基于蚁群算法优化支持向量机的边坡位移预测

Forecasting Slope Displacement Based on Support Vector Machine Optimized by Ant Colony Algorithm

由于复杂工程地质条件和环境因素的综合影响,边坡变形呈现复杂非线性演变特征。借助支持向量机(SVM)可有效解决小样本、高维数、非线性等问题的优点,对边坡实测位移进行数据挖掘,预测边坡变形趋势。为了避免人为选择支持向量机模型参数的盲目性,提高模型预测精度和泛化能力,引入改进的蚁群算法(ACO)对模型参数进行寻优,结合位移时序滚动预测方法,建立了适合边坡变形预测的ACO-SVM模型。将该模型应用于2个边坡的位移预测,研究结果表明,ACO-SVM预测精度高,模型建立正确。与遗传算法、粒子群算法优化SVM的预测结果相比,ACO-SVM模型预测精度更高,具有更强的泛化能力,预测结果更加合理,在边坡变形预测中具有一定的工程应用价值。

Due to the combined influence of complex engineering geological conditions and environmental factors, the evolution of slope deformation is complicated and nonlinear. As support vector machine （SVM） could effectively solve the small sample, high dimension, and nonlinear problems, it is employed for the data mining of measured displacements of slope and the forecasting of slope deformation trend. In order to avoid the blindness of human choice of SVM parameters and to improve the prediction accuracy and generalization ability of the model, ACO- SVM model is built by adopting improved ant colony algorithm （ACO） to optimize parameters in association with rolling forecasting method of displacement time-series. The model is applied to two engineering examples. The re- search results show that ACO-SVM model is correct with high accuracy. Compared with optimizing SVM based on genetic algorithm or particle swarm optimization, ACO-SVM model has higher accuracy of prediction and stronger generalization ability. The forecasting results are more reasonable. It is of practical value for slope deformation pre- diction.