Waste sugar solution polymer-derived N-doped carbon spheres with an ultrahigh specific surface area for superior performance supercapacitors

Waste sugar solution (WSS), a waste by-product of manufacturing vitamin C, contains abundant waste acids and organics. In this work, a N/O-enriched copolymer was synthesized via a facile polymerization via the hydrogen bonding of O-containing functional groups and melamine and the crosslinking of aldehyde groups. Subsequently, N-doped carbon spheres were prepared by a typical carbonation/activation method. Remarkably, benefiting from an ultrahigh specific surface area (3612 m²/g) and rich heteroatom content (4.3% for N, 8.8% for O), the carbon spheres deliver a high specific capacitance of 387 F/g at 50 mA/g and 283 F/g at 5 A/g with 6 M KOH in two-electrode system. The assembled symmetric electric double-layer capacitor exhibits high energy density of 10.83 Wh/kg at 11.10 W/kg. This research provides a facile method for preparing N/O-doped carbon spheres by WSS, and confirms the excellent electrochemical performance of WSS-derived carbons in energy storage applications.     

Effects of Sr2+ substitution on the crystal structure,Raman spectra,bond valence and microwave dielectric properties of Ba3-xSrx(VO4)2 solid solutions

The effects of Sr²⁺ substitution for Ba²⁺ on microwave dielectric properties and crystal structure of Ba_3-xSr_x(VO₄)₂ (0 ≤ x ≤ 3, BSVO) solid solution were investigated. Such Sr²⁺ substitution contributes to significant reduction in sintering temperature from 1400 °C to 1150 °C. Both permittivity (∑_r) and quality factor (Q × f) values decreased with increasing x value, which was determined to be related with the descending values of average polarizability and packing fraction, whereas the increase in τ_f value was explained by the decreased average VO bond length, A-site bond valence. BSVO ceramics possessed encouraging dielectric performances with ∑_r = 12.2–15.6 ± 0.1, Q × f = 44,340 - 62,000 ± 800 GHz, and τ_f = 24.5–64.5 ± 0.2 ppm/°C. Low-temperature sintering was manipulated by adding B₂O₃ as sintering additive for the representative Sr₃V₂O₈ (SVO) ceramic and only 1 wt.% B₂O₃ addition successfully contributed to a 21.7% decrease in sintering temperature to 900 °C, showing good chemical compatibility with silver electrodes, which render BSVO series and SVO ceramics potential candidates in multilayer electronic devices fabrication.     

Room-temperature multiferroic characteristics and unique vortex domain structures of h-Yb1−xInxFeO3 solid solutions

Hexagonal rare-earth ferrites (h-RFeO₃) have attracted much scientific attention due to their room-temperature multiferroicity. However, it is still a hard job to obtain h-RFeO₃ bulk materials because of the meta-stability of such hexagonal phase, and the evaluation of room-temperature ferroelectric and magnetoelectric characteristics in such materials is also a challengeable issue. In the present work, Yb_1−xIn_xFeO₃ ceramics with the stable hexagonal structure were obtained by introducing chemical pressure, where the unique ferroelectric domain structures of sixfold vortex combined with tenfold vortex with a typical size of ~400 nm were determined. Symmetry of the present system evolved from centrosymmetric orthorhombic Pbnm (x = 0–0.4) to non-centrosymmetric hexagonal P6₃cm (x = 0.5 and 0.6) with a ferroelectric polarization up to 3.2 μC/cm², and finally to centrosymmetric hexagonal P6₃/mmc (x = 0.7 and 0.8). The Curie point decreased monotonically from 723 K to a temperature below room temperature with increasing x, and the antiferromagnetic phase transition above room temperature was determined for all compositions. Meanwhile, a large linear magnetoelectric coefficient (α_ME) up to 0.96 mV/cm Oe was obtained at room temperature, and this indicated the great application potential for magnetoelectric devices.     

基于 GA-SVM 算法的通信网络入侵信号自动识别技术

针对当前通信系统入侵行为自动识别技术存在入侵信号样本识别成功率较低、误识别率和漏识别率较高的问题,提出基于 GA-SVM 算法的通信网络入侵信号自动识别技术。利用混沌原理提取通信网络入侵的非平稳信号时域特征,并凭借自回归模型提取对应频域特征,捕捉邻域入侵信号间的非线性时空动作频率,评价相邻行为间的状态关联性,预测入侵信号后续行为,完成入侵信号的识别。实验表明,所提方法识别精度高、误识别率较低,漏识别率非常低,具有可应用于实际的理论价值。     

基于语言模型的预训练技术研究综述

预训练技术当前在自然语言处理领域占有举足轻重的位置。尤其近两年提出的ELMo、GTP、BERT、XLNet、T5、GTP-3等预训练模型的成功,进一步将预训练技术推向了研究高潮。该文从语言模型、特征抽取器、上下文表征、词表征四个方面对现存的主要预训练技术进行了分析和分类,并分析了当前自然语言处理中的预训练技术面临的主要问题和发展趋势。     

基于区间犹豫模糊语言集的水资源多目标决策

以鄂北水资源配置工程为对象,研究大型调水工程背景下的多水源多目标优化配置。将鄂北地区水库分为充蓄水库、补偿水库与在线水库等3类,根据各类水库特点对鄂北水资源配置工程进行系统概化,建立以年缺水量最小和调水成本最低为目标函数的多水源多目标优化配置模型。采用权重系数法进行求解得到非劣解集。最后采用基于区间犹豫模糊语言与TOPSIS的多目标决策方法对各方案的缺水量与调水量作出评价并遴选出最优方案。研究结果表明:对鄂北水资源配置工程而言,缺水量与调水量的权重比为7∶3的方案综合评价最优,重视缺水的同时兼顾调水成本;基于区间犹豫模糊语言集与TOPSIS的水资源多目标决策方法经本文实例验证具有较高的稳定性与可行性。     

基于预训练语言模型的关键词感知问题生成

问题生成任务是指根据给定的文本段落和答案来自动生成对应的问题。针对现有问题生成方法存在的误差累积现象以及问题生成任务固有的“一对多”情况，提出一种带有关键词感知功能的问题生成方法。在预训练语言模型的基础上，实现关键词分类模型与问题生成模型的网络结构设计。输入文本段落中蕴含关键词，为使所生成的问题中包含同样的关键词以保证问题与段落的语义一致性，利用关键词分类模型提取出文本段落中的关键词，将关键词与非关键词的区分特征融入问题生成模型的输入中，该特征作为问题生成过程的全局信息，用以消除问题生成模型仅依赖局部最优解的弊端，减少误差累积与“一对多”情况的发生。在SQuAD数据集上的实验结果表明，该方法能够提升问题生成的质量，其BLEU-4指标值可达24，优于带有复制机制、带有语义监督的问题生成模型，目前已经借助百度百科数据平台实现了大规模工业应用。