基于transformer自适应特征向量融合的图像分类

针对目前基于transformer的图像分类模型直接应用在小数据集上性能较差的问题,本文提出了transformer自适应特征向量融合网络,该网络在特征提取器中将不同阶段的特征进行融合,减少特征信息丢失的同时获得更多不同感受野下的信息,同时利用最大池化来去除特征中的冗余信息,从而使提取的特征更具有判别性。此外,为了充分利用图像的各级特征信息来进行分类预测,本文将网络各阶段产生的特征向量进行融合,使融合后的特征向量更具有表征能力,从而减少网络对大数据集的依赖,使网络在小数据集中也能获得很好的性能。实验表明,本文提出的 算法在数据集Mini-ImageNet-100、CIFAR-100和ImageNet-1k上的TOP-1准确率分别达到了74.22%、85.86%和81.4%。在没有增加计算量的情况下,在baseline上分别提高了6.0%、3.0%和0.1%,且参数量减少了18.3%。本文代码开源在“https://github.com/xhutongxue/afvf”。     

抑制间接矩阵变换器共模电压的调制策略

间接矩阵变换器在运行过程中会产生较大的共模电压,对电机系统造成严重的电磁干扰,影响系统正常运行。针对这一问题,提出一种新的空间矢量调制策略降低共模电压大小,改善输出电压性能。首先,将整流级空间矢量重新划分区域,根据参考电流矢量所在区域选择3个相邻的有效电流矢量,将有效矢量下的平均直流母线电压维持在一个恒定值,保证整流级的最大输出电压。其次,逆变级采用无零矢量调制策略,通过使用3个有效电压矢量合成所需要的输出电压,将共模电压峰值抑制到输入相电压幅值的57.7%。与传统调制策略相比,提出的调制策略不仅可以减轻系统计算负担,还可以减少系统输入输出谐波,改善输入输出波形质量。最后,通过Matlab/Simulink和实验平台搭建相应模型,仿真和实验结果验证了所提出策略的有效性。     

弱电网下基于谐波状态空间模型的光储-虚拟同步发电机稳定性分析与优化控制研究

虚拟同步发电机(virtual synchronous generator, VSG)技术因提供阻尼和惯量而被广泛用于新能源并网逆变器。光储并入弱电网易发生频率耦合等问题,因而VSG并网稳定性分析变得更加复杂。针对光储并入弱电网系统,考虑频率耦合,在dq域中建立基于谐波状态空间(harmonic state space, HSS)理论的多时间尺度虚拟同步发电机模型。采用归一化参数灵敏度分析法,揭示功率环、电压环和低通滤波中虚拟惯量J、虚拟阻尼Dp、下垂系数Ku、比例系数kpu和截止频率ωc等关键控制参数根轨迹的特性和电网强度对系统稳定性的影响。并通过调整优化控制参数,提高系统稳定性。最后在Matlab/Simulink仿真平台验证谐波状态空间模型精确度和关键参数根轨迹特性。     

变压器绕组松动故障的混沌特征分析方法

为了更加有效地对变压器绕组松动故障进行监测与识别,提出了一种变压器绕组松动故障的混沌特征分析方法。首先,针对振动信号的混沌动力学特性,采用互信息量法和G-P算法分别确定延迟时间和嵌入维数,对变压器振动信号进行相空间重构;其次,通过判断最大Lyapunov指数是否为正,进而证明变压器振动信号的混沌特性,在此基础上分析不同程度的绕组松动故障对相空间轨迹变化的影响;最后,将关联维数、Kolmogorov熵和最大Lyapunov指数作为一组混沌特征用以量化变压器绕组发生松动故障前后振动信号的混沌特性。结果表明：变压器振动信号的最大Lyapunov指数均大于0,证实了其具有混沌特性,所得到的混沌特征能够有效反映变压器绕组松动故障。研究结果为变压器绕组松动状态监测提供了一种理论依据。     

基于空间矢量复合判断指标的变电站动力电缆漏电检测算法

为提高变电站长段动力电缆漏电监测水平,提出了一种基于空间矢量复合判据的变电站长段动力电缆漏电检测算法。首先对现有方法中无法检测三相漏电故障的问题进行了分析。然后,引入空间矢量的概念,将剩余电流数据转换为空间矢量圆,提出了剩余电流差流、A相漏电流以及空间矢量圆半径变化率3个漏电状态判断指标。并建立了针对电缆不平衡漏电及三相漏电故障的类型判断机制,实现了对长段动力电缆漏电状态及类型的准确检测。最后,通过仿真及江苏电网某500 kV变电站实际数据对所提方法的有效性与优越性进行验证。结果表明,该方法能够准确实现对不平衡漏电故障、故障演变过程以及三相漏电故障的判断,丰富了电缆漏电故障的诊断信息,有效提高了对变电站漏电问题的分析处理效率。     

基于转移阻抗模型的功率回路和信号回路间串扰建模与试验验证

由于列车和船舶上的空间有限,互连非屏蔽线缆的空间非常小,导致非屏蔽线缆之间存在严重的串扰问题。其中功率回路对信号回路的串扰会严重影响信号的质量和系统的正常运行,因此有必要估计功率回路中共模和差模电磁干扰对信号回路的影响,为此提出了一种可扩展的级联多端口网络模型,可对线缆间共模和差模串扰进行精准建模,准确估计出功率回路对信号回路的影响。最后,搭建了试验台架进行验证,实验结果验证了该方法的有效性。     

面向方向图保形的共口径基站天线去耦技术

为了节省基站天线所占空间资源,降低其运营成本,现代移动通信需要将多个天线阵列紧密地排列在一起,构成多频共口径基站天线。多天线的紧密排列势必造成天线间强烈的耦合,不仅使得天线阻抗失配、隔离度变差,同时也造成辐射方向图的严重变形。因此,近年来多频共口径基站天线的去耦成为工业界和学术界研究的热点,早期的去耦技术主要是面向改善天线的阻抗匹配和隔离度,而对面向方向图保形的去耦技术研究相对较少。文章在简要介绍面向方向图保形的基站天线去耦技术的研究现状的基础上,重点阐述了褚庆昕教授天线射频团队基于感应电流抵消、耦合场抵消和辐射阻断等原理,提出的几种新型的去耦技术。这些技术不仅可以有效地改善基站天线的阻抗匹配和隔离,更重要的是很好地实现了天线方向图保形。仿真和测试结果验证了原理的正确性和设计的可行性。一些技术已被用于5G基站天线产品。     

基于FMCW雷达的呼吸模式分类技术研究

针对当前使用调频连续波雷达的呼吸模式分类算法准确度不高的问题,本文提出一种基于一维卷积神经网络（1DCNN）结合长短时记忆（LSTM）网络的多呼吸模式分类方法。方法共分为四步：第一步,对雷达提取的呼吸信号进行预处理;第二步,使用快速傅里叶变换（FFT）与连续小波变换（CWT）提取呼吸信号特征;第三步,根据呼吸特征对五种呼吸模式信号（正常呼吸、呼吸过速、呼吸过缓、呼吸深大、呼吸暂停）打标签制作数据集;第四步,使用数据集训练网络得到模型,并使用新数据测试模型。实验结果表明,此方法分类准确度要比现有使用CNN网络方法高5%左右。     

Web Page Recommendation Using Distributional Recurrent Neural Network

In the data retrieval process of the Data recommendation system, the matching prediction and similarity identification take place a major role in the ontology. In that, there are several methods to improve the retrieving process with improved accuracy and to reduce the searching time. Since, in the data recommendation system, this type of data searching becomes complex to search for the best matching for given query data and fails in the accuracy of the query recommendation process. To improve the performance of data validation, this paper proposed a novel model of data similarity estimation and clustering method to retrieve the relevant data with the best matching in the big data processing. In this paper advanced model of the Logarithmic Directionality Texture Pattern (LDTP) method with a Metaheuristic Pattern Searching (MPS) system was used to estimate the similarity between the query data in the entire database. The overall work was implemented for the application of the data recommendation process. These are all indexed and grouped as a cluster to form a paged format of database structure which can reduce the computation time while at the searching period. Also, with the help of a neural network, the relevancies of feature attributes in the database are predicted, and the matching index was sorted to provide the recommended data for given query data. This was achieved by using the Distributional Recurrent Neural Network (DRNN). This is an enhanced model of Neural Network technology to find the relevancy based on the correlation factor of the feature set. The training process of the DRNN classifier was carried out by estimating the correlation factor of the attributes of the dataset. These are formed as clusters and paged with proper indexing based on the MPS parameter of similarity metric. The overall performance of the proposed work can be evaluated by varying the size of the training database by 60%, 70%, and 80%. The parameters that are considered for performance analysis are Precision, Recall, F1-score and the accuracy of data retrieval, the query recommendation output, and comparison with other state-of-art methods.     

Association Rule Mining Frequent-Pattern-Based Intrusion Detection in Network

In the network security system, intrusion detection plays a significant role. The network security system detects the malicious actions in the network and also conforms the availability, integrity and confidentiality of data information resources. Intrusion identification system can easily detect the false positive alerts. If large number of false positive alerts are created then it makes intrusion detection system as difficult to differentiate the false positive alerts from genuine attacks. Many research works have been done. The issues in the existing algorithms are more memory space and need more time to execute the transactions of records. This paper proposes a novel framework of network security Intrusion Detection System (IDS) using Modified Frequent Pattern (MFP-Tree) via K-means algorithm. The accuracy rate of Modified Frequent Pattern Tree (MFPT)-K means method in finding the various attacks are Normal 94.89%, for DoS based attack 98.34%, for User to Root (U2R) attacks got 96.73%, Remote to Local (R2L) got 95.89% and Probe attack got 92.67% and is optimal when it is compared with other existing algorithms of K-Means and APRIORI.