A computer vision-assisted method for identifying wheel loads of moving vehicles from dynamic bridge responses

Abstract

Bridge weigh-in-motion (BWIM) technology, which identifies traffic information such as the axle weights, wheelbases, and velocities of vehicles passing over a bridge from bridge responses, can be beneficial for bridge design, management, and maintenance. However, conventional BWIM methods simplified the coaxial wheel loads of a vehicle using one concentrated force, commonly known as the axle load. Such simplification would lead to two problems. Firstly, it fails to calculate the wheel weights of moving vehicles, which are essential for evaluating the service status of roads and is required to be measured by national weigh-in-motion specifications, such as ASTM E1318-09. Secondly, the transverse distance between the coaxial wheels (track width), and unequal load distributions between left and right coaxial wheels (unbalance) were ignored in the conventional BWIM methods. However, these two factors represented the actual characteristics of the vehicle load applied on the bridge. Hence, ignoring these two factors could lead to significant errors in estimating the vehicle weight when the track width or the unbalance ratio of the test vehicle was different from that of the calibration vehicle. To address the above two problems, this study proposed a new BWIM method, where two concentrated loads with fixed transverse distance were used to represent the coaxial wheel loads of the truck in analyzing the interactions between the truck and the bridge. By doing so, the wheel weights of moving vehicles can be calculated, and both the effects of the unbalance and the track width can be considered in estimating the axle weights of the vehicle. Both experiment results and simulation results demonstrated that the proposed method can effectively identify the wheel weights of moving vehicles, and provide more accurate axle weights than conventional BWIM methods under complex traffic scenarios.