基于支持向量机的钢筋混凝土桥梁损伤识别

为了克服现有方法存在的一些不足,提出基于小波包和支持向量机的混凝土桥梁损伤识别方法。采用小波包对环境振动下的信号进行分解,获得各个频带上的能量,该向量对损伤敏感,可以作为模型识别的输入向量。利用支持向量机强大的分类功能,提出根据频带能量建立支持向量机并进行损伤模式识别的方法。应用该方法对一座三跨连续梁桥进行了损伤识别分析。结果表明经过训练的支持向量机可以较准确地识别出损伤位置和程度。对小波频带能量进行主成分分析后建立的支持向量机会获得更好的识别效果。获得更精确的实际信号特征将进一步提高有限元模型精度和实际应用效果。

Damage Identification of Reinforced Concrete Bridge Based on Support Vector Machine

A novel damage identification method was proposed based on WPD and SVM to overcome some deficiency of existing methods.Wavelet packet decomposition(WPD) was applied to the structural response signals under ambient vibration and the energy spectrum of different frequency bands was obtained.The WPD energy spectrum varies with different structural damage locations and degrees for its sensitivity to the change of structural dynamic characteristics,hence,the spectrum was selected as the feature vectors in damage pattern recognition.Support vector machine(SVM) was used for pattern recognition for the high accuracy and good generalization capability.This method was used for finite element analysis of three-span prestressed concrete bridge.The results from SVM show that the damage patterns could be recognized accurately though the number of trained samples is small.Further more,a better result could be gained if dimension reduction is used for the energy spectrum by principal component analysis(PCA).The application effect could be enhanced once the accurate frequency features of the actual signals are obtained.