WA联合ELM与OS-ELM的滑坡位移预测模型

摘要

The curve landslide cumulative displacement is usually nonlinear.Hence,it is challenging to build predictive models with less error.In this paper,we propose a new methodology of embedding wavelet analysis with basic extreme learning machine(ELM)and online sequential extreme learning machine(OS-ELM)to predict the cumulative displacement.Firstly,by wavelet transformation,the cumulative function of displacement is discretized into periodic displacement and trend displacement.Secondly,basic ELM and OS-ELM are selected to predict the periodic displacement and trend displacement.Lastly,the cumulative displacement function is computed by ensembling the predicted periodic and trend displacement values.For basic ELM,a sigmoid function is selected as the kernel function and a single hidden layer with 33 nodes performs best.For OS-ELM,the prediction error reaches its minimum with 1 00 hidden nodes when the RBF function is selected as the kernel function.RMSE for ELM is 0.1 423 and for OS-ELMis 0.1 31 5.This methodology with high predictive accuracy performs better in comparison with other methods.%滑坡累积位移监测曲线往往呈现出复杂的非线性增长特性，对此建立了不少相关的预测模型，而以往的预测模型存在着许多不足。本文基于小波函数（Wavelet Analysis，WA），ELM与 OS-ELM，提出一种名为 WA 联合 ELM、OS-ELM的预测方法。首先，该方法基于小波函数，将滑坡累积位移分解成受内部地质条件影响的趋势项和受外部影响因子影响的周期项；然后，基于 ELM与 OS-ELM分别对趋势项和周期项进行预测；最后将趋势项和周期项的预测值叠加得到累积位移的预测值。结果表明，小波函数得到的趋势项展现出良好的趋势性，而周期项也展现出良好的周期性；以 Sigmoid 方程为核函数，隐含层神经元个数为33的 ELM模型能准确高效对趋势项进行预测，而以 RBF 方程为核函数，隐含层神经元个数为100的 OS-ELM模型能准确高效对周期项进行预测；累积位移预测数据的 RMSE 分别为0．1423和0．1315，预测结果相对较好，能够在滑坡位移预测领域发挥一定的作用。