分级加载下岩土流变的神经网络模型

针对分级加载下用线性叠加原理来描述岩土非线性流变存在的问题,采用人工神经网络方法代替传统数学力学方法,建立了分级加载下岩土流变的神经网络模型,并且对具体岩石的蠕变试验曲线进行了模拟。结果表明,该模型能较好地描述岩土的非线性流变,具有较强的泛化能力,为研究岩土的流变特性特别是分级加载下岩土的非线性流变特性提供了一条新途径。     

新近系红层软岩流变特性试验研究

流变是控制岩体稳定的主要因素之一。流变试验结果显示,新近系红层软岩蠕变和松弛曲线具有明显的三阶段变化特征。对比研究新近系粉砂质泥岩和泥质粉砂岩的流变结果,得出粉砂质泥岩较泥质粉砂岩强度低,蠕变量大,达到稳定需要的时间长。经过模型辨识,认为红层软岩流变符合伯格斯模型（Burgers model）。利用改进后的麦夸托（Marquardt）算法求取了红层软岩的流变参数。结果显示,反映岩石流变特性的两个主要指标弹性模量和黏滞系数远低于硬岩的同类指标,充分说明新近系红层软岩具有显著的流变性。获取的流变参数可用于伯格斯模型计算。     

二维气固两相后台阶流动的大涡模拟

采用大涡模拟、Smagorinsky亚格子应力模型及SIMPLER 算法,数值模拟了气固两相流体后台阶流动中各物理量的输运特性。首先,讨论了后台阶流动中分离涡旋结构和回流区长度随入口速度分布、台阶上游长度的变化情况;然后,用颗粒追踪法,计算了不同入流速度、不同粒径的颗粒在后台阶流场中的扩散行为。计算结果表明,选择抛物线型入口速度分布,能比较好地模拟真实的后台阶流动;在气固两相流动中,颗粒运动是气相涡旋和固相颗粒惯性共同作用的结果,随着颗粒粒径的增加,颗粒对气体涡旋的跟随性逐渐变弱,而其惯性特性表现得越来越显著。当大量颗粒由不同粒径组成时,颗粒的扩散将具有选择性,显示出多尺度耦合现象。所得结果有助于解释风成地貌形态的形成机理,也可以为风沙流风洞模拟实验设计提供一定的理论指导。     

长江三峡库区木鱼包滑坡地表变形规律分析

三峡库区木鱼包滑坡自2006年实施专业监测以来,一直持续变形,对三峡大坝工程和长江航道造成巨大威胁。通过多次野外地质调查资料、长期现场巡查、人工GPS位移监测数据、近1年的全自动监测数据等,深入分析该滑坡在库水涨落及降雨条件下的变形特征、演化规律及变形机制。结果表明,滑坡坡体结构、岩性及地质构造等地质因素控制了木鱼包滑坡的变形,库水位是主要的驱动因素。库水位上升过程中,库水位由145 m升到155 m左右,月位移量为最小值;动水压力向坡内,滑坡变形最小;库水位155 m上升至175 m期间,库水入渗前部坡体,对滑坡前部抗滑段形成浮托减重效应,变形有所增加。库水位由175 m下降到170 m左右,累积位移形成阶跃,坡受向坡外动水压力和浮托减重效应作用,月位移达最大值。库水位由170 m降到145 m期间,浮托减重效应作用减小,月位移量降低。目前,木鱼包滑坡变形趋势减小,产生大规模滑动的可能性较小,但须进一步加强监测和机制研究。     

How turbidity current frequency and character varies down a fjord‐delta system: Combining direct monitoring,deposits and seismic data

Submarine turbidity currents are one of the most important processes for moving sediment across our planet; they are hazardous to offshore infrastructure, deposit petroleum reservoirs worldwide, and may record tsunamigenic landslides. However, there are few studies that have monitored these submarine flows in action, and even fewer studies that have combined direct monitoring with longer‐term records from core and seismic data of deposits. This article provides one of the most complete studies yet of a turbidity current system. The aim here is to understand what controls changes in flow frequency and character along the turbidite system. The study area is a 12 km long delta‐fed fjord (Howe Sound) in British Columbia, Canada. Over 100 often powerful (up to 2 to 3 m sec^?1) events occur each year in the highly‐active proximal channels, which extend for 1 to 2 km from the delta lip. About half of these events reach the lobes at the channel mouths. However, flow frequency decreases rapidly once these initially sand‐rich flows become unconfined, and only one to five flows run out across the mid‐slope each year. Many of these sand‐rich, channelized, delta‐sourced flows therefore dissipated over a few hundred metres, once unconfined, rather than eroding and igniting. Upflow migrating bedforms indicate that supercritical flow dominated in the proximal channels and lobes, and also across the unconfined mid‐slope. These supercritical flows deposited thick sand beds in proximal channels and lobes, but thinner and finer beds on the unconfined mid‐slope. The distal flat basin records far larger volume and more hazardous events that have a recurrence interval of ca 100 years. This study shows how sand‐rich delta‐fed flows dissipate rapidly once they become unconfined, that supercritical flows dominate in both confined and unconfined settings, and how a second type of more hazardous, and much less frequent event is linked to a different scale of margin failure.     

阶区对走滑型地震地表破裂带传播与终止行为的影响

系统收集了国内外28个走滑地震破裂带上阶区与地震破裂行迹的资料,利用数理统计的方法分析了阶区的类型、尺度与地震破裂的关系。结果表明,不同震级档的走滑地震阶区限制破裂传播的止裂尺度是不同的,震级MS为6.5~6.9,阶区的最小止裂宽度约为3km;震级MS为7.0~7.5,阶区的最小止裂宽度约为4km;震级MS为7.5~8.0,阶区的最小止裂宽度约为6km;震级MS为8.0~8.5,阶区的最小止裂宽度约为8km;且拉分阶区比挤压阶区更容易被破裂所贯通。上述给出的阶区的最小止裂宽度可作为判定地震破裂止裂尺度的重要标志,是进行破裂分段的前提和基础,对地震危险性分析具有重大的实用价值。     

Operational flood control of a low-lying delta system using large time step Model Predictive Control

The safety of low-lying deltas is threatened not only by riverine flooding but by storm-induced coastal flooding as well. For the purpose of flood control, these deltas are mostly protected in a man-made environment, where dikes, dams and other adjustable infrastructures, such as gates, barriers and pumps are widely constructed. Instead of always reinforcing and heightening these structures, it is worth considering making the most of the existing infrastructure to reduce the damage and manage the delta in an operational and overall way. In this study, an advanced real-time control approach, Model Predictive Control, is proposed to operate these structures in the Dutch delta system (the Rhine–Meuse delta). The application covers non-linearity in the dynamic behavior of the water system and the structures. To deal with the non-linearity, a linearization scheme is applied which directly uses the gate height instead of the structure flow as the control variable. Given the fact that MPC needs to compute control actions in real-time, we address issues regarding computational time. A new large time step scheme is proposed in order to save computation time, in which different control variables can have different control time steps. Simulation experiments demonstrate that Model Predictive Control with the large time step setting is able to control a delta system better and much more efficiently than the conventional operational schemes.     

The reffection of regular and irregular waves by a partially perforated caisson breakwater on a step bed

This study examines the reffection of regular and irregular waves from a partially perforated caisson breakwater located on a step bed. The step bed is treated as an idealized rubble mound foundation. Based on the linear potential theory, an analytical solution is developed to calculate the reffection coefficient of the structure subjected to regular waves. The matched eigenfunction expansion method is used for the solution. The regular wave method is also extended to irregular waves using a linear transfer...     

基于TDD的科考船航迹分段方法研究

"雪龙"号极地科考船是推动我国极地科学考察事业发展的重要工具,"雪龙"号在数十次的极地科考过程中累积了大量的航迹数据,其中蕴含的巨大价值亟须挖掘。针对科考船的航迹分段是将科考船移动轨迹分为停留与行驶两部分,合理的分段方法可以分离出信息更丰富的航迹段,有利于航迹知识提取。然而,由于原始航迹信息密度分布不均等原因,现有的航迹分段方法往往会造成分段过多等问题,结果并不理想。本文针对该问题,提出了一种针对科考航迹整体的时域差分(Time Domain Difference, TDD)分段方法。本方法基于时间域对航速进行差分处理,有效降低了因为航速波动频繁对分段结果的影响。同时,考虑到该方法的差分步长在航迹处理过程中的不明确性,本文将差分后航迹的路程损失和航速波动幅值进行归一化处理,提出了航迹差分时间步长的动态确定方法,并以速率阈值对航迹进行分段。最后本文以第29次南极科考航迹数据为例,将本方法与经典的具有噪声的基于密度的聚类方法 DBSCAN(Density-Based Spatial Clustering of Applications with Noise)进行了比较,实验结果表明本文提出的方法可有效降低航迹分段时分段过多的问题,在分段准确性和时间效率等方面结果更优。     

Fractal geochronology — dating of the past, planning for the future

Dates for regional-scale geological events are typically determined by analysis of radioactive decay processes within micron-scale closed systems, demonstrating that geochronology information is “fractal” — that is, largely independent of the scale of magnification. This has important implications for the way geochronology data should be stored, retrieved and processed in order to maximize extraction of useful geological information.