학술논문
考虑异常监测数据影响的桥梁拉索振动频率识别方法研究 / Vibration frequency identification method of bridge cable considering abnormal monitoring data
Document Type
Academic Journal
Author
Source
中南大学学报（自然科学版） / Journal of Central South University（Science and Technology). 54(12):4870-4881
Subject
Language
Chinese
ISSN
1672-7207
Abstract
针对桥梁健康监测系统中包含大量异常监测数据的现象,提出考虑异常监测数据影响的桥梁拉索振动频率识别方法.首先,根据正常监测数据的功率谱密度函数的分布特征,确定拉索振动频率的近似频带区间,进而采用峰值拾取法自动化获取近似频带区间内的拉索振动频率初始识别值.其次,利用前三阶频率建立三维空间密度聚类模型,进而采用聚类模型检测并剔除频率初始识别值中的异常值.利用外滩大桥的拉索加速度监测数据对所提方法进行验证.分析不同类型异常监测数据对拉索频率识别值的影响,考察不同维密度聚类模型对频率异常识别值的检测准确率.研究结果表明:异常监测数据严重干扰了拉索振动频率的准确识别;三维空间密度聚类模型对拉索振动频率异常识别值的检测准确率达到了98%以上,且剔除异常识别值后的拉索频率与环境温度呈现合理的相关性.
Aiming at the phenomenon that the bridge health monitoring systems contain a large number of abnormal monitoring data,an identification method for vibration frequencies of bridge cables with the influence of abnormal monitoring data was proposed.Firstly,the approximate band interval of each vibration frequency of the cables was determined according to the distribution characteristics of power spectral density function of normal monitoring data.Peak picking method was adopted to extract the initial identified results of the vibration frequencies of bridge cables automatically in the approximate band intervals.Secondly,a three-dimensional spatial density clustering model was established based on the first three order frequencies.Then,the abnormal values were detected and eliminated from the initial identified results of the vibration frequencies by using the clustering model.The proposed method was verified by using the cable acceleration monitoring data of the Waitan Bridge.The influence of different abnormal monitoring data on identification results of cable frequencies was analyzed.And the detection accuracy of density clustering models with different dimensions on abnormal values of the identified cable frequencies was also investigated.The results show that frequency identification of bridge cables is seriously interfered by abnormal monitoring data.The detection accuracy of three-dimensional spatial density clustering model on the abnormal identified frequencies is more than 98%.In addition,the cable frequency after removing the abnormal identification values shows a reasonable correlation with the ambient temperature.
Aiming at the phenomenon that the bridge health monitoring systems contain a large number of abnormal monitoring data,an identification method for vibration frequencies of bridge cables with the influence of abnormal monitoring data was proposed.Firstly,the approximate band interval of each vibration frequency of the cables was determined according to the distribution characteristics of power spectral density function of normal monitoring data.Peak picking method was adopted to extract the initial identified results of the vibration frequencies of bridge cables automatically in the approximate band intervals.Secondly,a three-dimensional spatial density clustering model was established based on the first three order frequencies.Then,the abnormal values were detected and eliminated from the initial identified results of the vibration frequencies by using the clustering model.The proposed method was verified by using the cable acceleration monitoring data of the Waitan Bridge.The influence of different abnormal monitoring data on identification results of cable frequencies was analyzed.And the detection accuracy of density clustering models with different dimensions on abnormal values of the identified cable frequencies was also investigated.The results show that frequency identification of bridge cables is seriously interfered by abnormal monitoring data.The detection accuracy of three-dimensional spatial density clustering model on the abnormal identified frequencies is more than 98%.In addition,the cable frequency after removing the abnormal identification values shows a reasonable correlation with the ambient temperature.