Neural network study of the nuclear ground-state spin distribution within a random interaction ensemble

The distribution of the nuclear ground-state spin in a two-body random ensemble(TBRE)was studied using a general classification neural network(NN)model with two-body interaction matrix elements as input features and the corresponding ground-state spins as labels or output predictions.The quantum many-body system problem exceeds the capability of our optimized NNs in terms of accurately predicting the ground-state spin of each sample within the TBRE.However,our NN model effectively captured the statistical properties of the ground-state spin because it learned the empirical regularity of the ground-state spin distribution in TBRE,as discovered by physicists.

Fragility of disconnect switch subjected to random earthquake ground motions

A fragility calculation scheme is estabtished in this paper for porcelain-type equipments subjected to random earthquake ground motions. All steps of the method are illustrated by the seismic damage analysis of GW4-110 disconnect switch. The model of the equipment is built applying the finite element method with flexible joints, and the seismic response of the equipment is analyzed using elastic time history method. On the base, according to the strength damage index and Monte-Carlo Method, the seismic damage ratios are counted and the seismic fragility curves are presented. Then the seismic damage of GW4-110 disconnect switch can be predicted.

基于点云特征的改进RANSAC地面分割算法

针对室外复杂场景下,轻量级和地面优化的激光雷达里程计与测图(LeGO-LOAM)算法由于地面分割不精确而导致算法定位精度降低的问题,提出一种基于改进随机一致性采样(RANSAC)的多线程地面分割算法:相较于传统RANSAC算法,该算法舍弃从全部原始数据中随机选取种子点拟合地面模型的迭代方式,首先利用点云高程、曲率等点特征信息挑选出所有小于高程、曲率等阈值的种子点以构建种子点集合,并根据种子点集合中的种子点数量判断是否需要多线程处理;然后根据判断结果从种子点集合中选择种子点子集进行地面拟合;最后比较各地面模型所包含的点云数量以获得最优地面模型参数以及地面点云集;地面分割精度的提高有效地降低了LeGO-LOAM算法的定位误差。实验结果表明,在室外复杂场景下所提出的地面分割算法分割效果更好,杂点更少;相较于原LeGO-LOAM算法,改进算法的定位误差降低至3.73 m,平面均方根误差降低了20.8%。

近断层地震动作用下土石坝加速度分布研究

近断层地震动较远场地震动具有频率高、速度快、能量更加集中的特点,在短时间内会释放大量能量,对工程结构具有严重危害。规范提供的加速度分布系数不适用于高度超过150 m的大坝,随着技术不断进步,我国超过150 m的土石坝越来越多,规范已不能满足现有需求。基于中国台湾集集近断层地震动记录,建立随机脉冲型地震动模型,分别对高度为40 m、70 m、150 m和200 m的土石坝进行随机动力分析,研究坝体的动态分布系数。结果表明,近断层地震动的低频速度脉冲使坝体的加速度反应谱有较宽的敏感段,易覆盖坝体自振周期,显著增加坝体的加速度。该研究通过总结近断层区域土石坝的加速度分布规律,建议了近断层脉冲型地震动区域的土石坝坝体动态分布系数,为近断层高地震风险情况下土石坝的抗震理论和方法提供参考。

海底长周期地震动预测模型:以日本相模湾海域为例

地震动反应谱预测模型是地震危险性分析的基础,基于日本K-NET台网相模湾地区的892组海域地震动数据和4 033组陆域地震动数据,考虑震级、震源距、震源深度以及震源类型和场地,建立了PGA和0.05~15.0 s的海域地震动加速度反应谱预测模型。通过残差分析以及相关性分析检验了模型的合理性,并分析了震源类型以及震源深度对反应谱的影响,对比了海陆地震动预测模型的差异。结果表明:震源类型以及震源深度对模型预测结果影响显著,水平向海域模型以及长周期(T≥1.0 s)竖向海域模型所预测反应谱大于陆域模型所预测的反应谱值,且水平向和长周期竖向模型的场地放大系数均大于1。该模型可为长周期海域地震动模型的建立和概率地震危险性分析提供参考,为长周期海域工程的抗震设防提供依据。

基于约束随机场的基坑开挖变形概率评估方法

为综合考虑岩土参数的空间变异性以及前期地质勘查所获取钻孔数据的确定性,提升基坑开挖变形风险评估的准确性,提出基于约束随机场的基坑开挖变形概率评估方法。所提方法利用克里金(Kriging)理论和随机场理论,生成考虑地勘钻孔数据的约束随机场,并结合数值分析和蒙特卡洛(Monte-Carlo)方法建立基坑开挖的概率评估模型,分析参数空间变异性对基坑开挖引起的环境效应和围护结构变形的影响规律。与传统确定性分析和基于传统随机场的概率分析进行对比,结果表明:基于约束随机场的概率分析方法能够很好地避免传统确定性分析低估风险以及传统随机场方法高估风险的问题,从而更合理地评估工程风险。研究成果可以为进一步制定基坑开挖风险的应对措施提供有益参考。

深厚覆盖层上沥青混凝土心墙坝非平稳随机地震动响应分析

由于良好的环境适应能力沥青混凝土心墙坝是我国西南强震区深厚覆盖层场地的优选坝型,然而对于沥青混凝土心墙坝非平稳随机地震作用下的抗震安全评价研究尚有不足。采用改进的Clough-Penzien功率谱结合随机函数,充分考虑地震动频率与强度的非平稳特征,生成了一系列与反应谱拟合的非平稳随机地震动集合,以某深厚覆盖层上沥青混凝土心墙坝为例,对其坝顶水平加速度及竖向永久变形等响应的均值及变化范围等进行了统计分析及分布检验,揭示了覆盖层与坝体响应均值及变异性沿高程变化规律,并通过概率密度演化方法,展示了坝顶水平加速度概率密度沿时间的演化过程,同时对生成的55条非平稳随机地震动进行调幅,结合多条带分析法易损性分析方法开展了深厚覆盖层上沥青混凝土心墙坝非平稳随机地震动易损性分析,给出了以震陷率为破坏等级控制指标的坝体易损性曲线,为深厚覆盖层上沥青混凝土心墙坝的抗震安全评价提供了新的方法和思路。

近断层地震下连续梁桥随机振动与易损性分析

由于桥梁建设场地难免会靠近地震活动断层,为了评定近断层地震动作用下连续梁桥的地震安全性水平以及提高抗震性能,亟待发展连续梁桥非线性随机地震动力响应与易损性分析的统一、准确方法。考虑到近断层地震动是典型的随机过程激励,本文提出了基于直接概率积分法的连续梁桥结构非线性随机振动响应与地震易损性准确、通用分析的新框架。发展直接概率积分法,准确高效地计算连续梁桥结构的随机振动响应;结合连续梁桥多种失效状态,根据动力可靠度的首次超越破坏准则,利用直接概率积分法对连续梁桥进行了地震易损性分析。最后,展示了近断层随机地震动激励作用下四跨非线性连续梁桥算例结果,验证了直接概率积分法的准确性特点。比较了近断层脉冲型和无脉冲地震动对非线性连续梁桥地震易损性曲线的影响。结果表明:近断层地震动的速度脉冲显著加剧了连续梁桥结构受到的地震破坏,增大了桥梁结构进入不同破坏状态的超越概率。

基于超参数优化随机森林算法的森林生物量遥感反演

【目的】准确地估测森林地上生物量(above ground biomass,AGB)对大区域森林资源调查和管理至关重要,机器学习算法能实现森林AGB高精度估测,但超参数的设置能直接影响模型效果。为了提升模型的构建效率和预测精度,研究通过构建超参数优化的机器学习算法进行森林AGB估测,并比较不同超参数下的模型误差变化。【方法】以西藏自治区江达县天然林为研究对象,利用森林资源调查数据提取实测森林AGB数据,结合Sentinel-2多光谱影像提取遥感变量。采用逐步回归法和Boruta法分别进行遥感变量筛选,构建多元线性回归模型、支持向量机模型和随机森林模型进行森林AGB反演。此外,对支持向量机模型和随机森林模型进行超参数优化,以提高模型反演精度。【结果】1)随机森林模型在所有反演模型中实现了最佳的估测精度,模型决定系数达到了20.63,同时实现了最低的均方根误差和相对均方根误差,分别为28.06 t/hm和23.03%。均方根误差相比多元线性回归模型和支持向量机模型分别降低了22.2%和12.1%。2)超参数优化可以有效地提高模型估测精度。通过分析不同参数组合下的误差变化趋势,确定最佳的参数组合,能有效地降低模型估测误差。3)较高的森林AGB值主要分布在东部、南部和东南部地区,中部地区和北部部分地区森林AGB值较小。超参数优化的随机森林模型森林AGB反演结果与研究区实际森林分布情况具有较好的一致性,整体反演效果较好。【结论】利用超参数优化的随机森林模型结合Sentinel-2遥感影像能实现较好的森林AGB反演效果,能为森林资源动态监测提供有效参考。

强震动随机性对重力坝动水压力影响研究

针对现行水工抗震规范推荐的动水作用计算方法未考虑地震动随机性的问题,研究了强震动随机性对动水压力的影响。首先以Westergaard级数解为基础,探讨声固耦合法模拟动水压力的可行性;然后以选取的25条天然波和生成的25条人工波为基础,研究强震动随机性对重力坝动水压力分布规律和最大值的影响。结果显示,Westergaard级数解仅适用低频激振力,声固耦合法对不同频率的激振力均适用;在动水压力分布规律方面,刚性地基情况下与Westergaard附加质量法规律一致,弹性地基情况下与规范推荐的动水压力系数规律相似;在强震动随机性影响方面,人工波随机性带来的动水压力最大值对均值的偏离程度为16.1%~22.2%,天然波随机性带来动水压力最大值对均值的偏离程度为39.4%~49.8%,天然波随机性对动水压力的影响是人工波的2.0~2.5倍。

基于随机场理论的隧道开挖后地表及隧道变形分析

天然土体由于受到多因素的影响,土体参数存在着内在的不确定性。为了更好地反映实际土体参数的分布情况,基于随机场理论,在考虑黏聚力和内摩擦角空间变异性的情况下,利用乔列斯基分解和随机有限元模型相结合的方法,对地表和隧道变形进行分析。研究结果表明:地表沉降曲线簇的波动范围随着黏聚力和内摩擦角相关距离的增大而增大,其竖直相关距离对地表沉降变形的影响更明显;黏聚力和内摩擦角的变异系数对地表沉降变形的影响强于其相关距离对地表沉降变形的影响;隧道变形值的分布随着黏聚力和内摩擦角变异系数的增大而逐渐离散,其中内摩擦角变异系数对隧道变形值分布的影响更明显;不考虑黏聚力和内摩擦角的空间变异性将偏小地预测地表沉降和隧道变形,使设计偏于危险。

MIMO探地雷达随机步进频信号旁瓣抑制

针对MIMO探地雷达中随机步进频信号的距离旁瓣较高的问题,提出了一种基于快速二阶锥优化问题的旁瓣抑制方法。该方法首先通过建立随机步进频信号模型对回波信号进行预处理,然后基于回波信号的旁瓣特点将旁瓣抑制问题转化为以信号的峰值旁瓣构建目标函数,以主峰损失、峰值旁瓣和积分旁瓣联合约束的二阶锥优化问题,最后使用内点法求解二阶锥规划问题,以此得到最优的旁瓣滤波器系数。仿真和实测数据处理结果表明,设计的失配滤波器能够在较低主峰损耗的情况下,有效降低峰值旁瓣电平和积分旁瓣电平,提高信号的主旁瓣比。该方法为MIMO探地雷达中的距离旁瓣问题提供了有效解决思路,有望在工程应用中提高探地雷达的性能。

顾及地形因素的S-RVOG模型和PD相干最优算法联合反演植被高度

针对极化干涉SAR植被高度反演中RVOG模型未考虑地形影响,且三阶段算法受到地面相位估计误差和纯体相干性估计误差影响,提出了一种植被高度反演思路,采用考虑地形因素的S-RVOG模型作为反演模型校正地形影响,同时引入PD相干最优算法用于改善三阶段算法中直线拟合地表相位估计和纯体相干性估计精度。为验证算法的有效性,首先采用欧空局提供的PolSARpro软件模拟了不同地形坡度水平的PolInSAR数据进行仿真试验,然后采用德国宇航局提供的E-SAR机载全极化SAR数据进行真实植被场景测试,并进行了定性和定量分析。结果表明,本文方法对于不同坡度水平数据,均能有效改善传统RVOG反演模型中地形影响和三阶段算法自身误差影响,反演精度更高。

基于S-RVoG模型的PolInSAR森林高度非线性复数最小二乘反演算法

针对经典的PolInSAR森林高度三阶段几何反演算法在单基线条件容易受到地体幅度比假设以及地形坡度影响的问题,从测量平差角度提出了基于S-RVoG模型的PolInSAR非线性复数最小二乘森林高度反演算法。该算法不再需要假设某一个极化通道地体幅度比为零,且采用考虑地形坡度影响的S-RVoG模型作为平差模型。为了验证算法,本文采用欧空局BioSAR2008项目提供的3景P波段极化干涉SAR数据进行两组单基线森林高度反演试验。结果表明,在单基线条件下,基于RVoG模型的非线性复数最小二乘算法反演结果优于三阶段几何反演算法,而基于S-RVoG模型的非线性复数最小二乘算法进一步提高反演精度,对于坡度较大区域(坡度>10°),精度平均提高了18.48%。

烂泥山水库沥青混凝土心墙坝抗震安全分析

沥青混凝土心墙坝是一种水利工程的坝型设计,其核心部分是使用沥青混凝土作为心墙。为了提高沥青混凝土心墙坝的抗震安全性,以贵州省铜仁市碧江区烂泥山水库工程为研究对象,结合坝体材料空间变异性与随机地震动进行随机响应分析。实验结果表明,综合材料空间变异性和随机地震动两种随机因素的坝顶水平峰值加速度更大,分别为4.48 m/s^(2)、5.68 m/s^(2)。研究方法对烂泥山沥青混凝土心墙坝抗震安全分析具有一定的参考价值。

Rupture directivity and hanging wall effect in near field strong ground motion simulation

A random synthesis procedure based on finite fault model is adopted for near field strong ground motion simulation in this paper. The fault plane of the source is divided into a number of sub-sources, the whole moment magnitude is also divided into more sub-events. The Fourier spectrum of ground motion caused by a sub-event in given sub-source, then can be derived by means of taking the point source spectrum, attenuation with distance, energy dissipation, and near surface effect, into account. A time history is synthesized from this amplitude spectrum and a random phase spectrum, and being combined with an envelope function. The ground motion is worked out by superposition of all time histories from each sub-event in each sub-source, with time lags determining by the differences between the triggering times of sub-events and distances of the sub-sources. From the example of simulations at 21 near field points in a scenario earthquake with 4 dip angles of the fault plane, it is illustrated that the procedure can describe the rupture directivity and hanging wall effect very well. To validate the procedure, the response spectra and time histories recorded at three near fault stations MCN, LV3 and PCD during the Northridge earthquake in 1994, are compared with the simulated ones.

SEISMIC RANDOM VIBRATION ANALYSIS OF STOCHASTIC STRUCTURES USING RANDOM FACTOR METHOD

Seismic random vibration analysis of stochastic truss structures is presented. A new method called random factor method is used for dynamic analysis of structures with uncertain parameters, due to variability in their material properties and geometry. Using the random factor method, the natural frequencies and modeshapes of a stochastic structure can be respectively described by the product of two parts, corresponding to the random factors of the structural parameters with uncertainty, and deterministic values of the natural frequencies and modeshapes obtained by conventional finite element analysis. The stochastic truss structure is subjected to stationary or non-stationary random earthquake excitation. Computational expressions for the mean and standard deviation of the mean square displacement and mean square stress are developed by means of the random variable＇s functional moment method and the algebra synthesis method. An antenna and a truss bridge are used as practical engineering examples to illustrate the application of the random factor method in the seismic response analysis of random structures under stationary or non-stationary random earthquake excitation.

Ground motion record simulation for structural analysis by consideration of spectral acceleration autocorrelation pattern

A novel approach is introduced to generate simulated ground motion records by considering spectral acceleration correlations at multiple periods. Most of the current reliable Ground Motion Record(GMR) simulation procedures use a seismological model including source, path and site characteristics. However, the response spectrum of simulated GMR is somewhat different when compared with the response spectrum based on recorded GMRs. More specifi cally, the correlation between the spectral values at multiple periods is a characteristic of a record which is usually different between simulated and recorded GMRs. As this correlation has a signifi cant infl uence on the structural response, it is needed to investigate the consistency of the simulated ground motions with actual records. This issue has been investigated in this study by incorporating an optimization algorithm within the Boore simulation technique. Eight seismological key parameters were optimized in order to achieve approximately the same correlation coeffi cients and spectral acceleration between two sets of real and simulated records. The results show that the acceleration response spectra of the synthetic ground motions also have good agreement with the real recorded response spectra by implementation of the proposed optimized values.

Seismic Response of Base-Isolated Structures underMulti-component Ground Motion Excitation

An analysis of a base-isolated structure for multi-component random ground motion is presented. The mean square respond of the system is Obtained under different parametric variations. The effectiveness of main parameters and the torsional component during an earthquake is quantified with the help of the response ratio and the root mean square response with and without base isolation. It is observed that the base isolation has considerable influence on the response and the effect of the torsional component is not ignored.

Ground Ozone Level Prediction Using Machine Learning

Because of the increasing attention on environmental issues, especially air pollution, predicting whether a day is polluted or not is necessary to people’s health. In order to solve this problem, this research is classifying ground ozone level based on big data and machine learning models, where polluted ozone day has class 1 and non-ozone day has class 0. The dataset used in this research was derived from the UCI Website, containing various environmental factors in Houston, Galveston and Brazoria area that could possibly affect the occurrence of ozone pollution [1]. This dataset is first filled up for further process, next standardized to ensure every feature has the same weight, and then split into training set and testing set. After this, five different machine learning models are used in the prediction of ground ozone level and their final accuracy scores are compared. In conclusion, among Logistic Regression, Decision Tree, Random Forest, AdaBoost, and Support Vector Machine (SVM), the last one has the highest test score of 0.949. This research utilizes relatively simple methods of forecasting and calculates the first accuracy scores in predicting ground ozone level;it can thus be a reference for environmentalists. Moreover, the direct comparison among five different models provides machine learning field an insight to determine the most accurate model. In the future, Neural Network can also be utilized to predict air pollution, and its test scores can be compared with the previous five methods to conclude the accuracy of Neuron Network.