Study on Prediction of Tailings Dam Deformation Based on MHA-BiLSTM
Tailings dam deformation is influenced by multiple factors.In response to the limitations of traditional prediction methods due to data complexity and non-linear relationships,which result in inadequate prediction accuracy,a method combining Multi-Head Attention mechanism and Bidirectional Long Short-Term Memory(BiLSTM)is proposed for predicting tailings dam displacement.In the prediction process,the original data is first processed using Z-score and Savitzky-Golay filtering techniques to eliminate disturbances caused by outliers and noise.Subsequently,the Grey Relational Analysis method is utilized to determine the factors influencing dam displacement.Finally,the MHA-BiLSTM model is employed to predict the dam displacement.Taking the measured data of a tailings pond in Liaoning Province as an example,the performance of the proposed model is compared with the traditional BiLSTM model.The results show that this method can predict the displacement of dam more accurately.
NETL
NSTL
万方数据
