基于判别分析法的地震砂土液化预测研究

将距离判别分析方法应用于砂土液化的预测问题中，建立了砂土液化预测的距离判别模型。选用震级、研究深度、震中距、标贯击数、地下水位及地震持续时间等6项指标作为判别因子，以大量的工程实例数据作为学习样本进行训练，建立了线性判别函数对待评样本进行了评价。研究结果表明，距离判别分析模型判别砂土液化效果良好，预测准确度高，回判估计误判率低，可望成为砂土液化预测的有效手段。

Discriminant analysis model for prediction of sand soil liquefaction during earthquake

Based on statistics theory, Mahalanobis distance discriminant analysis model for prediction of sand soil liquefaction is established, including the main six factors that have great influence on sand soil liquefaction as follows: seismic intensity, epicenter distance, groundwater table, sand depth, blow number of standard penetration test, duration. Linear discriminant function is obtained through training a large set of sand soil liquefaction samples. Then the model is applied to the practical engineering. The results show that the prediction model of diatance dicriminant analysis has excellent performance,high prediction accuracy, and might be referred to as an effective technique for prediction of sand soil liquefaction.