A novel VMD-LHPO-KELM machine learning-based TBM boring parameter prediction.
Saved in:
| Title: | A novel VMD-LHPO-KELM machine learning-based TBM boring parameter prediction. |
|---|---|
| Authors: | Lu, Zhipeng1 (AUTHOR), Shi, Kebin1 (AUTHOR) xndsg@sina.com |
| Source: | Earth Science Informatics. Sep2023, Vol. 16 Issue 3, p2925-2938. 14p. |
| Subject Terms: | *Tunnel design & construction, *Penetration mechanics, *Engineering models, *Machine learning, *Optimization algorithms, *Goodness-of-fit tests, *Statistical correlation |
| Abstract: | Scientific and reasonable prediction of tunneling parameters is the premise to ensure the safety of the project. To further improve the accuracy and reliability of the prediction of tunneling parameters. To address the shortcomings of the kernel limit learning machine parameter selection affecting the prediction ability and the characteristics of volatility and non-stationarity of load data, a prediction model based on variational modal decomposition with a hunter-prey algorithm optimized for the kernel limit learning machine improved by the population chaos strategy and the Levy flight strategy is proposed. Relying on the Xinjiang YEGS tunnel project, 6,900 tunneling data sets were selected after data processing, and the data sets were divided according to the construction sequence. Grey correlation analysis was used to select four dimensions of cutter torque, cutter speed, penetration degree, and total thrust to predict the tunneling speed of full-section tunnel boring machines. The results show that under the same conditions, the VMD-LHPO-KELM model shows a 385% and 28% improvement in RMSE value simulation and 50% and 47.7% improvement in MAPE simulation, respectively, compared to the whale algorithm and the unimproved hunter-prey algorithm seeking algorithm, showing a better iteration rate. Compared with the most commonly used neural network VMD-LSTM model in engineering simulation, the goodness of fit coefficient is increased by 9.2%. The established VMD-LHPO-KELM model can better restore the real environmental impact of TBM construction. [ABSTRACT FROM AUTHOR] |
| Database: | Energy & Power Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | Scientific and reasonable prediction of tunneling parameters is the premise to ensure the safety of the project. To further improve the accuracy and reliability of the prediction of tunneling parameters. To address the shortcomings of the kernel limit learning machine parameter selection affecting the prediction ability and the characteristics of volatility and non-stationarity of load data, a prediction model based on variational modal decomposition with a hunter-prey algorithm optimized for the kernel limit learning machine improved by the population chaos strategy and the Levy flight strategy is proposed. Relying on the Xinjiang YEGS tunnel project, 6,900 tunneling data sets were selected after data processing, and the data sets were divided according to the construction sequence. Grey correlation analysis was used to select four dimensions of cutter torque, cutter speed, penetration degree, and total thrust to predict the tunneling speed of full-section tunnel boring machines. The results show that under the same conditions, the VMD-LHPO-KELM model shows a 385% and 28% improvement in RMSE value simulation and 50% and 47.7% improvement in MAPE simulation, respectively, compared to the whale algorithm and the unimproved hunter-prey algorithm seeking algorithm, showing a better iteration rate. Compared with the most commonly used neural network VMD-LSTM model in engineering simulation, the goodness of fit coefficient is increased by 9.2%. The established VMD-LHPO-KELM model can better restore the real environmental impact of TBM construction. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 18650473 |
| DOI: | 10.1007/s12145-023-01043-2 |
