A COMBINED MODEL FOR PREDICTING LANDSLIDE DISPLACEMENT
Keywords:
Landslide displacement, landslide monitoring, neural network forecasting, Mailuu-Suu, continuum mechanics, hydrogen factorAbstract
This article discusses a deterministic-stochastic model for predicting landslide displacements, based on decomposing the total deformation vector into a long-term trend and a periodic component. Using the example of monitoring the Sary-Be landslide (Mailuu-Suu), the authors propose a method for adapting a combined model that accounts for the influence of climatic, hydrogeological, and seismic factors. To forecast the periodic component, an artificial neural network (MLP 9-4-1) with a backpropagation algorithm was applied. The results of the study show that the use of multidimensional neural network analysis allows for a satisfactory approximation of displacement dynamics; however, to improve forecast accuracy under the non-stationary regime of “jerky” deformations, it is necessary to continuously retrain the network using current monitoring data.
References
1. Du Y., Lacasse S. Displacement prediction in colluvial landslides// Landslides. Original papper. 2012.
2. Gruden D.M., Masumzadeh S. Accelerating creep of the slopes of a coal mine//Rock Mech. Eng. 1987. №20. P. 123-135.
3. Kawamura K. Methodology for landslide prediction//Proceeding of 11 th international conference on soil mechanics und foundation engineering. San-Francisco. 1985 p. 1155-1158.
4. Li T., Chen M., Wang L., Zhou Y. Time prediction of landslide using Verhulst inverse function model//Proceeding 7th Jnt. Symp. On Landslides. Trondheim. 1996. V2. P. 1289-1293
5. Mayoraz F., Vulliet L. Neural networks for slope movement prediction// Jnt. J. Geomech. 2002. №2. P. 153-173.
6. Алешин Ю.Г. Алгоритмизация оползневого прогнозирования на основе нейросетевогомоделирования// Известия НАН КР 2010. №2. С. 19-27.
7. Васильев В.И. Распознающие системы. Киев: Науковая думка. 1983. 422с.
