基于遗传算法优化神经网络权值的损伤识别

针对传统BP神经网络存在着容易陷入局部极小点、训练时间太长等缺点,采用基于浮点编码的遗传算法对BP神经网络的初值空间进行遗传优化.用基于浮点编码的遗传算法来优化BP神经网络的权值,得到最佳初始权值矩阵,并按误差前向反馈算法沿负梯度方向搜索进行网络学习.以弹簧质量系统作为算例,用结构的模态频率变化作为网络的输入向量,结构的损伤位置作为输出向量.对网络进行训练,仿真结果表明:遗传BP神经网络的收敛和诊断能力优于传统BP神经网络,可有效运用到结构的模态参数识别中.     

梁裂缝损伤检测的模态应变能法及试验研究

结构裂缝存在将使其局部刚度减少,而局部刚度下降与其固有频率变化紧密相关。应变能对局部损伤较固有频率更为敏感。把结构遭受裂缝损伤的应变能变化量作为损伤指标,提出一种全新的无损裂缝识别方法。并把影响损伤指标因素与假设检验相结合,利用参数假设检验进行识别结果验证。用简支梁模型进行模态试验,将梁损伤前、后的模态参数提取,并用该方法进行裂缝位置及裂缝深度参数的识别。结果表明,该损伤检测方法识别结果精度较高,将其用于裂缝损伤检测是可行的。     

裂纹对空间框架固有频率影响的数值分析

利用有限元方法,对不同位置、不同深度和不同方位裂纹引起的空间框架结构固有频率改变进行了比较分析,将有限元分析得到的梁单元模态应变能力分解为拉压、扭转和弯曲模态应变能,并对结构固有频率改变量随裂纹位置变化的曲线与这些模态应变能分布曲线进行了对比分析。     

基于模态应变能与神经网络的地下拱形结构损伤诊断

地下拱形结构是人防工程的重要结构形式,其损伤诊断是人防工程进行安全性评定的一个重要环节。基于地下拱形结构损伤前后动力特性的分析,证明模态应变能变化对单元损伤的敏感性。定义单元模态应变能变化率,并将其作为人防结构损伤诊断的标识量,通过标识量对损伤进行初步定位,然后采用BP神经网络来确定受损位置的损伤程度。以一典型的人防工程为算例,通过对其在不同损伤情况下计算结果的分析和讨论,说明本文所提出的方法是有效可行的,它可以准确完成地下拱形结构的损伤诊断。     

基于遗传BP神经网络反演楔体参数的研究

为了获得未知楔体的参数,建立了遗传算法和反向传播(Back Propagation,BP)神经网络结合的反演模型。仿真得到不同角度、密度、杨氏模量下楔体导波的频散曲线。采用反对称第一阶模态相速度数据作为遗传BP神经网络反演模型的输入变量;利用遗传算法改进BP神经网络获得优化的初始权值和阈值,并对BP神经网络进行训练;最后将实测的楔体一阶模态相速度代入训练好的网络进行参数反演。结果表明,通过该反演模型可同时反演出楔体的角度、密度、杨氏模量,并且较单一BP神经网络具有收敛速度快、精度高的优点。     

基于缩聚模态应变能与频率的结构损伤识别

在结构损伤识别中,测试振型往往是不完整的,这阻碍了结构损伤识别技术的应用效率。提出了一种采用缩聚应变能与频率相结合的损伤识别定性与定量方法。采用基于Neumann 级数展开的方法对有限元模型进行缩聚,定义了单元缩聚模态应变能,并证明了缩聚模态应变能的变化对损伤的敏感性;将单元缩聚应变能变化率作为标识量来识别损伤的位置;在损伤位置初步判定的基础上,采用特征值灵敏度法求解损伤程度。以一简支钢梁为例证明了所提出的方法。结果表明:在测量模态数据有误差的情况下也能较好地给出损伤识别结果。     

基于斜拉索振动测量与神经网络技术的斜拉桥损伤位置识别方法

以香港汲水门大桥为背景,探讨了斜拉索张力指标和神经网络技术相结合对桥面结构损伤的定位或分类方法。基于高精度三维有限元模型,采用BP网络,模拟了12种可能的损伤情况的定位。以不同损伤程度下斜拉索张力指标作为神经网络的训练与测试输入,由神经网络的输出来指示损伤位置。该方法的突出优点是只用到少量斜拉索的局部模态的基频,就能获得较好的识别结果。而对少量斜拉索的局部模态的基频测量要比其它面向损伤检测的测量容易得多。另外,该方法可十分方便地推广应用于悬索桥的桥面损伤定位。因此,该方法具有重要的实用价值。     

基于模态应变能和小波变换的结构损伤识别研究

针对单一方法对结构同时发生多处不同程度损伤识别的不敏感性缺陷,本文结合小波变换在时域、频域内表征信号局部特性且能够聚焦到信号或函数的任意细节进行处理的能力,提出了一种基于单元模态应变能和小波变换的结构损伤识别方法。在单元模态应变能基础上,利用小波变换系数的变化和分布情况构建单元模态应变能小波变换结构损伤指标,通过对简支梁的数值模拟和斜拉桥模型试验研究的结果与单元模态应变能平均变化率作为损伤识别指标的计算结果进行对比,结果表明该方法能有效确定结构同时发生多处不同程度损伤的位置和估计损伤程度,为实际工程应用奠定了基础。     

利用模态试验参数识别结构损伤的神经网络法

利用结构位移模态试验和应变模态试验参数和神经网络方法对结构损伤定位和定量辨识问题进行了研究。为获得对结构损伤更加敏感的结构损伤识别指标,在分析现有识别指标的基础上,提出了用于神经网络方法的六种基于结构模态试验参数的损伤识别指标,并对它们进行了实例识别和比较研究。它们均能对结构的损伤进行预报,其中应变类型的损伤识别指标对结构损伤的敏感度比位移类型的损伤识别指标高。     

柱壳结构损伤识别方法

摘 要:对柱壳结构的损伤部位和程度进行识别是柱壳结构进行安全性评定的一个重要环节。本文对柱壳结构损伤前后动力特性进行了分析,采用摄动法推导了由于单元损伤引起的各阶模态振型改变系数;推导了模态应变能变化与单元损伤之间的关系;采用单元模态应变能的变化率作为柱壳结构损伤诊断的标识量,证明了单元模态应变化率对损伤的敏感性。对柱壳结构损伤识别方法进行了研究,所提出的方法仅需要量测的低阶模态信息和刚度矩阵就能完成结构的损伤识别。最后,以一混凝土柱壳结构为工程实例,对其在不同损伤情况下的损伤进行了分析,结果表明,所提出的损伤识别方法能够很好地识别不同损伤组合下的损伤部位和程度,同时能较好地识别结构小损伤,因而证明了本文所提出的方法的正确性和有效性。 关键词:柱壳结构;损伤识别;动力特性;损伤标识量;模态应变能     

Free vibration analysis of uniform and stepped cracked beams with circular cross sections

This paper presents a novel numerical technique applicable to analyse the free vibration analysis of uniform and stepped cracked beams with circular cross section. In this approach in which the finite element and component mode synthesis methods are used together, the beam is detached into parts from the crack section. These substructures are joined by using the flexibility matrices taking into account the interaction forces derived by virtue of fracture mechanics theory as the inverse of the compliance matrix found with the appropriate stress intensity factors and strain energy release rate expressions. To reveal the accuracy and effectiveness of the offered method, a number of numerical examples are given for free vibration analysis of beams with transverse non-propagating open cracks. Numerical results showing good agreement with the results of other available studies, address the effects of the location and depth of the cracks on the natural frequencies and mode shapes of the cracked beams. Modal characteristics of a cracked beam can be employed in the crack recognition process. The outcomes of the study verified that presented method is appropriate for the vibration analysis of uniform and stepped cracked beams with circular cross section.     

Modal analysis of multi-cracked beams with circular cross section

This paper proposes a numerical model that combines the finite element and component mode synthesis methods for the modal analysis of beams with circular cross section and containing multiple non-propagating open cracks. The model virtually divides a beam into a number of parts from the crack sections and couples them by flexibility matrices considering the interaction forces that are derived from the fracture mechanics theory. The main feature of the presented approach is that the natural frequencies and mode shapes of a beam with an arbitrary number of cracks and any kind of two end conditions can be conveniently determined with a reasonable computational time. Three numerical examples are given to investigate the effects of location and depth of cracks on the natural frequencies and mode shapes of the beams. Moreover, it is shown through these examples that the evaluation of modal data obtained by the proposed model gives valuable information about the location and size of defects in the beams.     

Ritz analysis of discontinuous beams using local trigonometric functions

The objective of the current paper is to present a Ritz-type analytical model for predicting the behavior of discontinuous beams such as thin-walled beams with cracks and multiply-stepped beams. The beam is discretized in the cracked as well as the un-cracked domains for a cracked thin-walled beam and in uniform beams for a multiple-stepped beam. A set of local trigonometric trial functions is used to define the twist angle for the cracked domain and the un-cracked domains, as well as to define the displacement field for uniform domains. A global equation system of unknown Ritz coefficients is derived by minimizing the Lagrangian functional or the total potential energy. In the present Ritz model, the interface continuity conditions between sub-domains are investigated and enforced into the global equation system using the condensation procedure or the Lagrange multipliers. Examples are presented to illustrate the effectiveness of the current model for free vibration and torsional analysis. Results obtained from the current model are found to agree well with those obtained using a detailed finite element method or with existing results in literature. The proposed model offers an efficient approach to reduce the modeling efforts and computational time required to analyze complex beams with cracks or multiple steps.     

简支梁裂纹位置识别的一种简单方法

由等效线弹簧来模拟裂纹引起的软化效应,基于铁摩辛柯梁理论得到含裂纹简支梁横向振动的频率计算式,由此获得识别裂纹位置的一种近似方法。文中利用梁的二维有限元模态分析数据进行裂纹位置的识别,结果表明该法在较宽的高跨比范围内,有好的效果;裂纹的深度对识别精度影响不大。     

基于摄动法的多条裂纹欧拉梁特征模态分析

基于摄动理论推导了带多条开口裂纹的欧拉梁的特征模态参数的理论计算公式。采用最直接的方式将梁开口裂纹模拟成梁微段内的横截面折减并用δ函数表达了带开口裂纹的梁沿轴线的截面惯矩和线质量等物理参数。基于此,建立了裂纹梁动力微分方程,并采用一阶摄动理论推导得到了梁的模态频率和振型计算公式。简支梁及悬臂梁算例研究表明,该方法具有很好的精度,与有限元模拟结果及实验结果都能很好地吻合。并采用此方法分析了裂纹深度和位置对带多条开口裂纹梁的特征模态参数的影响。结果表明,裂纹对各阶模态频率虽然影响有限,但其引起的各阶频率变化有着明显的模式,可用于结构损伤定位;裂纹对模态振型影响不明显,但对模态曲率影响比较大,可用于结构损伤位置和程度的诊断。     

梁损伤小波包分析和神经网络识别

针对柔性悬臂梁裂缝损伤问题进行损伤位置和损伤程度的识别研究。首先用有限元法建立系统动力学模型。然后对系统的动力响应信号进行小波包分解,建立基于小波包能量谱的损伤指标。把损伤指标作为改进BP神经网络的输入特征参数,用分步识别方法进行损伤位置和损伤程度的识别。最后进行了数值仿真研究。仿真结果表明,利用小波包分析和改进的BP神经网络可以精确地识别出柔性梁的损伤位置和损伤程度。     

Modal parameter identification and vibration based damage detection of a multiple cracked cantilever beam

This paper presents a detailed investigation on the modal parameter identification and vibration based damage detection of a multiple cracked cantilever beam with hollow circular cross-section. To consider multiple crack effects, a cantilever beam including cracks is considered for six damage scenarios. Finite element models are constituted in ANSYS software for numerical solutions. The results are validated by experimental measurements. Ambient vibration tests are performed to extract the dynamic characteristics using Enhanced Frequency Domain Decomposition (EFDD) and Stochastic Subspace Identification (SSI) methods. Calculated and measured natural frequencies and mode shapes for undamaged and damaged beams are compared with each other. Automated model updating is carried out using the modal sensitivity method based on Bayesian parameter estimation to minimize the differences for damage detection. In addition, modal assurance criterion (MAC) and coordinated modal assurance criterion (COMAC) factors are obtained from the mode shapes and two set of measurements to establish the correlation between the measured and calculated values for damage location identification.     

基于谱元法的复合材料裂纹梁Lamb波传播特性研究

基于谱元法提出了一种弹簧元来模拟复合材料梁由于横向裂纹导致的轴弯耦合效应,分析复合材料裂纹梁中Lamb波的传播特性。由断裂力学的相关知识求得弹簧元的刚度,建立复合材料裂纹梁的损伤谱元模型。通过模拟复合材料裂纹梁内Lamb波传播,并和传统的有限元结果进行比较,验证了所提出模型的可行性和有效性。推导了频域内Lamb波各模态的能量计算公式,裂纹处的能量守恒证明了所提出模型的正确性,同时计算表明复合材料梁中裂纹处反射与透射的Lamb波各模态能量随着裂纹深度的变化规律具有单调性,结论可以为定量识别复合材料梁裂纹提供实用依据。     

火箭蒙皮上梁结构的损伤识别方法研究

针对火箭蒙皮上梁结构损伤识别问题,首先利用基于模态应变能的损伤识别指标,对火箭蒙皮上梁结构进行损伤位置识别,再利用基于广义柔度矩阵的损伤敏感函数并结合遗传算法,对火箭蒙皮上梁结构进行损伤程度大小的计算。该损伤识别方法以结构损伤导致模态参数变化为依据,通过对比损伤前后模态应变能变化构建损伤识别指标。在数值仿真中,对T型梁进行了方法验证,发现该方法能够准确识别损伤位置和大小。最后根据火箭蒙皮上T型梁结构特点,设计试验并研究了方法的可行性,发现在T型梁结构上,该方法能够有效地识别损伤位置和大小,误差均在5%以下。     

Identification of crack size via an energy approach

An energy method for identifying the size of a crack at given location in structures using one measured eigencouple of the cracked structures is presented. The method utilizes the maximum strain energies of the structures both with and without a crack and the additional strain energy induced by the crack to construct the energy balance equation from which the size of the crack is evaluated through an iteration procedure. A pair of measured vibration frequency and mode shape of the cracked structure is used in a free vibration analysis to derive for its maximum strain energy. The maximum inertia force of the cracked structure is then applied to the uncracked structure with known stiffness and the resulting strain energy of the uncracked structure is obtained in a finite element analysis. Fracture mechanics is used to derive for the additional strain energy induced by the crack. Experimental investigations of several cracked free-free beams are performed to validate the proposed method. Examples of the identification of crack sizes for a number of damaged beam structures are given to further illustrate the feasibility and effectiveness of the present method. Overall the results are encouraging showing that the present method has the prospect of becoming an alternative approach for crack size detection.