液化判别中基于人工神经网络的粘粒含量取值研究

岩土工程勘察实践中,得到正确的粘粒含量是进行液化判别的前提与基础。利用静力触探成果进行液化判别时,粘粒含量往往采用场地平均值或邻近钻孔相应深度处土样的粘粒含量值,与判别点处粘粒含量实际值间存在一定差别,影响了采用静力触探液化判别的准确性。采用人工神经网络可建立起需液化判别土层各点粘粒含量与相应空间坐标间的关系,从而得到场地内需液化判别土层粘粒含量的分布规律,为液化判别提供更为准确的粘粒含量。

Research on the adoption of clay particle content during seismic liquefaction with artificial neural network

During geotechnical engineering investigation,adopting the true clay particle content is basic and important to deal with seismic liquefaction.The average clay particle content of the field or those of the neighbor soil sample at corresponding depth is used to carry seismic liquefaction based on SPT result,which is usually different from the true clay particle content.In this paper,a methodology is proposed to provide clay particle content closer to the truth by the relationship between clay particle content and the space coordinate,which is established by artificial neural network.