基于多种小波基的桥梁损伤信号分析方法研究

发布时间：2018-01-22 16:53

本文关键词： 小波变换去噪固有频率小波基函数　出处：《吉林建筑大学》2017年硕士论文　论文类型：学位论文

【摘要】：桥梁损伤的无损检测技术是近几年桥梁研究的热点问题,通过振动信号分析识别结构的模态参数是进行无损检测的基础,因此,如何准确的进行结构模态参数识别也吸引了广大学者的关注。传统的傅里叶变换是在频域内对信号进行分析,得到连续的频谱函数,分析精度高,但仍然存在去噪效果不明显的缺点,特别是对于短时高强的噪声,傅里叶变换将其能量平均到整个频谱,对频率的识别造成了干扰。基于小波变换的频率识别是一种时频域的识别方法,其时间定位能力和多分辨率特性使小波分析得到了广泛的应用,但是由于小波基的选择问题制约了小波变换在实际应用中的推广。本文的研究内容主要包括:第一,利用不同的小波基对信号进行去噪处理并分析各小波基的去噪效果,确定对振动信号去噪效果最好的小波基,并利用傅里叶变换的频谱图对去噪后的信号进行频率识别。第二,介绍小波时频图识别桥梁各阶频率的原理,采用不用的小波基对实测信号进行小波时频图分析,识别桥梁各阶固有频率,并比较有限元解、傅里叶频谱图识别结果与小波时频图识别结果。简而言之,本文是利用小波分析的去噪能力和时频分析能力,有效去除信号中各种噪声对固有频率识别的影响,使测得的频率更加准确可信,为之后的损伤识别定位打下基础。
[Abstract]:Non-destructive detection of bridge damage is a hot issue in recent years. Modal parameters identification by vibration signal analysis is the basis of nondestructive testing. How to accurately identify the structural modal parameters has attracted the attention of many scholars. The traditional Fourier transform is to analyze the signal in the frequency domain to obtain a continuous spectrum function, the analysis accuracy is high. However, there is still a disadvantage that the denoising effect is not obvious, especially for the short-time high-strength noise, the Fourier transform equalizes its energy to the whole spectrum. Frequency recognition based on wavelet transform is a time-frequency domain recognition method, its time location ability and multi-resolution characteristics make wavelet analysis widely used. However, the selection of wavelet bases restricts the extension of wavelet transform in practical applications. The main research contents of this paper are as follows: first. Different wavelet bases are used to Denoise the signal and analyze the denoising effect of each wavelet basis to determine the best denoising effect for the vibration signal. The frequency of the de-noised signal is recognized by Fourier transform spectrum. Secondly, the principle of wavelet time-frequency map recognition of bridge frequency is introduced. The wavelet time-frequency analysis of the measured signal is carried out by using the unneeded wavelet basis to identify the natural frequencies of each order of the bridge, and to compare the finite element solution, the Fourier spectrum recognition result and the wavelet time-frequency pattern recognition result. In this paper, the denoising ability of wavelet analysis and the ability of time-frequency analysis are used to effectively remove the influence of all kinds of noise in the signal on the recognition of natural frequency, so that the measured frequency is more accurate and reliable. It will lay a foundation for the later damage identification and localization.
【学位授予单位】：吉林建筑大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U446

【相似文献】