结构损伤诊断与系统时域辨识研究综述

Review of Structural Damage Detecting Based on Dynamics System Identification in Time Domain

对结构时域辨识理论及其应用进行系统研究和综述。研究表明:环境激励的非平稳与不确定性、响应信息的不完备性、反问题的不适定性以及测试噪声等是影响系统时域辨识理论在结构损伤诊断中应用的关键因素。以影响因素为主线,系统阐述结构时域辨识理论、方法的发展和研究现状,并对其优、缺点进行比较。分析指出,基于系统识别的结构损伤诊断技术适合于从整体层次上把握结构的健康状况,只有将其与局部损伤检测方法结合才有可能对大型复杂结构的健康状况作出正确的评估。展望结构时域辨识技术的发展趋势认为:随机子空间法等先进的系统辨识方法、EMD非平稳信号处理技术以及神经网络智能算法将成为未来研究和发展的主流。

A review of structural damage detecting based on dynamics system identification in time domain is given. Research shows that non-stationary uncertain ambient excitation, incomplete response information, ill-condition of inverse problem and testing noise are the key factors which impact the application of theory of system identification on structural damage diagnosis. According to the main line of affecting factors, the development and status of theories on structural identification in time domain is introduced sys- tematically, and their merits and defects are analyzed and compared. Finally, it is pointed out that the technology of structural damage detecting based on dynamics system identification applies to the health diagnosis at the overall structure level, and the local damage detecting methods are more efficient in identifying the damage of component parts of structure. If the two types of identification methods are combined, it is possible to evaluate rightly the health state of structure. The future development trend on structural iden- tification in time domain is prospected, and Stochastic Subspace Identification, Empirical Mode Decomposition and Neural Networks are expected to become the mainstream.