基于EEMD和SVR的单自由度结构状态趋势预测  

Trend prediction for a single-degree of freedom structure's state based on EEMD and SVR

作　　者：刘义艳[1] 贺栓海[2] 巨永锋[1] 段晨东[1] 

机构地区：[1]长安大学电子与控制工程学院,西安710064 [2]长安大学公路学院,西安710064

出　　处：《振动与冲击》2012年第5期60-64,共5页Journal of Vibration and Shock

基　　金：国家科技支撑计划项目:高大空间建筑工程安装维护设备技术与产业化开发(2008BAJ09B06);中国博士后基金:斜拉桥损伤识别和健康状态预测技术研究(20110491637)

摘　　要：为了解决结构早期损伤难以正确识别的问题,结合聚类经验模式分解(EEMD)解决随机不确定性问题和支持向量机(SVM)解决预测问题这两者的优势,提出了一种基于EEMD特征提取的支持向量机回归(SVR)结构状态趋势预测方法。先对单自由度结构渐进损伤的加速度振动信号进行EEMD,再进行希尔伯特变换(HT),计算瞬时频率,然后用回归支持向量机对反映结构健康状态的瞬时频率进行趋势预测。研究表明:对于渐变损伤该方法可以准确地、高精度地预测结构状态趋势。In order to solve a difficult problem to correctly identify early damages of a structure, a trend prediction model of support vector machine （SVM） based on ensemble empirical mode decomposition （EEMD） was proposed. In the model, EEMD processing stochastic and uncertain signals and SVM regression （SVR） solving small-sample pattern recognition problems were integrated. Firstly, the acceleration vibration signals of a single degree of freedom structure model were processed by using EEMD, the intrinsic mode functions （IMFs） containing structural damage information were selected; then, the selected IMFs were transformed by using Hilbert transformation （HT） and instantaneous frequencies （IFs） were calculated; finally, the trend prediction based on SVR of IFs was realized. The prediction of structural engineering simulation data showed that the proposed method can predict trends of structure conditions correctly and precisely.

关 键 词：聚类经验模式分解 支持向量机回归 单自由度结构 瞬时频率 趋势预测 

分 类 号：TN911.72[电子电信—通信与信息系统] TU311.3[电子电信—信息与通信工程]