基于VMD和BSA-KELM的高陡边坡位移预测模型研究

边坡位移的时间序列曲线存在复杂的非线性特性,传统的预测模型精度不足以满足预测要求。为此提出了基于变分模态分解的鸟群优化-核极限学习机的预测模型,并用于河北省某水泥厂的边坡位移预测。该方法首先采用VMD把边坡位移序列分解为一系列的有限带宽的子序列,再对各子序列分别采用相空间重构并用核极限学习机预测,采用鸟群算法优化相空间重构的嵌入维度和KELM中惩罚系数和核参数三个数值,以取得最优预测模型。最后将各个子序列预测值叠加,得到边坡位移的最终预测值。结果表明：和KELM、BSA-KELM、EEMD-BSA-KELM模型相比,基于VMD的BSA-KELM预测精度更高,为边坡位移的预测提供一种有效的方法。

Study on high and steep slope displacement prediction model based on VMD and BSA-KELM

The time series curve of slope displacement has complex nonlinear characteristics and the accuracy of traditional prediction model is not enough to meet the requirement of prediction. A prediction model based on variational mode decomposition of bird swarm optimization-nuclear limit learning machine is proposed and applied to the slope displacement prediction of a cement plant in Hebei province. In this method, the slope displacement sequence is decomposed into a series of sub-sequences with limited bandwidth by VMD, and then each sub-sequence is reconstructed by phase space and predicted by nuclear limit learning machine. In order to obtain the optimal prediction model, the bird colony algorithm is used to optimize the three values of the embedded dimension and the penalty coefficient and the kernel parameter in the KELM of phase space reconstruction. Finally, the predicted value of each sub-sequence is superimposed to obtain the final predicted value of slope displacement. The results show that compared with KELM, BSA-KELM and EEMD-BSA-KELM models, the prediction accuracy of BSA-KELM based on VMD is higher, which provides an effective method for the prediction of slope displacement.