论文部分内容阅读
基于Fisher判别理论建立了砂土液化可能性的Fisher判别分析(FDA)模型。在分析砂土液化影响因素的基础上,选取烈度、震中距、地下水位、砂层埋深、标贯击数、平均粒径、不均匀系数、剪应力比等8个实测特征指标作为FDA模型的预测指标。利用砂土液化的实测数据作为训练样本进行训练,建立FDA模型对砂土液化进行预测,并用其他未参加训练的实测数据进行了验证。研究结果表明,FDA模型简便可行、预测精度高,是解决砂土液化预测问题的有效方法之一。
Fisher discriminant analysis (FDA) model was established based on Fisher discriminant theory to evaluate the possibility of sand liquefaction. Based on the analysis of influencing factors of sand liquefaction, eight measured characteristics such as intensity, epicentral distance, groundwater level, sand depth, number of standard penetration, average particle size, non-uniform coefficient and shear stress ratio were selected as FDA model Predictors. The measured data of sand liquefaction were used as training samples for training, the FDA model was established to predict the liquefaction of sand and the data were verified by other measured data without training. The results show that the FDA model is simple and feasible and the prediction accuracy is high. It is one of the effective methods to solve the problem of sand liquefaction prediction.