基于递归矩阵奇异熵的损伤识别方法

摘要

对递归图方法在土木工程结构损伤检测中的适用性进行研究,对环境激励的结构响应信号采用多变量融合的方法,并扩展到高维相空间中,研究信号在高维空间中临近点的分布规律和运动特点,采用递归图理论,结合奇异熵理论提出一个新的递归量化指标——递归矩阵的奇异熵,将其作为损伤敏感特征指标,用于结构的损伤识别.数值算例验证了该指标的有效性.%The practicability of using recurrence plot's recognizing performance for structure damage detecting was discussed. Signals, recorded at different locations on the structure, were sampled to reconstruct the phase space and subsequently generate the multivariate recurrence matrix. The multivariate recurrence matrix was then used to construct a novel feature which combines with singular entropy to identify structural damages. Numerical experimentation verifies this feature's validity.