基于BP神经网络的薄互层储层预测

根据地震属性来进行储层预测的研究,对于寻找油气具有十分重要的意义。而由于地下地质情况千变万化,不确定的因素太多,对这种多自变量与因变量的复杂关系模拟问题,神经网络技术是目前较成熟、实际应用效果也较好的方法之一。用BP神经网络来预测薄互层储层厚度,它可以建立属性参数与预测目标之间的高度非线性映射,并应用于具体的实例,为薄互层储层预测提供了新的思路。     

级配碎石填料大三轴试验及累积塑性应变预测模型

级配碎石作为重载铁路基床表层的主要填料,其受列车荷载的影响最大。因此,研究级配碎石在循环荷载作用下的动力行为及累积塑性应变演化特征变得尤为重要。首先,通过制备不同细粒含量的级配碎石填料,开展一系列大型动三轴试验,探究细粒含量、围压及动应力幅值对循环荷载作用下试样累积塑性应变的耦合影响机制。其次,基于塑性安定理论,确定不同应力水平下试样的动力行为,得到考虑围压及细粒含量参数的塑性蠕变状态临界动应力计算模型。最后,结合试验数据,建立考虑应力水平及细粒含量参数的塑性蠕变动力行为累积塑性应变预测模型,并明确各参数的物理意义。其研究成果可为既有重载铁路路基健康状态评估及考虑强度、变形综合控制的路基结构设计提供参考。     

基于金矿规格单元数据的机器学习方法在成矿建模分析中的应用

现代金矿勘察主要是通过综合地球化学和地质测量等数字化方法对深部矿床进行研究,所需要的人力物力成本较高。而通过分析积累的金矿规格单元数据,可以建立金矿成矿情况与相关成矿元素含量之间的非线性关系,从已有的勘查数据中寻找金矿成矿的一般规律。本文基于与金矿相关的成矿元素含量数据,分别采用逻辑斯蒂回归、随机森林和决策树方法对原始数据和重采样数据进行训练,综合运用召回率、精确率和准确率对模型进行评价。通过对比发现,在训练和测试原始数据过程中,由于每组之间数据量的巨大差距,导致成矿数据被淹没;而在训练重采样数据过程中,随机森林在召回率和准确率方面均有较好的表现,分别达到了90.63%和70.78%;并最终分析了随机森林模型中不同分类边界对于金矿成矿情况预测结果的影响。利用不同的测量指标对模型进行评价分析,使模型更适用于金矿成矿预测,可有效地提高金矿勘察的效率。     

Seismic fracture evaluation of diaphragm joints in welded beam-to-box column moment connections

Steel box columns are widely used in steel building structures in Taiwan due to their dual strong axes. To transfer the beam-end moment to the column, diaphragm plates of the same thickness and elevations as the beam flanges are usually welded inside the box column. The electro-slag welding (ESW) process is widely used to connect the diaphragms to the column flanges in Taiwan because of its convenience and efficiency. However, ESW may increase the hardness of the welds and heat-affected zones (HAZs), while reducing the Charpy-V notch strength in the HAZ. This situation can cause premature fracture of the diaphragm-to-column flange welds before a large plastic rotation is developed in the beam-to-box column joints. To quantify the critical eccentricity and the effectiveness of fracture prediction, this study uses fracture prediction models and finite element model (FEM) analysis to correlate the test results. In this study, two beam-to-box column connection subassembly tests are conducted with different loading protocols and ESW chamber shapes. To implement a fracture prediction model, the material parameters are established from circumferential notched tensile tests and FEM analysis. Test results indicate that the fracture instances can be predicted on the basis of the cumulative plastic deformation in the HAZs. Analytical results indicate that fracture instances and locations are sensitive to the relative locations of the ESW joints and beam flange. Tests also confirm that the possible fracture of the diaphragm-to-column flange joints can be mitigated by enlarging the chamber of the ESW joint.     

The tunnel seismic advance prediction method with wide illumination and a high signal-to-noise ratio

Tunnel seismic prediction is widely used in the field of tunnel seismic advance detection. The illumination of the target and the signal-to-noise ratio of the data are two key factors affecting the precision of data interpretation. Current seismic prospecting has shortcomings on sites: (1) The lighting shots are solely towards one side of the tunnel wall, (2) the geophones are placed far away from the tunnel face and (3) the surface waves from the tunnel wall dominate over the reflection waves, lowering the signal-to-noise ratio of the data at the tunnel wall. This paper proposes a tunnel symmetrical geometry to tackle the above challenges. The arrangement is to place 12 sources uniformly on each side of the tunnel wall and six geophones on the tunnel wall and face. Results of simulated data and measured data show that the proposed method enables (1) broad illumination of the target body, (2) the enhancement of illumination energy of the target body, and (3) higher data signal-to-noise ratio. The proposed symmetrical geometry method provides better interpretation in terms of broader coverage, higher quality and greater distance of investigation.     

Water quality monitoring at a virtual watershed monitoring station using a modified deep extreme learning machine

ABSTRACT

A new deep extreme learning machine (ELM) model is developed to predict water temperature and conductivity at a virtual monitoring station. Based on previous research, a modified ELM auto-encoder is developed to extract more robust invariance among the water quality data. A weighted ELM that takes seasonal variation as the basis of weighting is used to predict the actual value of water quality parameters at sites which only have historical data and no longer generate new data. The performance of the proposed model is validated against the monthly data from eight monitoring stations on the Zengwen River, Taiwan (2002–2017). Based on root mean square error, mean absolute error, mean absolute percentage error and correlation coefficient, the experimental results show that the new model is better than the other classical spatial interpolation methods.     

FluBiDi – a model to estimate flood based on runoff: validation using extreme and natural basin conditions

FluBiDi is a two-dimensional model created to simulate real events that can take days and months, as well as short events (minutes or hours) and inclusive laboratory tests. To verify the robustness of FluBiDi, it was tested using a previous study with both designed and real digital elevation models. The results highlight good agreement between the models (i.e. Mike Flood, SOBEK, ISIS 2D, and others) tested and FluBiDi (around 90% for a specific instant and 95% for the complete time simulation). In the simulated hydrographs, the discharge peak value, time to peak, and water level results were accurate, reproducing them with an error of less than 5%. The velocity differences observed in a couple of tests in FluBiDi were associated with very short periods of time (seconds). However, FluBiDi is highly accurate for simulating floods under real topographical conditions with differences of around 2 cm when water depth is around 150 cm. The average water depth and velocities are precise, and the model describes with high accuracy the pattern and extent of floods. FluBiDi has the capability to be adjusted to different types of events and only requires limited input data.     

Artificial neural network simulation for prediction of suspended sediment concentration in the River Ramganga,Ganges Basin,India

The relation between the water discharge (Q) and suspended sediment concentration (SSC) of the River Ramganga at Bareilly, Uttar Pradesh, in the Himalayas, has been modeled using Artificial Neural Networks (ANNs). The current study validates the practical capability and usefulness of this tool for simulating complex nonlinear, real world, river system processes in the Himalayan scenario. The modeling approach is based on the time series data collected from January to December (2008-2010) for Q and SSC. Three ANNs (T1-T3) with different network configurations have been developed and trained using the Levenberg Marquardt Back Propagation Algorithm in the Matlab routines. Networks were optimized using the enumeration technique, and, finally, the best network is used to predict the SSC values for the year 2011. The values thus obtained through the ANN model are compared with the observed values of SSC. The coefficient of determination (R2), for the optimal network was found to be 0.99. The study not only provides insight into ANN modeling in the Himalayan river scenario, but it also focuses on the importance of understanding a river basin and the factors that affect the SSC, before attempting to model it. Despite the temporal variations in the study area, it is possible to model and successfully predict the SSC values with very simplistic ANN models.     

基于ShakeMap_CNST的2017年精河6.6级地震灾害快速评估

基于考虑区域地震动衰减关系、场地效应及震中破裂等多因素的烈度快速评估模型,结合震害预测方法,研发了一套震害预测系统,并以2017年8月9日精河M_S6.6地震的震害预测为例,将对其的预测结果与当前主流系统进行对比。结果表明,采用ShakeMap_CNST系统能对地震的影响范围和烈度进行更为准确的估计,在人员伤亡、受灾人口估计、紧急安置人数及经济损失评估等方面,相对于其他系统,本系统的结果与现场调查的结果更为接近。     

基于遗传优化神经网络的高速公路路基沉降量预测

控制路基沉降是公路工程中的一个关键技术问题,而路基沉降与其影响因素之间存在着线性、非线性关系。当输入自变量较多时,用传统神经网络建模容易出现过拟合现象,导致网络模型预测精度较低。针对此问题,本文用遗传算法对神经网络模型的权值和阈值进行优化,同时讨论遗传参数的设定对输出结果的影响。通过对成南高速的实测数据进行仿真,试验结果表明:优化后的BP神经网络具有较高的预测精度,预测效果明显优于传统神经网络模型的输出结果,该预测方法可作为高速公路路基长期沉降预测的一种有效辅助手段。