Tunnel settlement time sequence prediction method based on CEEMDAN-BiLSTM

Abstract

The invention discloses a tunnel settlement time sequence prediction method based on CEEMDAN-BiLSTM, and the method comprises the following steps: collecting ground surface settlement data above a tunnel, sequentially recording the settlement value of each collection point according to the time sequence, obtaining a single-dimensional tunnel settlement time sequence, and carrying out the preprocessing of the single-dimensional tunnel settlement time sequence; carrying out complete ensemble empirical mode decomposition of adaptive noise on the preprocessed tunnel settlement time sequence to obtain n stable intrinsic mode function IMF1-IMFn components with different scales and a residual error Res component; determining a time scale, reconstructing the decomposed IMF components and errors Res, normalizing unified dimensions, and determining a training set and a test set; establishing a bidirectional long-term and short-term memory network prediction model for each training set and each test set, and predicting a tunnel settlement subsequence; and reversely normalizing the tunnel settlement sub-sequences under different scales, superposing the tunnel settlement sub-sequences to obtaina final tunnel settlement result, and evaluating the prediction effect and stability of the model according to different evaluation indexes.

Description

technical field [0001] The invention relates to a tunnel settlement prediction method, in particular to a tunnel settlement time series prediction method based on CEEMDAN-BiLSTM. Background technique [0002] Tunnel settlement not only affects the development of urban rail transit, but also poses a great threat to the safety of life and property of urban residents. Therefore, it is of great significance to accurately predict the settlement of tunnels. Scholars at home and abroad have done a lot of research on tunnel settlement prediction. Research methods can be roughly divided into two categories: theoretical calculation empirical method and measured data analysis method. The empirical method of theoretical calculation is represented by the Peck empirical formula method, including numerical analysis method, numerical simulation method, semi-theoretical analysis method and stochastic theoretical model. [0003] Measured data analysis methods are divided into statistical m...

AI Technical Summary

Problems solved by technology

[0007] In view of this, the present invention proposes a fully ensemble empirical mode decomposition (CEEMDAN) based on adaptive noise and a two-way long-short-term memory network in the case of strong random time series single-dimensional data and complex nonlinear regression problems (BiLSTM) tunnel settlement time series prediction method is used to solve the problem of low prediction accuracy in the above circumstances existing in the prior art

Images