砂土地震液化的神经网络预测

PREDICTION OF SAND SOIL LIQUEFACTION DURING EARTHQUAKES BY THE BP NEURAL NETWORK

摘要:

BP网络具有很强的非线性映射和自适应学习功能,可用于模式识别和预测评估等领域.在简要分析BP算法的基础上,选取砂土的平均粒径(d₅₀/mm)、相对密度(D_r/%)、标准贯入击数(N_63.5/击)、上覆有效压力(σ_v/kPa)、地震烈度(I₀)作为指标,应用BP神经网络的理论与方法,预测砂土在地震作用下液化的可能性,取得了较好的预测效果.说明将BP网络用于沙土液化预测是可行的.

Abstract:

The BP neural network is applied to predicting the potential of sand soil liquefaction.The mean diameter,relative density,standard penetration number and effective over burden pressure are selected as the indices.The forecasting results show that the method is feasible.