| 摘要: | 自從Seed和Idriss成功在1971年發表「液化評估法」後，時至今日，已有眾多學者嘗試利用當地的土壤液化資料庫，建立多種不同的液化評估法，而本研究以SPT-N為基礎，進行液化評估法的相關研究。首先，本研究將使用CSR7.5與(N1)60，建立適用於台灣的羅吉斯迴歸模型，之後我們也將進行CSR7.5與(N1)60,cs散佈圖繪製，並以台灣的土壤液化數據為基礎，建立每位學者提出之液化評估法的混淆矩陣，再透過四種模型評估指標，比較各方法的差異。
本研究使用的液化評估法，參考Hwang et al. [1]提及的方法，其中包含：HBF法、B&I法、Cetin法、NCEER法、AIJ法與JRA法，而本研究所使用的數據，同樣是由Hwang et al. [1]提供，使用以SPT-N為基礎的土壤液化數據庫，作為數據來源，並以CSR7.5與(N1)60,cs的計算結果，建立不同方法的混淆矩陣，以利進行後續分析。
除此之外，由於我們也發現地震矩規模修正因子(Magnitude scaling factor, MSF)可能對液化評估法的計算結果，造成一定程度之影響，所以本研究也將透過文獻提供的509筆，來自於台灣的土壤液化數據，分別使用多位不同學者提出之MSF，替換HBF法中的MSF，並將計算結果各別建立混淆矩陣與模型評估指標，評估MSF的變化對液化評估法之影響大小。
最後，本研究將最佳化擁有MSF的液化評估法，以四種模型評估指標的最大值為最佳化目標，使用Excel Solver尋找各自方法之最佳化參數，以進行MSF的模型參數最佳化。
;Since Seed and Idriss published the "Simplified Procedure" in 1971, numerous scholars have attempted to develop various simplified procedures using local soil liquefaction databases. In this study, simplified procedures based on SPT-N value will be used for the relevant research. Initially, this study will establish a logistic regression model in Taiwan using CSR7.5 and (N1)60, followed by the generation of scatter diagrams of CSR7.5 and (N1)60,cs. Subsequently, based on soil liquefaction data in Taiwan, the confusion matrix of the simplified procedures proposed by each scholar will be established, and the differences between the methods will be compared using four model validation indices.
The simplified procedures used in this study are provided by Hwang et al. [1], including the HBF method, B&I method, Cetin method, NCEER method, AIJ method, and JRA method. Furthermore, Hwang et al. [1] have provided the data used in this study based on the SPT-N value soil liquefaction database. After the calculation of CSR7.5 and (N1)60,cs, the confusion matrices of different methods will be established for subsequent analysis.
Additionally, we have found that the magnitude scaling factor (MSF) may impact the calculation results of the simplified procedure. Therefore, this study will use the 509 data provided by the literature from the soil liquefaction data in Taiwan to verify the MSF. The MSF proposed by different scholars will be used to replace the MSF in the HBF method. The confusion matrix and model validation index will be respectively established to evaluate the impact of the change of MSF on the simplified procedure.
Finally, this study will optimize the simplified procedures with the maximum value of the four model validation indices as the optimization goal. To achieve the objective, Excel Solver will be utilized to determine the optimal parameters of each method to optimize the model parameters of MSF. |
