Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Research on dam foundation deformation prediction based on VMD optimized temporal convolutional network
The deformation behavior of the dam foundation directly determines the overall safety of the dam body, especially in the impoundment period and early operation stage of the dam, so it is particularly important to predict and study the displacement of the dam foundation. In order to further improve the accuracy and reliability of dam foundation deformation prediction, VMD (Variational Mode Decomposition) is used to decompose the displacement monitoring time series, and the trend component of displacement and induced component of displacement are obtained. Through the analysis method of time-lag cross-correlation, the dynamic MIC (Maximal information coefficient) between input and output sequences is calculated, the problem of time-lag of displacement monitoring data in the learning and training process is solved, and the length of input time-series variables is determined. Through the iterative learning and training of the trend component and the induced component by TCN (Temporal Convolutional Network), the fitting model functions of the two are obtained respectively, and then the prediction components of the two can be obtained. The classification sum of the two is the final displacement prediction value. The results of case study showed that VMD-TCN prediction model has higher prediction accuracy of dam foundation displacement than ARIMA (Auto Regressive Integrated Moving Average) prediction model, and can better evaluate the future evolution state of dam foundation deformation.
Research on dam foundation deformation prediction based on VMD optimized temporal convolutional network
The deformation behavior of the dam foundation directly determines the overall safety of the dam body, especially in the impoundment period and early operation stage of the dam, so it is particularly important to predict and study the displacement of the dam foundation. In order to further improve the accuracy and reliability of dam foundation deformation prediction, VMD (Variational Mode Decomposition) is used to decompose the displacement monitoring time series, and the trend component of displacement and induced component of displacement are obtained. Through the analysis method of time-lag cross-correlation, the dynamic MIC (Maximal information coefficient) between input and output sequences is calculated, the problem of time-lag of displacement monitoring data in the learning and training process is solved, and the length of input time-series variables is determined. Through the iterative learning and training of the trend component and the induced component by TCN (Temporal Convolutional Network), the fitting model functions of the two are obtained respectively, and then the prediction components of the two can be obtained. The classification sum of the two is the final displacement prediction value. The results of case study showed that VMD-TCN prediction model has higher prediction accuracy of dam foundation displacement than ARIMA (Auto Regressive Integrated Moving Average) prediction model, and can better evaluate the future evolution state of dam foundation deformation.
Research on dam foundation deformation prediction based on VMD optimized temporal convolutional network
Yang, Jun (Autor:in) / Wangjia (Autor:in) / Lv, Zhibin (Autor:in)
06.11.2021
534216 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
A Spatio-Temporal Graph Convolutional Network for Air Quality Prediction
| DOAJ | 2023