温度—静水压力下非线性位移时序模型及应用

Nonlinear Displacement Time-Series Model and Application under Temperature and Hydrostatic Pressure

岩土结构的位移一般是在多种内外因素的共同作用下产生的,而目前模型大多仅考虑时序对岩土结构位移的影响,忽略了外界因素的变化对岩土结构位移的影响。考虑温度、静水压力外界因素的变化对岩土结构位移的影响,提出一种复杂环境影响下的非线性位移时间序列建模方法。该方法用人工神经网络建模取代传统的分析方法,与遗传算法结合,自动确定输入时步长度和神经网络模型结构,建立温度、静水压力等外界因素影响下的非线性位移时间序列模型。例证表明,该模型具有较好的预测与外推预测功能。

Modeling of nonlinear evolvement of displacement is great important in geotechnical engineering. The displacement caused under effect of several kinds of factors together is commonly characterized by nonlinear kinetic evolution. To offer more significant and valuable parameters for practical engineering, a modeling method in which the temperature, hydrostatic pressure and time-series are considered is proposed. This method describes the characteristic of nonlinear evolvement of displacement using ANN （the artificial neural network） whose structure and time-series is automatically searched by GA （the genetic algorithm）. Each sample of ANN is made of the temperature and the hydrostatic pressure as input and the displacement as output. The practical example shows that the model obtained by this algorithm has more accurate predicting result and is globally optimal.