基于小波分解与TCN-BP的地铁基坑沉降预测模型OA

To address the challenges of strong nonlinearity,high dynamics,and difficult accurate prediction in the subway foundation pit settlement process,this paper proposed a settlement prediction model based on wavelet decomposition and tem-poral convolutional network(TCN)-back propagation(BP).First,wavelet decomposition was used to separate the vertical displacement data into low-frequency and high-frequency components.Second,TCN and BP neural networks were used separately for prediction.Finally,the results were fused to obtain the comprehensive settlement prediction values.Verified by the measured data from a subway foundation pit project,the model achieved an average absolute error of 0.44 mm,which was significantly lower than the 0.60 mm of the wavelet-optimized BP neural network and the 1.36 mm of the autoregressive integrated moving average(ARIMA)model.The results indicate that the proposed model can effectively characterize the complex dynamic features of foundation pit settlement,with high prediction accuracy and strong reliability,and it provides a more effective technical means for foundation pit safety monitoring.