利用卷积神经网络检测地震的方法与优化

本文以西昌台阵观测的8 321次近震数据为例,详细介绍了利用深度卷积神经网络检测地震的数据处理流程,包括数据预处理、模型训练、波形长度、网络层数、学习率和概率阈值等关键参数对检测结果的影响,并将训练得到的最优模型,应用于事件波形和连续波形的检测。研究表明,数据预处理和数据增强可以提升模型的检测精度和抗干扰能力。用于模型训练的波形窗口长度可近似于S-P到时差的最大值。不同网络层数（5—8层）的检测结果差别不大。对于地震检测,学习率设为10?4—10?3较为合适。卷积神经网络检测出的地震数量与选择的概率阈值有关,通过绘制精确率-召回率变化曲线,可以为选择合适的概率阈值提供参考。本文为进一步利用深度学习算法提高地震检测效果提供了参考。      

基于改进的风险矩阵法的生命线网络地震脆弱性分析

生命线网络的脆弱性不单单只表示地震发生后对网络作用而产生的后果,还应该包括网络的连通情况。本文在重新确定生命线网络脆弱性定义的基础上,运用风险评估理论中的风险矩阵方法综合考虑生命线网络的连通性能和失效后果两个方面来评价生命线网络的脆弱性,并以一个供气管网为例说明改进的风险矩阵法评价生命线网络脆弱性的有效性和合理性,找出供气管网中脆弱性等级最高的节点,分析其脆弱性等级最高的原因,以便于重点保护,并降低网络的脆弱性。     

基于GRACE RL05数据研究智利地震长期震后重力变化特征

利用GRACE卫星重力可对地震引起的大范围重力变化进行观测,并从重力数据中发现主要的变化特征.发生于2010年的MW8.8智利地震震级较高,可观测到震中附近广泛的同震和震后长期重力变化.本文基于GRACERL05Level-2时变重力场数据,对2010年智利地震的同震和震后长期变化进行了计算.对同震变化的计算发现,智利地震引起的同震变化极值达-5μGal,而本文为减小水文信号的干扰而采用的3年平均的方法可以获得良好的效果.在对震后重力变化的计算中发现,智利地震震后在2011-2016年间的重力变化存在先增大后逐渐衰减的过程.对震后变化的拟合表明,智利地震震中附近有约1μGal的震后重力变化,震后变化的特征时间约1.1年.同时,在智利地震中未出现较明显的两个震后变化阶段(短期、长期).     

Evaluation of the MODIS snow cover fraction product

Eleven years of daily 500 m gridded Terra Moderate Resolution Imaging Spectroradiometer (MODIS) (MOD10A1) snow cover fraction (SCF) data are evaluated in terms of snow presence detection in Colorado and Washington states. The SCF detection validation study is performed using in‐situ measurements and expressed in terms of snow and land detection and misclassification frequencies. A major aspect addressed in this study involves the shifting of pixel values in time due to sensor viewing angles and gridding artifacts of MODIS sensor products. To account for this error, 500 m gridded pixels are grouped and aggregated to different‐sized areas to incorporate neighboring pixel information. With pixel aggregation, both the probability of detection (POD) and the false alarm ratios increase for almost all cases. Of the false negative (FN) and false positive values (referred to as the total error when combined), FN estimates dominate most of the total error and are greatly reduced with aggregation. The greatest POD increases and total error reductions occur with going from a single 500 m pixel to 3×3‐pixel averaged areas. Since the MODIS SCF algorithm was developed under ideal conditions, SCF detection is also evaluated for varying conditions of vegetation, elevation, cloud cover and air temperature. Finally, using a direct insertion data assimilation approach, pixel averaged MODIS SCF observations are shown to improve modeled snowpack conditions over the single pixel observations due to the smoothing of more error‐prone observations and more accurately snow‐classified pixels. Copyright © 2012 John Wiley & Sons, Ltd.     

震后紧急供电的电力网络安全检测研究

当前震后紧急供电安全检测方法忽略了电网节点的三相电流不对称问题,未计算电力网络节点功率、支路电流及网损,导致电力网络进行节点连通安全检测准确率低。据此提出震后紧急供电的电力网络安全检测方法,根据连通度矩阵电力网络节点的工作参数,电力网络潮流计算不同节点的电压向量。利用快速耦合法优化牛顿拉夫逊法的计算速度,求出节点功率、支路电流及网损数值。以所构建的电力网络节点连通安全检测模型为基础,采用蒙特卡洛方法完成震后紧急供电的电力网络安全检测。经实验证明,所提方法可有效检测震后紧急供电的电力网络节点安全性,当震动加速度分别为0.05 gal、0.20 gal、0.80 gal时,实验对象连通安全检测结果分别为较高、中等和极低状态,与实际结果均一致,说明所提方法能准确检测电力网络节点连通的安全性。     

Impact of sensor measurement error on sensor positioning in water quality monitoring networks

This paper studies the impact of sensor measurement error on designing a water quality monitoring network for a river system, and shows that robust sensor locations can be obtained when an optimization algorithm is combined with a statistical process control (SPC) method. Specifically, we develop a possible probabilistic model of sensor measurement error and the measurement error model is embedded into a simulation model of a river system. An optimization algorithm is used to find the optimal sensor locations that minimize the expected time until a spill detection in the presence of a constraint on the probability of detecting a spill. The experimental results show that the optimal sensor locations are highly sensitive to the variability of measurement error and false alarm rates are often unacceptably high. An SPC method is useful in finding thresholds that guarantee a false alarm rate no more than a pre-specified target level, and an optimization algorithm combined with the thresholds finds a robust sensor network.     

A probabilistic approach for the classification of earthquakes as ��triggered�� or ��not triggered��

The occurrence time of earthquakes can be anticipated or delayed by external phenomena that induce strain energy changes on the faults. ??Anticipated?? earthquakes are generally called ??triggered??; however, it can be controversial to label a specific earthquake as such, mostly because of the stochastic nature of earthquake occurrence and of the large uncertainties usually associated to stress modelling. Here we introduce a combined statistical and physical approach to quantify the probability that a given earthquake was triggered by a given stress-inducing phenomenon. As an example, we consider an earthquake that was likely triggered by a natural event: the M?=?6.2 13 Jan 1976 Kópasker earthquake on the Grímsey lineament (Tj?rnes Fracture Zone, Iceland), which occurred about 3?weeks after a large dike injection in the nearby Krafla fissure swarm. By using Coulomb stress calculations and the rate-and-state earthquake nucleation theory, we calculate the likelihood of the earthquake in a scenario that contains only the tectonic background and excludes the dike and in a scenario that includes the dike but excludes the background. Applying the Bayes?? theorem, we obtain that the probability that the earthquake was indeed triggered by the dike, rather than purely due to the accumulation of tectonic strain, is about 60 to 90?%. This methodology allows us to assign quantitative probabilities to different scenarios and can help in classifying earthquakes as triggered or not triggered by natural or human-induced changes of stress in the crust.     

内蒙古区域地震台网监测能力研究

国际上新近发展的“基于概率的完整性震级”(PMC)方法,具有可考察地震定位中由于台站人为选择等造成的台网监测能力下降,以及避免传统基于G-R关系的统计算法因地震数目过少而无法评估等优点.本研究利用PMC方法,计算得到内蒙古区域地震台网39个台站对周边地震事件的检测概率及台网检测概率.单台检测概率结果显示:PMC方法能够客观地反映39个台站对地震事件的检测能力;因台网布局等影响,内蒙古区域地震台网中西部和中东部地区的台站检测能力较强,而靠近蒙古、俄罗斯边境的台站, 阿拉善右旗附近地区的台站,以及邻近吉林、黑龙江等地区的台站检测能力较低.合成检测概率结果显示,由于邻省台站的引入,全区80%的地区基于概率的最小完整性震级M_P达到2.2左右,其余地区M_P达到3.3左右.为提高地震台网监测能力,建议在监测能力较弱的中蒙交界地区、东北部地区,以及阿拉善左旗以西地区适度加密台站,进一步优化台网布局.      

2016年12月25日智利M7.6地震球型自由振荡检测

智利M7.6的地震激发了地球自由振荡,利用中国数字台网(CDSN)的10个台站记录到7 d的VHZ波形数据,采用功率谱密度估计方法,在没有滤波的情况下,提取了2016年12月25日智利地震激发的地球球型自由振荡基频振型_0S_4～_0S₅₃和一系列的谐频振型,并与PREM模型的理论自由振荡频率峰值进行对比,发现2.5～4.5×10¹³Hz的频率值与PREM模型预测的球型自由振荡周期符合得很好,说明本次地震确实激发了全球范围内的自由振荡,而4.5～6×10¹³ Hz的频率值稍微有偏离,这是由于地壳的横向不均匀所致,可以用来研究地球内部的构造。     

Vulnerability of floating tunnel shafts for increasing earthquake loading

Fragility curves constitute the cornerstone in seismic risk evaluations and performance-based earthquake engineering. They describe the probability of a structure to experience a certain damage level for a given earthquake intensity measure, providing a relationship between seismic hazard and vulnerability. In this paper a numerical approach is applied to derive fragility curves for tunnel shafts built in clays, a component that is found in several critical infrastructure such as urban metro networks, airport facilities or water and waste water projects. The seismic response of a representative tunnel shaft is assessed using tridimensional finite difference non-linear analyses carried out with the program FLAC^3D, under increasing levels of seismic intensity. A hysteretic model is used to simulate the soil non-linear behavior during the seismic event. The effect of soil conditions and ground motion characteristics on the soil-structure system response is accounted for in the analyses. The damage is defined based on the exceedance of the concrete wall shaft capacity due to the developed seismic forces. The fragility curves are estimated in terms of peak ground acceleration at a rock or stiff soil outcrop, based on the evolution of damage with increasing earthquake intensity. The proposed fragility models allows the characterization of the seismic risk of a representative tunnel shaft typology and soil conditions considering the associated uncertainties, and partially fill the gap of data required in performing a risk analysis assessment of tunnels shafts.     

极值理论在福建沿海地区地震预测中的研究与应用

利用陈培善等人对极值理论修改后的极值分布函数模型对1971年1月-2012年9月福建沿海地区的地震资料进行统计分析,单位时间内最小与最大地震的确定分别考虑相关区域内台网的监控能力以及删除余震后的实际情况.根据修正后的极值理论统计计算出该地区地震的复发周期及截至2015年12月可能发生的相应地震的次数与发震概率,并对有关结果进行映震能力分析,为今后福建沿海地区地震趋势分析提供可遵循的理论依据.     

Multiparticle statistical model of the solar wind

A multiparticle statistical approach to plasma (gas) modeling is presented, in which the fact that the macroscopic parameters are measured with finite resolution scales is taken into account [Minkova, 2004; 2005; 2007]. This approach is based on the Liouville theorem formulated for a stationary open system in the approximation of detailed dynamic balance with its surroundings. When the finite resolution scales of measuring instruments are taken into account, the plasma (gas) is described by multiparticle distribution functions. The latter are used to derive the probability distribution functions of fluctuations and the average values of macroscopic parameters. The multiparticle statistical approach allows a stationary solar wind model to be constructed under a number of simplifying assumptions. Its results for the average values of macroscopic parameters coincide with those of the two-particle kinetic model [Vasenin et al., 2003] and agree with inecliptic observational data.     

地震动参数速报仪的研制

在出现区域大震后,为了迅速展开有效的应急救援行动,需要及时准确了解烈度分布.目前震动图的获得有震后人工调查、震源参数计算、地震监测台网、地震烈度速报台网等几种方式,其中最有效的是在重点监护区建立烈度速报台网.但烈度速报台网建设往往沿用地震监测台网的模式,建设成本高、台站密度有限、实时数据传输量大.本文研制的地震动参数速报仪采用MEMS加速度计为测震传感器和ARM+Linux嵌入式计算机技术,具有体积小、成本小、功耗低、一体化、智能化的特点;其内置地震信息实时处理算法,能够自动判别地震事件并计算地震动参数;在实际应用中安装简便.通过大量密集布设这种小型仪器而组建的地震动参数速报网络,具有数据传输量小、分布式计算、组成的速报网络可靠性高、能够快速产出高分辨率的精细震动图等特点.     

Application on Anomaly Detection of Geoelectric Field Based on Deep Learning

The deep learning method has made nurnerials achievements regarding anomaly detection in the field of time series. We introduce the speech production model in the field of artificial intelligence, changing the convolution layer of the general convolution neural network to the residual element structure by adding identity mapping, and expanding the receptive domain of the model by using the dilated causal convolution. Based on the dilated causal convolution network and the method of log probability density function, the anomalous events are detected according to the anomaly scores. The validity of the method is verified by the simulation data, which is applied to the actual observed data on the observation staion of Pingliang geoeletric field in Gansu Province. The results show that one month before the Wenchuan M_S8.0, Lushan M_S7.0 and Minxian-Zhangxian M_S6.6 earthquakes, the daily cumulative error of log probability density of the predicted results in Pingliang Station suddenly decreases, which is consistent with the actual earthquake anomalies in a certain time range. After analyzing the combined factors including the spatial electromagnetic environment and the variation of micro fissures before the earthquake, we explain the possible causes of the anomalies in the geoelectric field of before the earthquake. The successful application of deep learning in observed data of the geoelectric field may behefit for improving the ultilization rate both the data and the efficiency of detection. Besides, it may provide technical support for more seismic research.     

Application of open tools and datasets to probabilistic modeling of road traffic disruptions due to earthquake damage

The disruption of a transportation network can have a high social and economic impact on the welfare of a society, as it can significantly affect the daily routines of a community. Although many studies have focused on the estimation of physical risk in the components that compose these networks, only a limited number have analyzed their interconnections and impact in the traffic flow. The present study analyzes how earthquake damage can disrupt the road network in an urban environment, and how this will influence the ability of the population to travel. Traffic due to daily commutes is modeled for different layouts of the network, corresponding to possible disruptions caused by earthquake damage. The duration and length of each trip were calculated both for the undamaged network conditions and for the disrupted network. The increase in the median duration and length of each trip allows estimating the economic loss for each event due to drivers' delay. By combining the probability of a specific road being blocked with its number of users, the average number of affected vehicles was estimated, and the most critical segments identified. The methodology was applied to a case study concerning the road network of the area around the Italian city of Messina in Sicily. The results were calculated for both a repetition of the well-known historical event of 1908 and a set of simulated earthquakes consistent with the national probabilistic seismic hazard model of Italy.     

高速铁路与国家地震台网信息接入及资源共享方式研究

本文对高速铁路与国家地震台网信息接入和资源共享方式进行了讨论。通过对组网模式和时延的分析及测试,在遵循目前地震系统信息传输模式的前提下,对于实时波形数据,可以根据实际情况采取国家中心型或者省局中心型的方式进行共享,用于高铁系统台站的补充,在地震预警发布阶段使用;对于地震速报信息,可以采取国家中心型的方式进行共享,用于地震信息的确认,在地震预警解除阶段使用。为了减小实时数据传输的时延以适应预警需求,应对数据采集器进行升级,并更新相应的流服务器的仪器适配器。     

一种基于经验分布的大地震复发概率计算方法

以历史重演原则和构造类比原则为基础,提出了一种基于经验分布的大地震复发概率计算方法,该方法不作任何复发概率分布的强假定,直接通过对大量地震序列数据的蒙特卡罗随机抽样来模拟未来大地震的复发规律,进而统计得到未来一段时间内的大地震发生概率,并以鲜水河断裂带炉霍段和道孚段为实例,利用本文给出的复发概率计算方法得出炉霍段和道孚段未来50年大地震发生概率分别为0.15和0.31。      

高灵敏度电子地震预警器故障检测方法研究

传统基于时间序列数据挖掘的故障检测方法,缺少高灵敏度电子地震预警器灵敏度分析,故障检测率低。提出新的高灵敏度电子地震预警器故障检测方法,根据高灵敏度电子地震预警器的结构,采用质量块、铰链和延长梁构建其力学模型,得到电子预警器的灵敏度表达式,根据该表达式建立多信号模型相关矩阵,得到故障优先概率的故障源耗费,采用基于故障模式故障概率的改进多信号模型检测方法（包括故障模式概率均分法和故障模式概率优先法）,计算得到电子地震预警器故障概率表达式,实现电子地震预警器故障检测。实验结果表明,所提方法对电子地震预警器TE过程中G和I两个指标的故障检测率分别为0.989、0.905,对PL过程的故障检测时刻为180 s、故障检测率为0.412 8,都高于传统基于时间序列数据挖掘的故障检测方法,说明所提方法具有较高的故障检测性能。     

Diversity of the 2014 Iquique’s foreshocks and aftershocks: clues about the complex rupture process of a Mw 8.1 earthquake

We study the foreshocks and aftershocks of the 1 April 2014 Iquique earthquake of Mw 8.1. Most of these events were recorded by a large digital seismic network that included the Northern Chile permanent network and up to 26 temporary broadband digital stations. We relocated and computed moment tensors for 151 events of magnitude Mw?≥?4.5. Most of the foreshocks and aftershocks of the Iquique earthquake are distributed to the southwest of the rupture zone. These events are located in a band of about 50 km from the trench, an area where few earthquakes occur elsewhere in Chile. Another important group of aftershocks is located above the plate interface, similar to those observed during the foreshock sequence. The depths of these events were constrained by regional moment tensor (RMT) solutions obtained using the records of the dense broad band network. The majority of the foreshocks and aftershocks were associated to the interplate contact, with dip and strike angles in good agreement with the characteristics of horst and graben structures (>2000 m offset) typical of the oceanic Nazca Plate at the trench and in the outer rise region. We propose that the spatial distribution of foreshocks and aftershocks, and its seismological characteristics were strongly controlled by the rheological and tectonics conditions of the extreme erosive margin of Northern Chile.     

2010年智利8.8级地震在北京房山岩体附近的动态触发活动

本文采用波形分析和β统计方法,基于首都圈地区数字测震台网和部分流动地震台的观测资料,对2010年2月27日M_w8.8智利大地震在北京房山岩体附近地区的动态触发活动进行了分析讨论.结果表明,在背景地震活动相对较弱的北京房山岩体附近,我们检测出至少5次小震活动事件在智利地震面波到达时发生.这些触发活动显著地改变了房山岩体地区这一地震活动相对平静地区的地震活动性,但是对北京地区地震活动性的影响并不明显.NKY地震台记录的智利地震触发活动的最大动态应力与之前的研究结果相比要小,约为7 kPa.这可能与房山岩体附近地区的背景地震在智利地震前一直较为平静有关.此外,相对有利的面波入射方向,以及在对蹠点上的前两组面波叠加的效应等因素综合作用,使得本研究能在约2万公里极远处观测到远震触发小震活动.然而,在随后的多次面波叠加期间并没有观测到明显的触发地震活动.在智利地震10 h前发生的琉球M_w7.0地震也未在房山地区触发相关的微震活动,这可能与它们引起的动态应力变化太过微弱有关.