一、前言 谷本喜一把统计判别方法用于建立预测砂层地震液化势公式的基本前提是假定判别函数为线性函数,而这一假定是有局限性的。在实际资料中,某些变量与液化势之间的关系往往呈现出非线性性质,这就给我们提出了研究建立非线性判别函数的必要性。 I. INTRODUCTION The basic premise of a statistical discriminant method used by Ryuichi Matsumoto to establish a formula for predicting the liquefaction potential of a sand layer is to assume that the discriminant function is a linear function, and this assumption is limited. In the actual data, the relationship between some variables and the liquefaction potential tends to be nonlinear, which gives us the necessity to study the establishment of nonlinear discriminant functions.