臺灣博碩士論文加值系統

結構參數修正法已經發展多年，大多數是以模態、頻率已知的狀況進行結構參數修正。本研究提出以部分的模態以及量測點位不完全的狀況下，能夠有效率的進行結構參數修正，並且把答案的穩定性也考慮在內。本方法是由兩個階段所組成，第一個階段為勁度修正指標階段，第二階段為菁英策略之非監督式模糊類神經網路。第一階段會將案例在全域的範圍隨機佈點，在每一次疊代中，會計算每個案例各自的勁度修正指標，每個案例再由勁度修正指標往目標點進行修正，第一階段結束後，可以使得答案與目標點差異在10%左右，也不會有過於冗長的疊代次數。非監督式模糊類神經網路在好的初始點狀況下能夠快速的找到精準的答案，因此第二階段以第一階段輸出之結果當作非監督式模糊類神經網路的初始點，每次疊代會以目前找到的最佳解為中心，在其附近隨機佈點，以與目標較為相近的案例根據非監督式模糊類神經網路的重心法求最佳解。菁英策略為在每次非監督式模糊類神經網路的疊代中，保留上一次疊代部分好的案例，使當下疊代時有更多好的案例可以參考，並且在第一頻率增加權重，讓答案的穩定性及精確度增加。為了測試本研究的精確度及穩定性，會以三個數值模型來做測試，分別為勁度不同6層樓及9層樓之剪力構架。破壞位置檢測則是以6層樓剪力構架之數值模型及3層樓與8層樓之剪力構架實驗模型來做測試。測試結果顯示本研究方法可以使用部分頻率、模態得到穩定且精度在工程可接受範圍的答案，也能準確地識別出結構的破壞位置。

The method of structural parameter updating has been developed for decades. In most modal, frequency is the most known conditions for structural parameters updating. In this study, it is proposed that the efficiency of structural parameters’ modification and the stability of the solution are both taken into account in the case of partial modal data and incomplete measurement points. There are two stages in this method: the first stage is the stiffness revise index stage, and the second stage is the unsupervised fuzzy neural network of the elite strategy. The first stage will randomize the case in the global scope. During each iteration, it will calculate the respective stiffness index of each case, and each case is revised by the stiffness revise index to the target point. The difference between the solution and the target point can be around 10%, and there will not be too many iterations after the end of the first phase. Unsupervised fuzzy neural networks can quickly find accurate solutions at good initial conditions, so the second stage takes the results of the first stage output as the initial point of the unsupervised fuzzy neural network. The best solution will be the center in each iteration, and it will randomly distribute case in its vicinity. Elite strategy in each iteration of unsupervised fuzzy neural network keeps good examples from prior iteration for the sample of next iteration. This process can increase the weight at the first frequency to increase accuracy and stability of the solution. 6-story and 9-story shear-type structures are employed to verify the accuracy and stability of the proposed approach. The damage detection is based on the numerical model of the 6-story shear-type structures and the experimental model of the 3-story and 8-storey shear-type structures. The result revealed that the proposed approach can accurately and stably identify the damage locations by partial modal data.

摘要 I
Abstract II
誌謝 III
目錄 IV
表目錄 VI
圖目錄 IX
第一章　緒論 - 1 -
1.1 研究背景 - 1 -
1.2 研究動機 - 2 -
1.3 研究目的 - 3 -
1.4 研究步驟 - 3 -
1.5 論文架構 - 4 -
第二章 文獻回顧 - 5 -
2.1 Guyan模態擴張法 - 5 -
2.2 雲線擬合法 - 6 -
2.3 以近似柔度法做破壞位置檢測 - 7 -
2.4 非監督式模糊類神經網路 - 9 -
第三章　研究方法 - 14 -
3.1 勁度修正指標 - 14 -
3.2 不完全量測勁度修正指標修正條件設定及準確性 - 15 -
3.3 全域搜尋與區域搜尋之勁度修正指標效率比較 - 23 -
3.4 菁英策略之非監督式模糊類神經網路 - 28 -
3.5 權重與適應值 - 31 -
3.6 演算法流程 - 32 -
第四章　數值模擬 - 36 -
4.1 建立模型 - 36 -
4.2 結構模型修正準確性與穩定性 - 46 -
4.2.1 模型1(無雜訊) - 46 -
4.2.2 模型1(包含雜訊) - 48 -
4.2.3 模型2(包含雜訊) - 51 -
4.2.4 模型3(包含雜訊) - 60 -
4.3 破壞檢測 - 70 -
第五章　實驗案例測試 - 80 -
5.1 三層樓剪力構架 - 80 -
5.2 八層樓剪力構架 - 84 -
第六章 結論與建議 - 88 -
6.1 結論 - 88 -
6.2 建議與未來展望 - 89 -
參考文獻 - 90 -

[1] Mottershead, J. E. and Friswell, M. I. (1993), “Model updating in structural dynamics: a survey”, Journal of sound and vibration, 167(2), 347-375.
[2] Brownjohn, J. M. W., Xia, P. Q., Hao H., Xia Y. (2001), “Civil structure condition assessment by FE model updating: methodology and case studies”, Finite Elements in Analysis and Design, 37(10), 761-775.
[3] Teughels, A., Roeck, G. D. (2005), “Damage detection and parameter identification by finite element model updating”, Archives of Computational Methods in Engineering, 12(2), 123-164.
[4] Perera, R. and Torres, R. (2006), “Structural damage detection via modal data with genetic algorithms”, Journal of Structural Engineering, 132(9), 1491-1501.
[5] Kang, F., Li, J. J. and Xu, Q. (2012), “Damage detection based on improved particle swarm optimization using vibration data”, Applied soft computing, 12(8), 2329-2335.
[6] 吳維珉，「應用勁度修正指標與非監督式模糊類神經網路於剪力構架之結構勁度參數修正」，國立交通大學，碩士論文，民國105年。
[7] Guyan, R. J. (1965), “Reduction of Stiffness and Mass Matrices”, AIAA Journal, 3(2), 380-381.
[8] Kano, H., Fujioka, H., Egerstedt, M., Martin, C. F. (2005), “optimal smoothing spline curves and contour synthesis”, IFAC Proceedings Volumes, 38(1), 297-302.
[9] Erb, W. (2016), “Bivariate Lagrange interpolation at the node points of Lissajous curves –the degenerate case”, Applied Mathematics and Computation, 289, 409-425.
[10] Sung, S. H., Koo, K. Y., Jung, H. J. (2014), “Modal flexibility-based damage detection of cantilever beam-type structures using baseline modification”, Journal of Sound and Vibration, 333(18), 4123-4138.
[11] Koo, K. Y., Lee, J. J., Yun, C. B., Kim, J. T. (2008), “Damage detection in beam-like structures using deflections obtained by modal flexibility matrices”, Smart structures and systems, 4(5), 605-628.
[12] 葉怡成，類神經網路模式應用與實作，二版，儒林圖書有限公司，台北，民國八十二年八月。
[13] Brooks, H., Tucker, N. (2015), “Electrospinning predictions using artificial neural networks”, Polymer, 58, 22-29.
[14] Wang, J. H., Wang, H. Y., Chen, Y. L., Liu, C. M. (2015), “A constructive algorithm for unsupervised learning with incremental neural network”, Journal of Applied Research and Technology, 13(2), 188-196.
[15] Chef, S., Jacquir, S., Sanchez, K., Perdu, P., Binczak, S., Gan, C. L. (2015), “Unsupervised learning for signal mapping in dynamic photon emission”, Microelectronics Reliability, 55(9), 1564–1568.
[16] Hung, S. L. and Jan, J. C. (1999), “Machine learning in engineering analysis and design an integrated fuzzy neural network learning model”, Computer-aided civil and infrastructure engineering, 14(3), 207-219.
[17] Hung, S. L., Jan, J. C., Wen, C. M. and Huang, C. S. (2007), “Unsupervised fuzzy neural networks for damage detection of structures”, Structural Control and Health Monitoring, 14(1), 144–161.