dc.description.abstract | 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. | en_US |