A METHOD OF SAND LIQUEFACTION PROBABILISTIC ESTIMATIONBASED ON RBF NEURAL NETWORK MODEL

摘要

Based on the 344 liquefaction site data of 25 strong earthquakes in the world, through training and testing the neural network of radial basis function (RBF), the nonlinear relation between corrected blow count of standard penetration test (N_1)_(60) and cyclic resistance ratio of saturated sand CRR is analyzed, and also empirical equation CRR_(cri)of sand liquefaction limit state curve is constructed. By statistic analysis, probability density functions of liquefaction and non-liquefaction as well as empirical equation between safety factor and liquefaction probability of saturated sand are given, then the empirical equation of cyclic resistance ratio of saturated sand CRR under different probability is deduced. When liquefaction probability level is equal to 50%, the method in this paper is consistent to traditional deterministic method of sand liquefaction estimation, and it's reliability for liquefaction and non-liquefaction estimation of saturated sand is 90. 4% and 81. 2%, respectively; The method in this paper makes the sand liquefaction probabilistic estimation on engineering sites is as easy and convenient as traditional deterministic method of sand liquefaction estimation.