利用健康监测的数据，反演结构的健康状态

aspen

如何利用健康监测的数据，反演结构的健康状态，大家说说自己的idea把。 无论是系统识别还是模式识别来识别结构的损伤几乎是不可能的，而且由于桥梁的工作环境不同，所以桥梁的正常状态应该是一个区间而不是一个状态点，所以美国国家重点实验室提出的统计模式识别（statistical patter recognition）可能是一种比较好的出路，都是应用到大型桥梁的损伤识别也有很大距离。 美国的Aktan教授在他的论文Condition assessment for bridge management 认为桥梁健康监测的目的在于整体工作状态评估，然后才是识别可能的DDD（defect，damage and deterioration),原文为Condition assessment should first focus on the evaluation of global health state, and then on regoinal and local DDD).所以，对当前来说，真正的监测系统的重点是否不应该放在损伤识别，而是监测整体性能（integrity）呢？ 另外，监测系统出了监测结构的性能之外，很重要的就是如何指导日常的养护管理工作，对这一点，大家有没有什么idea啊。 ==========

看了flutter 楼主的帖子，感觉很有道理，现在的桥梁结构的损伤检测在理论方面好像取得了一些结果，不过在实际的应用中好像还有许多难以预测的困难。我现在也在作桥梁结构的损伤识别方面的论文，不过现在遇到了一些难题，想向flutter 请教一下，望回复，问题如下： ※请问损伤识别用神经网络方法可行么？具体的损伤诊断方法哪种方法的可行性更好一些？ ===========

柔度可能是比较好的指标吧。去LANL下载一些论文看看http://www.lanl.gov/projects/ncsd/publications.htm 里面有大量的论文，有个review，还有两篇基于振动损伤识别方法的比较，98年时候的，还是很强。里面基本上包括了见几年各类损伤识别方法。 神经网络应该还是可以的，不过要网络节点和网络层数恰当，否则据说容易陷入局部最小值（这个偶不懂，呵呵）。 最近热门了一下支持向量机（Support Vector Machine）可以克服局部最小，但是寻找合适的模型参数计算量也是巨大。 小波、HHT都可以试试吧，没用过的说。 ===========

据本人了解，HHT是一种比较先进的信号处理方法，其核心是EMD，同Fourier谱相比较，Hilbert谱在客观性和分辨率方面具有明显的优越性。 ===========

看了一篇英文文章，个人认为重要的观点讨论列出一下。文章出处：By A.E. Aktan, Issues in health monitoring for intelligent infrastructure, Smart materials and structures, 1998, 672-692 （1）：Global objective condition assessment（整体客观状态评价）, structural identification（结构识别）, failure-model analysis（失效模式分析） and structural reliability （结构可靠性）are concepts that are currently foreign to most bridge engineers and managers, but they hold the promise for developing and implementing innovative strategies for effective, optimal management.－健康监测的目的主要是为了桥梁的维护和管理把，对维护其作用的是否是结构的可靠度（或者风险，＝可靠度×失效后果）呢？因此把监测系统和结构可靠度理论或风险分析挂钩如何？ While time-based structural reliability formulations are well suited for rare, extreme events which lead to damage and failure limit states, symptom-based reliability formulations have been shown to hold great promise for health monitoring at the serviceability limit state. （2）：The data imply that the following phenomena should be explored further: how the climactic and environmental conditions at the time of construction may affect life-cycle performance; how seasonal, monthly, weekly and daily environmental and ambient condition changes may impact damage/deterioration/defects; what are the combined effects of environmental and traffic responses which may be in phase or out of phase depending on time of occurrences.－桥梁施工环境如何影响桥梁的性能。 Obviously, the implementation of intelligent infrastructures will require the ability to navigate the subtleties of the response of the structure and its associated on different time scales, to different inputs, e.g., traffic, environmental, damage, deterioration, etc, in order to reach objective, reliable decisions. It is also observed that because of magnitudes and variability of forces, unless the changes in intrinsic responses of a bridge are monitored for several yeas, it may not be possible to differentiate the impact of deterioration and/or damage from natural changes in intrinsic responses. －桥梁的健康监测是个数据积累、自学习的过程，必须要有大量的数据才可能识别结构特性的变化是由损伤还是环境。 （3）Lessons learned from destructive testing of a constructed facility are some raw data at damage state of structures. We also realize that there may be many modes and manifestations of sensor malfunctions, some of these are discovered only by experimenting on actual constructed facilities in the field. This observation further highlighted the need for monitoring systems, especially those operating autonomously, to have self-checking mechanisms which are themselves ‘intelligent ’－传感系统的可靠性，错误数据的剔除 （4）文章最后提出的几个问题(1) Is it possible to detect structural damage by linearized indices, and what are the most appropriate linearized indices for expressing global health? (2) Can we find rational as well as statistical and heuristic relationships between measurable loads and response quantities, global structural condition indices and structural reliability? (3) Can we establish rational relationships between microscopic damage indices, local ‘meso-level’ defects, common visual damage and deterioration indices and ‘macro-level’ indices such as flexibility? (4) What are the optimum use modes and possible synergies between modal testing and instrumented monitoring for structural identification and condition assessment of constructed facilities? (5) Is it possible to assess structural condition without a baseline?

Izaac

顶一个！

张三分