节点文献
基于区间分析的不确定性结构损伤识别方法研究
Structural Damage Detection with Uncertainty Quantification Using Interval Analysis
【作者】 李坤;
【导师】 刘纲;
【作者基本信息】 重庆大学 , 结构工程, 2015, 硕士
【摘要】 结构损伤识别利用传感器测试数据,通过反演计算对结构损伤的有无、部位和程度进行识别,是土木工程结构健康监测中重要的一环。在将损伤识别应用于实际结构时,受荷载作用、环境温度、测试误差等不确定性因素的影响,易出现损伤的误判或漏判,导致结构健康监测失去应有的作用。本文基于区间分析理论,结合遗传算法和向量回归模型提出了损伤特征指标的区间求解算法,根据损伤特征指标的区间重叠率建立了结构损伤的两阶段识别方法。通过引入Information-gap(Info-gap)决策模型和决策评价标准,克服了需人为指定区间重叠率阈值的不足,从而实现了不确定性因素影响下结构损伤部位的识别。本文完成的主要工作以及研究结果概括如下:①提出了基于NSGA-Ⅱ的区间求解优化算法。分别选取数值算例和有限元模型算例,将区间优化算法和传统的区间组合法、区间截断法进行对比分析,结果表明区间优化算法能有效减少区间求解过程中的区间扩张问题,计算精度优于传统方法。分析了NSGA-Ⅱ中不同参数设置对区间求解优化算法的影响,为算法合理设置参数提供了参考。②根据测试的加速度响应信号建立向量自回归模型(Vector Auto-Regression,VAR),选取VAR模型系数主对角线元素的马氏距离作为损伤特征指标。在不确定性因素影响下,利用区间优化算法求解特征指标区间,并提出基于区间重叠率系数的损伤判别标准,建立了不确定性结构损伤识别两阶段法。六自由度数值算例结果表明,采用区间分析进行损伤初判定,然后通过区间重叠率进行损伤最终判定的两阶段方法可有效识别结构的损伤。③将Info-gap理论与区间分析融合,提出基于区间分析的Info-gap决策模型和稳健度模型,在分析这两个模型中参数指标物理意义的基础上,通过反馈函数建立了不确定性损伤识别决策评价模型。通过六自由度算例表明,基于Info-gap决策模型的不确定性损伤识别方法能准确识别结构的损伤。④在实验室搭建四层钢框架结构,通过增加质量、不同室温测试、不同VAR模型阶数来分别模拟质量、环境温度和损伤识别算法的不确定性。采用损伤两阶段判定方法的结果表明:基于区间分析的损伤识别方法能有效处理不确定损伤识别问题。
【Abstract】 Structural damage identification is a method to identify the position and the degree of the damage structure by using sensor test data and the inversion calculation, which is now an important part of the structural health monitoring. When the damage identification is applied to the actual structure, it will be affected by load, ambient temperature and the test error of the uncertainties. These uncertainty factors will be prone to false positives, and lead to invalidation of the structural health monitoring. Based on the theory of interval analysis, combined with the genetic algorithm and vector regression model, a solving algorithm of damage characteristics is proposed in this paper. According to the characteristics of the damage index interval overlap rate, the two stage method of structural damage identification is established. Through the introduction of Information-gap(Info-gap) decision model and decision evaluation criteria, we can identify the structural damage position and the degree of the uncertainties without specifying the interval overlap threshold.The main works are summarized as follows:① The interval optimization algorithm based on NSGA-Ⅱis put forward. By using numerical example and finite element model, the interval optimization algorithm are compared with the traditional interval combination method and the interval truncation method. The results show that the interval optimization algorithm can effectively reduce the extension interval in the solving process, and the former calculation accuracy is better than the latter. Moreover, the influences of different parameter on the interval optimization algorithm in NSGA-Ⅱis discussed, which provides a reasonable reference for parameter setting algorithm.② Vector Auto-Regression(VAR) is used to model the acceleration response signal. The main diagonal elements of the VAR model coefficients are selected as the damage characteristic index and They are combined with the interval optimization algorithm to solve the problem of uncertainty of structural damage and to set up a two stage decision process. Through six-story shearing numerical frame example, it verify that the proposed damage identification method is effective.③ With the combination of the Info-gap theory and the interval analysis, the decision model and robust degree model were proposed. Based on the analysis of the physical meaning of parameters in the two models, the Info-gap uncertainty decision evaluation model was established by the feedback function. And the evaluation model was verified by the six-story shearing frame model. The calculation results show that the proposed damage identification method based on Info-gap decision model can identify the structural damage accurately.④ Through increasing mass on a four story steel frame structure, testing in different environmental temperature and using different VAR model order, which simulate the uncertainty of quality, environment temperature and the damage index property extraction. The results obtained by the injury two stage decision method show that under the three kinds of uncertain factors the uncertainty of structural damage identification method can effectively identify the structure damage.
【Key words】 Damage identification; Uncertainty; Interval analysis; VAR model; Info-gap theory;