Landslide displacement prediction based on multi-source data fusion and sensitivity states

Liu, Yong; Xu, Chang; Huang, Biao; Ren, Xingwei; Liu, Chuanqi; Hu, Baodan; Chen, Zhe

doi:10.1016/j.enggeo.2020.105608

Cited by 72 publications

(34 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Chen et al [7,8] considered twelve landslide-related parameters for landslide susceptibility mapping, including slope angle, slope aspect, plan curvature, profile curvature, altitude, land use, distance to faults, distance to roads, distance to rivers, lithology, and rainfall. Liu et al [9,10] obtained the distribution of the geological strength index (GSI) using geostatistics-based methods to determine the spatial variability of the mechanical parameters. e mechanical parameters of a rock mass in mining engineering can be characterized by spatial variability and time decay, and these play an important role in slope stability analyses.…”

Section: Introductionmentioning

confidence: 99%

The Formation Mechanism of Surface Landslide Disasters in the Mining Area under Different Slope Angles

Zhao

Zhang

et al. 2021

Advances in Civil Engineering

View full text Add to dashboard Cite

Slope stability analysis is important for the safe mining of mineral resources. The collapse of goafs in loess gullies can lead to natural disasters such as surface landslides. In this context, this study analyzes monitoring data obtained from surface observation in the Shendong mining area of the Hanjiawan coal mine based on the geological conditions therein. The monitoring results show that the working face experiences a starting period, an active period, and a declining period, from the start of mining to the end of the working face. At the initial mining stage, there is no evident surface movement or deformation in the mining area. When the advance distance of the 12106 working face is between 13 m and 109 m, the surface movement and deformation vary significantly, and the maximum subsidence reaches 1963 mm, which is enough to cause landslides. We select the physical and mechanical parameters of the rock and soil in the mine and then simulate the formation mechanism of surface landslides under different slope angles of the mining area using FLAC3D software. Because of the collapse of the mined-out area, the overlying strata structure is destroyed, the subsidence basin is shifted to the center as a whole, and the slope mass is subjected to tensile and compression deformation, resulting in plastic damage, which develops downward along the crack and leads to a collapse because of the discontinuous movement and deformation of the surface; moreover, step-type ground fissures are produced. The results also show that when the slope angle is greater than 60°, the displacement of the slope mass is not uniform, and the rock stratum in a position with large displacement loses its support, leading to landslides; when the slope angle is less than 30°, the bedrock surface forms a sliding surface and develops to the surface, thus decreasing the possibility of landslides. Based on the stability analysis of the collapsed slope in the goaf of the loess gully, a scientific basis is provided for the effective prevention and control of geological disasters in the Shendong mining area.

show abstract

Section: Introductionmentioning

confidence: 99%

The Formation Mechanism of Surface Landslide Disasters in the Mining Area under Different Slope Angles

Zhao

Zhang

et al. 2021

Advances in Civil Engineering

View full text Add to dashboard Cite

show abstract

“…From the perspective of prediction model establishment, the development of landslide 95 prediction models has experienced a leap from empirical models, semiempirical models, and 96 mathematical-statistical models to nonlinear neural network models in recent decades (Saito, 1965;97 Fukuzono, 1985;Li et al, 2012;Du et al, 2013; Intrieri et al, 2019;Liu et al, 2020). Especially 98 with the rise of artificial intelligence technologies, various neural network models have been 99…”

Section: Introductionmentioning

confidence: 99%

Adaptive Hybrid Machine Learning Model For Forecasting The Step-Like Displacements of Reservoir Colluvial Landslides: A Case Study in The Three Gorges Reservoir Area, China

Miao

et al. 2021

Preprint

View full text Add to dashboard Cite

Constructing an accurate and stable displacement prediction model is essential to build a capable early warning system for landslide disasters. To overcome the drawbacks of previous displacement prediction models for step-like landslides, such as the incomplete or excessive decompositions of cumulative displacements and input factors and the redundancy or lack of input factors, we propose an adaptive hybrid machine learning model. This model is composed of three parts. First, candidate factors are proposed based on the macroscopic deformation response of landslides. Then, the landslide displacement and its candidate factors are adaptively decomposed into different displacement and factor components by applying optimized variational mode decomposition (OVMD). Second, in the gray wolf optimizer-based kernel extreme learning machine (GWO-KELM) model, the global sensitivity analysis (GSA) of the prediction results of different displacement components to each decomposed factor is analyzed based on the PAWN method. Then, the decomposed factors are reduced according to the GSA results. Third, based on the reduced factors, the optimal GWO-KELM models of the different displacement components are established to predict the displacement. Taking the Baishuihe landslide as an example, we used the raw data of three representative monitoring sites from June 2006 to December 2016 to verify the validity, accuracy, and stability of the model. The results indicate that the proposed hybrid model can effectively determine the displacement decomposition parameters. In addition, this model performed well over a three-year forecast with low model complexity.

show abstract

“…Nowadays, with the development of high-speed computers, various machine learning models have been widely used in landslide displacement prediction, and many breakthroughs have been achieved. At present, the commonly used models for the displacement prediction include Back Propagation Neural Network (BPNN) (Chen and Zeng, 2013;Fu et al, 2021;Zhang et al, 2021), Support Vector Regression (SVR) (Liu et al, 2020;Dong et al, 2021), Extreme Learning Machine (ELM) , Evaluating Machine Learning (EML) (Goetz et al, 2015), Kernel Extreme Learning Machine (KELM) (Zhou et al, 2018;Li et al, 2021), Long Short-Term Memory (LSTM) (Xu et al, 2018;Yang et al, 2019), and so on. And many algorithms are used to optimize parameters for the prediction models, including Genetic Algorithm (GA) (Li and Kong, 2014), Grid Search algorithm (GS) (Miao et al, 2018b), Particle Swarm Optimization (PSO) (Zhou et al, 2016), Grey Wolf Optimizer (GWO) (Guo et al, 2019), Fruit Fly Optimization Algorithm (FOA) (Wang et al, 2019), and so on.…”

Section: Introductionmentioning

confidence: 99%

An Integrated Approach for Displacement Prediction of Landslides Based on the Data Mining and VMD-FOA-BPNN Model

Miao

Xie

et al. 2021

Preprint

View full text Add to dashboard Cite

Landslide prediction is important for mitigating geohazards but is very challenging. The fluctuation of reservoir water level and rainfall are the main external triggering factors controlling the deformation of riverine landslides. In this paper, the Baishuihe landslide in the Three Gorges Reservoir area, which has a typical “step-like” behavior, is taken as the study case, and an integrated approach for landslide displacement prediction combining data mining and Variational Mode Decomposition, Fruit Fly Optimization Algorithm, Back Propagation Neural Network (VMD-FOA-BPNN) is proposed. Nine triggering factors including the reservoir level and rainfall are extracted. First, triggering factors and monthly velocity of the landslide are clustered into several categories by Two-step Clustering (TSC). Then, Apriori algorithm is used to mine the association rules between triggering factors and monthly velocity, and comprehensive contribution of each factor is calculated based on the data mining results. Next, the displacement of monitoring point ZG93 and triggering factors are decomposed by VMD based on the time series analysis of the landslide. Last, the trend term displacement is trained and predicted by the subsection functions, and FOA-BPNN models are used to train and predict the periodic and random term. The prediction results show that, compared with the current popular prediction models, the proposed model can effectively improve the prediction accuracy, which has high practicability and application value in the study of landslide displacement prediction.

show abstract

Landslide displacement prediction based on multi-source data fusion and sensitivity states

Cited by 72 publications

References 32 publications

The Formation Mechanism of Surface Landslide Disasters in the Mining Area under Different Slope Angles

The Formation Mechanism of Surface Landslide Disasters in the Mining Area under Different Slope Angles

Adaptive Hybrid Machine Learning Model For Forecasting The Step-Like Displacements of Reservoir Colluvial Landslides: A Case Study in The Three Gorges Reservoir Area, China

An Integrated Approach for Displacement Prediction of Landslides Based on the Data Mining and VMD-FOA-BPNN Model

Contact Info

Product

Resources

About