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.

Testing Analysis of Composite Ground with Grouting Piles and Deep Mixing Piles

This paper discusses a new technique to improve soft ground with grouting piles and deep mixing piles. The bearing capacity of composite ground and the stress ratio between piles and soil is discussed by means of the static test. Based on Mindlin solution and Boussinesq solution, the additional stress and settlement of the composite ground are acquired.Compared the practical value with calculation, a better calculating method is confirmed.

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.

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.

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.

Soil-cement mixture properties and design considerations for reinforced excavation

soil-cement is a mixture produced by grouting or mixing cement with soils. This paper reviews and discusses the general classifications of grouting techniques and the suitability of their applications.The mechanical properties of soil-cement mixture and the influence of sodium silicate added are discussed. Design considerations for deep soil mixed wall(DSMW) for excavation support and vault arch for tunnelling stabilisation are presented. Parameters for the numerical analysis of soil-cement mixture are evaluated and recommended.

Bearing Behaviors of Stiffened Deep Cement Mixed Pile

A series of investigations were conducted to study the bearing capacity and load transfer mechanism of stiffened deep cement mixed （SDCM） pile. Laboratory tests including six specimens were conducted to investigate the frictional resistance between the concrete core and the cementsoil. Two model piles and twenty-four full-scale piles were tested to examine the bearing behavior of single pile. Laboratory and model tests results indicate that the cohesive strength is large enough to ensure the interaction between core pile and the outer cement-soil. The full-scale test results show that the SDCM piles exhibit similar bearing behavior to bored and cast-in-place concrete piles. In general, with the rational composite structure the SDCM piles can transmit the applied load effectively, and due to the addition of the stiffer core, the SDCM piles possess high bearing capacity. Based on the findings of these experimental investigations and theoretical analysi , a practical design method is developed to predict the vertical bearing capacity of SDCM pile.

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.

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。

V型Transformer的遥感影像障碍物提取方法

遥感影像中的障碍物是地震采集观测系统变观的重要依据之一。传统的人工提取障碍物方法效率低,且易受人为因素影响,难以保证结果的一致性,不适用于复杂地表环境及数量庞大的障碍物。当前通用的卷积神经网络自动提取障碍物方法,由于卷积核的尺寸受限,无法直接进行远距离的语义交互,也不能准确提取具有较大跨度且部分被遮蔽的障碍物(乡间道路、河流等)。为此,提出了基于V型全自注意力网络(MTNet)提取遥感影像障碍物的方法。首先,MTNet采用端到端的V型编码器—解码器结构,通过跳跃连接实现信息交互;其次,用具有远距离建模能力的Mix-Transformer模块取代传统卷积层,提取和重建更准确的障碍物多尺度特征;最后,用轻量的块扩展层取代转置卷积,实现上采样和图像分割,重建障碍物信息。实验结果表明,该网络分割障碍物的精度和速度显著优于现有方法,尤其在道路识别方面,优势更明显。

加强融合表情和语音的抑郁症检测模型

抑郁症患者的表情和语音具有直观、易于获取等优点,已被广泛应用于抑郁症检测,但现有研究存在忽略表情变化过程包含的信息在抑郁症检测中的作用,未能将动态表情包含的信息与静态表情、语音有效结合,识别准确度不高等问题。针对上述问题,提出一种用动态表情和语音加强融合静态表情特征的抑郁症检测模型。在语音特征提取模块中加入Bi-LSTM网络,挖掘语音的时序信息,用情感语音迁移学习,再用抑郁症语音训练。表情特征提取模块采用双通道结构,利用混合注意力机制分别提取动态表情和静态表情特征,特征更具判别性。特征加强融合模块用语音和动态表情加强融合静态表情,特征信息互补加强。实验结果表明,所提方法在AVEC2014数据集上检测的RMSE和MAE降低到8.21和6.03,优于目前使用语音和表情检测抑郁症的方法。

钻井法凿井壁后充填大比重水泥浆研制及其耐久性能

为提升西部地区深大钻井井筒第一段高壁后充填置换效果,提出采用大比重水泥浆作为新型壁后充填材料。首先,通过对比试验,优选重晶石粉作为加重剂、硅酸镁铝为悬浮剂。然后,采用正交试验和极差分析法,分析了水灰比、重晶石粉、硅酸镁铝和膨胀剂掺量对大比重水泥浆的密度、初凝时间、流动度、3 d及28 d抗压强度的影响,采用综合平衡法确定了大比重壁后充填材料的最优配合比为水泥∶水∶重晶石粉∶硅酸镁铝∶膨胀剂∶减水剂=1∶0.55∶0.5∶0.03∶0.025∶0.015。硫酸盐溶液侵蚀对比试验表明,与常用的水泥浆充填材料相比,大比重水泥浆充填材料具有可靠的耐久性能。最后,通过XRD和SEM分析了优选组加重水泥浆在3 d和28 d的水化产物和微观结构,揭示其优异抗渗性和耐久性能的机理。研究结果可为西部地区深大钻井井筒壁后充填提供大比重、高强度、低渗透以及耐久性好的新型壁后充填材料。

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。

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。

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。

基于多Agent深度强化学习的无人机协作规划方法

人机协作控制是多无人机任务规划的重要方式。考虑多无人机任务环境协同解释和策略控制一致性需求,提出基于多Agent深度强化学习的无人机协作规划方法。依据任务知识和行为状态,构建基于任务分配Agent的任务规划器,生成人机交互的相互依赖关系;设计一种深度学习强化方法,解决群体行为最优策略和协同控制方法,并利用混合主动行为选择机制评估学习策略。实验结果表明:作为人机交互实例,所提方法通过深度强化学习使群体全局联合动作表现较好,学习速度和稳定性均能优于确定性策略梯度方法。同时,在跟随、自主和混合主动3种模式比较下,可以较好地控制无人机飞行路径和任务,为无人机集群任务执行提供了智能决策依据。

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。