【作者】 张纪平；

【导师】 熊诗波；

【作者基本信息】 太原理工大学 ， 机械电子工程， 2010， 博士

【摘要】 随着科学技术的进步、现代工业的发展以及人类生活日益增长的需求,现代机械装备与结构正在向大型化、复杂化与自动化方向发展。机械结构在使用过程中,在承受工作载荷的作用以及各种突发性外在因素影响下可能发生局部损伤。当局部损伤积累到一定程度时,会导致局部或整体结构的失效。对于重要结构,损伤所导致的机械装备与结构的故障停机将造成巨大的经济损失,乃至引发重大的灾难性事故。世界上不乏由于机械损伤导致灾难性事故的实例。对机械结构进行损伤监测与失效预警,将在很大程度上消除事故隐患和避免发生灾难性事故。因此,结构损伤监测与诊断方法的研究具有重要意义。基于振动的结构损伤检测与诊断技术可实现在线损伤(特别是早期损伤)的定量检测与诊断,从而保证在役设备与结构的运行安全,并为近代预知维修提供可靠信息。因此,近年来这种技术得到迅速发展。其中以模型为基础的方法能够实时、敏感地发现由于结构损伤所造成的结构动态特性的本质变化,并有可能对损伤作定量评估,从而受到国内外研究者的广泛关注。在基于结构振动特性的损伤检测方法中,常用和有效的一种方法是利用结构定量模型(如模态模型)的参数检测方法,如利用结构固有频率、振型、模态曲率、应变模态、传递函数、模态置信判据(MAC)、坐标模态置信判据(COMAC)等模态参数的变化进行结构损伤检测。本文采用了基于模态模型参数辨识的一种模型损伤检测与诊断方法,解决了将此方法应用于复杂机械结构损伤检测的一系列问题,提出了基于结构易损伤部位及关键位置预测、结构损伤-动力学响应特征.模态参数变化之间关系预分析的少传感器结构损伤检测与诊断方法。提高了这种方法应用于复杂工程结构的可行性。提出了一种基于实验模态分析数据的有约束的贝叶斯估计有限元模型修改方法,以多段组合的门形框架结构为例,建立了修改后的模态模型。分析了这种框架结构在激振器激励下的动力学响应及引起结构损伤的敏感参数。在此基础上,采用了基于三维动力学仿真的疲劳分析,确定了门形框架结构的易损伤位置。为发生损伤后动力学行为变化的进一步分析提供依据。从理论上研究了损伤对结构模态频率和振型的影响,讨论了基于频率变化监测结构损伤的可行性。以门形框架结构为例,运用有限元方法,计算了疲劳裂纹以及法兰螺栓松动条件下门形框架结构模态频率的变化。用模态实验验证了有损伤门形框架结构的模态频率漂移特征。通过计算与实验结果比较,研究了归一化的模态频率漂移向量的一致性。运用频响函数结构综合方法,分析了组合结构的频响函数。讨论了几种特殊连接情况下,组合结构中的子结构的频响函数。结果表明,子结构损伤或刚度变化对结构动力学特征或频响函数的影响是复杂的,其影响取决于子结构间连接状况。对于刚性连接结构来说,组合结构中子结构的频响函数极点与整体结构的频响函数极点相同。通过子结构的频响函数极点选取,可以反映整体结构的模态频率变化。以基于工作模态分析的在线模态参数辨识为基础,分析了辨识结果的不确定性,改进了辨识算法。提出用可表征损伤特征的主模态对应的互功率谱幅值构建Fisher信息阵。通过逐步消除不敏感传感器直到余下的传感器数目为期望数值为止,得到了传感器优化布置(尽可能少的数量与最佳的配置位置)结果。减少了传感器布置数目、且适合于结构关键部位损伤监测。对有损伤的门形框架结构进行工作模态分析,验证了基于三维仿真与模型参数监测的机械结构局部损伤诊断方法。更多还原

【Abstract】 With the progress of science and technology, modern industrial development, as well as the growing demand for human life, modern machinery and equipment and the structure is developing towards large scale, complication and automation.Mechanical structure will be partially damaged during working process, which is caused by long-term load operation or other unexpected external factors. When the damage accumulates to a certain extent, the overall or local structure will lead to a sudden failure, which will cause emergent stop of equipments. This will bring large economic losses and even cause disastrous accidents. There are a lot of instances in the world. Structural damage detection and early warning is helpful to eliminating hidden dangers and avoiding catastrophic incidents. Therefore, research on structural damage monitoring and diagnosis is of great significance.Structural damage detection and diagnosis technique based on vibration can be used to conduct online damage (especially the early damage) and quantitative detection and diagnosis. With this technique, safe operation of in-service equipment and structure can be ensured, and reliable information can be provided for predictive maintenance. Thus, this technology is developing rapidly in recent years. Among them, model-based approach can be used to find the intrinsic changes of the structural dynamic characteristics real-timely and sensitively. Using this approach, it is possible to achieve quantitative assessment of damage. Structural damage detection and diagnosis technique based on model-based has attracted widespread attention by domestic and foreign researchers.The main technology of structural the damage detection is the diagnosis method based on structural dynamic characteristics parameters, including modal frequency, modal shape, modal curvature, strain mode, transfer function, MAC(Modal Assurance Criterion), COMAC(Coordinate Modal Assurance Criterion) and so on.In this paper, quantitative damage detection and diagnosis method based on modal parameters identification is presented, and problems aroused by the key techniques to apply this method on complicated mechanical structure are solved. The proposed method has following characteristics:prediction of the easily damaged local sections and key positions, analysis on relationship between structure damage and dynamic response features and less use of sensors. With these characteristics, the feasibility to apply this method on complex engineering structures is improved.A constrained Bayesian estimation method is proposed to update finite element model. Using this method, the finite element model of the multi-segment gantry is established. According to this model, the dynamic responses excited by work load and structure damage sensitive parameters are analyzed. On this basis, the easily damaged positions are defined, which provides foundation for dynamic analysis of the damaged structure.The modal frequencies and modal shapes changes of the damaged structure are theoretically studied and feasibility to monitor structure damage according to modal frequency changes is discussed. For a gantry, with finite element method, modal frequency changes are calculated under the condition of the fatigue crack and the flange bolts loose. To verify the calculated results, modal experiments are carried out. By comparison, the experimental results and the calculated results are consistent.The substructure’s FRFs(Frequency Response Functions) are deduced. And for several composite structures with special connection ways, the substructure’s FRFs are discussed. The results show that the impact on the structural dynamics characteristics or FRFs by substructure damage or stiffness change is complex and depends on connection condition of structures. For rigid connection structure, the poles of FRFs of composite structure are the same as the substructures’. The poles of the substructure FRFs can reflect the modal frequencies of composite structure.The inconsistent identification results using two modal parameter identification methods are analyzed. On this basis, the structure damage identification algorithm is improved. A new method to build the Fisher information matrix is developed, which use cross-power spectrum amplitudes of the modal of damage characteristics as its elements. The sensor layout is optimized through gradual elimination of sensor. Thus the number of sensor can be eliminated without effecting the damage detection of critical part. At last, the experiments are conducted, and the results show that the proposed technology is correct.更多还原