Answers are grounded in this paper and cite their sources.
Traffic load monitoring is essential for transportation system design and operation. Owing to the inefficiency and bias of traditional static weigh stations, weigh-in-motion technologies have been developed to acquire vehicle weights more effectively under real traffic conditions. Bridge weigh-in-motion (BWIM) treats the bridge as a weighing scale by recording structural deformation induced by passing vehicles and inferring axle loads from the measured responses. A key practical limitation of conventional BWIM is the need for a dedicated axle-detection subsystem to determine vehicle speed and entry/exit times, which are required to construct the inverse identification equations. To overcome this constraint, this paper proposes a novel BWIM framework incorporating a one-dimensional YOLO model for vehicle-event detection directly from bridge measurements. Field-test results demonstrate that vehicle-passing response intervals can be identified accurately, with precision and recall of 0.93 and 0.91, respectively. Based on the examined vehicle and test conditions, the proposed PINN-based BWIM approach achieves mean relative errors within 6.0% for individual axle weights and below 2.8% for GVW without knowing axle information. Furthermore, fleet-level analysis of 3,877 events identified as two-axle vehicles over nine monitored hours on three days shows repeatability of the inferred GVW distributions. Since the true fleet weights are unknown, these statistics do not validate absolute weighing accuracy or establish long-term stability.
No takes yet. Share an insight, caveat, or question.
Weng et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: