基于LS-SVM的桥梁挠度监测中温度效应分离  

Study on Separation of Bridge Deflection Temperature Effect Based on LS-SVM

作　　者：刘夏平[1] 杨红[2] 孙卓[1] 何清平[2] 王燕萍[2] 

机构地区：[1]广州大学土木工程学院,广东广州510006 [2]广州大学物理与电子工程学院,广东广州510006

出　　处：《铁道学报》2012年第10期91-96,共6页Journal of the China Railway Society

基　　金：国家自然科学基金项目(51078093);广东省科技计划项目(2011B010300026)

摘　　要：针对桥梁挠度监测中温度效应的分离问题,根据桥梁挠度的组分特性,利用最小二乘支持向量机(leastsquare support vector machines,LS-SVM)模型,提出桥梁挠度监测中精确分离温度效应的方法。温度和挠度温度效应具有确定的非线性关系。应用LS-SVM的逼近能力,将温度作为输入、挠度温度效应作为输出,建立温度和挠度温度效应的精确定量关系,在已知温度变化的情况下获取挠度温度效应的变化值,从而实现挠度温度效应准确分离。计算及分析结果表明,该方法能有效分离挠度监测信号中的温度效应,结合滤波方法,亦可有效分离出长期挠度,为结构的损伤识别提供了有效的数学手段和依据。Inview of the problem of separating temperature effects in bridge deflection monitoring,the method to accurately separate the temperature effects in bridge deflection monitoring was presented according to the constitutive properties of the bridge deflection and by use of the least square support vector machine（LS-SVM） model.The temperature and the deflection temperature effect were seen to be of the definite nonlinear relationship.Utilizing the LS-SVM approximation capability and taking the temperature as the input and the deflection temperature effect as the output,the precise quantitative relationship between the temperature and deflection temperature effect was established.In the case that the temperature change was known,the change of the deflection temperature effect can be determined and the deflection temperature effect can be accurately separated.Experimental and analytical results show that the method can effectively separate the temperature effect in deflection monitoring signal and also the long-term deflection by combing with filtering,thus providing the numerical basis for identification of structural damages.

关 键 词：最小二乘支持向量机 温度 挠度 温度效应 分离 

分 类 号：U441[建筑科学—桥梁与隧道工程]