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.     

基于滑坡灾害预警分级的应急处置措施

滑坡灾害应急处置能力是地质灾害减灾防灾的重要方面。目前,基于滑坡灾害预测和预警分级成果,系统性的应急措施分类研究还鲜有展开,因此,以三峡库区白水河滑坡为例,运用时间序列加法模型将滑坡累计位移分解为趋势项位移与周期项位移,并分别应用多项式拟合及自回归（AR）模型对2个分量进行预测,在此结果上采用聚类分析方法将滑坡变形分为匀速变形与加速变形阶段,综合判断滑坡灾害预警等级,开展了针对滑坡预警分级的应急措施研究。结果表明:白水河滑坡预警等级主要为蓝色和黄色2种类型,对处于不同的预警等级下的滑坡,可根据滑坡变形特征快速决策,基于滑坡灾害预测和预警分级结果能更有效地指导滑坡应急处置。      

一种基于M估计的Robust-ELM滑坡位移预测方法

采用传统ELM算法进行滑坡位移预测时，其网络输出权值由最小二乘估计得出，导致ELM抗差能力较差，从而造成网络训练参数不准确。为此，将M估计与ELM相结合，提出一种基于M估计的Robust-ELM滑坡变形预测方法。该方法利用加权最小二乘方法来取代最小二乘法计算ELM输出权值，以减少滑坡监测数据中粗差对ELM预测的干扰。分别以链子崖、古树屋滑坡体为例，将Robust-ELM进行了单维、多维粗差的抵御性验证。结果表明，该方法能够有效降低粗差对预测的影响，具有良好的抗差能力。     

顾及降雨及库水位因素的滑坡时滞分析与预测——以三峡库区新铺滑坡为例

降雨及库水位涨落是引起库岸滑坡形变失稳的主要诱发因素,但滑坡位移速率对此类诱发因素的响应具有一定的滞后性,影响人类对滑坡所处运动状态的判断与预测。针对常规预测模型中未考虑时滞效应的问题,利用三峡库区新铺滑坡的GNSS位移监测数据、奉节气象站降雨数据以及三峡库区库水位涨落数据,通过对监测区内9个GNSS监测点的位移速率序列与降雨量、库水位高程序列进行时滞互相关分析,确定时滞参数,进而应用多变量灰色系统理论方法,建立了时滞GM(1,3)预测模型,并对滑坡位移速率进行预测验证。结果表明:三峡库区新铺滑坡位移速率与降雨量显著相关,对降雨量的响应滞后时间约为5 d,滑体中后部受降雨影响比前缘更明显;位移速率与库水位高程高度相关,对三峡库区库水位涨落的响应滞后时间约为31 d,滑坡前缘受库水位涨落影响更明显,且离长江越近,滞后时间越短;利用加入时滞参数的时滞GM(1,3)模型进行预测,模型拟合优度达到0.702,相比GM(1,1)模型和未顾及时滞因素的GM(1,3)模型,预测精度分别提升了53.8%和58.3%,平均绝对误差百分比分别降低了7.19%和7.47%,在滑坡位移速率预测及库岸滑坡防灾减灾领域具有一定的工程应用价值。     

基于神经网络和多标度特征分析的古滑坡变形预测及趋势评价

提出一种新的古滑坡变形预测方法。首先结合集合经验模态分解(EEMD)和奇异值分解(SVD)对古滑坡变形数据进行分解,然后利用分项组合神经网络预测古滑坡复活区的变形,最后利用多重分形消除趋势波动分析(MF-DFA)进行古滑坡多标度趋势评价。以王家坡滑坡为例分析本文方法的有效性。结果表明,组合分解模型EEMD-SVD较单项分解模型具有更强的数据分解能力,可有效实现滑坡变形数据的信息分解;基于神经网络的分项组合预测模型适用于滑坡变形预测,所得预测结果的相对误差基本在2%左右,预测精度较高,且外推预测显示滑坡变形仍会进一步增加,增加速率为1.23~1.36 mm/周期;MF-DFA模型的多标度特征分析结果显示,滑坡变形具有多重分形特征,变形有进一步增加的趋势,这与预测结果较为一致,可佐证前述预测结果的准确性。     

滑坡变形预测灰色神经网络耦合模型的构建及适用性分析

结合灰色模型和神经网络的数据处理特点,提出串联、并联和混联式3种结构的灰色神经网络滑坡变形预测模型。串联式将滑坡变形位移时序分解为趋势项和随机项,采用灰色模型提取滑坡位移时序趋势,利用神经网络逼近随机波动;并联式以灰色模型和神经网络分别对滑坡预测,采用智能非线性组合,按照预测目标精度动态调整权重,从而获取最终组合预测结果;混联式通过增加灰白化层及灰模型群,对神经网络拓扑结构进行优化,达到弱化滑坡原始监测数据随机性、提高预测模型稳健性的目的。将3种模型应用于古树屋滑坡变形预测,并对其适用性进行讨论。结果表明,3种结构的灰色神经网络耦合模型均提高了预测精度,适用于复杂状况下滑坡体的变形预测。     

基于SSA-Adam-BP神经网络模型的堰塞坝稳定性预测

现有的堰塞坝稳定性预测模型多为线性模型, 无法充分考虑堰塞坝稳定性与其形态特征和水域条件之间的复杂非线性关系。鉴于此, 结合反向传播神经网络模型和樽海鞘优化算法, 提出了一种新型的堰塞坝稳定性预测模型SSA-Adam-BP。该模型通过网格搜索法选取确定模型结构的最佳超参数组合, 进而利用交叉验证和绘制ROC曲线的方式分别对采用不同优化算法的模型进行评估。使用开源数据库中的全球153例堰塞坝数据对模型的实际应用进行了说明及验证。与传统线性模型的对比表明神经网络模型预测准确率较高, 具有较低的误报率。将SSA与Adam优化算法结合提高了BP模型的全局搜索能力, 其平均交叉验证准确率达到了91.73%, 能够使用较少的参数实现对堰塞坝稳定性快速准确的预测。SSA-Adam-BP模型对近年来典型工程的稳定性能够准确预测, 具有一定的实用性和系统平台推广应用价值。      

基于SSA-Elman神经网络的电离层TEC短期预报模型

针对基于神经网络的电离层TEC短期预报存在精度较低、易陷入局部最优的问题,利用CODE中心提供的TEC数据及地磁活动指数,建立基于麻雀搜索算法(SSA)改进Elman神经网络的电离层TEC短期预报模型,并通过BP模型、Elman模型及SSA-Elman组合模型分别对电离层平静期和扰动期中低纬度TEC进行5 d连续预报....     

Predicting landslide scenes under potential earthquake scenarios in the Xianshuihe fault zone,Southwest China

Earthquake-induced landslides can seriously aggravate the earthquake’s destructive consequences and have caused widespread concern in recent decades. The Xianshuihe fault is a large active left-lateral strike-slip fault in the southeast margin of Qinghai-Tibet Plateau, Southwest China, where the frequent strong earthquakes have brought abundant geo-hazards. This study focuses mainly on exploring and predicting the landslide scenes induced by the potential earthquakes. Firstly, the sophisticated Newmark model is improved through landslide cases induced by the Ms7.9 Luhuo earthquake in 1973 to adapt the field seismotectonics of the Xianshuihe fault zone. Then, it is used to predict the landslide scenes under one speculated potential earthquake scenario with the similar focal mechanism with the Luhuo earthquake. The preliminary results show that the slope displacement resulted from Newmark model can reflect spatial distribution characteristics of earthquake-induced landslides. The predicted potential earthquake-induced landslide scenes present an obvious extending trend along the Xianshuihe fault. The landslide hazard is greater in the northeast regions than southwest regions of the Xianshuihe fault, where there are more complex topographic conditions. The study procedure will be a helpful demonstration for exploration and prediction of landslide scenes under potential earthquakes in the regions with high seismic activity.     

大坝变形的奇异谱分析预测

利用奇异谱分析法对大坝变形数据进行分析，提取趋势和周期分量，分析影响因子与各变形分量的相关性。结果表明，大坝变形主要与水位变化和时效因子有关，温度变化对大坝变形的周期成分贡献最大，其次为水位。另外在准确提取信号的基础上,利用奇异谱分析迭代法对大坝变形进行预测，并与多元回归分析方法和高斯过程模型进行对比，发现其预测精度明显高于后两者。     

基于SSA-LSTM的短期电离层TEC组合预报模型

针对电离层总电子含量(TEC)时间序列具有高噪声、非线性和非平稳的特性,在奇异谱分析基础上,融合长短期记忆神经网络模型构建短期电离层组合预报改进模型,并对磁暴期、磁平静期的电离层TEC预报精度进行分析。结果表明,在磁暴期和磁平静期,该模型预报3 d的TEC相对精度分别为91.17%和95.46%,比单一LSTM模型分别提高4.92百分点和3.17百分点。     

优化循环神经网络在滑坡位移预测中的应用

以中国典型黄土滑坡域甘肃黑方台党川6#滑坡体为例,基于滑坡体北斗和位移计时序监测数据,首先利用深度学习框架Tensorflow分别构建3种循环神经网络滑坡位移预测模型：简单循环神经网络(simple recurrent neural network,SimpleRNN)、长短期记忆网络(long short-term memory,LSTM)和门控循环单元(gated recurrent unit,GRU),并进一步针对循环神经网络在参数设置时多采用经验手动调参或采用网格搜索法,易造成人为主观影响较大和计算效率低下的突出问题,引入遗传算法（genetic algorithm,GA）优化循环神经网络参数的自动最佳化选取,分别构建3种基于遗传算法改进的循环神经网络滑坡位移高精度预测模型：GA-SimpleRNN、GA-LSTM、GA-GRU。研究结果表明,改进参数自动寻优后的3种循环神经网络预测模型具有更优的预测性能,特别是GA-GRU模型预测精度最高,更适用于滑坡体长时序位移的高精度预测。     

一种基于熵权法的小波去噪复合评价指标

传统的评价指标在真值未知的情况下不能满足小波去噪质量评价的要求。为此，借助变化率特征重新构建均方根误差变化量和平滑度变化量两个指标，利用熵权法定权将归一化后的两个指标线性组合，所得到的新指标即为复合评价指标。该方法借助指标的变化率随分解层数的增加表现出明显的收敛特性来确定去噪最优分解层数。实验表明，该方法能够在真值未知的情况下准确地指导小波分解，确定去噪最优分解层数，从而达到最优去噪效果。     

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.     

Landslide Susceptibility Mapping by Using Landslide RatioBased Logistic Regression:A Case Study in the Southern Taiwan

The object of the research is to compare the model performance and explain the error source of original logistic regression landslide susceptibility model(abbreviated as or-LRLSM) and landslide ratio-based logistic regression landslide susceptibility model(abbreviated as lr-LRLSM) in the Chishan watershed with a serious landslide disaster after 2009 Typhoon Morakot. The landslide inventory induced by 2009 Typhoon Morakot in South Taiwan is the main research material, while the Chishan watershed is the research area. Six variables, including elevation, slope, aspect, geological formation, accumulated rainfall, and bank erosion, were included in the two models. The performance of lr-LRLSM is better than that of or-LRLSM. The Cox & Snell R2, Nagelkerke R2 value, and the area under the relative operating characteristic curve(abbreviated as AUC) of lrLRLSM is larger than those of or-LRLSM, and the average correct ratio for the lr-LRLSM to predict landslide or non-landslide is larger than that of orLRLSM by 5.0%. The increase of the average correct ratio(abbreviated as ACR) difference from or-LRLSM to lr-LRLSM shows in slope, revised accumulated rainfall, aspect, geological formation and bank erosion variables, and only light decreases in elevation variable. The error sources of continuous variables in building the or-LRLSM is the dissimilarity between the distribution of landslide ratio and production of coefficient and characteristic values, while those of categorical variables is due to low correlation of landslide ratio and the coefficient value of each parameter. Using the classification of landslide ratio as the database to build logistic regression landslide susceptibility model(abbreviated as LRLSM) can revise the errors. The comparison of or-LRLSM and lr-LRLSM in the Chishan watershed also shows that building the landslide susceptibility model(abbreviated as LSM) by using lr-LRLSM is practical and of better performance than that by using the or-LRLSM.     

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。     

������С������ARMA���ģ�͵ĺ�ƽ���쳣Ԥ��

??????С??????????????????е?????????????????????????????в????????ARMA????????????????????????????????????1???????????0.29 cm???????????????0.63 cm????????????????????????????Ч???ж???SLA?????     

巨型古滑坡的稳定性评价及发展趋势分析

利用尖点突变模型进行滑坡稳定性评价,再以集合经验模态分解、GM(1,1)模型和支持向量机等方法为基础,构建滑坡变形预测模型。以变电站滑坡为例进行分析,结果表明,各监测点的突变特征值均大于0,即处于稳定状态;所得变形预测结果的平均相对误差均较小,验证了本文预测模型的有效性;通过外推预测,发现滑坡变形仍会进一步增加,稳定性变差。     

����ѧϰ���������Ӳ�Ԥ���е�Ӧ��

?????????????????????????????Sigmoidal??Sine??Hardlim??????????????????????????????????????????????????????б??????????????????????????????????????????????????????????????????????磬?????Sigmoidal????????????????????????     

导航卫星钟差预报的Elman神经网络算法研究

针对卫星钟差不能被精确模型化的问题,将具有较强记忆功能和强大计算能力的Elman神经网络运用到卫星钟差预报中,提出适用于卫星钟差预报的Elman模型。首先对原始钟差数据进行一次差处理,然后选择合适的神经网络结构建立预报效果最佳的Elman钟差预报模型,最后选用国际GNSS服务（IGS）提供的精密钟差数据进行GPS卫星钟差预报,并与二次多项式模型、附加周期项的多项式模型和灰色系统模型进行对比分析。结果表明,Elman模型进行1 d、7 d和30 d钟差预报的精度得到显著提高,分别达到亚ns、ns和μs级,表明该模型的钟差预报性能优于3种常用模型,在卫星钟差预报中具有可行性。