基于异构数据的电力短期负荷大数据预测方案

随着多种可再生能源电力的接入,电力系统正在向更智能、更灵活、交互性更高的系统过渡。负荷预测,特别是针对单个电力客户的短期负荷预测在未来电网规划和运行中发挥着越来越重要的作用。提出了一个基于异构数据的电力短期负荷大数据预测方案,该方案收集来自智能电表和天气预报的数据,预处理后将其加载到非关系型数据库中进行存储并做进一步的异构数据处理;设计并实现了一个长短期记忆递归神经网络模型,用于确定负荷分布并预测未来24 h的住宅小区用电量;最后利用一个住宅小区的智能电表数据集对提出的短期负荷预测框架进行了测试,并使用均方根误差和平均绝对百分比误差两个指标,对比了预测模型与两种经典算法的性能,验证了所提模型的有效性。     

An improved strategy for out-of-step oscillation separation devices based on apparent impedance angle applied to a series compensated line

Out-of-step oscillation separation devices based on apparent impedance angle trajectory are widely used in the power grid in China, and their reliability is of great importance. In this paper, the influence mechanism of series compensation on apparent impedance angle trajectory is analyzed. It reveals that series compensation capacitors reduce the equivalent impedance angle and consequently the apparent impedance angle cannot pass through all four zones, resulting in the risk of failure for the separation devices. A revised method of apparent impedance angle based on compensation principle is discussed, and then an improved out-of-step oscillation detection criterion is proposed. In view of the fact that the apparent impedance angle at the moment of maximum current is equal to the equivalent impedance angle during an oscillation process, a practical algorithm based on the compensated apparent impedance angle is proposed. RTDS tests are conducted to verify the effectiveness of the new criterion, considering transmission lines with a high series compensation level.     

基于相对生成对抗网络的低清小脸幻构

针对当前代表性低清小脸幻构方法存在的视觉真实感弱、网络结构复杂等问题,提出了一种基于相对生成对抗网络的低清小脸幻构方法(tfh-RGAN).该文方法的网络架构包括幻构生成器和判别器两个部分,通过像素损失函数和相对生成对抗损失函数的联合最小化,实现生成器和判别器的交替迭代训练.其中,幻构生成器结合了残差块、稠密块以及深度可分离卷积算子,保证幻构效果和网络深度的同时降低生成器的参数量;判别器采用图像分类问题中的全卷积网络,通过先后去除批归一化层、添加全连接层,充分挖掘相对生成对抗网络在低清小脸幻构问题上的能力极限.实验结果表明,在不额外显式引入任何人脸结构先验的条件下,该文方法能够以更简练的网络架构输出清晰度更高、真实感更强的幻构人脸.从定量角度看,该文方法的峰值信噪比相较之前的若干代表性方法可提高0.25～1.51 dB.     

Neural network-based robust finite-time attitude stabilization for rigid spacecraft under angular velocity constraint

Neural Computing and Applications - In this paper, the problem of attitude stabilization control of spacecraft under angular velocity constraint is investigated. A state-constrained finite-time...     

Defect‐Laden MoSe2 Quantum Dots Made by Turbulent Shear Mixing as Enhanced Electrocatalysts

A high density of edge sites and other defects can significantly improve the catalytic activity of layered 2D materials. Herein, this study demonstrates a novel top‐down strategy to maximize catalytic edge sites of MoSe₂ by breaking up bulk MoSe₂ into quantum dots (QDs) via “turbulent shear mixing” (TSM). The ultrasmall size of the MoSe₂ QDs provides a high fraction of atoms in reactive edge sites, thus significantly improving the catalytic activities. The violent TSM further introduces abundant defects as additional active sites for electrocatalytic reactions. These edge‐proliferated and defect‐laden MoSe₂ QDs are found to be efficient electrocatalysts for the hydrogen evolution reaction, and useful as counter electrodes in dye‐sensitized solar cells. The work provides a new paradigm for creating edge‐proliferated and defect‐rich QDs from bulk layered materials.     

无机固体电解质Li7La3Zr2O12的制备及掺杂改性研究进展

无机固体电解质由于其安全性能高、能量密度大等特点备受研究者的青睐。其中Garnet型锂离子无机固体电解质Li_7La_3Zr_2O_(12)具有较高的离子导电率,较低的界面电阻,优良的稳定性能和电化学性能,在未来的全固态锂离子电池、锂空气电池等领域有着广阔的应用前景。主要从Li_7La_3Zr_2O_(12)的晶体结构、制备工艺和掺杂改性等方面详细阐述无机固态电解质Li_7La_3Zr_2O_(12)的研究进展。     

多糖类可吸收材料降解产物分析方法研究进展

多糖类可吸收材料因来源丰富、生物相容性好、在人体内易被降解吸收而引起了人们的广泛关注,但国内外缺乏相关评价标准,其安全性评价存在较大难度,因而对多糖类可吸收材料降解产物的分析是不可或缺的。对多糖类可吸收材料及其降解产物进行了介绍,重点综述了多糖材料降解产物的体内外分析方法,并对当前存在的问题进行了展望,旨在为多糖类可吸收材料降解产物的分析提供一定的思路。     

应用差分GPS实现小麦精播机器人定位

应用差分GPS（Global Positioning System,全球定位系统）进行小麦精播机器人的定位,结合NI公司研发的Lab—view图形化编程软件开发了小麦精播机器人定位系统。首先分析了相对定位和绝对定位各种方法的优缺点;然后介绍了小麦精播机器人定位系统具体的工作流程,详细叙述了GPS数据的接收和提取、经纬度...     

Mean square exponential stability in high-order stochastic impulsive BAM neural networks with time-varying delays

In this paper, a class of stochastic impulsive high-order BAM neural networks with time-varying delays is considered. By using Lyapunov functional method, LMI method and mathematics induction, some sufficient conditions are derived for the globally exponential stability of the equilibrium point of the neural networks in mean square. It is believed that these results are significant and useful for the design and applications of impulsive stochastic high-order BAM neural networks.     

一种无外部持续激励的子空间闭环辨识算法

传统的闭环系统可辨识性受到外部激励和控制器结构的限制,然而在实际辨识应用中,由于过程操作条件的限制或者经济原因不希望存在外部持续激励。讨论了在无外部激励下基于快采样的多变量子空间模型闭环可辨识性,提出了无外部持续激励条件下的闭环快采样子空间模型辨识方法,在常规采样的可辨识性条件不满足或部分满足的情况下,通过对闭环系统控制对象的输入输出变量快采样来实现对象模型辨识,并通过仿真实例验证了基于快采样的辨识算法的有效性。