Engineering behaviour of in situ cored deep cement mixed marine deposits subjected to undrained and drained shearing

The deep cement mixing(DCM)is used to improve the capacity and reduce the settlement of the soft ground by forming cemented clay columns.The investigation on the mechanical behaviour of the DCM samples is limited to either laboratory-prepared samples or in-situ samples under unconfined compression.In this study,a series of drained and undrained triaxial shearing tests was performed on the in-situ cored DCM samples with high cement content to assess their mechanical behaviours.It is found that the drainage condition affects significantly the stiffness,peak and residual strengths of the DCM samples,which is mainly due to the state of excess pore water pressure at different strain levels,i.e.being positive before the peak deviatoric stress and negative after the peak deviatoric stress,in the undrained tests.The slope of the failure envelope changes obviously with the confining pressures,being steeper at lower stress levels and flatter at higher stress levels.The strength parameters,effective cohesion and friction angle obtained from lower stress levels(c′0 andφ′0)are 400 kPa and 58°,respectively,which are deemed to be true for design in most DCM applications where the in-situ stress levels are normally at lower values of 50-200 kPa.Additionally,the computed tomography(CT)scanning system was adopted to visualize the internal structures of DCM samples.It is found that the clay pockets existing inside the DCM samples due to uneven mixing affect markedly their stress-strain behaviour,which is one of the main reasons for the high variability of the DCM samples.

Reliability assessment for serviceability limit states of stiffened deep cement mixing column-supported embankments

The reliability and deterministic analyses of wood-cored stiffened deep cement mixing and deep cement mixing column-supported embankments(referred to as WSCSE and DCSE,respectively)considering serviceability limit state requirements are presented in this paper.Random field theory was used to simulate the spatial variability of soil
cement mixing(SCM)material in which the adaptive Kriging Monte Carlo simulation was adopted to estimate the failure probability of a column
supported embankment(CSE)system.A new method for stochastically generating random values of unconfined compressive strength(qu)and the ratio(Ru)between the undrained elastic modulus and qu of SCM material based on statistical correlation data is proposed.Reliability performance of CSEs concerning changes in the mean(μ),coefficient of variation(CoV),and vertical spatial correlation length(θv)of qu and Ru are presented and discussed.The obtained results indicate that WSCSE can provide a significantly higher reliability level and can tolerate more SCM material spatial variability than DCSE.Some performance of DCSE and WSCSE,which can be considered satisfactory in a deterministic framework,cannot guarantee an acceptable reliability level from a probabilistic viewpoint.This highlights the importance and necessity of employing reliability analyses for the design of CSEs.Moreover,consideration of only μ and CoV of qu seems to be sufficient for reliability analysis of WSCSE while for DCSE,uncertainties regarding the Ru(i.e.both μ and CoV)and θv of qu cannot be ignored.

Field testing of stiffened deep cement mixing piles under lateral cyclic loading

Construction of seaside and underground wall bracing often uses stiffened deep cement mixed columns （SDCM）. This research investigates methods used to improve the level of bearing capacity of these SDCM when subjected to cyclic lateral loading via various types of stiffer cores. Eight piles, two deep cement mixed piles and six stiffened deep cement mixing piles with three different types of cores, H shape cross section prestressed concrete, steel pipe, and H-beam steel, were embedded though soft clay into medium-hard clay on site in Thailand. Cyclic horizontal loading was gradually applied until pile failure and the hysteresis loops of lateral load vs. lateral deformation were recorded. The lateral carrying capacities of the SDCM piles with an H-beam steel core increased by 3-4 times that of the DCM piles. This field research clearly shows that using H-beam steel as a stiffer core for SDCM piles is the best method to improve its lateral carrying capacity, ductility and energy dissipation capacity.

Observed characteristics of flow,water mass,and turbulent mixing in the Preparis Channel

Preparis Channel is the very important exchange path of energy and materials between the northern Bay of Bengal and Andaman Sea(AS).A set of hydrographic measurements,a microstructure profiler,and a deep mooring were used to determine the characteristics of water masses,turbulent mixing,and flows in the Preparis Channel.The unprecedented short-term mooring data reveal that a deep current in the deep narrow passage(below 400 m)of the Preparis Channel flows toward the Bay of Bengal(BoB)with a mean along-stream velocity of 25.26 cm/s at depth of 540 m;above the deep current,there are a relatively weak current flows toward the AS with a mean along-stream velocity of 15.46 cm/s between 500 m and 520 m,and another weak current flows toward the BoB between 430 m and 500 m.Thus,a sandwiched vertical structure of deep currents(below 400 m)is present in the Preparis Channel.The volume transport below 400 m is 0.06 Sv(1 Sv=106 m^(3)/s)from the AS to the BoB.In the upper layer(shallower than 300 m),the sea water of the AS is relatively warmer and fresher than that in the BoB,indicating a strong exchange through the channel.Microstructure profiler observations reveal that the turbulent diffusivity in the upper layer of the Preparis Channel reaches O(10−4 m^(2)/s),one order larger than that in the interior of the BoB and over the continental slope of the northern AS.We speculate that energetic high-mode internal tides in the Preparis Channel contribute to elevated turbulent mixing.In addition,a local“hotspot”of turbidity is identified at the deep mooring site,at depth of about 100 m,which corresponds to the location of elevated turbulent mixing in the Preparis Channel.

Prediction of tree crown width in natural mixed forests using deep learning algorithm

Crown width(CW)is one of the most important tree metrics,but obtaining CW data is laborious and timeconsuming,particularly in natural forests.The Deep Learning(DL)algorithm has been proposed as an alternative to traditional regression,but its performance in predicting CW in natural mixed forests is unclear.The aims of this study were to develop DL models for predicting tree CW of natural spruce-fir-broadleaf mixed forests in northeastern China,to analyse the contribution of tree size,tree species,site quality,stand structure,and competition to tree CW prediction,and to compare DL models with nonlinear mixed effects(NLME)models for their reliability.An amount of total 10,086 individual trees in 192 subplots were employed in this study.The results indicated that all deep neural network(DNN)models were free of overfitting and statistically stable within 10-fold cross-validation,and the best DNN model could explain 69%of the CW variation with no significant heteroskedasticity.In addition to diameter at breast height,stand structure,tree species,and competition showed significant effects on CW.The NLME model(R^(2)=0.63)outperformed the DNN model(R^(2)=0.54)in predicting CW when the six input variables were consistent,but the results were the opposite when the DNN model(R^(2)=0.69)included all 22 input variables.These results demonstrated the great potential of DL in tree CW prediction.

Bearing Capacity Assessment of Collapsible Soils Improved by Deep Soil Mixing Using Finite Element Method

Problematic soils usually cause considerable problems to engineering projects. As an example, soil structure collapse caused by moisture increment or rising underground water level results in huge settlements. This type of problematic soil, named collapsible soil, can cause dramatic problems and should be amended where exists. Today, the use of different techniques for soil reinforcement and soil improvement is widely used to treat soil properties. One of these methods is Deep Soil Mixing (DSM) method. This method becomes more important in the cases of studying and examining collapsible soils. In this research, the settlement of amended collapsible soils, applying deep soil mixing method, is examined. The experiments show that soil amendment using this method, well prevents the settlement of collapsible soils giving rise to bearing capacity.

Mixed Noise Removal by Residual Learning of Deep CNN

Due to the huge difference of noise distribution,the result of a mixture of multiple noises becomes very complicated.Under normal circumstances,the most common type of mixed noise is to add impulse noise(IN)and then white Gaussian noise(AWGN).From the reduction of cascaded IN and AWGN to the latest sparse representation,a great deal of methods has been proposed to reduce this form of mixed noise.However,when the mixed noise is very strong,most methods often produce a lot of artifacts.In order to solve the above problems,we propose a method based on residual learning for the removal of AWGN-IN noise in this paper.By training,our model can obtain stable nonlinear mapping from the images with mixed noise to the clean images.After a series of experiments under different noise settings,the results show that our method is obviously better than the traditional sparse representation and patch based method.Meanwhile,the time of model training and image denoising is greatly reduced.

Bearing Behaviors of Stiffened Deep Cement Mixed Pile

一系列调查被进行学习适用的能力和混合的使硬的深水泥(SDCM ) 的负担转移机制堆积。实验试验包括六个标本被进行调查在具体核心和 thecement 土壤之间的摩擦抵抗。二模型堆积和 24 照原尺寸的堆积被测试检验单个堆积的适用的行为。实验室和模型测试结果显示连贯的力量足够大保证在核心堆积和外部水泥土壤之间的相互作用。full-scaletest 结果证明 SDCM 堆积展出类似的适用的行为到无聊并且 cast-in-placeconcrete 堆积。一般来说，与合理合成结构， SDCM 堆积能有效地播送应用负担，并且由于更生硬的核心的增加， SDCM 堆积拥有高适用的能力。基于这些试验性的调查和理论分析的调查结果，一个实际设计方法被开发预言 SDCMpile 的垂直适用能力。

基于K-means聚类和特征空间增强的噪声标签深度学习算法

深度学习中神经网络的性能依赖于高质量的样本,然而噪声标签会降低网络的分类准确率。为降低噪声标签对网络性能的影响,噪声标签学习算法被提出。该算法首先将训练样本集划分成干净样本集和噪声样本集,然后使用半监督学习算法对噪声样本集赋予伪标签。然而,错误的伪标签以及训练样本数量不足的问题仍然限制着噪声标签学习算法性能的提升。为解决上述问题,提出基于K-means聚类和特征空间增强的噪声标签深度学习算法。首先,该算法利用K-means聚类算法对干净样本集进行标签聚类,并根据噪声样本集与聚类中心的距离大小筛选出难以分类的噪声样本,以提高训练样本的质量;其次,使用mixup算法扩充干净样本集和噪声样本集,以增加训练样本的数量;最后,采用特征空间增强算法抑制mixup算法新生成的噪声样本,从而提高网络的分类准确率。并在CIFAR10、CIFAR100、MNIST和ANIMAL-10共4个数据集上试验验证了该算法的有效性。

全尺度密集卷积U型网络的视网膜血管分割算法

针对视网膜图像中血管尺度跨度大、细小血管与背景高度相似导致误分割和未分割等问题,提出一种全尺度密集卷积U型网络的视网膜血管分割方法。为提取更复杂的特征信息,构建级联卷积融合密集块(cascade convolutional fusion dense blocks, CCF-DB)作为U型网络的编解码器用以提取视网膜血管的特征信息;在网络最底端嵌入混合注意力级联卷积密集块(mixed attention cascaded convolutional dense block, MACC-DB),进一步提升感受野,获取更高维的语义特征信息;在模型的解码部分采用全尺度的跳跃连接,捕获不同尺度下的血管特征信息,提升模型的分割精度。实验结果表明,在DRIVE数据集上,相比于U-Net、U-Net3+、SA-Unet、FR-Unet等算法,此算法的AUC值达到了98.26%,准确率为95.82%;在CHASE-DB1数据集上,此算法的AUC值达98.84%,准确率达96.66%。采用此算法进行视网膜血管分割,分割的精度和鲁棒性均有不同程度的提升,对细小血管分割达到了优良的效果。

饱和地基水泥土复合桩近场主动隔振的BEM-FEM耦合分析

人工振动污染是当前城市发展过程中面临的一个难题,单排或多排桩是人工振动污染防治的一种重要方式。传统基桩施工容易造成泥浆污染、振动等各种负面环境问题,而水泥土复合桩是一种低振动、低噪声、无挤土、少排泥浆的新型基桩,特别适合在城区构建排桩屏障。针对饱和地基中单排水泥土复合桩的近场主动隔振问题,建立了动力机器基础环境振动影响的半解析BEM方程;在此基础上,分别采用半解析BEM对饱和地基、FEM对水泥土复合桩进行建模,根据饱和地基-水泥土桩接触面的平衡和相容条件,建立了水泥土复合桩近场主动隔振的半解析BEM-FEM耦合方程,给出了耦合方程稀疏矩阵的存储策略,并对单排水泥土复合桩的近场主动隔振效果进行了计算分析。研究结果表明:单排水泥土复合桩能够有效地对动力机器基础引起环境振动进行隔振;等长芯桩复合桩的隔振效果要优于短芯桩。单排水泥土复合桩的隔振效果随着桩数的增加而逐渐提高;较小桩间距并不一定取得更好的隔振效果,建议相邻桩桩间净距取2.0~2.5λ_(R)(Rayleigh波波长)。随着单排水泥土复合桩距振源距离的增加,屏障隔振效果逐渐降低,但降低幅度相对较小。此外,对水泥土复合桩而言,当内插预制桩的外轮廓尺寸相同时,预制芯桩型式对水泥土复合桩隔振效果影响相对较小,可根据实际工程条件选择合适的芯桩类型。

基于深度学习的弱反馈自混合干涉信号滤波方法

采集激光自混合干涉信号的过程中会受到环境和电路噪声的干扰,导致信号失真。为了去掉噪声,最大限度保留原信号特征,提出了基于深度学习的自混合干涉滤波方法,该方法适用于弱反馈条件。使用自编码器作为神经网络,用加入噪声的信号作为输入,未加噪声的信号作为输出来训练网络。仿真结果表明:该方法处理含噪声自混合干涉信号时不仅能提高含噪声信号的信噪比,还能很好地保留干涉条纹的波形特征,即条纹的倾斜方向。实验中,使用深度学习方法滤波,再用条纹计数法进行位移重构,结果表明该方法对弱反馈条件下的自混合干涉信号有较好的滤波效果。

水泥分解炉SNCR脱硝系统的深度强化学习多目标优化控制研究

选择性非催化还原(selective non-catalytic reduction,SNCR)脱硝过程的工艺参数优化可以有效减少水泥分解炉NO_(x)排放和脱硝运行成本。以某水泥分解炉为研究对象,建立基于LightGBM的NO_(x)浓度预测模型,以脱硝成本和NO_(x)浓度最小化为优化目标,采用深度确定性策略梯度(deep deterministic policy gradient,DDPG)算法对水泥分解炉掺烧污泥协同SNCR脱硝过程的相关工艺参数进行优化控制建模。结果表明,NO_(x)浓度预测模型均方根误差(root mean squared error,RMSE)为6.8,平均绝对百分比误差(mean absolute percentage error,MAPE)为3.48%;采用DDPG算法可以对相关工艺参数进行优化,喷氨量和污泥掺烧量分别为427.87 L/h和9.78 t/h时,NO_(x)排放浓度为225.99 mg/(Nm^(3)),脱硝运行成本为1 747.8元/h。该优化结果与其他优化算法结果和常规工况对比,NO_(x)排放浓度和脱硝运行成本均呈现不同程度下降;对模型进行仿真及效果验证可知,所建立模型能输出合理的喷氨量和污泥掺烧量组合,减少SNCR出口NO_(x)浓度波动,有效降低NO_(x)排放浓度和脱硝成本,可实现对SNCR脱硝系统的多目标优化控制。该结果可为基于智能算法的水泥分解炉SNCR脱硝的多目标优化控制设计提供一定参考。

选铁尾渣铜、钴、硫综合回收工艺研究及半工业调试

尾渣的二次资源化利用对缓解资源紧张、矿业可持续发展具有十分重要的意义。秘鲁某选铁尾渣量大,含有较高的Cu、Co、S等有价元素,其金属矿物在海水富离子介质条件下受溶解蚀变及上游选厂铁精矿脱硫降锌作业大量浮选药剂深度活化,分离回收难度较大。通过对矿石性质研究,采用阶段磨矿-混合浮选-铜与硫钴分离的工艺,在实验室小型试验的基础上进行半工业调试,取得较好的Cu、Co、S综合回收效果。

深度学习视域下大学英语写作教学中线上线下混合教育原则分析

写作是提升学生英语能力的重要途径之一,但是,现阶段大学英语写作教学并不尽如人意,如学生写作素材储备不足、写作内容漏洞百出等问题已成为制约大学生写作成长的主要因素。事实上,每个人与生俱来的表达愿望即是写作的最有效动力,当学生能将心中所想、所感以文字形式表达出来时,学生的文章自然会“花开蝶自来”。基于此,本文从大学英语写作教学入手,以深度学习理念为指导,从线上线下混合教育模式的应用出发,围绕写作教学的误区,分享几种较为可行的大学英语写作教学方法。

考虑训练样本分布不均衡的超短期风电功率概率预测

提出一种考虑训练样本分布不均衡的超短期风电概率预测方法。首先构建深度信念混合密度网络,通过深度信念网络独特的预训练和微调机制提取输入变量的隐特征,利用Beta混合概率分布的有界性准确表征风电预测功率的概率分布,实现隐特征与预测功率概率分布参数之间的非线性映射;然后引入训练样本分布平滑策略,其中特征分布平滑技术用于校准输入特征,标签分布平滑技术用于对各样本误差赋予差异化权重,从输入和输出两方面改善训练样本分布不均衡现象对预测结果的不利影响。实际算例结果表明,与常见风电功率概率预测模型相比,所提模型在点预测和概率预测方面均能获得较高的预测精度,尤其能有效提高低密度样本区域的预测精度。

高原深水湖泊抚仙湖溶解氧分层特征及驱动因素

深水湖泊特有的“表水层-温跃层-深水层”的热力学分层结构,决定了湖内溶解氧(DO)的垂直分布和混合交换,影响着湖内生态系统的健康.然而,目前对于深水湖泊耗氧与复氧过程及其驱动因素的研究还不够深入,尤其对高原深水湖泊的研究更为缺乏.为此,本研究于2021年1月-2022年2月进行了分层采样和逐月监测,探究了典型高原深水湖泊抚仙湖的溶解氧分层特征及驱动因素.结果表明:①抚仙湖热力学分层周期分为两期,即分层期和非分层期.分层期为3月下旬至11月上旬,混合期为11月下旬至3月上旬.②在湖泊热力学分层的驱动下,湖内DO垂直分层明显.表水层DO浓度年内变化范围为6.67~8.64 mg/L,而温跃层和深水层DO最低浓度分别可降至3.15和1.26 mg/L.③抚仙湖不同水层DO浓度变化的驱动因素及效应占比不同.表水层DO受到水温、光合作用、大气复氧和气象条件的综合影响;温跃层DO主要受分层强度和浮游植物生命活动的影响;深水层DO主要受分层强度、有机质沉降分解、沉积物有机质分解的影响.研究显示,抚仙湖底层水体长期处于厌氧状态,由此带来的湖泊生境的改变以及底部营养盐释放的风险值得关注.

深层水泥搅拌法加固软基模型试验研究

为确定水下深层水泥搅拌各种施工参数,现场试桩需花费较多的时间和成本。依托中交集团DCM(Deep Cement Mixing)法加固软基课题,研发了DCM模型试验装置,开展了DCM加固软基叶片改造、土层识别和成桩强度物理模型试验。结果表明:增加自由叶片后可有效防止水泥搅拌桩在黏土中糊钻问题。不同土层界面处扭矩会发生突变,通过不同深度处扭矩的变化可以感知不同土层。相同水泥掺量前提下不同土样水泥搅拌桩14,28,60,90 d无侧限抗压强度(UCS)规律为淤泥质土<黏土<砂土,在砂土中水泥土强度较大且增长速度较快,砂层中可以适当减少水泥喷浆量。模型试验研究成果,可对深层水泥搅拌桩施工提供参考。

劲芯水泥土桩复合地基承载路堤失稳破坏模型试验

劲芯水泥土桩可有效提高软土地基承载力,减小地基沉降和提高边坡稳定性。目前对于劲芯水泥土桩的研究主要集中在其承受竖向荷载条件下的工程特性上,对于路堤荷载下桩体失稳破坏模式与计算方法的认识尚不全面。开展1g重力场模型试验,对比研究了路堤荷载下天然地基、水泥土桩复合地基、劲芯水泥土桩复合地基路基变形、桩土应力、桩体应变、桩体破坏形态及路堤失稳模式,并评价了现有刚性桩复合地基稳定性计算方法对于劲芯水泥土桩复合地基承载路堤稳定性分析的适用性。结果表明:相较于水泥土桩,劲芯水泥土桩能有效约束地基土变形,路面沉降约为水泥土桩复合地基的61%~71%,坡外隆起约为12%~46%,进而显著提高路堤的稳定性。在路堤荷载下,劲芯水泥土桩表现出渐进式失稳破坏特征,桩体可发生纯压破坏、压弯破坏和弯曲破坏,且与桩体刚度和相对于路堤位置有关,桩体断裂位置与滑动面位置并未重合。等效抗剪强度法对于劲芯水泥土桩承载路堤稳定性分析具有一定的适用性和应用价值。

基于数据驱动知识显式嵌入的配电网最优需求响应策略

充分挖掘多元需求侧资源的灵活性,对于提升分布式新能源广泛接入背景下的新型配电网运行可靠性和经济性具有重要意义。然而,目前关于需求响应策略的解析化方法大多基于较为理想的用户行为观测和参数假设,纯数据驱动方法难以兼顾电网侧运行的复杂约束,策略的可用性存疑。为此,文章提出基于数据驱动知识显式嵌入的需求响应策略,首先,考虑到需求侧资源灵活性的强时段耦合特性,提出需求侧资源动态模型,定量分析需求侧灵活性资源的响应特性;其次,提出数据驱动知识的显式解析方法,将需求侧灵活性描述为混合整数线性模型并嵌入至配电网优化运行模型中,实现灵活实用的新型配电网供需交互与协调运行。最后,通过仿真算例验证所提方法兼具解析模型和数据驱动方法的优势,为不完全信息观测条件下源网荷协调运行提供较为实用化的解决方案。