据地震液化的实测资料 ,即地震荷载下实测的各项土的物理力学参数 ,利用BP人工神经网络理论 ,建立粉土液化等级评别的反向神经网络模型进行判别 .与现有的方法比较 ,具有通用性、客观性、科学性 .因人工神经网络具有良好的非线性映射能力 ,可以灵活方便地对多成因的复杂的未知数进行高度的建模 .神经网络兼有概念直观 ,公式简洁 ,标准化编程和易于实现等优点 According to the measured data of seismic liquefaction, that is, the physical and mechanical parameters of each soil measured under earthquake load, the artificial neural network theory of BP was used to establish the inverse neural network model of silt liquefaction grade evaluation.Compared with the existing methods , Which has versatility, objectivity and scientificness.Because artificial neural network has a good ability of nonlinear mapping, it can flexibly and conveniently model the complicated unknowns with multiple causes.NNs are both intuitive in concept, concise in formula, Standardized programming and easy to implement and so on