透過您的圖書館登入
IP:3.236.145.110
  • 學位論文

應用巨量數據縮減及二向度視化技術於結構健康診斷

Structural Damage Detection Using High Dimension Data Reduction and Visualization Techniques

指導教授 : 羅俊雄

摘要


近年來結構健康監測方面的研究發展迅速,至今已經開發出許多不同技術於結構損傷檢測方面,例如隨機子空間識別法或零子空間損傷識別…等。為了能更精準的判定結構物特性並迅速評估其結構物之狀態,可藉由安裝許多感測器於結構物上並蒐集量測所得之訊號,用於分析及提取結構損傷特徵。其中,分析方法可分為兩大類:Centralize data fusion technique 以及 Pattern level fusion technique,在此研究中,運用此二種方法經由資料縮減以及數據可視化之技術以進行結構損傷評估。 在本研究中,提出了一些結構健康監測方法用於損傷識別和判定損傷位置,其主要是由主成分分析概念延伸而來。首先,以頻率響應做為損傷評估之初始資料,並分別透過主成分分析以及多變量自回歸模型重建數據並將其轉換至 Sammon圖上呈現於二維坐標平面上。 Sammon map 是一種非線性二維縮減算法,此方法原理為嘗試盡可能地保留數據矩陣中的每一列間的歐幾里得距離。比對不同數據結構之間在 Sammon 圖上的結果可用於檢測損傷定位。 除此之外,不同測量中數據的時頻分析(小波包封轉換之能量譜)的相關性也可用於檢測損傷。國家地震工程研究中心採用振動台試驗方法,對兩種結構進行了驗證:一種是一雙塔鋼結構,其主要頻率變化於低頻部分,另一種是一六層樓且具有切割構件的鋼結構,其破壞主要於較高頻部分。另一方面,實際結構物採用國立中興大學土木與環境工程大樓收集到的歷年地震反應數據也被用於驗證這些方法。分析結果表明,大多數方法可用於識別損傷檢測和定位。

並列摘要


Structural health monitoring (SHM) is considered as an incentive multi-disciplinary technology for conditional assessment of infrastructure system. Numerous techniques from the disciplines of multivariate statistics and pattern recognition in the field of structural damage detection have been developed, such as stochastic subspace identification or null-space and subspace damage identification. In order to obtain the accurate and quick structural damage assessment of a structure, the most convenient way for structural health monitoring is to use the measurement from all sensing nodes in the structure to extract damage features. Due to large amount of data directly from measurement, considering both centralized and pattern-level damage assessment techniques, dimension reduction and data visualization techniques for damage assessment need to be investigated so as to have a quick safety assessment on the current structural state. In this study the structural health monitoring methods for damage identification and localization, which were developed from the principal component analysis (PCA) of measurement data. First, the frequency response function (FRF)-based damage assessment will be investigated. Based on either the PCA or Multivariate autoregressive model (MV-AR model) to reconstruct data and present them on Sammon map for visualization from all measurements will be investigated. Sammon’s mapping is a nonlinear dimesional reduction algorithm which seeks to preserve Euclidean distances between each row in the data matrix as far as possible. Similarity among different data structure can be used to detect damage localization. Besides, the correlation of time-frequency analysis of data (wavelet-based energy spectrum from WPT) among different measurement is also used to detection damage. Two structures were used to verify the proposed methods which were operated on shaking table tests in National Center for Research on Earthquake Engineering: One is a twin-tower structure which focus on the damage of lower vibration modes and the other one a 6-story steel structure with cutting members which will focus on the damage of high frequency modes. On the other hand, seismic response data collected from an actual structure, which is Civil & Environmental Engineering Building in National Chun Hsing University, was also used to verify these methods. The analysis results show that the most of the methods can be used to identify the damage severity and also to locate the damage.

參考文獻


[1] Loh, C. H. and Chao, S. H., Centralized vs. Pattern-Level Feature Extraction for Structural Damage Detection. The 37th National Conference on Theoretical and Applied Mechanics, pp. 1-10, 2012.
[2] Liu, Y. C., Loh, C. H. and Ni, Y. Q., Stochastic Subspace Identification for Output-only Modal Analysis: Application to Super High-rise Tower under Abnormal Loading Condition. Earthquake engineering & Structural dynamics, vol. 42, pp. 477-498, 2013.
[3] Ruotolo, R. and Surace, C., Using SVD to detect damage in structures with different operational conditions. Journal of Sound and Vibration, Vol. 226(3), pp. 425-439, 1999.
[4] Vu, V. H., Thomas, M., Lakis, A. A. and Marcouiller, L., Operational modal analysis by updating autoregressive model. Mechanical Systems and Signal Processing, vol. 25, pp. 1028-1044, 2011.
[5] K. Krishnan Nair, Anne S. Kiremidjian, and Kincho H. Law, Time series-based damage detection and localization algorithm with application to the ASCE benchmark structure. Journal of Sound and Vibration, vol. 291, pp.359-368, 2006.

延伸閱讀