Time series analysis and gated recurrent neural network model for predicting landslide displacements

ABSTRACT

An effiecient landslide displacement prediction is important for early warning system of landslides. Based on time series method, the cumulative deformation of a landslide is decomposed into periodic and trend ones. A cubic polynomial is employed to forecast the trend deformation. Considering the periodic changes in rainwater and reservoir levels, the proposed model combines a convolutional neural network (CNN) with a gated recurrent unit (GRU) neural network to forecast periodic deformations. CNN effectively identifies the characteristics of the raw data, and GRU automatically controls the impact of historical information by adjusting the weights of the reset and update gates. C-GRU performance in predicting the periodic displacement is compared with GRU, and a backpropagation neural network optimised using particle swarm optimisation (PSO-BP). Monitoring points of Baishuihe landslide are employed to compare the performance of the various models. The findings show that the proposed model has strong data mining performance and deals with time series data efficiently. The new model can incorporate historical information more effectively than PSO-BP. Compared with GRU, the proposed model better captures the input data characteristics and improves the prediction accuracy. C-GRU achieves a low mean square error, representing a significant improvement in the accuracy of landslide predictions.

KEYWORDS:

• Time series
• landslide displacement prediction
• gated recurrent unit
• convolutional neural network

Disclosure statement

No potential conflict of interest was reported by the author(s).

Funding

The research reported in this paper was supported by the National Natural Science Foundation of China (U20A20314 and 42277183), the Chongqing Natural Science Foundation of China (cstc2020jcyj-jqX0006, cstc2022ycjh-bgzxm0086), and the Fundamental Research Funds for the Central Universities, CHD (300102262507).

Data availability

All data generated or analyzed during this study are included within the article.