透過您的圖書館登入
IP:3.137.178.133
  • 學位論文

有效隨機維度與邊坡穩定可靠度分析

Effective Random Dimension in Slope Stability

指導教授 : 卿建業

摘要


可靠度指標是在分析時考慮了分析誤差和變數不確定性的指標。分位數值法(Quantile Value Method, QVM)相較於部分係數設計法和工作應力分析, 提供了更高的精度, 並且在效率上勝過蒙地卡羅模擬(Monte Carlo simulation,MCS)。但當要分析的案例涉及到多變量分析,QVM方法的表現並不理想。因此,卿建業等人在2014年提出加入有效隨機維度的概念,發展出一套結合了QVM的效率和MCS的準確度的新方法。 有效隨機維度反映了有多少隨機變量真正參與在分析中。我們預期能採用隨機維度來連接確定值分析(deterministic)和機率(probabilistic)分析的結果。如果我們能夠建立一個完整的數學模型來描述有效隨機維度,以後我們將可以單利用精確解反算出具有不確定性的結果,而不需要經過繁複的MCS。 此研究是關於水平分層之邊坡穩定分析。以土壤各層的不排水剪力強度為隨機變數。本文首先會先闡述有效隨機維度、機率門檻和可靠度設計等重要名詞, 之後會介紹研究中使用的水平分層邊坡和選用的參數, 最後在研究中採用線性回歸和數學模型建立的二進制信息準則(BIC),希望能找到一組適當的參數組合去描述有效隨機維度和該邊坡案例的關係。

並列摘要


Reliability index is the index that considers the errors and uncertainties in the analysis. The Quantile Value Method (QVM) provides better accuracy to determine the reliability index in reliability analysis than partial factor design method and working stress analysis; have better efficiency than Monte-Carlo simulation. But when it comes to multivariate analysis, QVM method doesn't perform stably. Ching et al. developed a new method called 'ERD-based QVM', that combines the QVM’s efficiency and the MCS’s accuracy with Effective Random Dimension (ERD). The effective random dimension is an index that reflects how many random variables are really considered in the analysis. We expect using effective random dimension to connect the deterministic analysis and the probabilistic result. If we can form a complete mathematical model to describe the effective random dimension, we can back calculate the probabilistic solution from the only deterministic solution without doing Monti-Carlo simulation. The research is about the layered slope stability analysis. The undrained shear strength of each layer of soil is random. We use linear regression and the binary information criterion (BIC) on math model building and hope to find a appropriate combination of parameters to determine the effective random dimension.

並列關鍵字

QVM reliability index ERD layered slope

參考文獻


[6] 楊志堅. 有效隨機維度在大地工程可靠度中的角色. 國立Ѡ灣大學土木工程學
[1] W.B. Wei Y.M. Cheng, T. Lansivaara. Two-dimensional slope stability analysis by
137–150, 2007.
Geoenvironmental Engineering, 136(4), 2010.
[4] Jian Ji and Bak Kong Low. Stratified response surfaces for system probabilistic evaluation

延伸閱讀