模拟桥梁结构故障声发射检测技术研究

发布时间：2018-03-28 09:15

本文选题：模拟桥梁结构　切入点：局部损伤故障识别　出处：《沈阳理工大学》2015年硕士论文

【摘要】：随着桥梁在交通枢纽中的广泛应用,对桥梁实时承载情况进行监测和故障诊断得到广泛关注。桥梁的工作环境通常比较恶劣,在时变载荷作用下,桥梁的内部和外部结构容易产生破损。此外,桥梁分布地域宽广且无专人值守,对其潜在的结构故障进行检测和诊断存在技术困难,因此开展模拟桥梁结构的局部损伤故障检测技术的研究,对提高桥梁建设质量、在役桥梁安全管理都具有现实意义。本文从分析局部结构损伤产生声发射现象的原理入手,阐述几种常见的局部损伤故障声发射信号产生的原因。针对桥梁早期故障信号具有微弱、时变、非平稳等特点,本文提出了利用在时频域具有良好分辨率的小波变换结合具有非线性映射能力的神经网络的故障类型识别方法。通过对连续小波变换及其离散化进行分析,提出可以消除噪声干扰的小波阈值消噪方法,并且利用Matlab进行了仿真验证;针对桥梁各故障状态机理的非线性特性,从模式识别的角度,应用BP神经网络对桥梁各个故障状态进行识别;为了降低BP神经网络结构的复杂性,利用统计分析的方法从经过小波阈值消噪后的信号中提取特征量,作为BP神经网络的输入;设计了基于小波阈值消噪及神经网络分类器的模拟桥梁结构的局部损伤故障类型识别系统。利用声发射信号检测平台对桥梁局部损伤故障发生时产生的声发射信号进行检测,最后通过故障类型识别系统对信号进行分析,比较准确地实现了对其故障类型的识别和分类。
[Abstract]:With the wide application of bridges in transportation hubs, the monitoring and fault diagnosis of bridge real-time loading are paid more and more attention. The working environment of bridges is usually very bad, which is affected by time-varying loads. The internal and external structures of bridges are easily damaged. In addition, there are technical difficulties in detecting and diagnosing the potential structural faults of bridges due to their wide distribution and lack of dedicated personnel. Therefore, it is of practical significance to carry out the research of local damage fault detection technology of simulated bridge structure to improve the quality of bridge construction and the safety management of in-service bridges. This paper begins with the analysis of the principle of acoustic emission phenomenon caused by local structure damage. This paper expounds the causes of acoustic emission signals of several common local damage faults, aiming at the weak, time-varying and non-stationary characteristics of the early fault signals of bridges. In this paper, a fault type identification method based on wavelet transform with good resolution in time-frequency domain and neural network with nonlinear mapping ability is proposed. The continuous wavelet transform and its discretization are analyzed. A wavelet threshold de-noising method which can eliminate noise interference is proposed, and the simulation is carried out by Matlab, and the nonlinear characteristics of each fault state mechanism of bridge are analyzed from the view of pattern recognition. In order to reduce the complexity of BP neural network structure, the statistical analysis method is used to extract the characteristic quantity from the signal after wavelet threshold de-noising as the input of BP neural network. Based on wavelet threshold de-noising and neural network classifier, a local damage fault identification system for bridge structure is designed. Acoustic emission signal detection platform is used to detect the acoustic emission signal generated when the bridge local damage fault occurs. Finally, the signal is analyzed by fault type recognition system, and the fault type recognition and classification are realized accurately.
【学位授予单位】：沈阳理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U446

【参考文献】