Measurement uncertainty quantification for computer vision-based structural dynamic displacement monitoring

Abstract

The structural dynamic displacement response is a crucial indicator for assessing the condition and performance of structures. Computer vision-based structural dynamic displacement measurement (CV-SDDM) has emerged as a promising non-contact technique. However, a key aspect of promoting any measurement technology is the scientific assessment of its measurement uncertainty. This study develops a measurement uncertainty quantification model for CV-SDDM systems and proposes specific quantification methods. The proposed approach is validated through a bridge shaker experiment, demonstrating that the measurement errors of CV-SDDM consistently fall within the estimated uncertainty bounds. Quantitatively, the highest exceedance rate of measurements beyond the estimated uncertainty bounds was 2.03 %, while in most cases it remained below 1 %. Furthermore, this study analyzes the effects of hardware parameters and software algorithms on measurement uncertainty. CV-SDDM systems offer flexible hardware configurations for different monitoring scenarios. This study discusses how these configurations affect system performance and provides practical guidelines for researchers and engineers.