基于最小二乘支持向量机的桥梁应变丢失数据重构

Lost Strain Data Reconstruction Based on Least Squares Support Vector Machine

针对桥梁工程结构健康监测系统中,测量信号的物理量类型不同,提出了一种新型的信号关联度计算方法,用来分析与传感器丢失数据相关联的信号,由最小二乘支持向量机分别对丢失信号数据实现同类型变量和不同类变量重构,以恢复丢失数据。桥梁监测数据的试验仿真分析表明,所提出的关联度计算法和重构方法是正确、有效的,且在实际监测系统中具有很强的参考和应用价值。

As a deferent type of sensors in a structural health monitoring system,a novel algorithm is proposed to calculate the correlativity between two sensors.According to the correlativity,the sensors signals is used to reconstruct lost sampling datum of others sensors by the least square support vector machine in the two cases of the same type sensors signals and the different type sensors signals.The proposed data reconstruction algorithm is tested in the case of the bridge structural health monitoring system,and the simulation results indicate the proposed algorithm is correct and efficient,and has a potential for reference and application.