基于局部离群因子的装配式梁横向联系病害监测分析

Monitoring and Analysis of Deteriorations in Transverse Connections of Precast Segmental Girders Based on Local Outlier Factors

为快速筛查大规模装配式梁的横向联系病害,以某大型区域性桥梁集群监测系统中的32座装配式梁桥为背景,提出了一种基于局部离群因子的装配式梁横向联系病害识别方法。该方法通过对车致应变监测数据进行观测周期和统计周期的切分,提取同一截面测点的应变监测数据最大值集合及其二层统计值(最大值和中位数)集合,再遍历集合计算车致应变二层统计值的无量纲指标——局部离群因子,通过对比各单梁的局部离群因子与单位1的偏离程度分析该点位是否为离群点,实现病害的定位和量化。实桥检测验证了该方法对装配式梁横向联系病害快速识别的可靠性。基于该方法开发了大规模城市装配式梁集群横向联系病害自动分析程序,对监测系统装配式梁桥进行分析,结果显示该系统中有6座桥梁横向联系异常,横向联系的薄弱位置均为边梁和次边梁。

This study presents a method based on local outlier factors that can rapidly identify deteriorations in the transverse connections of the large-scale precast segmental girders.The precast segmental girders involved in a large-scale monitoring system of 32 bridges in a particular region were used as cases.By using this method,the monitored vehicle-induced strain data are categorized in accordance with the observing periods and statistical periods,the maximum set of monitored strain data as well as its two-layer statistical set(maximum and median)from measurement points in the same cross-section are extracted.Subsequently,dataset traversal was carried out to calculate the local outlier factors of dimensionless indexes of the two-layer statistical data of vehicle-induced strains.The local outlier factors of individual girders are compared to check the deviation degrees against 1 and analyze if the measurement point is an outlier point,and at last,achieving the positioning and quantification of the deteriorations.The application of the method in a real bridge verified the reliability of the presented method in rapidly identifying the deteriorations in the transverse connections of the precast segmental girders.On the basis of this method,a deterioration automatic analysis procedure for the transverse connections in the large-scale bridge clusters is developed,to analyze the girders with monitoring system.Six bridges in the system revealed abnormities,and the vulnerable locations were in the side and secondary side spans.