基于支持向量机的桥梁应变噪声数据重构

Noise strain reconstruction based on support vector machine in bridge

针对桥梁结构健康监测系统中应变数据存在噪声干扰现象,提出一种邓氏灰度关联度计算方法,分析与计算传感器噪声数据之间的关联度,选择关联度大的信号作为支持向量机的输入向量,实现基于支持向量机的噪声数据重构,以恢复测量数据,并以均方根和频域功率谱来评估重构数据的效果.由桥梁监测数据的试验仿真分析表明,所提出关联度计算和基于支持向量机的数据重构方法是正确、有效的.

In the case of strain sample spoiled by noise in structural health monitoring system for the bridge,a novel algorithm is proposed to calculate the correlation between two sensors by Dengs＇ degree of gray incidence,and select the sensor signal of larger corrleation degree as input vector for support vector machine to reconstruct the other sensor signal,which is spoiled by noises.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 with the index for mean square error and power spectrum in frequency domain.