基于解析模态分解的环境温度作用下结构动力特性研究

Due to the effects of environments factors, the structures in long term service are systems with fluctuant dynamic properties. Changing temperature is the most significant environmental effect accounting for the variation in modal parameters, i.e., the temperature effect causes modal parameters’ variation and further make structure damage identification based on vibration data unreliable. Therefore, it is necessary to consider temperature effect in structure health monitoring system. The proposed project attempts to process structure dynamic response signal with analytical model decomposition (AMD) method. Both low frequency temperature response and high frequency dynamic load response will be decomposed from the measured signal. With wavelet transform method; the high frequency dynamic load response of the structure can be used to calculate temporal modal parameters. In the next step, the relationship of obtained time-varying modal parameters and temperature response will be analyzed by regression method, which aims to build a mathematical model that can predict new modal parameters in new situations. The results are reliable references for damage identification. The research is focused on the realization and optimization of the AMD algorithm with reasonable time-varying bisecting frequencies. The aim of this research is to analysis the relationship of the modal parameters and the temperature condition. Finally this method will be verified by modal test and field test data.

服役状态下的工程结构物受环境因素的影响，动力特性具有较大变异性。时变温度是影响结构动力特性最显著的环境因素之一，它造成的结构模态参数变化会使基于振动数据的结构损伤识别可靠性降低，因而有必要在结构健康监测过程中考虑温度效应。本项目拟从信号处理的角度，采用解析模态分解（AMD）理论，将结构动态响应信号分解成慢变的温度响应和快变的动荷载响应两部分。并将滤除了温度响应后的动荷载响应用小波变换方法识别出桥梁的模态参数。随后利用回归分析方法研究温度响应与时变的模态参数间的相关关系，以建立两者间的数学模型，使之能预测新的温度分布条件下的模态参数，为结构损伤识别提供可靠的依据。研究重点在于解析模态分解算法的实现和优化、时变分隔频率的确定、时变模态参数的识别及揭示温度变化对模态参数的影响规律和机理，并通过模型试验及实际运营桥梁监测数据对所提出的算法进行检验。

由于服役状态下的工程结构物受环境因素的影响，动力特性具有较大变异性。项目提出从信号处理的角度，将结构动态响应信号分解成慢变的温度响应和快变的动荷载响应两部分。研究实现了利用解析模式分解算法及基于广义解调和希尔伯特变换的信号分解算法。滤除了温度响应后的动荷载响应可用于识别出桥梁的模态参数。研究重点在于解析模态分解算法的实现和优化、时变分隔频率的确定、时变模态参数的识别及揭示温度变化对模态参数的影响规律和机理，并通过模型试验及实际运营桥梁监测数据对所提出的算法进行检验。在研究过程中，提出了信号瞬时频率精确计算方法，并发展了一套利用峰值密度聚类算法实现盲源信号分离进而识别模态振型的方法。项目成果达到预期要求，发表或接收SCI收录期刊论文2篇，Ei收录期刊论文2篇，培养研究生5名。