基于MFE-BERT与FNNAttention的心理医学知识图谱构建

针对心理医学领域文本段落冗长、数据稀疏、知识散乱且规范性差的问题, 提出一种基于多层级特征抽取能力预训练模型(MFE-BERT)与前向神经网络注意力机制(FNNAttention)的心理医学知识图谱构建方法. MFE-BERT在BERT模型基础上将其内部所有Encoder层特征进行合并输出, 以获取包含更多语义的特征向量, 同时对两复合模型采用FNNAttention机制强化词级关系, 解决长文本段落语义稀释问题. 在自建的心理医学数据集中, 设计MFE-BERT-BiLSTM-FNNAttention-CRF和MFE-BERT-CNN-FNNAttention复合神经网络模型分别进行心理医学实体识别和实体关系抽取, 实体识别F1值达到93.91%, 实体关系抽精确率达到了89.29%, 通过融合文本相似度与语义相似度方法进行实体对齐, 将所整理的数据存储在Neo4j图数据库中, 构建出一个含有3652个实体, 2396条关系的心理医学知识图谱. 实验结果表明, 在MFE-BERT模型与FNNAttention机制的基础上构建心理医学知识图谱切实可行, 提出的改进模型所搭建的心理医学知识图谱可以更好地应用于心理医学信息管理中, 为心理医学数据分析提供参考.     

标签运动状态下的RFID系统反碰撞算法

针对现实应用中标签经常处于运动状态的情况,对传送带模型下的无线射频识别(RFID)系统进行了分析,并使用Matlab做了相应的仿真实验。结果证明,当标签密度D与传送带速度V保持不变时,识别率P在帧长度N与作用范围内未识别标签数n相等时最高。当时隙持续时间一定时,识别率P只与V和D有关,而与作用区域内传送带长度L无关。同时针对该模型,对现有的标签估算方法进行了修正,在此基础上提出了一种新的估算方法,该方法的准确度在标签数目较大时有着明显的提高。     

Research on dynamic RWA algorithm supporting service-differentiated provision

In the multi-service requirement of the next generation of optical networks, differentiated services should be provided, and the transmission quality of signal and the path reliability are two important parameters of service class differentiation. A new dynamic routing and wavelength assignment (RWA) algorithm supporting service-differentiated provision, which takes account of both requirements of the signal transmission quality and the path reliability, is proposed. The numerical results from simulation show that the algorithm can not only overcome the impact of impairment on signal transmission quality and guarantee lightpath reliability, but also offer the service-differentiated lightpath according to its class.     

A study of green wavelength-division multiplexed optical communication systems using cascaded fiber bragg grating

This paper studies the performance analysis of wavelength-division multiplexed optical communication systems (WDM). First, flat-gain erbium doped fiber amplifiers (EDFAs) are seriously needed to obtain proper and equal amplification of all channels. Such amplifiers can be designed by intrinsically modifying the host material or extrinsically using proper filters. In this research, we benefit from both the intrinsic and extrinsic methods to achieve sharp flat EDFA output gain using cascaded fiber Bragg gratings (FBGs). Second, the performance of our technique has been evaluated through calculating the bit error rate (BER) and signal-to-noise ratio (SNR) of a WDM system embedded with the reported EDFA flattening system. The parametric simulations of the FWHM of FBGs, SNR, optical power and the transmission distance have shown a noticeable improved performance. Sending data via an optical WDM system will be proven from comprehensive simulations to achieve high quality signal transmission spectrums, increased transmission distances and low power consumption. By extension, the reported design using cascaded FBGs can also be generalized to equalize the gain of any arbitrary profile.     

A highly nonlinear photonic quasi-crystal fiber with low confinement loss and flattened dispersion

We proposed a novel photonic quasi-crystal fiber with near-zero flattened dispersion, highly nonlinear coefficient, and low confinement loss by using the dual concentric core structure. By optimizing the structure parameter, the proposed photonic quasi-crystal fiber can achieve a nonlinear coefficient larger than 33 W⁻¹ km⁻¹ and near-zero flatten dispersion of 0 ± 3.4 ps/nm/km with a near-zero dispersion slope of 8.5 × 10⁻³ ps/nm²/km at the wavelength of 1550 nm. Near-zero flattened dispersion and low confinement loss in the ultralow order of 10⁻⁷ dB/m are simultaneously obtained in the wavelength range from 1373 to 1627 nm. Furthermore, two zero dispersion wavelengths can be achieved in a wide wavelength ranger from 1373 to 1725 nm. From the point of view of practical fabrication, the influence of deviation of each air hole diameter within 3% of imperfections on dispersion, nonlinearity, and is discussed to verify the robustness of our design.     

MRCD: a genetic algorithm for multiobjective robust control design

A genetic algorithm (GA) for the class of multiobjective optimization problems that appears in the design of robust controllers is presented in this paper. The design of a robust controller is a trade-off problem among competitive objectives such as disturbance rejection, reference tracking, stability against unmodeled dynamics, moderate control effort and so on. However, general methodologies for solving this class of design problems are not easily encountered in the literature because of the complexity of the resultant multiobjective problems. In this paper, a recently developed class of GAs, multiobjective GAs, are used to solve robust control design problems. Here, a new algorithm, called multiobjective robust control design, has been proposed. The structure and operators of this algorithm have been specifically developed for control design problems. The performace of the algorithm is evaluated by solving several test cases and is also compared to the standard algorithms used for the multiobjective design of robust controllers.     

Multipath affinage stacked—hourglass networks for human pose estimation

Recently, stacked hourglass network has shown outstanding performance in human pose estimation. However, repeated bottom-up and top-down stride convolution operations in deep convolutional neural networks lead to a significant decrease in the initial image resolution. In order to address this problem, we propose to incorporate affinage module and residual attention module into stacked hourglass network for human pose estimation. This paper introduces a novel network architecture to replace the stacked hourglass network of up-sampling operation for getting high-resolution features. We refer to the architecture as an affinage module which is critical to improve the performance of the stacked hourglass network. Additionally, we also propose a novel residual attention module to increase the supervision of upsample process. The effectiveness of the introduced module is evaluated on standard benchmarks. Various experimental results demonstrated that our method can achieve more accurate and more robust human pose estimation results in images with complex background.     

新的桶形畸变的点阵样板校正方法

为了校正广角镜头的桶形畸变,提出一种新的桶形畸变数字校正方法。它使用点阵样板校正的方法,根据畸变图和理想图中圆点的位置关系,得出畸变图像素在X轴和Y轴方向上的偏移量曲面,采用三次B插值函数对曲面插值;由曲面插值获取像素点的偏移量,对各像素进行坐标转换得到校正图像;然后对图像进行了双线性插值的灰度重建。仿真结果表明,该方法使图像的坐标位置和灰度都得到很好的校正。     

多用户正交频分复用加权比例公平调度

传统的正交频分复用(OFDM)调度进行资源分配时没有考虑用户间速率比例公平性。针对这一问题,提出了一种新的适合于混合业务的多用户OFDM系统比例公平调度方案,其用户队列承载混合类型业务。该方案在用户间速率成比例约束条件下最大化系统权重容量和;授予用户队列中不同类型的分组不同的权重因子,并通过权重因子计算用户的权重;在子载波分配时不仅定义了信道优先级因子,而且在用户间速率成比例公平约束条件下通过该因子为用户分配子载波;最后推导出一种线性的功率分配方式。仿真结果与分析显示,该方案可以较好地满足用户速率和业务时延的需求,且可以在有效提高系统容量的基础上,严格地保证用户容量公平性。     

热镀锌电磁封流技术的研究综述

热镀锌是镀锌板生产过程中必不可少的步骤,而热镀锌电磁封流技术相对于传统热镀锌技术具有带钢表面无划伤、沉没辊和稳定辊不被腐蚀、维修频率低和工作效率高的优势,具有重要的工程应用价值。首先,介绍了热镀锌电磁封流技术的特点和意义。其次,分别介绍了电磁泵热镀锌电磁封流技术、CVGL热镀锌电磁封流技术、直流磁场热镀锌电磁封流锌技术和永磁感应密封热镀锌电磁封流技术的工作原理、研究现状、技术组成及其优缺点。再次,分析了热镀锌电磁封流锌工作过程中,温度场、流场、固体结构力学场和电磁场之间复杂耦合的特点,并且依据热镀锌电磁封流过程的技术特点,指出了热镀锌电磁封流过程中带钢振动的抑振技术、锌液不稳定流动抑制技术、温度稳定性控制技术、张力控制等关键技术。最后,提出了电磁装置结构参数优化、带钢振动抑制技术和基于大数据的过程控制技术是提高热镀锌电磁封流锌工作过程的关键技术。