A novel load testing method for condition assessment of network-level highway bridges using moving artificial truck fleets in an open traffic environment

Load testing is the most reliable method for bridge condition assessment. Conventional load testing requires traffic closures, involves intensive labor, and lacks timeliness, presenting significant challenges for condition assessment of network-level bridges under open traffic conditions. This study proposes a novel load testing methodology for condition assessment of bridges within freeway networks, using moving artificial truck fleets in an open traffic environment. The methodology integrates four core information technologies: (1) recognition of traffic load sequences within monitoring regions via computer vision and data fusion, (2) reproduction of spatiotemporal traffic loads beyond monitoring regions using a hybrid virtual-real traffic simulation approach, (3) precise spatiotemporal mapping of traffic loads to measured responses through a modified dynamic time warping algorithm, and (4) assessing health conditions of bridges using aligned theoretical and measured load effects. The methodology was rigorously validated through field experiments on seven long-span bridges within a freeway network, completed in just seven hours while crossing 384 km. This load testing methodology demonstrated high accuracy and efficiency, presenting a viable alternative to conventional load testing methods that rely on static truck fleets requiring traffic closure. By accurately identifying the spatiotemporal distribution of traffic loads across the entire bridge deck (with averaged weighted longitudinal location matching errors ≤ 0.5 m and lane label matching errors ≤ 4 %) and achieving precise time-history alignments between theoretical and measured traffic load effects (with Pearson correlation coefficients ≥ 0.985), this study introduces an innovative input-output-based structural health monitoring approach for assessing the condition of network-level bridges.