Although machine learning models have achieved high enough accuracy in predicting shield position deviations,their“black box”nature makes the prediction mechanisms and decision-making processes opaque,leading to wea...Although machine learning models have achieved high enough accuracy in predicting shield position deviations,their“black box”nature makes the prediction mechanisms and decision-making processes opaque,leading to weaker explanations and practicability.This study introduces a novel explainable deep learning framework comprising the Informer model with enhanced attention mechanisms(EAMInfor)and deep learning important features(DeepLIFT),aimed at improving the prediction accuracy of shield position deviations and providing interpretability for predictive results.The EAMInfor model attempts to integrate channel attention,spatial attention,and simple attention modules to improve the Informer model's performance.The framework is tested with the four different geological conditions datasets generated from the Xiamen metro line 3,China.Results show that the EAMInfor model outperforms the traditional Informer and comparison models.The analysis with the DeepLIFT method indicates that the push thrust of push cylinder and the earth chamber pressure are the most significant features,while the stroke length of the push cylinder demonstrated lower importance.Furthermore,the variation trends in the significance of data points within input sequences exhibit substantial differences between single and composite strata.This framework not only improves predictive accuracy but also strengthens the credibility and reliability of the results.展开更多
The presented research introduces a novel hybrid deep learning approach for the dynamic prediction of the attitude and position of super-large diameter shields-a critical consideration for construction safety and tunn...The presented research introduces a novel hybrid deep learning approach for the dynamic prediction of the attitude and position of super-large diameter shields-a critical consideration for construction safety and tunnel lining quality.This study proposes a hybrid deep learning approach for predicting dynamic attitude and position prediction of super-large diameter shield.The approach consists of principal component analysis(PCA)and temporal convolutional network(TCN).The former is used for employing feature level fusion based on features of the shield data to reduce uncertainty,improve accuracy and the data effect,and 9 sets of required principal component characteristic data are obtained.The latter is adopted to process sequence data in predicting the dynamic attitude and position for the advantages and potential of convolution network.The approach’s effectiveness is exemplified using data from a tunnel construction project in China.The obtained results show remarkable accuracy in predicting the global attitude and position,with an average error ratio of less than 2 mm on four shield outputs in 97.30%of cases.Moreover,the approach displays strong performance in accurately predicting sudden fluctuations in shield attitude and position,with an average prediction accuracy of 89.68%.The proposed hybrid model demonstrates superiority over TCN,long short-term memory(LSTM),and recurrent neural network(RNN)in multiple indexes.Shapley additive exPlanations(SHAP)analysis is also performed to investigate the significance of different data features in the prediction process.This study provides a real-time warning for the shield driver to adjust the attitude and position of super-large diameter shields.展开更多
基金supported by the National Natural Science Foundation of China(Grant Nos.52378392,52408356)the Foal Eagle Program Youth Top-notch Talent Project of Fujian Province,China(Grant No.00387088).
文摘Although machine learning models have achieved high enough accuracy in predicting shield position deviations,their“black box”nature makes the prediction mechanisms and decision-making processes opaque,leading to weaker explanations and practicability.This study introduces a novel explainable deep learning framework comprising the Informer model with enhanced attention mechanisms(EAMInfor)and deep learning important features(DeepLIFT),aimed at improving the prediction accuracy of shield position deviations and providing interpretability for predictive results.The EAMInfor model attempts to integrate channel attention,spatial attention,and simple attention modules to improve the Informer model's performance.The framework is tested with the four different geological conditions datasets generated from the Xiamen metro line 3,China.Results show that the EAMInfor model outperforms the traditional Informer and comparison models.The analysis with the DeepLIFT method indicates that the push thrust of push cylinder and the earth chamber pressure are the most significant features,while the stroke length of the push cylinder demonstrated lower importance.Furthermore,the variation trends in the significance of data points within input sequences exhibit substantial differences between single and composite strata.This framework not only improves predictive accuracy but also strengthens the credibility and reliability of the results.
基金financially supported by the National Natural Science Foundation of China(Grant Nos.52078304,51938008,52090084,and 52208354)Guangdong Province Key Field R&D Program Project(Grant Nos.2019B111108001 and 2022B0101070001)+1 种基金Shenzhen Fundamental Research(Grant No.20220525163716003)the Pearl River Delta Water Resources Allocation Project(CD88-GC022020-0038).
文摘The presented research introduces a novel hybrid deep learning approach for the dynamic prediction of the attitude and position of super-large diameter shields-a critical consideration for construction safety and tunnel lining quality.This study proposes a hybrid deep learning approach for predicting dynamic attitude and position prediction of super-large diameter shield.The approach consists of principal component analysis(PCA)and temporal convolutional network(TCN).The former is used for employing feature level fusion based on features of the shield data to reduce uncertainty,improve accuracy and the data effect,and 9 sets of required principal component characteristic data are obtained.The latter is adopted to process sequence data in predicting the dynamic attitude and position for the advantages and potential of convolution network.The approach’s effectiveness is exemplified using data from a tunnel construction project in China.The obtained results show remarkable accuracy in predicting the global attitude and position,with an average error ratio of less than 2 mm on four shield outputs in 97.30%of cases.Moreover,the approach displays strong performance in accurately predicting sudden fluctuations in shield attitude and position,with an average prediction accuracy of 89.68%.The proposed hybrid model demonstrates superiority over TCN,long short-term memory(LSTM),and recurrent neural network(RNN)in multiple indexes.Shapley additive exPlanations(SHAP)analysis is also performed to investigate the significance of different data features in the prediction process.This study provides a real-time warning for the shield driver to adjust the attitude and position of super-large diameter shields.
