边坡位移时序预测的组合模型研究

将边坡变形预测看作一个特殊的凸二次规划问题，以加权一阶局域法（AOLMM）、Lyapunov指数预报法以及神经网络预测方法（ANN）为基础，建立了边坡变形预测的组合模型，应用动态规划方法求解组合预测模型的最优解，以达到有效利用各种预测方法提供的信息和提高模型预测精度的效果．通过工程实例研究表明，该组合预测模型较单一预测模型精度有较大提高，表明组合预测模型的可行性及有效性．

Study of Combinatorial Model for Slope Deformation Prediction

Deformation prediction is treated as a convex quadratic programming problem. Basing on weighted one--rank local region method, Lyapunov exponent prediction method and chaos-artificial neural network,the combinatoriat modelfor prediction of slope deformation is established. And it applies dynamic programming vided study show method to find the optimal solution for the combinatorial model, so as to exploit all the information provided by different method and improve the accuracy of the prediction model. Also, the results of the case study show that the combinatorial model is much more efficient in improving prediction accuracy, so as to show that the model is feasible and effective.