地铁施工段地表沉降组合预测模型研究

为解决地铁施工段地表沉降随机波动较大对预测模型造成扰动的问题，提出了一种组合预测模型。首先利用Mallat算法对沉降序列进行分解和重构，分解并重构出非平稳时间序列中的平滑分量和细节分量；然后，对平滑分量用灰色模型进行拟合与初步预测，对细节分量则引入马尔可夫模型进行预测；最后，将各模型的预测结果进行叠加，从而得到原始沉降序列的预测值。该方法能充分拟合平滑分量数据，而且可避免对细节分量的过拟合，防止随机波动较大值造成模型的整体偏离。通过实例验算得出该模型具有很好的预测精度。

Combined Prediction Model for Subsidence of Metro Construction Unit

Considering the subsidence＇ s random fluctuation at the metro construction unit, a combination prediction model is provid- ed. By using wavelet decomposition and reconstruction, the non - stationary time series can be decomposed into smoothing and detail components. The smoothing components can be predicted by grey model, and detail components can be predicted by markov model. The prediction is the combination of the respective prediction. The combination model can fit the smoothing components better, and a- void the fitting of the details components excessively. The result acquired from the example shows that the model has high prediction accuracy.