Displacement prediction of Jiuxianping landslide using gated recurrent unit (GRU) networks
Citation
Zhang, W., Li, H., Tang, L., et al. (2022). Displacement prediction of Jiuxianping landslide using gated recurrent unit (GRU) networks. Acta Geotechnica, 17: 1367-1382. Link to paper
Abstract
This study applies gated recurrent unit (GRU) networks to displacement prediction for the Jiuxianping reservoir landslide in Chongqing, China. The authors compare GRU against ANN, random forest regression, and multivariate adaptive regression splines, showing that static models struggle to reproduce local peaks in testing data and cannot fully capture deformation dynamics. By using sequential historical information, the GRU model better represents periodic displacement evolution with fewer outliers and improved generalization. Results support dynamic recurrent models as effective tools for landslide deformation forecasting and early-warning workflows.