With the large-scale integration of distributed generation, the operational characteristics of distribution networks have become increasingly diverse. Frequent changes in network topology and complex variations in fault feature distributions reduce the reliability and accuracy of existing protection and fault location methods. To address this issue, this paper proposes a topology-adaptive fault location method for distribution networks. First, multidimensional fault information in traveling waves is extracted via a wavelet transform to generate panoramic traveling-wave patterns, and the impact of topology changes on traveling-wave feature distributions is analyzed in the time, frequency, and spatial domains. Second, a Transformer-based fault-location model is developed, where graph random-walk and graph Laplacian encodings are employed to represent topological information, and the Transformer encoder learns multi-point correlations to capture the relationships among panoramic patterns, system topology, and fault locations. Finally, by embedding topology variations into the featureextraction process, the model learns essential fault features that are insensitive to topology changes yet indicative of the actual fault location, thereby enabling accurate fault localization. Simulation results demonstrate that, across datasets containing multiple topology types, the proposed method achieves a fault-location accuracy exceeding 99% and an average absolute location error of less than 20 m, showing strong adaptability to different topology variations. Moreover, it exhibits high sensitivity to weak faults and accurately locates 10 k Omega grounding faults.

In flexible DC distribution networks, DC-side short-circuit faults can result in severe overcurrent problems, making fault current-limiting essential to mitigate the impact on the grid. However, current-limiting measures may compromise the effectiveness of conventional protection methods that rely on fault current magnitude or rate-of-change. To address this challenge, a novel protection method for flexible DC distribution networks based on high-frequency integrated impedance is proposed in this paper. Firstly, based on the characteristics of fault current, a current-limiting control strategy based on virtual-resistance is adopted to suppress the fault current. Secondly, a protection criterion is constructed using the ratio of the summed high-frequency voltage fault components to the summed high-frequency current fault components at both ends of the line after a fault occurs. Finally, a flexible DC distribution network model is established in MATLAB/Simulink for simulation verification. Simulation results demonstrate that the proposed method is simple to set, does not require strict communication synchronization, and remains unaffected by transition resistance and noise.

The accuracy and reliability of single-ended fault location methods are often affected by boundary elements and transition resistances. This paper proposes a single-ended fault location method for DC transmission lines based on the initial traveling wave dominant frequency (ITWDF) attenuation characteristics. Firstly, the attenuation characteristics of the initial traveling wave frequency component are analyzed, and the attenuation law and distribution differences between the frequency component and fault distance are revealed. Secondly, to accurately extract and describe the trend of the initial traveling wave frequency component with the change of fault distance, the energy of the initial traveling wave is constructed, and the mathematical expression of the ITWDF related to the fault distance and unit length distortion coefficient is obtained. Theoretical analysis verifies that the ITWDF exhibits a monotonic attenuation mapping relationship with increasing fault distance. In addition, the attenuation characteristics of the ITWDF are not affected by fault impedance. Finally, a fault sample set containing the ITWDF and reference unit length distortion coefficients at different positions is constructed. The fault location is calculated using the ITWDF and the reference unit length distortion coefficient, with continuous wavelet transform and inverse distance weighting interpolation algorithms applied, respectively. Simulation results show that the proposed method can reliably locate faults under high-impedance conditions of up to 800 Omega , and the absolute error is less than 200m. Moreover, the proposed method can be applied to DC lines with various boundary structures, improving the reliability of single-ended fault location.

To address the challenges in existing distribution network fault location methods, such as difficulties in surge identification, reliance on precise time synchronization, and limited adaptability to overhead-cable hybrid lines, a precise fault location method based on spatiotemporal panoramic information is proposed. First, fault information from multiple detection points in a complex distribution network is integrated to construct spatiotemporal panoramic fault data, enabling comprehensive observation of fault signals in the time-frequency-space domain. Based on this, a 3D convolutional feature extraction module is employed to mine fault information from the spatiotemporal panoramic data. An improved SE attention mechanism and a multi-scale fault feature fusion module are then used to selectively extract large-, medium-, and small-scale fault features with assigned importance weights. Finally, a classification-regression output module maps the extracted features to determine the fault section and fault distance, achieving precise fault location in the distribution network. Simulation results demonstrate that the proposed method achieves a fault section location accuracy of 98.2%, with an average location error of only 77.3m.

Affected by factors such as high impedance faults (HIF), traveling wave head detection becomes difficult, leading to low reliability in fault location methods. Fault traveling waves are broadband signals with time-frequency characteristics, containing comprehensive fault information. This paper first analyzes the time-frequency characteristics of traveling waves and proposes a representation of fault traveling waves based on traveling wave full waveform (TWFW) characteristics. To address the severe cross-term interference in Wigner-Ville distribution (WVD), a time-frequency analysis method combining WVD and Smoothed Pseudo Wigner-Ville distribution (SPWVD) is introduced, exploiting SPWVD’s advantage of avoiding false frequency components. By performing matrix operations on time-frequency matrices and setting threshold to correct element, the method effectively suppresses new cross-terms and those overlapping with auto-components. Simulations and field tests demonstrate that the method exhibits strong noise suppression and high time-frequency resolution. The results confirm its feasibility and applicability in fault location applications.

The frequent occurrence of motor faults has been a great disturbance to the development of production in various fields. Traditional fault diagnosis methods primarily use 1-D data or 2-D data. However, 3-D data hold significant promise for motor fault diagnosis due to its voluminous data and unique spatial information. This article aims to explore a motor fault diagnosis method leveraging 3-D data. Nonetheless, motor fault 3-D data exhibit the limitation of lacking geometric structure. To address this limitation, this article proposes a fault diagnosis method named SP-PointCNN. This method uses the proposed spherical projection (SP) method to mitigate the limitations present in the 3-D data of the motor. Meanwhile, a neural network based on PointCNN is constructed to enable the utilization of 3-D data in fault diagnosis. A series of experiments demonstrated the validity of the proposed method. After tested, the diagnosis accuracy of SP-PointCNN can reach 99.36%.

Short-term load forecasting is a crucial task within the power system. However, existing studies have overlooked the spatial-temporal relationships between multiple series loads. Accounting for this spatial-temporal adjacency can lead to more accurate forecasting in certain scenarios. In this paper, we propose a short-term load forecasting model named Parallel Spatial-Temporal Graph Attention Network (PST-GAT). The method encodes the multi-sequence loads as nodes and constructs the adjacency matrix using the Dynamic Time Warping (DTW) technique to form a fully connected graph of the load data. Combining the sliding window concept, the constructed fully connected graph is partitioned into a series of subgraphs, and the Graph Attention Network (GAT) performs feature extraction on each subgraph individually. To realize learning at multiple scales, PST-GAT adopts a parallel multi-branching approach, where each branch splits the subgraphs with varying lengths. Finally, the feature vectors extracted by each branch are concatenated, and forecasting is accomplished using a fully connected layer. Moreover, the unique structural design of PST-GAT allows for the simultaneous prediction of multiple sequence loadings. Experimental results based on real-world load data validate the superior prediction accuracy of this method compared to existing algorithms.

Influenced by weak fault signal and noise interference in distribution network, it is difficult to extract and detect the downlink wavefront in the case of high imped ance fault（HIF）, which results in the low reliability of the traveling-wave-based HIF detection method. To solve the above problems, a novel HIF detection method based on traveling wave full waveform fault feature self-identification is proposed. Firstly, the time-frequency difference between high-resistance ground fault and normal transient disturbance voltage traveling waveform is analyzed with the help of traveling waveform panorama waveform; Then, the CBAMCNN model is built to make it more anti-interference than the traditional CNN model, and the traveling wave full waveform is input into the convolution network in the form of gray image to obtain the feature representation with more anti-interference ability, so as to realize the extraction and utilization of multi-dimensional fault features. Finally, a 10 kV distribution network model is built on PSCAD for simulation analysis under various fault conditions. The results show that the proposed method can reliably detect the HIF with good anti-noise performance, and is not affected by fault location, transition resistance, inception angle, which greatly improves the reliability and sensitivity of the HIF detection method based on traveling wave signals.

High impedance ground fault (HIF) occurs in a distribution network, the fault features are weak and susceptible to factors such as noise interference. Extracting and detecting the wavefronts in this case can be challenging, leading to low reliability in HIF detection methods based on waveforms. To address these issues, this paper proposes an HIF detection method based on self-recognition of fault features using traveling wave full waveform (TWFW) characteristics. Firstly, it analyzes the differences between HIF and normal transient disturbances wave signals in the time-frequency domain based on TWFW. Then, it utilizes a CBAM-CNN neural network model to explore fault features from TWFW grayscale images from multiple dimensions, enhancing the method's fault detection and interference resistance capabilities. Simulation results demonstrate that the proposed method achieves a high detection accuracy of up to 99.8%, reliably detects HIF, exhibits excellent noise resistance, and is unaffected by factors such as fault location and fault impedance.

The conventional traveling wave protection (TWP) for flexible DC transmission lines suffer from low-sensitivity problem in far-end high-impedance faults. Also, they are lack of theoretical basis in threshold setting. In this paper, the initial traveling wave is analysed considering the frequency attenuation law and discrepancy between internal and external faults. Then, the frequency domain expressions of the initial traveling wave are analysed deeply. By enhancing the influence of high-frequency components in energy calculation reasonably, the initial traveling wave dominant frequency (ITWDF) is constructed, which can reflect the fault location and the attenuation characteristics of the boundary. The expressions of the ITWDF show that the ITWDF decreases with the increasing fault distance, and the ITWDF is significantly different between the internal faults and the external faults. On this basis, a single-ended TWP method based on the attenuation characteristics of the ITWDF is proposed. The method is immune to the fault impedance, and the expressions of the ITWDF provide a theoretical basis for protection threshold setting as well. The simulation results show that the proposed method can fast and correctly identify line faults, even with 1000 omega fault impedance. The ITWDF can reflect the fault location effectively, and the threshold setting is simple. Moreover, the proposed method is high-adaptability in different network topologies.

Conventional single-ended traveling wave fault location methods, which commonly depend on partial features of traveling wavefronts, may lead to fault location failure, especially for weak-signal faults such as high impedance or zero-crossing faults and close-in faults. To tackle it, this article presents a fault location method according to the extracted panoramic features of traveling wave full waveform (TWFW) in time-frequency domains. Firstly, two rules of the TWFW features are probed, that is, the wavefront arrival sequence is varied from fault sections, as well as the frequency distribution of wavefronts is strongly related to the precise fault distance. Ergo, the mapping relationships between the TWFW and fault distance are qualitatively confirmed in order to demonstrate the uniqueness of the TWFW subsequently. Next, a LeNet-5-based convolution neural network (CNN) model is constructed to quantitatively evaluate such mapping relationships. In this model, various TWFW features will be extracted to the convolution channels when the CNN parameters are optimized adaptively, and the mapping relations can be formed in light of these sensitive features to estimate the fault distance. Finally, a Grad-CAM visualization method is deployed in the case study, and the accuracy along with the robustness of the proposed method in various fault conditions can be validated consequently.

针对电压行波传感器二次侧故障行波信号不能真实反映电网一次行波波形特征的问题,提出了一种基于L1正则化反演的电压行波高精度检测方法.首先,分析了行波传感器的非理想传变特性,揭示了一、二次行波信号的波形差异性.在此基础上,提出利用小波包变换对观测信号进行多尺度分解,并对各频段信号分别进行反演的方法,从而减小由行波传感器引起的畸变误差.其次,在反演模型中引入L1正则化约束对模型进行稀疏性刻画,使反演结果更能体现真实故障波形特征.最后,利用快速迭代收缩阈值算法(fast iterative shrinkage-thresholding algorithm,FISTA)进行迭代求解,将各分量的反演波形线性叠加,实现故障行波信号的精确还原.仿真和实验结果表明:与直接反演相比,所提方法能够实现故障行波在时域和频域上的高精度真实测量,在微弱故障和噪声环境下也能获得较为精确的反演结果,具有一定的工程应用价值.

配电网结构复杂、分支线路众多,利用较少的行波定位装置实现全网故障定位具有显著经济效益.基于此,提出了一种基于全网故障可观测的行波定位装置优化配置方法.首先,引入树形结构图,定义配电网线路各节点间的层级关系.并根据行波传输特性,提出了装置配置基本条件:在目标配电网树形结构图中,邻接节点与其父节点至少有一个节点配置装置.将该基本条件作为约束,建立以装置数量最少化为目标函数的数学模型.模型中引入权重系数,量化线路长度、历史故障率对装置配置需求程度的影响.最后,使用改进灰狼算法求解模型最优解,形成最优配置方案.仿真算例结果表明,基于所提优化配置方法,在满足定位精度的前提下可显著减少行波定位装置数量、降低投资成本,故障定位误差小于100 m.且在线路N-1情况下具有一定的适应性,满足实际工程需求.

受高阻接地故障、过零点故障和高频噪声等因素的影响,行波波头检测困难,导致行波保护和故障定位方法可靠性不高.由于故障行波具有全时频特性,检测一定时间窗内时频域行波波形将包含全景故障信息,从而实现故障特征可观测.融合故障行波时频域信息,提出了基于全波形信息的故障行波表现形式.在此基础上,提出了一种基于变分模态分解(VMD)和Wigner-Ville分布(WVD)的行波时频分析方法.对故障行波进行VMD分解,得到多个单一成分的固有模态分量.利用Wigner-Ville分布分解各固有模态分量,将分解结果线性叠加即可得到行波时频域波形,即行波全波形.利用该方法对模拟和实测行波信号进行分析,结果表明:该方法具有良好的噪声抑制作用,同时具有较高的时频分辨率和聚集性,实现了行波全波形的真实、准确提取.应用于1:1真型配电网实验平台中,可实现准确故障定位,且不受高阻接地故障、过零点故障、噪声干扰和采样率等因素的影响,有望大大提高行波保护和故障定位方法的可靠性.

With the construction of a new generation of electric power systems with new energy as the main body, there are major changes in the traditional electrical engineering undergraduate teaching system. On the one hand, the dynamic operating characteristics of the power system dominated by new energy sources and dominated by power electronic equipment will be significantly different from the traditional power system dominated by synchronous generators. Therefore, there is an urgent need to reform the curriculum system to train electrical engineering professionals who can adapt to the future development trend of the power system. On the other hand, with the rapid advancement of advanced information technology in recent years, the application of cutting-edge technologies such as cloud computing, big data, and artificial intelligence in the field of electrical engineering has become more and more in-depth. Therefore, there is an urgent need to cultivate advanced technical experts who possess more advanced information theory and cybernetics knowledge and master the application of cutting-edge information technology in the field of electrical engineering. From the perspective of student training and professional construction needs, this paper systematically analyzes the lag of the traditional electrical engineering curriculum system, and analyzes the knowledge needs of electrical engineering students adapting to the development of the new generation of power systems. From the perspective of teaching schemes, curriculum systems, and evaluation systems, a plan for restructuring the teaching system was proposed. The work in this article can provide a reference for the training of college students majoring in electrical engineering, and it can also provide a more feasible improvement idea for the training of college students in other majors.

The transition resistance of high impedance fault (HIF) in distribution network can reach thousands of ohms. The fault characteristics are weak, and it is easy to be confused with normal transient disturbances (NTD), such as no-load line switching, load switching, capacitor switching. Therefore, the fault detection is very difficult. In this paper, the difference of traveling wave full waveform energy distribution between HIF and NTD is deeply analyzed: the energy distribution of HIF is relatively uniform, and the degree of sparseness is small; the energy distribution of NTD is relatively concentrated, and the degree of sparseness is large. Accordingly, a HIF detection method based on the energy distribution sparsity of the traveling wave full waveform is proposed. By calculating the sparseness factor of the time-frequency energy spectrum matrix of the traveling wave full waveform, the sparsity of the energy distribution is represented, and the HIF detection criterion is constructed to reliably identify the and NTD. Simulation analysis shows that this method can sensitively and reliably detect HIF in distribution networks up to 5kΩ.

该文深入剖析故障暂态信息产生、传输、传变的机理,提出一种基于多源暂态信息融合的输电线路单端故障定位方法.根据故障暂态信号的传输过程和折反射机理,定性分析不同故障位置暂态信号的时-频差异性,揭示故障位置和时频特征量之间的内在联系,挖掘暂态信息中能够充分体现时频域差异性的五种典型故障特征量:各次暂态浪涌到达时间差、幅值、极性,各次暂态浪涌高低频电压幅值比和主频分量波速度.利用五种故障特征量,构建暂态信息融合矩阵,量化分析不同故障条件下,暂态信息矩阵的相似性和差异性,进而实现电力线路的准确故障定位.该文所提方法融合多源暂态特征量,有效地克服了现有故障定位方法对单一特征量准确提取的严重依赖,具备较强的容错性能.理论研究和大量算例分析结果表明:所提定位方法受故障条件影响小,在线路发生末端高阻故障或者电压过零点故障时,仍能准确定位故障位置,可靠性高.

现有单端行波保护方法提取的故障特征量难以综合反映故障位置和边界特性,易受过渡电阻等不同故障工况的影响,导致保护整定困难.该文深入分析了区内外故障初始反行波的传输路径差异,研究了线路参数及线路边界对行波幅值衰减的影响,确定以初始反行波各频段幅值衰减特性为依据提取故障特征量.在此基础上,利用连续小波变换分解包含初始反行波的行波信号,得到行波全波形,并构造随故障位置动态变化的行波全波形主频分量.理论分析发现:区内故障时行波全波形主频分量随故障距离增大而减小;线路边界导致区内外故障主频分量差异显著.据此提出一种基于行波全波形主频分量衰减特性的输电线路快速保护方法.该方法以动态变化的行波全波形主频分量设置保护判据,避免不同网络拓扑结构、故障工况对保护整定值的影响.仿真结果表明所提方法能快速、可靠判别区内外故障,在线路末端经300??过渡电阻故障时仅需1ms时间窗采样数据也能可靠动作,在不同故障工况下具有较强适应性.

配电网发生高阻接地故障(high impedance fault,HIF)时,故障特征微弱且易与空载线路投切,负荷投切,电容投切等正常暂态扰动混淆,检测困难.针对上述问题,该文定性和定量分析HIF与正常暂态扰动信号在时频域能量分布上的差异,发现前者在时频域上能量分布相对均匀,稀疏程度小,后者在时频域上能量分布相对集中,稀疏程度大.据此,提出一种基于行波全波形能量分布特征的HIF灵敏、可靠检测方法.首先,利用连续小波变换时频局部化特性,将行波波形的时频域故障信息可视化呈现,得到行波全波形,基于此,构建时频能谱矩阵;通过行波全波形能量突变构造启动判据;引入稀疏因子量化时频能谱矩阵的能量分布均匀程度,构造HIF检测判据,可靠辨识HIF和正常暂态扰动.仿真分析及现场实验结果表明:该方法可以灵敏、可靠检测高达5k??的配电网HIF.

为提高配电网单相接地故障选线的准确性和可靠性,提出一种基于时频域行波全景波形的配电网故障选线方法,融合全景故障特征量实现可靠配电网选线.首先,利用S变换提取各条线路时频域行波全景波形,真实展现故障行波全景特征量,详细分析故障线路和健全线路行波全景波形中幅值、频率和极性的差异性和相似性;然后,利用波形相似度原理,对故障线路和健全线路检测的行波全景波形的差异性和相似性进行量化分析;最后,针对各条线路构建相似度关联系数矩阵,通过计算各条线路综合相关系数,放大故障线路和健全线路行波全景波形的差异性,实现准确、可靠配电网故障选线.所提故障选线方法无需人工设置阈值,大量仿真分析结果表明:该方法具有较高的算法适应性,在3k?高阻接地故障下,仍能实现可靠配电网故障选线.