Diffraction deep neural network-based classification for vector vortex beams

The vector vortex beam(VVB)has attracted significant attention due to its intrinsic diversity of information and has found great applications in both classical and quantum communications.However,a VVB is unavoidably affected by atmospheric turbulence(AT)when it propagates through the free-space optical communication environment,which results in detection errors at the receiver.In this paper,we propose a VVB classification scheme to detect VVBs with continuously changing polarization states under AT,where a diffractive deep neural network(DDNN)is designed and trained to classify the intensity distribution of the input distorted VVBs,and the horizontal direction of polarization of the input distorted beam is adopted as the feature for the classification through the DDNN.The numerical simulations and experimental results demonstrate that the proposed scheme has high accuracy in classification tasks.The energy distribution percentage remains above 95%from weak to medium AT,and the classification accuracy can remain above 95%for various strengths of turbulence.It has a faster convergence and better accuracy than that based on a convolutional neural network.

Multi-User MmWave Beam Tracking via Multi-Agent Deep Q-Learning

Beamforming is significant for millimeter wave multi-user massive multi-input multi-output systems.In the meanwhile,the overhead cost of channel state information and beam training is considerable,especially in dynamic environments.To reduce the overhead cost,we propose a multi-user beam tracking algorithm using a distributed deep Q-learning method.With online learning of users’moving trajectories,the proposed algorithm learns to scan a beam subspace to maximize the average effective sum rate.Considering practical implementation,we model the continuous beam tracking problem as a non-Markov decision process and thus develop a simplified training scheme of deep Q-learning to reduce the training complexity.Furthermore,we propose a scalable state-action-reward design for scenarios with different users and antenna numbers.Simulation results verify the effectiveness of the designed method.

Stability influence factors analysis and construction of a deep beam anchorage structure in roadway roof

Deep beam anchorage structures based on spatial distribution analysis of the cable prestressed field have been proposed for roadway roof support, Stability and other factors that influence deep beam structures are studied in this paper using mechanical calculations, numerical analysis and field measurements, A mechanical model of deep beam structure subjected to multiple loading is established, including analysis of roof support in the return airway of S1203 working face in the Yuwu coal mine, China, The expression of maximum shear stress in the deep beam structure is deduced according to the stress superposition criterion, It is found that the primary factors affecting deep beam structure stability are deep beam thickness, cable pre-tension and cable spacing, The variation of maximum shear stress distribution and prestressed field diffusion effects according to various factors are analyzed using Matlah and FLAC3DTM software, and practical support parameters of the S1203 return airway roof are determined, According to the observations of rock pressure, there is no evidence of roof separation, and the maximum values of roof subsidence and convergence of wall rock are 72 and 48 mm, respectively, The results show that the proposed roof support design with a deep beam structure is feasible and achieves effective control of the roadway roof,

Comparison on construction of strut-and-tie models for reinforced concrete deep beams

With consideration of the differences between concrete and steel,three solutions using genetic evolutionary structural optimization algorithm were presented to automatically develop optimal strut-and-tie model for deep beams.In the finite element analysis of the first method,the concrete and steel rebar are modeled by a plane element and a bar element,respectively.In the second method,the concrete and steel are assigned to two different plane elements,whereas in the third method only one kind of plane element is used with no consideration of the differences of the two materials.A simply supported beam under two point loads was presented as an example to verify the validity of the three proposed methods.The results indicates that all the three methods can generate optimal strut-and-tie models and the third algorithm has powerful capability in searching more optimal results with less computational effort.The effectiveness of the proposed algorithm III has also been demonstrated by other two examples.

Analysis and seismic tests of composite shear walls with CFST columns and steel plate deep beams

A composite shear wall concept based on concrete filled steel tube （CFST） columns and steel plate （SP） deep beams is proposed and examined in this study. The new wall is composed of three different energy dissipation elements： CFST columns; SP deep beams; and reinforced concrete （RC） strips. The RC strips are intended to allow the core structural elements - the CFST columns and SP deep beams - to work as a single structure to consume energy. Six specimens of different configurations were tested under cyclic loading. The resulting data are analyzed herein. In addition, numerical simulations of the stress and damage processes for each specimen were carried out, and simulations were completed for a range of location and span-height ratio variations for the SP beams. The simulations show good agreement with the test results. The core structure exhibits a ductile yielding mechanism characteristic of strong column-weak beam structures, hysteretic curves are plump and the composite shear wall exhibits several seismic defense lines. The deformation of the shear wall specimens with encased CFST column and SP deep beam design appears to be closer to that of entire shear walls. Establishing optimal design parameters for the configuration of SP deep beams is pivotal to the best seismic behavior of the wall. The new composite shear wall is therefore suitable for use in the seismic design of building structures.

Flexural behavior of hybrid fiber reinforced high performance concrete deep beam

In order to reveal the flexural behavior of hybrid fiber reinforced high-performance concrete deep beam, 16 high-performance concrete deep beams of different fiber volume content have been tested according to the state standards and testing methods. The effects of hybrid fiber on the yield moment and bending bearing capacity of the cross-section have been analyzed, the calculation method for the bending capacity＇ is discussed and the propositional formula are provided as well. Results shoxv that the flexural properties increased obviously when add ≤1.0% of volume content steel fibers and ≤0.11% of volume content polypropylene fibers in to deep beam. The results are useful to the further amendments of fiber reinforced concrete structure technical regulation （CECS 38：2004）.

Nonlinear Finite Element Analysis of Steel Fiber Reinforced Concrete Deep Beams

By the nonlinear finite element analysis （FEA） method, the mechanical properties of the steel fiber reinforced concrete （SFRC） deep beams were discussed in terms of the crack load and ultimate bearing capacity. In the simulation process, the ANSYS parametric design language （APDL） was used to set up the finite element model; the model of bond stress-slip relationship between steel bar and concrete was established. The nonlinear FEA results and test results demonstrated that the steel fiber can not only significantly improve the cracking load and ultimate bearing capacity of the concrete but also repress the development of the cracks. Meanwhile, good agreement was found between the experimental data and FEA results, if the unit type, the parameter model and the failure criterion are selected reasonably.

Use of a handheld slit beam intraoperatively to assist in big bubble formation during deep anterior lamellar keratoplasty

Deep anterior lamellar keratoplasty(DALK)is preferred over conventional penetrating keratoplasty(PKP)for the treatment of anterior corneal opacities or ectasias due to decreased risk of endothelial rejection.However,DALK remains surgically challenging,largely due to challenges associated with achieving consistent pneumo-dissection of posterior stroma from the underlying pre-Descemet’s or Descemet’s membrane(DM).Air must be injected at sufficient depth in the corneal stroma in order to achieve successful pneumo-dissection,but advancing a needle too deep into the cornea can lead to perforation of DM.We describe here a novel technique using a handheld slit lamp(Eidolon model 510L,Eidolon Optical LLC,Natick,MA,USA)to assist in creation of the big-bubble in DALK surgery.Use of a handheld slit beam intraoperatively is a safe,relatively inexpensive,and effective technique for increasing the success of big-bubble formation in DALK procedures.

Deep Learning Applied to Computational Mechanics:A Comprehensive Review,State of the Art,and the Classics

Three recent breakthroughs due to AI in arts and science serve as motivation:An award winning digital image,protein folding,fast matrix multiplication.Many recent developments in artificial neural networks,particularly deep learning(DL),applied and relevant to computational mechanics(solid,fluids,finite-element technology)are reviewed in detail.Both hybrid and pure machine learning(ML)methods are discussed.Hybrid methods combine traditional PDE discretizations with ML methods either(1)to help model complex nonlinear constitutive relations,(2)to nonlinearly reduce the model order for efficient simulation(turbulence),or(3)to accelerate the simulation by predicting certain components in the traditional integration methods.Here,methods(1)and(2)relied on Long-Short-Term Memory(LSTM)architecture,with method(3)relying on convolutional neural networks.Pure ML methods to solve(nonlinear)PDEs are represented by Physics-Informed Neural network(PINN)methods,which could be combined with attention mechanism to address discontinuous solutions.Both LSTM and attention architectures,together with modern and generalized classic optimizers to include stochasticity for DL networks,are extensively reviewed.Kernel machines,including Gaussian processes,are provided to sufficient depth for more advanced works such as shallow networks with infinite width.Not only addressing experts,readers are assumed familiar with computational mechanics,but not with DL,whose concepts and applications are built up from the basics,aiming at bringing first-time learners quickly to the forefront of research.History and limitations of AI are recounted and discussed,with particular attention at pointing out misstatements or misconceptions of the classics,even in well-known references.Positioning and pointing control of a large-deformable beam is given as an example.

弹性环梁支撑下圆形深基坑支护结构变形解析解

将环梁和支护结构分别视为弹性圆环和弹性圆柱薄壳,基于环梁和圆柱壳的变形协调,得到了具有任意数目弹性环梁支撑圆形深基坑支护结构变形的解析解.在验证解析解合理性和可靠性的基础上,针对某一实际圆形基坑工程,比较了在刚性和弹性环梁支撑下支护结构变形和内力的差异,并分析了支护结构底部边界条件、环梁弹性模量、尺寸和数量以及位置等参数对支护结构变形和内力分布的影响.研究结果表明:相对于刚性环梁支撑,弹性环梁支撑处内力变化较小,环梁弹性模量和尺寸的改变只对基坑挖掘面以上支护结构变形和内力影响显著,而对挖掘面以下的支护结构变形和内力几乎没有影响.该研究成果为圆形基坑支护结构设计提供了理论依据和指导.

基于决策性能评估的多波束低地球轨道卫星网络资源分配算法

为了解决多波束低地球轨道(LEO)卫星波束间同频干扰、频谱短缺、业务量分布不均等问题,针对单一决策网络缺乏自我修正能力、容易陷入局部最优解、无法充分考虑长期影响等弊端,提出了一种基于决策性能评估的资源分配算法。该算法引入不同用户的业务满足指数来衡量系统的公平性,在考虑公平性的前提下优化系统的吞吐量性能,并将该优化问题建模为多目标优化问题。将具有时间相关性的连续资源分配过程建模为马尔可夫过程,提出基于决策性能评估的网络资源分配算法来解决该问题。所提算法可以根据评估网络的评估结果调整决策网络参数,从而优化资源分配方案,同时更新评估网络自身参数。通过迭代优化的方式,实现决策网络的准确预测。仿真结果表明,所提算法在吞吐量性能和公平性方面优于传统资源分配算法。

面向用户多样化业务需求的多波束卫星系统动态资源分配算法

多波束卫星通信系统由于其高吞吐量和高资源利用率而受到广泛关注.已有研究主要考虑多波束卫星通信系统的信道或功率分配问题,但较少考虑用户分组和动态资源分配策略的联合优化设计,导致系统性能受限.此外,现有研究往往假设固定的波束覆盖半径,忽略了波束覆盖半径可变性对波束覆盖性能提升的影响.本文研究了多波束卫星通信系统中的用户分组和资源分配问题,提出了一种两阶段资源管理方案.针对动态和多样化的用户服务需求,首先设计一种基于Voronoi图的迭代用户分组算法以实现分组之间的负载均衡,然后将子信道和功率分配问题建模为系统平均效用函数最大化问题.为解决该问题,将每个波束视为一个智能体,采用一种基于多智能体深度Q网络(Deep Q Network,DQN)的算法来确定子信道和功率分配策略.仿真结果表明,与K-均值用户分组方案相比,本文所提出的基于Voronoi图的迭代用户分组算法对应的用户组负载差异值可降低49.2%,体现了本文所提算法在实现用户组间负载均衡方面的优势.此外,本文所提两阶段资源管理方案与现有文献中所提算法相比,系统所提供容量与用户需求差值可降低83.43%,体现了本文所提算法在实现系统资源高效利用及用户服务需求保障方面的性能优势.

深度学习赋能波束管理:现状、挑战与机遇

随着载波频率的不断提高和大规模天线阵列的广泛部署,基于模拟移相器的波束赋形成为下一代无线通信的标志性技术之一。此时,波束管理被用于获取和维护基站和用户端具有最大接收功率的最优波束对,以保障可靠的无线通信服务。传统波束管理方法往往依赖于海量搜索。同时,传统数学模型无法全面的、准确刻画非线性的波束的内在关联和高维无线环境特征,因而难以取得令人满意的波束增益性能。近年来,得益于深度学习强大的自适应拟合能力,深度学习赋能波束管理得到了国内外广泛关注。本文总结了深度学习赋能波束管理的研究进展,并展望了未来的研究方向。首先,阐述了深度学习应用于波束管理的典型场景和潜在优势;随后,从空/时/频域切入,讨论当前深度学习赋能波束管理的主要研究路线和代表性工作;最后,面向更大规模的无线网络、更多元的波束管理功能和更鲁棒的深度学习模型,阐述未来的研究挑战与机遇。

基于DDPG的智能反射面辅助无线携能通信系统性能优化

针对智能反射面(IRS, intelligent reflecting surface)辅助的多输入单输出(MISO, multiple input singleoutput)无线携能通信(SWIPT, simultaneous wireless information and power transfer)系统,考虑基站最大发射功率、IRS反射相移矩阵的单位膜约束和能量接收器的最小能量约束,以最大化信息传输速率为目标,联合优化了基站处的波束成形向量和智能反射面的反射波束成形向量。为解决非凸优化问题,提出了一种基于深度强化学习的深度确定性策略梯度(DDPG, deep deterministic policy gradient)算法。仿真结果表明,DDPG算法的平均奖励与学习率有关,在选取合适的学习率的条件下,DDPG算法能获得与传统优化算法相近的平均互信息,但运行时间明显低于传统的非凸优化算法,即使增加天线数和反射单元数,DDPG算法依然可以在较短的时间内收敛。这说明DDPG算法能有效地提高计算效率,更适合实时性要求较高的通信业务。

移动场景下基于深度学习的图像辅助毫米波波束预测方案

针对移动环境下毫米波大规模MIMO通信系统下行链路的快速波束预测问题,提出了一种基于深度学习的图像辅助波束预测方案.该方案将基站采集的RGB图像上传至MEC服务器,通过Faster RCNN目标检测模型与DNN神经网络结合,预测通信环境中用户图像与毫米波下行链路波束向量的高维非线性关系.仿真结果表明:该方案预测下行链路波束向量的可达速率接近理论最优,在模型复杂度和高天线数低信噪比情况下的性能等方面均优于基线算法.

桩撑式支护基坑通用弹性地基梁分析法

弹性地基梁法常用于模拟分析桩撑(锚)式支护基坑在开挖、设撑等施工工况中变形和受力的变化。目前利用该方法对某工况基坑分析时,首先要将整个支护桩在支撑处划分为若干个桩段,然后利用相邻桩段交界处的位移连续条件和受力平衡条件,推导出相应的待定参数方程,之后再利用求出的待定参数分析支护体系(基坑)的变形和受力。由于不同施工工况对应的支撑数和桩段数不同,因此该方法必须根据不同工况重新推导各自的参数方程,从而导致计算过程繁杂,难以形成通用计算公式和方法,也不易编程等问题。通过将支撑与桩段归纳为3种基本连接形式(即支撑分别相连于桩顶、相邻两桩段之间和坑底),本文建立了相邻两个桩段间待定参数的递推公式,进而推导出了适用于内支撑和锚索支撑形式、任意支撑道数、开挖或设撑工况的通用待定参数方程。该方程仅含4个参数,远少于现有方法且易于求解。文中算例结果验证了本文方法的合理性和可行性。

基于CBCT的深度学习辅助解剖结构分割在口腔种植中的应用

应用深度学习进行口腔解剖结构分割相比手动分割及传统算法分割可高效获得精准、一致性良好的分割结果。该方法可以快速获得术区解剖结构信息,进行口腔种植手术及口腔修复方案的设计。本文拟对基于锥形束计算机体层成像的深度学习在口腔种植领域解剖结构分割方面的研究进展做一综述。

深部开采环境下底板隔水关键层深梁力学分析

【目的】随着矿井开采深度增加,来自高承压岩溶水威胁增大,导致煤层工作面出现涌水、突水等水害现象,分析深部开采环境下煤层底板隔水关键层抵抗水压力强度是解决这一现象的重要内容之一。【方法】为解决此问题,将隔水底板简化为岩梁模型,并运用深梁理论解决深部开采突水预测中的岩梁模型问题,采用理论分析和数值模拟相结合的方法,根据深梁弯曲力学特点,结合前人研究成果,将深梁条分成浅梁,通过弹性力学单根浅梁受力分布形式假定层间挤压应力σy为三次函数,给出深梁弯曲应力求解的条分技术,并将计算结果与弹性力学解、FLAC3D模拟结果进行对比。【结果和结论】结果表明,深梁条分解得到的应力及位移与数值解结果趋势更为接近,计算结果更为精确,相对误差均在10%以内,同时,随着条分层数增加,精度也增加,但提高幅度逐渐降低,因此,工程应用中针对深部岩梁模型条分到一定程度即可;随着高跨比不断增加,精度误差也在增加,说明条分层宽度不宜过大,否则造成误差增加;底板隔水关键层实例表明,当隔水关键层高跨比大于0.2时,为典型的深梁问题,常规弹性力学的最大拉应力求解结果误差较大,相对误差达到40.6%,给正确判定关键层突水危险性带来不利影响,此时采用深梁条分法求解应力精度较高,可为深部煤层底板突水预测研究起到重要的指导作用。

面向无蜂窝通感一体化系统的智能波束扫描方法

传统的信息处理流程中,基于蜂窝网络的通信功能和基于无线电信号的感知功能是相互独立的。而未来的无蜂窝通感一体化网络采用了以用户为中心的理念,不再局限于传统的小区边界,以确保所有用户在服务范围内获得一致的覆盖和性能。同时感知和通信将被整合在一起,通信信号在数据传输的同时可以被用来实现对潜在目标的持续感知,从而实现更高效、更智能的信息处理和交互。本文在毫米波频段无蜂窝通感一体系统场景下,设计了一种智能的收发端联合探测波束码字选择方法。首先对接收端接入点的信息进行预处理,获得路径损失和目标的估计信息,并通过构造的统计量监测目标是否存在。随后,通过记录多次发射波束码字选择与回声信号的反馈信息,并利用强化学习算法探索最优码字与强反馈信息之间的映射关系,获得一种高效的波束码字探索策略。最后,通过不断调整通信环境和目标特性,基于深度强化学习的波束扫描模型能够排除对环境中先验信息的依赖,显著提高模型的泛化性能。仿真实验表明,相比于传统的波束扫描算法,所提算法探索到最优收发波束对需要的探索次数显著减少,这种优势在大规模波束组合的情况下更为明显。此外,即使在低信噪比情况下,所提算法依然能够通过少量尝试选择出最优的收发波束对。

基于CBAM-CNN和压电悬臂梁的温度解耦质量感知方法

悬臂梁结构广泛用于微小质量测量,而温度变化会引起测量结果漂移。传统测量方法需要在温度稳定的环境中进行,但实际应用中通常难以满足此要求,且温度漂移对测量的影响难以直接解耦。本文提出了一种基于数据驱动,CBAM-CNN和压电悬臂梁的自适应温度解耦质量感知方法。首先,搭建谐振式压电悬臂梁温控测量平台采集不同质量负载下的阻抗响应信号,设计自适应加权预处理方法以增强结构特征并突出有限样本中的关键信息;其次,设计基于混合领域注意力机制的CBAM-CNN网络来评估信号中多个谐振峰的相对关系,实现温度解耦和质量感知。实验结果表明,该方法在25℃至55℃的温度范围内的对0.1~1 g的质量感知准确率高达99.70%,无需进行温度补偿即可实现大跨度温度下的精确质量感知。