【作者】 项长生；

【导师】 李慧； 杜永峰；

【作者基本信息】 兰州理工大学 ， 结构工程， 2021， 博士

【摘要】 结构性态识别与损伤诊断作为结构健康监测的重要手段,离不开先进的结构损伤识别方法,本文在对动力指纹类指标、信号处理技术、人工智能算法和信息融合的结构损伤识别方法做了详细的介绍,在此基础上将信息熵理论进入到结构的损伤识别领域中,对混凝土桥梁结构的局部损伤、裂缝损伤和铰缝损伤进行定位和定量分析,为实际工程中混凝土桥梁结构的健康监测与评估提供理论与技术上的指导,本文主要工作有:1)改进传统的铰接板铰缝受力模型,考虑横向不均匀传力情况,对铰缝的受力和变形进行简化,建立了基于弹簧-铰接杆体系的铰接板桥结构计算模型,并基于该模型提出了一种基于铰缝损伤度的铰缝损伤识别方法。通过数值模拟分析和桥梁静载试验分析对该方法的正确性和实用性进行了验证分析。2)利用应变能对损伤的高灵敏度和强抗噪声特性,提出了一种用于铰缝和铰接板局部损伤识别的应变能信息熵指数。并根据铰节点单元及其周围单元的应变能,构造了新的局部应变能熵函数损伤指标。采用简支梁模型试验验证了所提出指标的有效性。3)基于信息熵理论,提出以模态曲率效用信息熵为指标的结构损伤识别方法,该方法通过计算互重变化矩阵和权重-概率系数,综合反映结构的损伤位置和损伤程度。分别采用简支梁算例和简支梁试验对指标进行验证,结果表明:模态曲率效用信息熵充分利用了模态曲率对损伤位置较敏感的优势,且能有效地克服其不足之处,能够准确地定位结构的损伤位置,定性反应损伤程度且具有良好的抗噪性,为梁结构的损伤识别提供了有效的理论借鉴。4)结合广义柔度矩阵对低阶损伤模态的敏感性与信息熵对系统非线性显著的突显作用,提出了广义柔度曲率信息熵指标。利用结构广义柔度曲率矩阵的对角向量元素与对角元素和之的比值作为整体概率函数,来构造熵函数指标用于结构损伤识别。采用简支梁和连续梁两种不同形式的梁结构算例,对所提指标的损伤识别能力进行了分析验证,结果表明,所提出的指标可以有效地识别出结构单点损伤、对称位置损伤以及多点损伤的位置及其程度。5)基于连续小波变换的奇异性检测方法和BP神经网络的并行处理与非线性衍射能力,提出联合连续小波变换和神经网络数据融合处理位移响应的损伤识别方法,该方法可以实现损伤精确定位,且数据融合后损伤定量结果更精确。6)将基于位移的广义局部信息熵的概念进行引申,“广义”采用点i的曲率模态来表示该点具备的信息量,以“局部”反映该点具有的信息量占附近节点信息量的比重。构造计算点i的广义局部曲率模态信息熵的局部概率p_i,推导了广义局部曲率模态信息熵指标,利用该指标并结合BP神经网络训练对损伤位置和程度进行了判别。更多还原

【Abstract】 Structural condition recognition and damage detection as an important means of structural health monitoring depend on the advanced method of structural damage identification.In this article,the dynamic fingerprint(natural frequency and modal vibration mode,the modal curvature,flexibility matrix and modal strain energy),signal processing technology(the wavelet transform and Hilbert-Huang transform),artificial intelligence algorithm and the information fusion of structure damage identificati on method are introduced in detail.Then,the information entropy theory is introduced into the field of structural damage identification,in positioning and quantitative analysis of the partial damage of concrete bridge structure,fracture and damage of t he hinge joints,to provide theoretical and technical guidance of the health of the concrete bridge structure in practical project monitoring and evaluation.This article main research contents are as follows:1)Improving the traditional model of hinge jo int of hinged slab,by considering the transverse uneven force,and simplification of the stress and deformation of the hinge joints,a hinged slab bridge structure calculation model is established based on the spring-hinged rod system,then based on this model,a hinge joints damage identification method is deduced.Through numerical simulation analysis and bridge static test,the correctness and practicability of the analysis of this method are verified.2)By applying the high sensitivity and strong anti-noise characteristics of strain energy to damage,a strain energy information entropy index is proposed for identifying of local damage in hinge joints and hinged slabs.Moreover,a new damage index of the local strain energy entropy function is construct ed according to the strain energy of a hinge joint element and its surrounding elements.Lastly,the effectiveness of the proposed indicators is verifyied by the simply supported beam model test.3)Based on the information entropy theory,a modal curvature utility information entropy index is proposed,the index comprehensively reflects the damage state of the structure by calculating mutual retransformation matrix and the weight-probability coefficient.The simply supported beam example simulation and model test are established to verify the index,t he results show that the modal curvature utility information entropy index can take advantage of the merits of the modal cu rvature index which is sensitive to damage and can overco me its shortcomings effectively,it can identify the damage location and damage degree accurately of structure and has certain noise immunity,which provides an effective damage identification method for beam structures.4)Based on the merit of generalized flexibility matrix,which can be obtained more accurately from low-order modal data and the truncation error caused by truncating higher order modes is reduced.a new index named generalized flexib ility curvature information entropy is proposed,which taking the ratio of each diagonal element to the sum of the total diagonal elements in the generalized flexibility curvature matrix as the probability function.The damage identification capability of the proposed index is verified by using two different types of beam structures:simply supported beam and continuous beam.The results show that the generalized flexibility curvature information entropy index can effectively identify the damage at single,symmetric and multiple locations of beam structures.5)Based on the singularity detection theory of continuous wavelet transform and the parallel processing and non-linear diffraction capabilities of BP neural network,a damage identification method is proposed by combining continuous wavelet transform and neural network data fusion processing for displacement response.The results show that the indicators and methods used in this paper can accurately locate the damage of bridge structures,and the quantitative results of damage after data fusion are more accurate.6)Extending the concept of the displacement of generalized partial information entropy,the"generalized"means applying the curvature modal of point i as its amount of information,the"local"means proportion of the information of point i to the amount of information of nearby nodes.The local probability p_i of generalized local curvature modal information entropy is constructed,to deduce the index of generalized local curvature modal information entropy,then to identify the damage location and degree combined with BP neural network training work.更多还原