基于地貌特征的滑坡堰塞坝形成敏感性研究

滑坡堰塞坝是由斜坡失稳堵塞河道而形成的天然坝体,且易溃坝诱发洪水,对沿岸群众生命财产构成巨大的威胁。为提升主动减灾防灾能力,急需构建了一种快速预测与判断滑坡堵江成坝能力的方法。通过文献资料查阅,结合遥感技术,提取了70处典型滑坡的地貌特征参数,其中50处为堵江成坝滑坡。运用K-S检验和M-W U检验方法分析了滑坡地貌特征因子的敏感性,利用Boruta算法确定了因子重要度,筛选了滑坡体积、面积、高差、长度及河宽共5个地貌特征参数。基于此,利用Bayes判别法与逻辑回归方法,分别建立了滑坡堰塞坝形成的预测模型,准确率超过90%。选取高重要度且差异显著的因子,利用比值法建立了滑坡堵江成坝阈值判据,实现了滑坡堰塞坝形成的快速判定。统计不同诱因下滑坡地貌特征,对比V-W_r经验公式,确定了滑坡堰塞坝形成与诱因间的关系,为进一步构建不同诱因下滑坡堰塞坝形成预测模型提供了技术支撑。      

基于MEA-WNN的大坝变形预测

文章将思维进化优化算法引入大坝变形预测领域,提出基于思维进化法优化小波神经网络(MEA-WNN)的大坝变形预测模型。通过算例验证,并与WNN、GA-WNN对比分析,认为该模型能够克服多数进化算法的不足,提高算法的整体搜索效率,同时能够确保较优的局部预测值和较好的全局预测精度,具备快速的收敛能力,验证MEAWNN预测模型在大坝变形预测中的可行性和实用性。     

滑坡堰塞坝堆积形态影响因素离散元分析

滑坡是形成堰塞坝的最主要原因,在地震、降雨、冰雪融水等作用下均可形成滑坡堰塞坝,而滑坡堰塞坝的堆积形态、范围等对评价堰塞坝的稳定性有着重要的影响。通过离散元方法(DEM),系统分析了三维条件下滑动距离、滑面出口宽度、滑面倾角、河床倾角、河谷形状对堰塞坝堆积形态的影响。研究结果表明:滑动距离和出口宽度对坝体高度影响最大;随出口宽度和坡面倾角的增加,坝长和坝宽分别呈线性增大和减小趋势;滑动距离可以有效控制滑体速度,进而影响堆积角大小;河床倾角主要影响坝长;对坝高、坝长、上下游绝对倾角正切值和堆积角正切值进行回归分析表明,数学模型契合程度高,说明其形态可以预测;引入2个参数λ和χ,对堰塞坝堆积特征进行了描述;河谷形状的影响主要体现在随着河谷底部宽度的增大,滑体爬高爬坡能力增强。研究成果对根据实际地形预测滑坡堰塞坝堆积形态进而评估坝体的安全性具有重要意义,可以为进一步开展堰塞湖溃决研究提供一定的参考。      

改进哈里斯鹰优化算法与BP神经网络组合的滑坡位移高精度预测模型

开展滑坡位移高精度预测研究对于滑坡灾害的防灾预警具有重要意义。针对哈里斯鹰优化算法(HHO)搜索精度低且会陷入局部最优的问题,对其进行改进并进一步与BP神经网络组合,同时有效兼顾滑坡外部影响因子,发展了一种改进哈里斯鹰优化算法(IHHO)与BP神经网络组合(IHHO-BP)的滑坡位移高精度预测模型。结合我国典型黄土滑坡——甘肃黑方台党川滑坡HF08、HF05和HF09等3个监测点的北斗/GNSS实测数据,验证了IHHO-BP模型在3个实测数据集中的位移预测精度均优于单一BP神经网络模型,以及哈里斯鹰优化算法、麻雀搜索算法(SSA)、粒子群算法(PSO)、遗传算法(GA)与BP神经网络组合的预测模型。结果表明：引入Levy变异、局部增强和随机化Halton序列种群初始化策略的改进哈里斯鹰优化算法,可有效解决哈里斯鹰优化算法搜索精度低且会陷入局部最优的问题;IHHO-BP模型具有更好的泛化能力,可有效提升滑坡位移的预测精度,该组合预测模型具有更好的推广应用价值。     

优化组合预测模型在桥梁变形预测中的应用

为提高桥梁变形预测的精度,探讨不同预测模型在桥梁变形预测中的效果,结合桥梁变形监测数据及组合预测思路,构建桥梁的MC误差修正优化组合预测模型。通过实例验证得出,组合预测较单项预测具有更高的预测精度及稳定性,其中以RBF神经网络组合的预测精度最高;同时,误差优化修正模型进一步减小了预测误差,优化后预测结果的相对误差期望值为0.86%,方差值为0.097 3 mm2,准确预测了桥梁变形,验证了该思路的有效性。     

灰线性加权非等距GM(1,1)形变预测模型

结合加权非等距GM(1,1)模型与线性回归理论,构建灰线性加权非等距GM(1,1)预测模型,并给出对模型预测精度起决定性作用的灰指数v和参数m的优化方法。与加权非等距GM(1,1)模型和线性回归预测模型相比,灰线性加权非等距GM(1,1)预测模型的精度更高,预测有效时间更长,模型的稳定性更好。优化v和m后,灰线性加权非等距GM(1,1)预测模型的实用性、稳定性进一步提高。     

基于改进鲸鱼算法优化神经网络的GPS高程拟合方法

采取混沌映射和自适应惯性权重结合的策略对标准鲸鱼算法进行改进,从而提高算法的全局寻优能力和收敛速度,并针对BP神经网络的劣势,利用改进鲸鱼算法对BP神经网络进行优化处理。在此基础上建立改进鲸鱼算法优化BP神经网络的GPS高程异常拟合预测模型,并通过两组不同地形特征工程中的GPS数据对模型进行验证。结果表明,利用改进鲸鱼算法优化的BP模型进行GPS高程拟合时可取得更高的精度和稳定性。     

通过改进自适应PSO优化BP网络预测大坝变形

提出自适应粒子群神经网络（ADPPSO-BP）算法,并加入自适应变异算子,提高粒子跳出局部搜索的能力,实现对坝体位移的精准预测。以丰满大坝为例进行验证,建立PSO-BP（粒子群神经网络）、LPSO-BP（线性粒子群神经网络）、改进ADPPSO-BP（改进自适应粒子群神经网络）3种模型,对大坝进行变形预测。结果表明,ADPPSO-BP预测误差最小。     

分数阶傅里叶变换联合支持向量机的建筑物变形预测

提出一种分数阶傅里叶变换(fractional Fourier transformation, FrFT)与支持向量机(support vector machine, SVM)相结合的建筑物变形组合预测模型。首先利用FrFT对变形时间序列进行多尺度分析,将复杂时间序列分解为一系列结构较为简单的子序列;然后利用SVM对每个子序列分别建立预测模型,通过将各个子序列的预测结果进行综合叠加,得到最终预测结果;同时考虑到SVM模型参数选择的难题,提出一种改进果蝇优化算法(improved fruit fly optimization algorithm, IFOA)对其进行全局寻优,提升预测性能。以西南地区某混凝土坝变形实测数据为例开展验证实验,结果表明,本文组合预测模型能够充分挖掘数据中隐含的趋势性和规律性信息,获得较高的预测精度。     

基于深度全连接神经网络的大坝变形预测研究

将深度全连接神经网络引入大坝变形预测领域,结合大坝多源监测数据的训练样本,建立基于深度全连接神经网络的大坝变形预测模型.利用几种常见的深度优化学习算法对模型进行优化训练,通过对比各损失函数的变化曲线选取最优学习算法,进一步构建基于最优学习算法的深度全连接神经网络大坝变形预测模型;最后结合大坝多源监测数据的测试样本对模型...     

Rapid geometry analysis for earthquake-induced and rainfall-induced landslide dams in Taiwan

Stability analysis of the dam is important for disaster prevention and reduction. The dam＇s geometry plays an important role in understanding its stability. This study develops a rapid landslide dam geometry assessment method for both earthquake-induced and rainfall-induced landslide dams based on nine real cases collected in Chinese Taipei and 214 cases collected worldwide. For simplification purposes, a landslide dam is classified into triangular or trapezoidal. The rapid landslide dam geometry assessment method in this paper uses only satellite maps and the topographic maps to get landslide area, and then analyze the dam geometry. These maps are used to evaluate the area of the landslide and the slope of the river bed. Based on the evaluation information, the proposed method can calculate dam height, the length of the dam, and the angles of the dam in both upstream and downstream directions. These geometry parameters of a landslide dam provide important information for further dam stability analysis. The proposed methodology is applied to a real landslide dam case at Hsiaolin Village. The result shows that the proposed method can be used to assess the landslide dam geometry.     

Analysis of landslide dam geometries

The geometry of a landslide dam is an important component of evaluating dam stability. However, the geometry of a natural dam commonly cannot be obtained immediately with field investigations due to their remote locations. A rapid evaluation model is presented to estimate the geometries of natural dams based on the slope of the stream, volume of landslides, and the properties of the deposit. The proposed model uses high resolution satellite images to determine the geometry of the landside dam. These satellite images are the basic information to a preliminary stability analysis of a natural dam. This study applies the proposed method to two case studies in Taiwan. One is the earthquake-induced Lung-Chung landslide dam in Taitung, and the second is the rainfall-induced Shih-Wun landslide dam in Pingtung.     

Effects of river flow velocity on the formation of landslide dams

Natural dams are formed when landslides are triggered by heavy rainfall during extreme weather events in the mountainous areas of Taiwan.During landslide debris movement, two processes occur simultaneously: the movement of landslide debris from a slope onto the riverbed and the erosion of the debris under the action of high-velocity river flow. When the rate of landslide deposition in a river channel is higher than the rate of landslide debris erosion by the river flow, the landslide forms a natural dam by blocking the river channel. In this study, the effects of the rates of river flow erosion and landslide deposition(termed the erosive capacity and depositional capacity, respectively) on the formation of natural dams are quantified using a physics-based approach and are tested using a scaled physical model.We define a dimensionless velocity index vde as the ratio between the depositional capacity of landslide debris(vd) and the erosive capacity of water flow(ve).The experimental test results show that a landslidedam forms when landslide debris moves at high velocity into a river channel where the river-flow velocity is low, that is, the dimensionless velocity index vde 54. Landslide debris will not have sufficient depositional capacity to block stream flow when the dimensionless velocity index vde 47. The depositional capacity of a landslide can be determined from the slope angle and the friction of the sliding surface, while the erosive capacity of a dam can be determined using river flow velocity and rainfall conditions. The methodology described in this paper was applied to seven landslide dams that formed in Taiwan on 8 August 2009 during Typhoon Morakot,the Tangjiashan landslide dam case, and the YingxiuWolong highway K24 landslide case. The dimensionless velocity index presented in this paper can be used before a rainstorm event occurs to determine if the formation of a landslide dam is possible.     

Influence of inflow discharge and bed erodibility on outburst flood of landslide dam

Accurate prediction of the hydrographs of outburst floods induced by landslide dam overtopping failure is necessary for hazard prevention and mitigation. In this study, flume model tests on the breaching of landslide dams were conducted. Unconsolidated soil materials with wide grain size distributions were used to construct the dam. The effects of different upstream inflow discharges and downstream bed soil erosion on the outburst peak discharge were investigated. Experimental results reveal that the whole hydrodynamic process of landslide dam breaching can be divided into three stages as defined by clear inflection points and peak discharges. The larger the inflow discharge, the shorter the time it takes to reach the peak discharge, and the larger the outburst flood peak discharge. The scale of the outburst floods was found to be amplified by the presence of an erodible bed located downstream of the landslide dam. This amplification decreases with the increase of upstream inflow. In addition, the results show that the existence of an erodible bed increases the density of the outburst flow, increasing its probability of transforming from a sediment flow to a debris flow.     

Characteristics and interpretation of the seismic signal of a field-scale landslide dam failure experiment

Outburst floods caused by breaches of landslide dams may cause serious damages and loss of lives in downstream areas; for this reason the study of the dynamic of the process is of particular interest for hazard and risk assessment. In this paper we report a field-scale landslide dam failure experiment conducted in Nantou County, in the central of Taiwan.The seismic signal generated during the dam failure was monitored using a broadband seismometer and the signal was used to study the dam failure process.We used the short-time Fourier transform(STFT) to obtain the time–frequency characteristics of the signal and analyzed the correlation between the power spectrum density(PSD) of the signal and the water level. The results indicate that the seismic signal generated during the process consisted of three components: a low-frequency band(0–1.5 Hz), an intermediate-frequency band(1.5–10 Hz) and a highfrequency band(10–45 Hz). We obtained the characteristics of each frequency band and the variations of the signal in various stages of the landslide dam failure process. We determined the cause for the signal changes in each frequency band and its relationship with the dam failure process. The PSD sediment flux estimation model was used to interpret the causes of variations in the signal energy before the dam failure and the clockwise hysteresis during the failure. Our results show that the seismic signal reflects the physical characteristics of the landslide dam failure process. The method and equipment used in this study may be used to monitor landslide dams and providing early warnings for dam failures.     

基于特征筛选与差分进化算法优化的滑坡危险性评估方法

突发性地质灾害危险性评估对灾害防治与风险管理具有重要意义。由于不同地区影响灾害发生的因子各不相同,实际评估过程中难以全面客观地选取适宜的评估因子。机器学习对处理灾害系统的高维非线性问题独具优势,但因模型难以调优而评估效果有限。本文尝试提出一种双向优化的滑坡危险性评估方法：在构建因子敏感性指数开展定量敏感性分析的基础上,结合重要性分析、相关性分析、共线性分析构建四维(Four-Dimensional, 4D)特征筛选法用于评估因子综合优选;为克服模型难以调优的问题,引入差分进化(Differential Evolution, DE)算法优化支持向量机(Support Vector Machine, SVM)与多层感知机(Multi-Layer Perceptron, MLP) 2种推广能力较强的机器学习模型。最后,以福建省滑坡为例,开展评估方法研究。研究表明：4D特征筛选法能更加客观全面地选取适宜性更高的危险性评估因子,从而降低数据维度、减少信息冗余以提升评估模型性能;DE算法对SVM与MLP具有显著的优化效果,有益于增强模型滑坡危险性的评估准确度,DE-SVM、DE-MLP相较于未优化前模型的AUC值分别提升了4.43%与4.37%;基于双向优化的滑坡危险性评估结果表明,降雨与土地利用类型对福建省滑坡发生具有重要影响作用,福建省滑坡极高危险区普遍年均降雨较高、地形复杂多变,极低危险区主要位于东南沿海一带及闽江流域两侧。本研究为滑坡危险性评估中的影响因子客观选取与机器学习模型调优提供了一定思路。     

Analysis on dam-breaking mode of Tangjiashan barrier dam in Beichuan county

Tangjiashan landslide is a typical high-speed consequent landslide of medium-steep dip angle. This landslide triggered by earthquake took place in about semi-minute. The relative sliding displacement is 900 meters, so average sliding speed is about 30 meters per second. The longitudinal length of barrier dam which is formed by high-speed landslide along river is 803.4 meters; and maximum width crossing river is 611.8 meters. And its volume is estimated about 20.37 million steres. Through detailed geological investigation of the barrier dam, together with early geological information before earthquake, geological structures of the barrier dam and its stability of upstream and downstream slopes are studied when water level reaches different elevations in condition of continual after shocks with seismic intensity of 7 or 8 Richter scale. On this basis, dam-breaking mode of barrier dam is discussed deeply. Thereby, analytic results provide significant guidance and advices to front headquarters of Tangjiashan barrier dam, so that some proper engineering measures can be implemented and flood discharge can be carried out well.