提出了一种基于相空间重构与支持向量机预测滑坡位移的新方法。首先,以滑坡位移时间序列的混沌特性为基础,对其应用互信息法计算最优时间延迟;然后,利用小波变换对滑坡位移序列数据进行频域分解,应用Cao氏方法对分解后的每个分量序列分别计算其最佳嵌入维数,在此基础上,对各个分量序列进行相空间重构,利用支持向量机对每个分量单独进行建模预测;最后,将各分量预测结果进行小波重构,得到最终预测结果。实例证明,该方法可以在滑坡位移预测中获得有效的应用。 A new method of predicting landslide displacement based on phase space reconstruction and support vector machine is proposed. Firstly, based on the chaotic characteristics of landslide displacement time series, the optimal time delay is calculated by using the mutual information method. Then, the wavelet transform is used to decompose the landslide displacement sequence data in frequency domain. Applying the Cao’s method to each of the decomposed Then, the phase space of each component sequence is reconstructed, and each component is separately modeled and predicted by using support vector machine. Finally, the result of wavelet transform Structure, get the final prediction result. The example proves that this method can be effectively applied in landslide displacement prediction.