Forecasting of Landslide Stability Based on Gradient Boosting Decision Tree Model

Fei Niu ； Lianwu Chen

《International Core Journal of Engineering》 5卷11期 (2019/10) Pp. 42-48

https://doi.org/10.6919/ICJE.201910_5(11).0008

摘要

In order to forecast landslide stability, an ensemble learning method in machine learning, Gradient Boosting Decision Tree, was used to forecast landslide stability. Combined with the collected landslide data, six parameters, including unit weight of landslide material, cohesion force, internal angle of friction, landslide angle, landslide height, pore pressure ratio, were utilized as the input parameters, while factor of safety was used the output parameter, establishing prediction model. The influence of the main parameters of the GBDT algorithm on the training results was analyzed, and the better parameters were selected. The prediction results were compared with the support vector machine regression and BP neural network. The results show that the prediction results of the Gradient Boosting Decision Tree model are reliable, the average relative error is 6.53%. The prediction accuracy is high and the feasibility is strong. As a means of landslide stability prediction, its application prospect is good.

關鍵字

Ensemble learning ； GBDT ； Landslide stability ； Parameter selection

參考文獻

Y. H. Zhao, G. Liu, et al. Sensitivity analysis for the stability of loess landslide Based on grey correlation degree, Journal of Yangtze River Scientific Research Institute, Vol. 32 (2015) No. 7, p.94-98.

C. Guo, Q. Xu, Y. Wei, et al. Analysis of Propagation and Deposit Characteristics and Liquidity of Jiangliu 4# Landslide in Southern Jingyang Plateau, Shaanxi Province, Science Technology and Engineering,Vol. 17 (2017) No.27, p.15-25.

L.B. Shao, H. Ma, T. X. Wen. Study on landslide stability prediction based on RF－ELM model, Journal of Safety Science and Technology, Vol. 11 (2015) No.03, p.93-98.

S. Su, Y.C. Song, L. B. Yan. Application of Gaussian process machine learning to landslide stability evaluation, Rock and Soil Mechanics, Vol. 30 (2009) No.03, p.675-679.

J .M. Yang, C. Gao, Z. Y. Qu, et al. Random Forest Classification Algorithm Based on Cost-sensitive for Imbalanced Data, Science Technology and Engineering, Vol. 18 (2018) No.06, p.285-290.