CSPC effect, which we term context-dependent PCE (CDPCE). With experimental designs that implement the abstract conditions properly, it is therefore possible to rule out temporal learning as the sole cause of stimulus-unspecific adaptations to PC, and to unequivocally attribute the latter, at least partially, to selectivity adjustments. Secondly, we evaluate methodological and theoretical aspects of the presented approach. Finally, we report two experiments, that illustrate both the promise of and a potential challenge to this approach.