基于迭代加权最小二乘的桥梁动态称重理论与试验研究

A bridge weigh-in-motion system based on iteratively reweighted least squares

目前商用桥梁动态称重系统(weigh-in-motion systems,BWIM)大多基于Moses算法,虽然能高效快速地识别行驶于桥梁的车辆轴重,但轴重识别精度偏低。为解决这一问题,提出基于迭代加权最小二乘的桥梁动态称重(iteratively reweighted least squares,IRLS)算法。与Moses算法不同,IRLS算法考虑了荷载响应中存在的多种不确定性因素,为每个荷载响应值提供一个合适的权重系数,区分不同荷载响应对轴重识别的贡献度。首先,将迭代加权最小二乘引入桥梁动态称重,推导出相应的轴重识别计算公式;然后,通过车桥数值仿真模型,分别用IRLS算法和Moses算法识别轴重,对比分析两种算法的精度及影响因素;最后,基于怀化舞水五桥引桥的车桥动力试验,进一步验证IRLS算法用于桥梁动态称重的有效性和准确性。结果表明,IRLS算法能较合理地分配不同荷载响应对轴重识别的贡献度,在一定程度上提高车辆轴重识别的精度。

At present,commercial bridge weigh-in-motion systems(BWIM)are generally based on the Moses algorithm.Although they can efficiently and quickly identify the axle weights of vehicles driving by bridges,the accuracy is low.To solve this problem,a novel algorithm that finds axle weights using the bridge response subjected to the drive-by vehicles was presented.Unlike the Moses’algorithm assuming that all the bridge responses have the same uncertainties,the proposed algorithm based on iteratively reweighted least squares(IRLS)considers the unequal uncertainties and can assign each observed response data proper amount of influence over the axle weight estimates.Firstly,the formula of axle weights identification was derived using IRLS;then,numerical simulations were conducted to verify the IRLS algorithm using a simply supported beam-vehicle interaction model;finally,based on afield test of Wushui Fifth Bridge approach bridge,the axle weight identification of the Moses algorithm and the IRLS algorithm were compared and analyzed.Results show that the IRLS algorithm reasonably allocates the contribution of different load responses to axle load identification and it can get more accurate axle weights than the Moses’algorithm.