论文部分内容阅读
根据青藏公路多年冻土区路基监测断面下获得的温度时间序列资料,通过小波多分辨分析,采用db6小波函数将该非平稳温度时间序列分解为多个高频分量序列和一个低频分量序列,分别运用时间序列的自回归滑动平均混合(ARMA)模型和BP神经网络模型进行建模和预测,最后将各个模型进行组合形成冻土温度的预测模型。与单用滑动平均混合模型或BP神经网络对原始温度时间序列进行建模预测相比,该方法具有较小的模拟误差和预测误差。结果表明小波分析方法预测多年冻土区路基下土体的温度是合理可行的,并且具有较好的精度和稳定性。
According to the temperature time series data obtained from subgrade monitoring sections of Qinghai-Tibet Highway in permafrost region, the non-stationary temperature time series is decomposed into multiple high frequency components and one low frequency component using wavelet multiresolution analysis with db6 wavelet function The time series autoregressive moving average mixture (ARMA) model and BP neural network model are used for modeling and forecasting. Finally, the models are combined to form the prediction model of the temperature of frozen soil. Compared with the single sliding average mixed model or the BP neural network modeling and forecasting of the original temperature time series, this method has less simulation error and prediction error. The results show that it is reasonable and feasible to predict the temperature of soil under subgrade in permafrost region by wavelet analysis method, and it has better accuracy and stability.