基于RBF多变量时间序列的滑坡位移预测研究

摘要

Slope is a chaotic dynamic system influenced by various factors. It' s difficult to establish the deterministic equation of slope displacement since it is highly uncertain as a macro expression of the internal mechanical behavior of slope. Landslide is a genetic type of slope which has the same characteristics. Apart from groundwater, the major external motivation of landslide displacement, it is under the control of remedial measures after its treatment. Chaotic time series of landslide displacement and its influential factors could reflect the history of landslide displacement. According to the observed multivariable time series and the mapping relation between variables reflected by adopting RBF neural network, the displacement is predicted by reconstructing the dynamic system of landslide displacement. Results show that multivariable time series model could effectively predict landslide displacement , and the accuracy is higher than that of single-variable time series model; multivariable time series model is of clearer sense of the physical mechanics and reflects the real characteristics of deformation evolution.%斜坡是一个受到多种因素影响的混沌动力系统,斜坡位移是其内部力学现象的宏观表现,具有很强的不确定性,从而导致难以建立斜坡位移的确定性方程.滑坡是斜坡的一种成因类型,具有相同的系统特性.滑坡经过防治后,其位移的主要外在动力因素除地下水外同时还受到防治设施的控制.滑坡位移及其影响因素所构成的混沌时间序列能够反映滑坡位移动力系统的历史行为.根据观测获得的多变量时间序列重构原滑坡位移动力系统,采用RBF神经网络实现变量间的映射关系,对滑坡位移进行了预测.预测结果对比分析表明:采用多变量时间序列预测模型能对滑坡位移进行有效预测,取得比单变量时间序列预测模型更好的预测效果；多变量时间序列预测模型具有更明确的物理力学意义,更能反映滑坡演化变形的实质特征.