Structural Health Monitoring with model based damage detection using nonlinear model adaption and Artificial Intelligence methods

使用非线性模型自适应和人工智能方法进行基于模型的损伤检测的结构健康监测

• 批准号: 501496870
• 负责人: Professorin Dr.-Ing. Martina Schnellenbach-Held
• 金额: --
• 依托单位: Institut für Massivbau
• 依托单位国家: 德国
• 项目类别: Priority Programmes
• 资助国家: 德国
• 项目状态: 未结题
• 来源: https://gepris.dfg.de/gepris/projekt/501496870?language=en
• 关键词: Structural; Health; Monitoring; model; based

The evaluation of the condition of existing structures usually comprises visual inspections and assessments by engineers as well as supplemented local test procedures if damage is suspected. This involves a high dependency on the expert’s knowledge and judgement. In contrast, automated monitoring systems enable a continuous objective condition recording and thus an early detection and evaluation of damage as well as a continuous documentation of the aging process. On that basis, a purposive condition-related maintenance is facilitated instead of a conventional cost-intensive servicing strategy. Comprehensive concepts for an automated permanent monitoring provide an identification of loads and structural conditions and thus enable a condition prognosis. Particularly for highly stressed solid structures, significant challenges have to be met. For this purpose, the use of artificial intelligence methods in a model-based permanent monitoring is a promising approach. Aim of this research project is the development of a closed approach for an automated damage diagnosis as part of a continuous condition monitoring of open-air massive constructions. The approach comprises the identification of systems and loads based on a realistic adaptation of numerical calculation models on statically damage-sensitive measurement quantities. On that basis, objective statements on the building condition are allowed that include reliable information about the location and the extent of detected damages. For model adaptation, an optimization method is applied that is based on the use of nonlinear calculations combined with discrete structurally characteristic values. Identified systems represent the structural condition at the respective measurement time. The damage diagnosis is based on the comparison of identified systems at different measurement times: Changes of the structural characteristic values enable the determination of damages together with the location and extent. For solving the highly complex optimization tasks, evolutionary algorithms are applied; Cluster analysis methods are used to evaluate the reliability of the optimization task solutions. The elaborated methods are prototypically implemented for the assessment of bridge structures. Bridges are highly important infrastructural objects that are exposed to steadily increasing traffic load demands. Additionally, they represent a high investment volume and frequently exhibit damages due to their age composition. Model-based investigations are intended to verify the performance capacity of the approach. Through conjunction of innovative information science techniques and modern structural analysis methods as well as their assignment to realistic engineering problems, novel solution approaches are rendered possible for efficient structural monitoring systems and efficient maintenance.

对现有结构状况的评估通常包括工程师的目视检查和评估，以及在怀疑损坏时补充的当地测试程序。这高度依赖专家的知识和判断。相比之下，自动监测系统可以连续记录客观状况，从而尽早检测和评估损坏，并持续记录老化过程。在此基础上，有目的地进行与状况相关的维护，而不是传统的成本密集型维护策略。自动永久监测的综合概念可识别载荷和结构状况，从而实现状况预测。特别是对于高应力的固体结构，必须应对重大挑战。为此，在基于模型的永久监控中使用人工智能方法是一种有前途的方法。该研究项目的目的是开发一种封闭的自动损伤诊断方法，作为露天大型建筑连续状态监测的一部分。该方法包括基于对静态损伤敏感测量量的数值计算模型的实际适应来识别系统和负载。在此基础上，允许对建筑物状况进行客观陈述，其中包括有关位置和检测到的损坏程度的可靠信息。对于模型自适应，应用了基于非线性计算与离散结构特征值相结合的优化方法。识别出的系统代表了相应测量时间的结构状况。损伤诊断基于不同测量时间已识别系统的比较：结构特征值的变化可以确定损伤以及位置和程度。为了解决高度复杂的优化任务，应用进化算法；采用聚类分析方法评估优化任务解的可靠性。所阐述的方法典型地用于桥梁结构的评估。桥梁是非常重要的基础设施，面临着不断增加的交通负荷需求。此外，它们代表着高投资额，并且由于其年龄构成而经常出现损害。基于模型的调查旨在验证该方法的性能能力。通过创新的信息科学技术和现代结构分析方法的结合以及它们对现实工程问题的分配，为高效的结构监测系统和高效的维护提供了新颖的解决方案。