基于曲率模态小波神经网络的框架结构损伤识别研究

发布时间：2019-05-23 19:55

【摘要】：框架结构形式在工程中被广泛采用,由于结构在使用过程中往往存在初始损伤或因使用年限的增加,外荷载频繁作用使结构损伤的积累,可能导致结构抗力的减弱,当结构的某一部分出现损伤时可能会致使结构的其它部分甚至整个结构出现破坏,所以研究框架结构的稳定性与安全性不仅可以防止社会财富的损失,更重要的是能够保障人民的生命安全。因此,监测结构的工作状况,研究框架结构的损伤诊断方法,有着十分重要的工程及现实意义。小波变换能够分析信号在时、频两域的局部特征,神经网络具有很好的自组织能力、很强的非线性映射能力,通过将小波分析与神经网络相结合的方法实现对框架结构的损伤位置和损伤程度的识别,运用的基本方法是：通过建立损伤框架结构的有限元模型并对其动力特征进行分析,将得到的曲率模态进行连续小波变换可以得到结构的小波系数,由小波系数模极大值确定损伤的位置。以损伤后结构的固有频率作为神经网络输入参数构造神经网络,从而实现对框架结构损伤程度的识别。本文以框架结构为研究对象,建立了框架结构的有限元模型(一层一跨多处损伤的框架结构、一层两跨多处损伤的框架结构),运用小波分析原理,采用Lanczos法得到框架结构的曲率模态,对其曲率模态进行连续小波变换可以得到结构的小波系数,由小波系数模极大值确定损伤的位置。以损伤后结构的固有频率作为神经网络输入参数构造神经网络,从而实现对框架结构损伤程度的识别,建立了一种基于小波神经网络算法的框架结构损伤识别方法。本文在简单的框架结构基础上,将上述方法应用到较复杂的两层一跨、两层两跨多处损伤框架结构上,建立了框架结构的有限元模型,由小波系数模极大值确定损伤的位置,由构造的神经网络来确定损伤的程度,验证了方法的有效性。本文提出的方法可供结构损伤诊断的工程应用参考。
[Abstract]:Frame structure is widely used in engineering. Because of the initial damage or the increase of service life of the structure, the accumulation of structural damage is caused by the frequent action of external load, which may lead to the weakening of structural resistance. When a part of the structure is damaged, it may cause damage to other parts of the structure or even the whole structure, so the study of the stability and safety of the frame structure can not only prevent the loss of social wealth. More importantly, it can ensure the safety of people's lives. Therefore, it is of great engineering and practical significance to monitor the working condition of the structure and study the damage diagnosis method of the frame structure. Wavelet transform can analyze the local characteristics of signal in time and frequency domains. Neural network has good self-organization ability and strong nonlinear mapping ability. By combining wavelet analysis with neural network to identify the damage location and degree of frame structure, the basic method is to establish the finite element model of damaged frame structure and analyze its dynamic characteristics. The wavelet coefficients of the structure can be obtained by continuous wavelet transform of the curvature modes, and the damage location can be determined by the modulus Maxima of wavelet coefficients. The natural frequency of the damaged structure is used as the input parameter of the neural network to construct the neural network, so as to realize the identification of the damage degree of the frame structure. In this paper, taking the frame structure as the research object, the finite element model of the frame structure (one story, one span and multiple damage frame structure, one layer, two span multiple damage frame structure) is established, and the wavelet analysis principle is used. The curvature mode of frame structure is obtained by Lanczos method. The wavelet coefficient of the structure can be obtained by continuous wavelet transform of curvature mode, and the damage position can be determined by the modulus Maxima of wavelet coefficient. The natural frequency of the damaged structure is used as the input parameter of the neural network to construct the neural network, so as to realize the identification of the damage degree of the frame structure, and a damage identification method of the frame structure based on wavelet neural network algorithm is established. In this paper, on the basis of simple frame structure, the above method is applied to the complex two-story one-span, two-story two-span multi-damage frame structure, and the finite element model of the frame structure is established, and the damage location is determined by the modulus Maxima of wavelet coefficients. The degree of damage is determined by the constructed neural network, and the effectiveness of the method is verified. The method proposed in this paper can be used as a reference for the engineering application of structural damage diagnosis.
【学位授予单位】：长沙理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TU317
，

本文编号：2484184