基于改进DE-LSSVM模型的边坡变形预测

Slope deformation prediction based on improved DE-LSSVM

针对目前最小二乘支持向量机(LSSVM)在预测算法中存在的不足,通过改变差分演化算法(DE)中的缩放因子个数、杂交概率的个数和变异策略来建立改进DE-LSSVM预测模型,利用某矿山的边坡观测数据。结果表明,基于改进DE-LSSVM预测模型有较优的预测能力。

Aiming at the deficiency of the least square support vector machine （LSSVM） in the prediction algorithm, the improved DE - LSSVM prediction model was established by changing the number of scaling factors, the number of hybridization probabilities, and the mutation strategy in the differential evolution algorithm （DE） using the slope observation data of a metal mine. Results showed that the improved DE - LSSVM prediction model had better predic- tion ability.