Dynamic intelligent prediction approach for landslide displacement based on biological growth models and CNN-LSTM

Influenced by complex external factors,the displacement-time curve of reservoir landslides demonstrates both short-term and long-term diversity and dynamic complexity.It is difficult for existing methods,including Regression models and Neural network models,to perform multi-characteristic coupled displacement prediction because they fail to consider landslide creep characteristics.This paper integrates the creep characteristics of landslides with non-linear intelligent algorithms and proposes a dynamic intelligent landslide displacement prediction method based on a combination of the Biological Growth model(BG),Convolutional Neural Network(CNN),and Long ShortTerm Memory Network(LSTM).This prediction approach improves three different biological growth models,thereby effectively extracting landslide creep characteristic parameters.Simultaneously,it integrates external factors(rainfall and reservoir water level)to construct an internal and external comprehensive dataset for data augmentation,which is input into the improved CNN-LSTM model.Thereafter,harnessing the robust feature extraction capabilities and spatial translation invariance of CNN,the model autonomously captures short-term local fluctuation characteristics of landslide displacement,and combines LSTM's efficient handling of long-term nonlinear temporal data to improve prediction performance.An evaluation of the Liangshuijing landslide in the Three Gorges Reservoir Area indicates that BG-CNN-LSTM exhibits high prediction accuracy,excellent generalization capabilities when dealing with various types of landslides.The research provides an innovative approach to achieving the whole-process,realtime,high-precision displacement predictions for multicharacteristic coupled landslides.

Forecasting step-like landslide displacement through diverse monitoring frequencies

The precision of landslide displacement prediction is crucial for effective landslide prevention and mitigation strategies.However,the role of surface monitoring frequency in influencing prediction accuracy has been largely neglected.This study examined the effect of varying monitoring frequencies on the accuracy of displacement predictions by using the Baijiabao landslide in the Three Gorges Reservoir Area(TGRA)as a case study.We collected surface automatic monitoring data at different intervals,ranging from daily to monthly.The Ensemble Empirical Mode Decomposition(EEMD)algorithm was utilized to dissect the accumulated displacements into periodic and trend components at each monitoring frequency.Polynomial fitting was applied to forecast the trend component while the periodic component was predicted with two state-of-the-art neural network models:Long Short-Term Memory(LSTM)and Gated Recurrent Unit(GRU).The predictions from these models were integrated to derive cumulative displacement forecasts,enabling a comparative analysis of prediction accuracy across different monitoring frequencies.The results demonstrate that the proposed models achieve high accuracy in landslide displacement forecasting,with optimal performance observed at moderate monitoring intervals.Intriguingly,the daily mean average error(MAE)decreases sharply with increasing monitoring frequency,reaching a plateau.These findings were corroborated by a parallel analysis of the Bazimen landslide,suggesting that moderate monitoring intervals of approximately 7 to 15 days are most conducive to achieving enhanced prediction accuracy compared to both daily and monthly intervals.

Data-augmented landslide displacement prediction using generative adversarial network

Landslides are destructive natural disasters that cause catastrophic damage and loss of life worldwide.Accurately predicting landslide displacement enables effective early warning and risk management.However,the limited availability of on-site measurement data has been a substantial obstacle in developing data-driven models,such as state-of-the-art machine learning(ML)models.To address these challenges,this study proposes a data augmentation framework that uses generative adversarial networks(GANs),a recent advance in generative artificial intelligence(AI),to improve the accuracy of landslide displacement prediction.The framework provides effective data augmentation to enhance limited datasets.A recurrent GAN model,RGAN-LS,is proposed,specifically designed to generate realistic synthetic multivariate time series that mimics the characteristics of real landslide on-site measurement data.A customized moment-matching loss is incorporated in addition to the adversarial loss in GAN during the training of RGAN-LS to capture the temporal dynamics and correlations in real time series data.Then,the synthetic data generated by RGAN-LS is used to enhance the training of long short-term memory(LSTM)networks and particle swarm optimization-support vector machine(PSO-SVM)models for landslide displacement prediction tasks.Results on two landslides in the Three Gorges Reservoir(TGR)region show a significant improvement in LSTM model prediction performance when trained on augmented data.For instance,in the case of the Baishuihe landslide,the average root mean square error(RMSE)increases by 16.11%,and the mean absolute error(MAE)by 17.59%.More importantly,the model’s responsiveness during mutational stages is enhanced for early warning purposes.However,the results have shown that the static PSO-SVM model only sees marginal gains compared to recurrent models such as LSTM.Further analysis indicates that an optimal synthetic-to-real data ratio(50%on the illustration cases)maximizes the improvements.This also demonstrates the robustness and effectiveness of supplementing training data for dynamic models to obtain better results.By using the powerful generative AI approach,RGAN-LS can generate high-fidelity synthetic landslide data.This is critical for improving the performance of advanced ML models in predicting landslide displacement,particularly when there are limited training data.Additionally,this approach has the potential to expand the use of generative AI in geohazard risk management and other research areas.

Short-term displacement prediction for newly established monitoring slopes based on transfer learning

This study makes a significant progress in addressing the challenges of short-term slope displacement prediction in the Universal Landslide Monitoring Program,an unprecedented disaster mitigation program in China,where lots of newly established monitoring slopes lack sufficient historical deformation data,making it difficult to extract deformation patterns and provide effective predictions which plays a crucial role in the early warning and forecasting of landslide hazards.A slope displacement prediction method based on transfer learning is therefore proposed.Initially,the method transfers the deformation patterns learned from slopes with relatively rich deformation data by a pre-trained model based on a multi-slope integrated dataset to newly established monitoring slopes with limited or even no useful data,thus enabling rapid and efficient predictions for these slopes.Subsequently,as time goes on and monitoring data accumulates,fine-tuning of the pre-trained model for individual slopes can further improve prediction accuracy,enabling continuous optimization of prediction results.A case study indicates that,after being trained on a multi-slope integrated dataset,the TCN-Transformer model can efficiently serve as a pretrained model for displacement prediction at newly established monitoring slopes.The three-day average RMSE is significantly reduced by 34.6%compared to models trained only on individual slope data,and it also successfully predicts the majority of deformation peaks.The fine-tuned model based on accumulated data on the target newly established monitoring slope further reduced the three-day RMSE by 37.2%,demonstrating a considerable predictive accuracy.In conclusion,taking advantage of transfer learning,the proposed slope displacement prediction method effectively utilizes the available data,which enables the rapid deployment and continual refinement of displacement predictions on newly established monitoring slopes.

Displacement characteristics and prediction of Baishuihe landslide in the Three Gorges Reservoir

In order to reach the designated final water level of 175 m, there were three impoundment stages in the Three Gorges Reservoir, with water levels of 135 m, 156 m and 175 m. Baishuihe landslide in the Reservoir was chosen to analyze its displacement characteristics and displacement variability at the different stages. Based on monitoring data, the landslide displacement was mainly influenced by rainfall and drawdown of the reservoir water level. However, the magnitude of the rise and drawdown of the water level after the reservoir water level reached 175 m did not accelerate landslide displacement. The prediction of landslide displacement for active landslides is very important for landslide risk management. The time series of cumulative displacement was divided into a trend term and a periodic term using the Hodrick-Prescott(HP) filter method. The polynomial model was used to predict the trend term. The extreme learning machine(ELM) and least squares support vector machine(LS-SVM) were chosen to predict theperiodic term. In the prediction model for the periodic term, input variables based on the effects of rainfall and reservoir water level in landslide displacement were selected using grey relational analysis. Based on the results, the prediction precision of ELM is better than that of LS-SVM for predicting landslide displacement. The method for predicting landslide displacement could be applied by relevant authorities in making landslide emergency plans in the future.

Prediction of landslide displacement with dynamic features using intelligent approaches

Landslide displacement prediction can enhance the efficacy of landslide monitoring system,and the prediction of the periodic displacement is particularly challenging.In the previous studies,static regression models(e.g.,support vector machine(SVM))were mostly used for predicting the periodic displacement.These models may have bad performances,when the dynamic features of landslide triggers are incorporated.This paper proposes a method for predicting the landslide displacement in a dynamic manner,based on the gated recurrent unit(GRU)neural network and complete ensemble empirical decomposition with adaptive noise(CEEMDAN).The CEEMDAN is used to decompose the training data,and the GRU is subsequently used for predicting the periodic displacement.Implementation procedures of the proposed method were illustrated by a case study in the Caojiatuo landslide area,and SVM was also adopted for the periodic displacement prediction.This case study shows that the predictors obtained by SVM are inaccurate,as the landslide displacement is in a pronouncedly step-wise manner.By contrast,the accuracy can be significantly improved using the dynamic predictive method.This paper reveals the significance of capturing the dynamic features of the inputs in the training process,when the machine learning models are adopted to predict the landslide displacement.

Landslide displacement prediction based on the ICEEMDAN,ApEn and the CNN-LSTM models

Landslide deformation is affected by its geological conditions and many environmental factors.So it has the characteristics of dynamic,nonlinear and unstable,which makes the prediction of landslide displacement difficult.In view of the above problems,this paper proposes a dynamic prediction model of landslide displacement based on the improvement of complete ensemble empirical mode decomposition with adaptive noise(ICEEMDAN),approximate entropy(ApEn)and convolution long short-term memory(CNN-LSTM)neural network.Firstly,ICEEMDAN and Ap En are used to decompose the cumulative displacements into trend,periodic and random displacements.Then,the least square quintic polynomial function is used to fit the displacement of trend term,and the CNN-LSTM is used to predict the displacement of periodic term and random term.Finally,the displacement prediction results of trend term,periodic term and random term are superimposed to obtain the cumulative displacement prediction value.The proposed model has been verified in Bazimen landslide in the Three Gorges Reservoir area of China.The experimental results show that the model proposed in this paper can more effectively predict the displacement changes of landslides.As compared with long short-term memory(LSTM)neural network,gated recurrent unit(GRU)network model and back propagation(BP)neural network,CNN-LSTM neural network had higher prediction accuracy in predicting the periodic displacement,with the mean absolute percentage error(MAPE)reduced by 3.621%,6.893% and 15.886% respectively,and the root mean square error(RMSE)reduced by 3.834 mm,3.945 mm and 7.422mm respectively.Conclusively,this model not only has high prediction accuracy but also is more stable,which can provide a new insight for practical landslide prevention and control engineering.

Auto machine learning-based modelling and prediction of excavationinduced tunnel displacement

The influence of a deep excavation on existing shield tunnels nearby is a vital issue in tunnelling engineering.Whereas,there lacks robust methods to predict excavation-induced tunnel displacements.In this study,an auto machine learning(AutoML)-based approach is proposed to precisely solve the issue.Seven input parameters are considered in the database covering two physical aspects,namely soil property,and spatial characteristics of the deep excavation.The 10-fold cross-validation method is employed to overcome the scarcity of data,and promote model’s robustness.Six genetic algorithm(GA)-ML models are established as well for comparison.The results indicated that the proposed AutoML model is a comprehensive model that integrates efficiency and robustness.Importance analysis reveals that the ratio of the average shear strength to the vertical effective stress E_(ur)/σ′_(v),the excavation depth H,and the excavation width B are the most influential variables for the displacements.Finally,the AutoML model is further validated by practical engineering.The prediction results are in a good agreement with monitoring data,signifying that our model can be applied in real projects.

Slope displacement prediction based on morphological filtering

Combining mathematical morphology (MM),nonparametric and nonlinear model,a novel approach for predicting slope displacement was developed to improve the prediction accuracy.A parallel-composed morphological filter with multiple structure elements was designed to process measured displacement time series with adaptive multi-scale decoupling.Whereafter,functional-coefficient auto regressive (FAR) models were established for the random subsequences.Meanwhile,the trend subsequence was processed by least squares support vector machine (LSSVM) algorithm.Finally,extrapolation results obtained were superposed to get the ultimate prediction result.Case study and comparative analysis demonstrate that the presented method can optimize training samples and show a good nonlinear predicting performance with low risk of choosing wrong algorithms.Mean absolute percentage error (MAPE) and root mean square error (RMSE) of the MM-FAR&LSSVM predicting results are as low as 1.670% and 0.172 mm,respectively,which means that the prediction accuracy are improved significantly.

A WD-GA-LSSVM model for rainfall-triggered landslide displacement prediction

This paper proposes a WD-GA-LSSVM model for predicting the displacement of a deepseated landslide triggered by seasonal rainfall,in which wavelet denoising(WD)is used in displacement time series of landslide to eliminate the GPS observation noise in the original data,and genetic algorithm(GA)is applied to obtain optimal parameters of least squares support vector machines(LSSVM)model.The model is first trained and then evaluated by using data from a gentle dipping(~2°-5°)landslide triggered by seasonal rainfall in the southwest of China.Performance comparisons of WD-GA-LSSVM model with Back Propagation Neural Network(BPNN)model and LSSVM are presented,individually.The results indicate that the adoption of WD-GA-LSSVM model significantly improves the robustness and accuracy of the displacement prediction and it provides a powerful technique for predicting the displacement of a rainfall-triggered landslide.

Dynamic prediction of landslide displacement using singular spectrum analysis and stack long short-term memory network

An accurate landslide displacement prediction is an important part of landslide warning system. Aiming at the dynamic characteristics of landslide evolution and the shortcomings of traditional static prediction models, this paper proposes a dynamic prediction model of landslide displacement based on singular spectrum analysis(SSA) and stack long short-term memory(SLSTM) network. The SSA is used to decompose the landslide accumulated displacement time series data into trend term and periodic term displacement subsequences. A cubic polynomial function is used to predict the trend term displacement subsequence, and the SLSTM neural network is used to predict the periodic term displacement subsequence. At the same time, the Bayesian optimization algorithm is used to determine that the SLSTM network input sequence length is 12 and the number of hidden layer nodes is 18. The SLSTM network is updated by adding predicted values to the training set to achieve dynamic displacement prediction. Finally, the accumulated landslide displacement is obtained by superimposing the predicted value of each displacement subsequence. The proposed model was verified on the Xintan landslide in Hubei Province, China. The results show that when predicting the displacement of the periodic term, the SLSTM network has higher prediction accuracy than the support vector machine(SVM) and auto regressive integrated moving average(ARIMA). The mean relative error(MRE) is reduced by 4.099% and 3.548% respectively, while the root mean square error(RMSE) is reduced by 5.830 mm and 3.854 mm respectively. It is concluded that the SLSTM network model can better simulate the dynamic characteristics of landslides.

Failure evolution and disaster prediction of rock under uniaxial compression based on non-extensive statistical analysis of electric potential

Rock failure can cause serious geological disasters,and the non-extensive statistical features of electric potential(EP)are expected to provide valuable information for disaster prediction.In this paper,the uniaxial compression experiments with EP monitoring were carried out on fine sandstone,marble and granite samples under four displacement rates.The Tsallis entropy q value of EPs is used to analyze the selforganization evolution of rock failure.Then the influence of displacement rate and rock type on q value are explored by mineral structure and fracture modes.A self-organized critical prediction method with q value is proposed.The results show that the probability density function(PDF)of EPs follows the q-Gaussian distribution.The displacement rate is positively correlated with q value.With the displacement rate increasing,the fracture mode changes,the damage degree intensifies,and the microcrack network becomes denser.The influence of rock type on q value is related to the burst intensity of energy release and the crack fracture mode.The q value of EPs can be used as an effective prediction index for rock failure like b value of acoustic emission(AE).The results provide useful reference and method for the monitoring and early warning of geological disasters.

Application of phase space reconstruction and v-SVR algorithm in predicting displacement of underground engineering surrounding rock

A new method for predicting the trend of displacement evolution of surroundingrock was presented in this paper.According to the nonlinear characteristics of displace-ment time series of underground engineering surrounding rock,based on phase spacereconstruction theory and the powerful nonlinear mapping ability of support vector ma-chines,the information offered by the time series datum sets was fully exploited and thenon-linearity of the displacement evolution system of surrounding rock was well described.The example suggests that the methods based on phase space reconstruction and modi-fied v-SVR algorithm are very accurate,and the study can help to build the displacementforecast system to analyze the stability of underground engineering surrounding rock.

A new method for predicting injection multiples of extreme displacement in waterflood reservoirs

The theoretical relationship between water injection multiple(i.e.injected pore volume)and water saturation is inferred from theoretical concepts of reservoir engineering.A mathematical model based on core displacement tests is established for the entire injection process that satisfies both initial displacement and extreme displacement,simultaneously.The results show that prior to the flooding,the water injection multiple has a linear relationship with the water saturation,and the utilization rate of the injected water is the highest.As water breakthrough at the production end,the water-cut increases,and the injection multiple increases exponentially while the utilization efficiency of the injected water gradually decreases.When the injection multiple approaches infinity,the utilization efficiency of the injected water gradually decreases to 0,by which time the water-cut at the production end is always 1.At this time,the water saturation no longer changes,and the water flooding recovery rate reaches its limit.Based on the experimental test data,a mathematical model of the entire process of injection multiple and water saturation is established,which has high fitting accuracy that can predict the injection multiple in the different stages of development of a mature oil reservoir.The dynamically changing index of the injection water utilization efficiency in reservoir development by reactive water flooding can be obtained through reasonable transformation of the mathematical model.This is of great significance in guiding evaluations of the effects of reservoir development and formulating countermeasures.

SEISMIC DAMAGE PREDICTION ON MULTI-STORYBRIOK BUILDINGS WITH TWO FRAME-SHEAR-WALL-SUPPORTED STORIES

This paper presents a method for searching the weak story by using the ultimate shear force coefficient on the multi-story brick buildings with two frame-shear-wall-supported stories. The method of seismic damage prediction is discussed according to different weak stories. When the first story is t theweak one,the damage state of the building can be determined by the displacement ratio. The prediction method is also used in a practical engineering project.

Determination of ultimate bearing capacity of uplift piles using intact and non-intact load−displacement curve

Based on the field destructive test of six rock-socketed piles with shallow overburden,three prediction models are used to quantitatively analyze and predict the intact load−displacement curve.The predicted values of ultimate uplift capacity were further determined by four methods(displacement controlling method(DCM),reduction coefficient method(RCM),maximum curvature method(MCM),and critical stiffness method(CSM))and compared with the measured value.Through the analysis of the relationship between the change rate of pullout stiffness and displacement,a method used to determine the ultimate uplift capacity via non-intact load−displacement curve was proposed.The results show that the predicted value determined by DCM is more conservative,while the predicted value determined by MCM is larger than the measured value.This suggests that RCM and CSM in engineering applications can be preferentially applied.Moreover,the development law of the change rate of pullout stiffness with displacement agrees well with the attenuation form of power function.The theoretical predicted results of ultimate uplift capacity based on the change rate of pullout stiffness will not be affected by the integrity of the curve.The method is simple and applicable for the piles that are not loaded to failure state,and thus provides new insights into ultimate uplift capacity determination of test piles.

A graph deep learning method for landslide displacement prediction based on global navigation satellite system positioning

The accurate prediction of displacement is crucial for landslide deformation monitoring and early warning.This study focuses on a landslide in Wenzhou Belt Highway and proposes a novel multivariate landslide displacement prediction method that relies on graph deep learning and Global Navigation Satellite System(GNSS)positioning.First model the graph structure of the monitoring system based on the engineering positions of the GNSS monitoring points and build the adjacent matrix of graph nodes.Then construct the historical and predicted time series feature matrixes using the processed temporal data including GNSS displacement,rainfall,groundwater table and soil moisture content and the graph structure.Last introduce the state-of-the-art graph deep learning GTS(Graph for Time Series)model to improve the accuracy and reliability of landslide displacement prediction which utilizes the temporal-spatial dependency of the monitoring system.This approach outperforms previous studies that only learned temporal features from a single monitoring point and maximally weighs the prediction performance and the priori graph of the monitoring system.The proposed method performs better than SVM,XGBoost,LSTM and DCRNN models in terms of RMSE(1.35 mm),MAE(1.14 mm)and MAPE(0.25)evaluation metrics,which is provided to be effective in future landslide failure early warning.

Prediction of maximum upward displacement of shield tunnel linings during construction using particle swarm optimization-random forest algorithm

During construction,the shield linings of tunnels often face the problem of local or overall upward movement after leaving the shield tail in soft soil areas or during some large diameter shield projects.Differential floating will increase the initial stress on the segments and bolts which is harmful to the service performance of the tunnel.In this study we used a random forest(RF)algorithm combined particle swarm optimization(PSO)and 5-fold cross-validation(5-fold CV)to predict the maximum upward displacement of tunnel linings induced by shield tunnel excavation.The mechanism and factors causing upward movement of the tunnel lining are comprehensively summarized.Twelve input variables were selected according to results from analysis of influencing factors.The prediction performance of two models,PSO-RF and RF(default)were compared.The Gini value was obtained to represent the relative importance of the influencing factors to the upward displacement of linings.The PSO-RF model successfully predicted the maximum upward displacement of the tunnel linings with a low error(mean absolute error(MAE)=4.04 mm,root mean square error(RMSE)=5.67 mm)and high correlation(R^(2)=0.915).The thrust and depth of the tunnel were the most important factors in the prediction model influencing the upward displacement of the tunnel linings.

Quantitative prediction of oil saturation of unconsolidated sandstone reservoir based on time-lapse seismic “relative difference method”: Taking Zeta oil field in West Africa as an example

In view of the disadvantage that the absolute difference of time-lapse seismic(the difference between monitoring data and base data) is not only related to the change of oil saturation, but also closely related to the thickness of reservoir, a time-lapse seismic "relative difference method"(the ratio of monitoring data to base data) not affected by the thickness of reservoir but only related to the change of fluid saturation, is proposed through seismic forward modeling after fluid displacement simulation. Given the same change of fluid saturation, the absolute difference of time-lapse seismic conforms to the law of "tuning effect" and seismic reflection of "thin bed", and the remaining oil prediction method based on absolute difference of time-lapse seismic is only applicable to the reservoirs with uniform thickness smaller than the tuning thickness or with thickness greater than the tuning thickness. The relative difference of time-lapse seismic is not affected by reservoir thickness, but only related to the change of fluid saturation. It is applicable to all the deep-sea unconsolidated sandstone reservoirs which can exclude the effect of pressure, temperature, pore type and porosity on seismic. Therefore, the relation between the relative difference of time-lapse seismic and the change of fluid saturation, which is obtained from seismic forward modeling after Gassmann fluid displacement simulation, can be used to quantitatively predict the change of reservoir water saturation and then the distribution of the remaining oil. The application of this method in deep sea Zeta oil field in west Africa shows that it is reasonable and effective.

基于机器学习方法的浅埋暗挖隧道地表位移预测研究

浅埋暗挖隧道的地表位移受众多因素影响,如隧道几何参数、地质力学参数和施工参数等。因此,建立一种能够应对复杂条件的地表位移智能预测方法,是浅埋暗挖隧道关注的核心问题,也是实现隧道智慧化施工的重要基础。采用4种机器学习方法,分别为支持向量回归、随机森林、反向传播神经网络和极限学习机,并结合粒子群优化算法确定4种机器学习模型的最优超参数或随机数,从而建立一种针对浅埋暗挖隧道地表位移的混合智能预测方法。以北京地铁12号线安华桥站—安贞桥站区间段为工程背景,构建了包含73组浅埋暗挖隧道地表位移数据的数据库,并基于3种性能评价指标(决定系数R2、平均绝对百分比误差MAPE和均方根误差RMSE)对4种混合智能预测模型的预测性能进行了对比分析。此外,采用SHAP模型研究输入特征对输出值的贡献,讨论了各影响因素对地表位移的相对重要性。结果表明,支持向量回归混合智能预测模型在测试集中性能最佳,R2为0.86,MAPE为0.0079,RMSE为1.94 mm;土的压缩模量对地表沉降影响最大,其次为隧道的中心埋深和隧道横截面的宽度。