采用自适应模糊神经网络建立了土石坝的沉降组合模型 ,并以某土石坝为实例进行了具体分析应用 ,将神经网络预测模型与各个子模型进行了对比分析研究 .结果表明 ,该组合模型具有较好的拟合和预报精度 .采用模糊神经网络进行组合预报 ,可以避免常规组合模型难以确定权系数的困难 The subsidence combination model of earth-rock dam is established by using adaptive fuzzy neural network, and an earth-rock dam is taken as an example to carry on the concrete analysis and application, the neural network prediction model and each sub-model are compared and analyzed.The results show that the combined model has Better fitting and forecast accuracy.Using the fuzzy neural network to forecast the combination can avoid the difficulty of determining the weight coefficient in the conventional combination model