基于环境激励的桥梁振动模态识别算法研究∗

摘要

在桥梁建造和维护过程中，需要对桥梁的振动模态进行在线、实时的分析，急需一种不需要人工激励进行快速模态分析的算法。通过研究自然激励技术NExT( Natural Excitation Technique)与自回归滑动平均模型ARMA( Auto￣Regressive and Moving Average Model)，在常规的自然环境模态分析算法的基础上构造出一种快速求解NExT￣ARMA模型的算法进行桥梁模态识别。相比于传统的环境激励模态参数计算方法，该算法不但降低了传统算法的复杂度，而且采用了反馈的方式提高了计算精度。采用ANSYS建立有限元模型并搭建简易实验系统分别对该算法进行仿真验证和实验验证，验证结果表明，该算法能够有效地在自然激励下提取出桥梁结构的各阶模态，其中对前三阶固有频率的识别相对误差降到1％左右。%An online,real￣time analysis of the bridge vibration mode is often required during a bridge construction and maintenance. So a modal analysis algorithm with no human excitation is in urgent need. By researching the mathematical sense of NExT( Natural Excitation Technique) algorithm and ARMA model( Auto￣Regressive and Mov￣ing Average Model),a quick solution for NExT￣ARMA model is proposed on the basis of conventional algorithms. Compared to the traditional ambient modal analysis,this algorithm not only reduces the complexity of traditional method,but also build a feedback loop to improve the accuracy. Build a simply supported beam model both in AN￣SYS and real experimental platform to verify the algorithm. It shows that each mode of the bridge structure can be i￣dentified under nature white noise excitation. The relative error of the first three natural frequencies is 1% or less.