一种基于时间序列与核岭回归的结构损伤定位方法

A structural damage detection method based on time series and kernel ridge regression

摘要:

结构健康监测的一个重要目的是实现结构损伤识别与定位,文章将结构监测数据的时间序列模型与机器学习中的核岭回归相结合,提出了一种新的结构损伤定位方法.先定义结构损伤识别矩阵,推导出结构损伤系数向量与损伤结构和未损伤结构的自回归系数向量差值的关联关系,结构的损伤识别矩阵可以通过机器学习中的核岭回归算法获得.对比其他回归算法,核岭回归的正则化、核函数特性可以大幅提高模型的拟合性能与泛化性能,更好地应用于结构损伤识别.然后通过一混凝土框架数值模型对该方法进行验证.结果表明该方法对结构的单损伤、多损伤均可进行有效识别,准确率较高.

Abstract:

One important purpose of the structural health monitoring is to detect and localize the structural damage. A new structural damage detection and localization method is proposed by combining the time series model of monitoring data with kernel ridge regression. First, a theory was proposed that the difference between the autoregressive coefficient vectors of damaged and undamaged structures is related to the damage coefficient vector, and the relation was included in the structural damage detection matrix. The damage detection matrix of structure can be obtained by the kernel ridge regression algorithm in machine learning. Then, the method was verified by using the numerical model of a concrete frame. The results show that compared with other regression algorithms, the proposed method can greatly improve the fitting and generalization performance of the model, thus it can be better applied to structural damage detection; the method can effectively detect the single damage scenario and multiple damage scenario, and the accuracy is high.