基于最近邻支撑向量特征线融合算法的核爆地震识别

为增强核爆地震模式分类器的泛化能力以提高对核爆炸事件的准确识别能力,论文提出了一种选择支撑向量样本集来表征训练样本集的最近邻支撑向量特征线分类算法,用以训练时扩展核爆地震的训练样本库,提高分类器的泛化能力.该算法用于核爆炸和地震的识别结果发现,和最近邻特征线分类器相比,提出的算法降低了计算复杂度,但识别能力却有些许降低.对新算法的分析发现,纯粹的支撑向量集不能完全代表原始样本空间集,支撑向量比例在其中有重要作用,为发挥支撑向量比例的作用以提高核爆分类器的识别能力,提出了最近邻支撑向量特征线融合算法.最后以核爆地震数据库对上述算法进行了检验和分析,理论分析和识别结果证实,在相同的训练样本选择条件下,最近邻支撑向量特征线融合算法对于核爆炸的识别来说具有较好的泛化能力,正确识别率达到90.3%,且优于支持向量机算法和最近邻特征线算法.     

路基病害识别算法与应用研究

本文通过对支持向量机原理以及路基病害特征的讨论和分析,提出了基于支持向量机的探地雷达回波信号识别算法;利用所提出的算法对探地雷达实测数据进行处理,试验结果表明该算法的识别效果优于神经网络算法,并且该算法克服了神经网络的过学习和局部极小值的缺点,是一种适合路基病害识别的高效算法.     

用决策方法识别地震和爆破

在矩阵决策方法的基础上,建立了识别地震与爆破的矩阵决策算法（MDA）．着眼于快速识别,研究了5个判据. 其中初动方向和振幅比判据的效果较好．用MDA算法对北京附近62个事件进行了识别和检验得到:用5个判据C检验的正确识别率达到97%, U检验也达到93%,识别效果较好;依次从5个判据中任选4个判据作决策识别和检验, 10个结果中有7个的正确识别率在93.3%以上.结果表明,本文建立的MDA算法和所选的特征能有效地识别地震与爆破,可应用于两者的快速识别．      

多算法协同分类在致密砂岩流体识别中的应用

储层含气时,岩石的弹性力学参数会发生改变,利用岩石的弹性性质可以对流体性质进行识别.引入流体敏感度评价参数,优选出反映储层含气性敏度高的5个弹性参数,并将这5个弹性参数作为流体识别算法的输入参数.分别利用贝叶斯算法、K近邻算法、广义神经网络算法、主成分分析算法、支持向量机算法对不同流体性质进行识别,并将识别结果采用投票从众的原则,构建多算法协同分类方法(BKGPS).采用BKGPS算法对苏里格气田西部盒8、山1致密砂泥岩储层6口井的75个样本进行判别,样本的识别符合率达到93%,BKGPS识别结果与5种单算法相比,能够提高流体识别的准确性与稳定性.对研究区致密砂岩储层实际资料处理,BKGPS算法处理结果与试气结论一致,验证了该方法的可靠性.     

蚁群算法在地表隐伏断裂识别中的应用的探讨——以九寨沟景区为例

在发震地区,准确识别地表断裂带可为地质与地震灾害预测提供基础数据。由于断裂带一般在地表出露是不连续的,在此对基于蚁群算法的地表断裂隐伏段的识别研究进行了尝试,基于遥感图像及航测数据,获取地形、地貌要素以及可以参考的其他要素,使用改进最短对角线方法提取断层表面的图像要素特征作为识别母体,在算法上依次将断裂带三角面片全部配对连接成四面体,再采用蚁群算法进行地表隐伏断裂空间识别。在九寨沟景区中进行了实例探讨,对2017年九寨沟地震的断裂进行尝试性识别,认为该方法可用来初步作为辅助识别地表断裂时的一个参考,同时基于本文方法与其他两种识别方法进行了分析对比。     

测井岩性识别新方法研究

为了更好地解决测井岩性识别问题,引入了一种基于粒子群优化的支持向量机算法.通过实际测井资料和岩性剖面资料进行学习训练支持向量机,并利用粒子群优化算法对支持向量机参数进行优化,建立了测井岩性识别的支持向量机模型,应用该方法对准噶尔盆地某井的测井岩性进行识别,并将该方法的识别结果与BP神经网络方法的识别结果进行了比较,结果表明该方法优于BP神经网络方法,具有识别正确率高、收敛速度快、推广能力强等优点.     

基于复合反演算法的结构损伤识别及地震动反演研究

针对输入激励信息未知、输出响应信息非完备的情况,进行了结构损伤识别算法研究。基于物理参数时域识别的复合反演算法,定义单元刚度参数变化率为损伤因子,并引入矩形窗方法剔除噪声异常数据,给出了时域内结构损伤识别复合反演算法(损伤识别和地震动反演)。以一个5层框架结构为例进行了数值仿真识别损伤,针对矩形窗长度、噪声水平不同等情况对比了识别结果(损伤识别精度、损伤位置确定)和收敛速度,并比较了反演的地震动与真实地震动时程曲线。研究表明,所提基于复合反演算法建立的结构损伤识别复合反演算法是非常有效的。     

云南省潜在震源区综合统计判定之一

本文将云南省地震局多年来积累的关于中强震发生的地质构造、地貌、地球物理场、地形变等特征基础资料,用于Cora-3算法的模式识别,得出云南省潜在震源区的背景指标,结果是比较理想的。“投票”结果表明,80%的学习集D类对象仍被识别为D类,85%的学习集N类对象仍被识别为N类,控制试验的结果也表明识别结果是比较稳定和可靠的。     

用算法复杂性分析地震活动演化特征

以非线性动力学理论为依据，用非线性动力学中分析时间序列的算法复杂性研究地震活动规律和构成地震活动复杂现象的动力学特征。提出了按震级计算的地震算法复杂性ＡＣ值和按发震时间计算的地震算法复杂性ＴＡＣ值。选取１９７０年以来中国东部ＭＳ≥６．０地震，西部ＭＳ≥６．５的强震，对ＡＣ值和ＴＡＣ值进行回顾性分析；利用１９７０￣１９８０年大华北地区展望性空区识别结果，对ＡＣ值和ＴＡＣ值进行展望性检验分析，并进行了     

基于改进Stacking算法的致密砂岩储层测井流体识别

致密砂岩储层物性差,测井响应对孔隙流体不敏感,应用传统测井解释图版划分流体类型精度较低.机器学习技术通过学习更多维度的特征,可以建立合适的流体识别模型.相较于单一算法,集成学习可以通过联合多个专家模型提升预测精度,但是不同的集成学习策略性能差距较大.本文提出了一种改进的Stacking算法,通过平均影响值法寻找敏感测井曲线作为输入,利用不同的特征集构建多个子模型,并使用不同集成策略将其组合为若干性能更佳的专家模型进行训练,同时引入独立专家避免过拟合,将专家模型的预测结果通过交叉验证的方式进行模拟预测,最后应用元学习器预测最终结果.将该方法用于库车坳陷迪北气藏致密砂岩储层流体识别,测试准确率可达93%,优于CatBoost模型和XGBoost模型,证明了该方法的有效性和适用性.为致密砂岩储层流体识别提供了新的思路.     

裂缝的双侧向测井响应的数学模型及裂缝孔隙度的定量解释

根据裂缝双侧向测井响应的特征，裂缝分３种状态：低角度裂缝、倾斜裂缝和高角度裂缝，给出了判断裂缝状态的方法；建立了裂缝的双侧向测井响应与裂缝孔隙度、裂缝内流体电导率、基岩电导率之间的函数关系；得出人工估算和计算机精确计算裂缝孔隙度的两套反演公式及相应计算方法．     

On‐line identification of non‐linear hysteretic structural systems using a variable trace approach

In this paper, an adaptive on‐line parametric identification algorithm based on the variable trace approach is presented for the identification of non‐linear hysteretic structures. At each time step, this recursive least‐square‐based algorithm upgrades the diagonal elements of the adaptation gain matrix by comparing the values of estimated parameters between two consecutive time steps. Such an approach will enforce a smooth convergence of the parameter values, a fast tracking of the parameter changes and will remain adaptive as time progresses. The effectiveness and efficiency of the proposed algorithm is shown by considering the effects of excitation amplitude, of the measurement units, of larger sampling time interval and of measurement noise. The cases of exact‐, under‐, over‐parameterization of the structural model have been analysed. The proposed algorithm is also quite effective in identifying time‐varying structural parameters to simulate cumulative damage in structural systems. Copyright © 2001 John Wiley & Sons, Ltd.     

利用动态时间规整方法实现不同时间长度地震观测资料的形态匹配

本文将动态时间规整方法引入到地震观测资料的形态匹配分析中,以解决因时间尺度不一致的两列观测数据无法定量比对的问题。基于动态时间规整技术方法原理,通过测试数据验证了动态时间规整方法的可行性,并利用云南西部地区的断层实际观测数据,分析了1996年丽江M_S7.0地震前的跨断层观测数据异常形态与当前数据异常形态的相似性问题。结果表明:① 动态时间规整方法可用于地震资料时间长度不一致时的相似性匹配;② 时间不一致的两列观测数据可用累积规整路径距离来定量表征,累积距离越短,曲线形态越一致;③ 动态时间规整方法可用于给定模板的前兆数据相似度的计算机自动提取,可提高当前仅依靠人工判别的工作效率;④ 从模式识别的角度考虑,当前下关跨断层水准观测数据变化形态与1996年丽江M_S7.0地震和2008年汶川M_S8.0地震前的水准数据变化形态较为一致。      

应用单台垂向记录进行区域地震事件实时检测和直达犘波初动自动识别

提出一种基于直达P波信号和其它背景噪声在能量、非高斯性、非线性和偏振特性的不同而进行区域地震事件实时检测的新方法信噪综合差异特征量方法（简写为EFGLP方法）,同时对比分析了应用信号的不同统计特性来精细识别震相初至的３种有效方法,其中的TOC AIC方法是新提出的．应用山东数字地震波资料处理的结果表明:①与常规的STA/LTA地震事件触发算法相比,EFGLP方法能够有效降低地震事件的错误报警率和漏报率;②与人机交互震相识别结果相比,当信噪比比较低、震相初至比较模糊时,3种震相精细识别方法中的TOC-AIC方法识别精度最高;当信噪比比较高、震相初至比较清晰时,基于VAR-AIC 和TOC-AIC方法所测量得到的震相初至识别基本一致．      

图像增强技术在X射线医学影像中的应用

目的:对X射线影像进行图像增强。方法:应用多种互补的图像增强方法,首先采用拉普拉斯法来突出图像中的细节,使用梯度法突出边缘,然后用平滑后的梯度图像掩蔽拉普拉斯图像,通过高提升滤波得到锐化图像,再使用灰度变换增大锐化图像的灰度动态范围。将本文算法的结果与传统图像增强算法的结果进行比较,该算法对X射线影像增强效果显著,相比传统算法有很大的优势。增强后的图像细节更加清晰,优于传统算法的实现效果。结论:本文算法对医学X射线影像具有一定的理论和实际应用价值。      

Application of the SP algorithm to the INTERMAGNET magnetograms of the disturbed geomagnetic field

The algorithmic system developed in the Laboratory of Geoinformatics at the Geophysical Center, Russian Academy of Sciences, which is intended for recognizing spikes on the magnetograms from the global network INTERMAGNET provides the possibility to carry out retrospective analysis of the magnetograms from the World Data Centers. Application of this system to the analysis of the magnetograms allows automating the job of the experts-interpreters on identifying the artificial spikes in the INTERMAGNET data. The present paper is focused on the SP algorithm (abbreviated from SPIKE) which recognizes artificial spikes on the records of the geomagnetic field. Initially, this algorithm was trained on the magnetograms of 2007 and 2008, which recorded the quiet geomagnetic field. The results of training and testing showed that the algorithm is quite efficient. Applying this method to the problem of recognizing spikes on the data for periods of enhanced geomagnetic activity is a separate task. In this short communication, we present the results of applying the SP algorithm trained on the data of 2007 to the INTERMAGNET magnetograms for 2003 and 2005 sampled every minute. This analysis shows that the SP algorithm does not exhibit a worse performance if applied to the records of a disturbed geomagnetic field.     

Recognition of disturbances with specified morphology in time series: Part 2. Spikes on 1-s magnetograms

Preliminary magnetograms contain different types of temporal anthropogenic disturbances: spikes, baseline jumps, drifts, etc. These disturbances should be identified and filtered out during the preprocessing of the preliminary records for the definitive data. As of now, at the geomagnetic observatories, such filtering is carried out manually. Most of the disturbances in the records sampled every second are spikes, which are much more abundant than those on the magnetograms sampled every minute. Another important feature of the 1-s magnetograms is the presence of a plenty of specific disturbances caused by short-period geomagnetic pulsations, which must be retained in the definitive records. Thus, creating an instrument for formalized and unified recognition of spikes on the preliminary 1-s magnetograms would largely solve the problem of labor-consuming manual preprocessing of the magnetic records. In the context of this idea, in the present paper, we focus on recognition of the spikes on the 1-s magnetograms as a key point of the problem. We describe here the new algorithm of pattern recognition, SPs, which is capable of automatically identifying the spikes on the 1-s magnetograms with a low probability of missed events and false alarms. The algorithm was verified on the real magnetic data recorded at the French observatory located on Easter Island in the Pacific.     

Times of Increased Probabilities for Occurrence of World’s Largest Earthquakes: 30 Years Hypothesis Testing in Real Time

Izvestiya, Physics of the Solid Earth - For the past 30 years, the M8 medium-term earthquake prediction algorithm has been applied globally every six months in real time for identifying the areas...     

Decoupling algorithm for evaluating multiple beam damages in steel moment‐resisting frames

Post‐earthquake safety evaluation of steel moment‐resisting frames mainly relies on the inspection of seismic damage to beam–column connections. Recently, in order to evaluate seismic damage of steel connections in a prompt and precise manner, a local damage evaluation method based on dynamic strain responses has been proposed and receives attention. In the evaluation method where strain responses are measured by piezoelectric strain sensors, a strain‐based damage index has been developed for evaluating individual seismic beam damage in a steel frame. However, for a steel frame suffering multiple beam damages, the damage index deteriorates its performance in identifying small damages with the presence of neighboring severe damages because of the moment redistributions induced by larger damages. This paper presents a decoupling algorithm that removes the issue of damage interaction and improves the performance of the damage index. The decoupling algorithm was derived on the basis of damage‐induced moment release and redistribution mechanism. The effectiveness of the decoupling algorithm was numerically and experimentally investigated using a nine‐story steel frame model and a large scale five‐story steel frame testbed that can simulate multiple fractures at beam ends. Copyright © 2016 John Wiley & Sons, Ltd.