技术领域Technical field
本发明涉及数据交互技术领域，尤其涉及一种人机终端画面数据交互方法及系统。The present invention relates to the field of data interaction technology, and in particular to a human-machine terminal screen data interaction method and system.
背景技术Background technique
电力系统中的设备故障可能对整个系统造成严重影响，需要快速、准确地进行故障诊断和预测。通过人机终端画面数据交互，运维人员可以利用机器学习和数据分析技术，实现对电力设备故障的早期诊断和预测Equipment failure in the power system may have a serious impact on the entire system, requiring rapid and accurate fault diagnosis and prediction. Through human-machine terminal screen data interaction, operation and maintenance personnel can use machine learning and data analysis technology to achieve early diagnosis and prediction of power equipment faults.
现有技术中，CN113726856A公开了一种基于微服务的调控画面综合数据轻量化交互方法及系统，方法包括如下步骤：对部署在客户端的单体应用程序下的功能进行分类，分成必要功能和非必要功能两类；将单体应用程序下的非必要功能按照功能特点划分为多个功能单一的微服务；对多个功能单一的微服务进行部署，部署到服务器端的多台服务器上，组成服务器集群；接收客户端发送的调用请求，并根据服务器集群的运行状态，选择运行请求服务的服务器；将服务器运行服务的结果返回到客户端。In the prior art, CN113726856A discloses a microservice-based comprehensive data lightweight interaction method and system for control screens. The method includes the following steps: classifying the functions deployed under the single application program on the client into necessary functions and non-essential functions. There are two categories of necessary functions; the non-essential functions under a single application are divided into multiple single-function microservices according to their functional characteristics; multiple single-function microservices are deployed and deployed to multiple servers on the server side to form a server Cluster; receives the call request sent by the client, and selects the server to run the requested service according to the running status of the server cluster; returns the result of the server running the service to the client.
综上，现有技术虽然能够实现人机终端画面数据的交互，但仅能实现预设的简单需求，无法对电网系统中的信息进行实时交互，因此需要一种方案解决现有技术中存在的问题。In summary, although the existing technology can realize the interaction of human-machine terminal screen data, it can only realize the preset simple requirements and cannot interact with the information in the power grid system in real time. Therefore, a solution is needed to solve the problems existing in the existing technology. question.
发明内容Contents of the invention
本发明实施例提供一种人机终端画面数据交互方法及系统，用于基于电网数据对电网故障进行诊断并通过人机终端画面数据实现交互。Embodiments of the present invention provide a human-machine terminal screen data interaction method and system, which are used to diagnose power grid faults based on power grid data and realize interaction through the human-machine terminal screen data.
本发明实施例的第一方面，提供一种人机终端画面数据交互方法，包括：A first aspect of the embodiment of the present invention provides a human-machine terminal screen data interaction method, including:
安装行波传感器，根据所述行波传感器获取故障电磁波，确定所述故障电磁波对应的时间戳和所述故障电磁波的行波传播速度，根据所述时间戳和所述行波传播速度，通过故障定位算法确定故障区域；Install a traveling wave sensor, acquire the fault electromagnetic wave according to the traveling wave sensor, determine the time stamp corresponding to the fault electromagnetic wave and the traveling wave propagation speed of the fault electromagnetic wave, and pass the fault based on the time stamp and the traveling wave propagation speed. The localization algorithm determines the fault area;
根据所述故障区域获取所述故障电磁波的行波信号数据，同时获取电网拓扑数据，提取所述行波信号数据中的行波特征和所述电网拓扑数据中的拓扑特征，根据所述拓扑特征确定结构特征，识别电网拓扑图中节点最短路径，根据所述节点最短路径结合预先引入的聚类系数进行特征融合，得到综合拓扑向量；Acquire the traveling wave signal data of the fault electromagnetic wave according to the fault area, obtain the power grid topology data at the same time, extract the traveling wave characteristics in the traveling wave signal data and the topology features in the power grid topology data, and extract the topological features according to the topology features. Determine the structural characteristics, identify the shortest path of the nodes in the power grid topology diagram, and perform feature fusion based on the shortest path of the node combined with the pre-introduced clustering coefficient to obtain a comprehensive topology vector;
基于预设的图神经网络对所述综合拓扑向量进行编码，并对所述行波特征和编码后的综合拓扑向量进行节点特征融合，生成初始节点特征，根据所述初始节点特征，结合预先引入的注意力机制，通过所述图神经网络更新所述初始节点特征，生成输入节点特征，根据所述输入节点特征，结合所述行波特征，通过所述图神经网络中的随机森林模块进行故障定位，确定故障位置。The comprehensive topology vector is encoded based on the preset graph neural network, and the node features are fused with the traveling wave features and the encoded comprehensive topology vector to generate initial node features. According to the initial node features, combined with the pre-introduced The attention mechanism updates the initial node features through the graph neural network and generates input node features. Based on the input node features and combined with the traveling wave features, faults are performed through the random forest module in the graph neural network. Locate and determine the location of the fault.
在一种可选的实施方式中，In an alternative implementation,
所述安装行波传感器，根据所述行波传感器获取故障电磁波，确定所述故障电磁波对应的时间戳和所述故障电磁波的行波传播速度，根据所述时间戳和所述行波传播速度，通过故障定位算法确定故障区域包括：The traveling wave sensor is installed, and the fault electromagnetic wave is acquired according to the traveling wave sensor, and the time stamp corresponding to the fault electromagnetic wave and the traveling wave propagation speed of the fault electromagnetic wave are determined. According to the time stamp and the traveling wave propagation speed, Determining the fault area through fault location algorithm includes:
在电力线路中安装行波传感器，根据所述行波传感器获取因故障引起的故障电磁波，并根据所述行波传感器获取所述故障电磁波的时间为行波检测数据添加时间戳，确定所述电力线路中的线路长度和导线类型，确定所述故障电磁波的行波传播速度；Install a traveling wave sensor in the power line, acquire the fault electromagnetic wave caused by the fault according to the traveling wave sensor, and add a time stamp to the traveling wave detection data according to the time when the fault electromagnetic wave is acquired by the traveling wave sensor, and determine the power The line length and conductor type in the line determine the traveling wave propagation speed of the fault electromagnetic wave;
根据所述时间戳和所述行波传播速度，判断故障发生区域中行波传感器的数量，若有两台及以上行波传感器，则通过比较所述故障电磁波的波形在不同行波传感器的到达时间，确定故障区域，若只有一台行波传感器，则通过反射行波的信息确定故障区域。According to the timestamp and the traveling wave propagation speed, the number of traveling wave sensors in the fault occurrence area is determined. If there are two or more traveling wave sensors, the arrival time of the waveform of the fault electromagnetic wave at different traveling wave sensors is compared. , determine the fault area. If there is only one traveling wave sensor, the fault area is determined by reflecting the traveling wave information.
在一种可选的实施方式中，In an alternative implementation,
所述根据所述故障区域获取所述故障电磁波的行波信号数据，同时获取电网拓扑数据，提取所述行波信号数据中的行波特征和所述电网拓扑数据中的拓扑特征包括：Obtaining the traveling wave signal data of the fault electromagnetic wave according to the fault area and acquiring the power grid topology data at the same time, and extracting the traveling wave characteristics in the traveling wave signal data and the topological features in the power grid topology data include:
根据所述故障区域，获取所述故障电磁波的幅值、频率、能量分布和相位角，记为行波数据信号，获取电网拓扑数据，包括节点信息、连接关系和线路参数；According to the fault area, the amplitude, frequency, energy distribution and phase angle of the fault electromagnetic wave are obtained, recorded as traveling wave data signals, and the power grid topology data is obtained, including node information, connection relationships and line parameters;
对所述行波数据信号进行预处理并提取所述行波数据信号中的幅值、能量和持续时间，并通过傅里叶变换分析所述行波数据信号的频率成分，并将提取得到的数据组合成行波特征；Preprocess the traveling wave data signal and extract the amplitude, energy and duration of the traveling wave data signal, analyze the frequency component of the traveling wave data signal through Fourier transform, and extract the obtained Data are combined into traveling wave features;
识别电网中的每个节点及节点对应的属性，根据节点的连接关系，确定连接类型和连接参数，将所述连接关系、所述连接类型和所述连接参数组合成拓扑特征。Identify each node in the power grid and the corresponding attributes of the node, determine the connection type and connection parameters according to the connection relationship of the nodes, and combine the connection relationship, the connection type and the connection parameters into topological features.
在一种可选的实施方式中，In an alternative implementation,
所述根据所述拓扑特征确定结构特征，识别电网拓扑图中节点最短路径，根据所述节点最短路径结合预先引入的聚类系数进行特征融合，得到综合拓扑向量包括：Determining structural features based on the topological features, identifying the shortest paths of nodes in the power grid topology diagram, performing feature fusion based on the shortest paths of nodes combined with pre-introduced clustering coefficients, and obtaining a comprehensive topology vector includes:
根据所述拓扑特征，对于电网中的每个节点，计算与当前节点直接相连的节点数量，记为当前节点的发散度，根据所述发散度，提取与当前节点直接相连的节点信息，并将与当前节点直接相连的节点记为当前节点的邻居节点；According to the topological characteristics, for each node in the power grid, the number of nodes directly connected to the current node is calculated and recorded as the divergence of the current node. According to the divergence, the node information directly connected to the current node is extracted and Nodes directly connected to the current node are recorded as neighbor nodes of the current node;
根据所述邻居节点，确定所述邻居节点与所述当前节点间的实际连接数，并计算所述邻居节点与所述当前节点间的聚类系数，遍历所述电网，计算所述电网中每个节点的聚类系数和电网的平均聚类系数，并通过预设的局部聚类算法计算当前节点的局部聚类系数，根据所述局部聚类系数将所述发散度，所述邻居节点的特征信息和节点间的路径信息进行融合，得到综合拓扑向量。According to the neighbor node, determine the actual number of connections between the neighbor node and the current node, calculate the clustering coefficient between the neighbor node and the current node, traverse the power grid, and calculate each node in the power grid. clustering coefficient of the nodes and the average clustering coefficient of the power grid, and calculate the local clustering coefficient of the current node through the preset local clustering algorithm. According to the local clustering coefficient, the divergence, the neighbor node's Feature information and path information between nodes are fused to obtain a comprehensive topology vector.
在一种可选的实施方式中，In an alternative implementation,
所述通过预设的局部聚类算法计算当前节点的局部聚类系数如下公式所示：The local clustering coefficient of the current node is calculated through the preset local clustering algorithm as shown in the following formula:
； ;
其中，Cwi表示节点i的局部聚类系数，△ijk表示节点i，j和k之间的三角关系，wij表示边ij的权重，wik表示边ik的权重，wjk表示边jk的权重，fi表示节点i的发散度。Among them,Cwi represents the local clustering coefficient of node i,△ijk represents the triangular relationship between nodesi , j and k,wij represents the weight of edge ij,wik represents the weight of edge ik,wjk represents the weight of edge jk The weight,fi represents the divergence of node i.
在一种可选的实施方式中，In an alternative implementation,
所述基于预设的图神经网络对所述综合拓扑向量进行编码，并对所述行波特征和编码后的综合拓扑向量进行节点特征融合，生成初始节点特征，根据所述节点特征，结合预先引入的注意力机制，通过所述图神经网络更新所述初始节点特征，生成输入节点特征包括：The graph neural network based on the preset encodes the comprehensive topology vector, and performs node feature fusion on the traveling wave features and the coded comprehensive topology vector to generate initial node features. According to the node features, combined with the preset The introduced attention mechanism updates the initial node features through the graph neural network, and generates input node features including:
获取所述综合拓扑向量，根据所述综合拓扑向量和所述行波特征初始化节点特征并构建当前节点的邻接矩阵并确定当前节点的邻居节点，根据所述邻居节点更新当前节点，记为初始节点特征，对于每个初始节点特征，收集所述邻居节点的节点信息，并根据预先引入的多头注意力机制计算每个邻居节点对当前节点的重要性，即注意力系数；Obtain the comprehensive topology vector, initialize the node characteristics according to the comprehensive topology vector and the traveling wave characteristics and construct the adjacency matrix of the current node and determine the neighbor nodes of the current node, update the current node according to the neighbor nodes, and record it as the initial node Features, for each initial node feature, collect the node information of the neighbor nodes, and calculate the importance of each neighbor node to the current node according to the pre-introduced multi-head attention mechanism, that is, the attention coefficient;
将每个邻居节点的节点信息进行合并，结合每个邻居节点对应的注意力系数通过激活函数加权计算邻居节点的综合特征，根据所述综合特征，更新当前节点的特征表示，生成输入节点特征。The node information of each neighbor node is merged, and the comprehensive features of the neighbor nodes are calculated by weighting the activation function in combination with the attention coefficient corresponding to each neighbor node. Based on the comprehensive features, the feature representation of the current node is updated to generate input node features.
在一种可选的实施方式中，In an alternative implementation,
所述根据所述输入节点特征，结合所述行波特征，通过所述图神经网络中的随机森林模块进行故障定位，确定故障位置包括：According to the input node characteristics, combined with the traveling wave characteristics, fault location is performed through the random forest module in the graph neural network, and the determination of the fault location includes:
根据所述行波特征初始化所述图神经网络中的随机森林模块，设置所述随机森林模块的超参数，即树的数量和深度，将所述输入节点特征作为输入信息添加至所述随机森林模块中；Initialize the random forest module in the graph neural network according to the traveling wave characteristics, set the hyperparameters of the random forest module, that is, the number and depth of trees, and add the input node characteristics as input information to the random forest in module;
对于所述随机森林模块中的每个树，随机生成训练集，根据所述训练集交叉验证不同树的数量下所述随机森林模块的性能，选择性能最佳时树的数量，对于每个树，增加最大深度，并根据所述训练集设置所述树中每个节点进行拆分的最小样本数，同时设置叶节点的最小样本数，在调整超参数的过程中观察所述随机森林模块的性能，当性能不再提升时，根据对应的超参数调整所述随机森林模块中树的深度和树的数量；For each tree in the random forest module, a training set is randomly generated. According to the training set cross-validation of the performance of the random forest module under different numbers of trees, the number of trees with the best performance is selected. For each tree , increase the maximum depth, and set the minimum number of samples for each node in the tree to be split according to the training set. At the same time, set the minimum number of samples for leaf nodes, and observe the performance of the random forest module during the process of adjusting the hyperparameters. Performance, when the performance no longer improves, adjust the depth of the trees and the number of trees in the random forest module according to the corresponding hyperparameters;
所述随机森林模块将所述输入节点特征发送至模块中的每个树中，所述树生成每个节点对应的预测故障概率，结合节点对应的行波信息判断所述预测故障概率的置信度，若所述置信度大于预设的置信度阈值，则认为当前节点存在故障，获取故障位置并返回。The random forest module sends the input node characteristics to each tree in the module. The tree generates the predicted failure probability corresponding to each node, and determines the confidence of the predicted failure probability based on the traveling wave information corresponding to the node. , if the confidence is greater than the preset confidence threshold, it is considered that the current node has a fault, the fault location is obtained and returned.
本发明实施例的第二方面，提供一种人机终端画面数据交互系统，包括：A second aspect of the embodiment of the present invention provides a human-machine terminal screen data interaction system, including:
第一单元，用于安装行波传感器，根据所述行波传感器获取故障电磁波，确定所述故障电磁波对应的时间戳和所述故障电磁波的行波传播速度，根据所述时间戳和所述行波传播速度，通过故障定位算法确定故障区域；The first unit is used to install a traveling wave sensor, obtain a fault electromagnetic wave according to the traveling wave sensor, determine the time stamp corresponding to the fault electromagnetic wave and the traveling wave propagation speed of the fault electromagnetic wave, and determine the time stamp corresponding to the fault electromagnetic wave and the traveling wave propagation speed of the fault electromagnetic wave. The wave propagation speed determines the fault area through the fault location algorithm;
第二单元，用于根据所述故障区域获取所述故障电磁波的行波信号数据，同时获取电网拓扑数据，提取所述行波信号数据中的行波特征和所述电网拓扑数据中的拓扑特征，根据所述拓扑特征确定结构特征，识别电网拓扑图中节点最短路径，根据所述节点最短路径结合预先引入的聚类系数进行特征融合，得到综合拓扑向量；The second unit is used to obtain the traveling wave signal data of the fault electromagnetic wave according to the fault area, obtain the power grid topology data at the same time, and extract the traveling wave characteristics in the traveling wave signal data and the topological features in the power grid topology data. , determine the structural features according to the topological features, identify the shortest paths of nodes in the power grid topology diagram, perform feature fusion based on the shortest paths of nodes combined with the pre-introduced clustering coefficients, and obtain a comprehensive topology vector;
第三单元，用于基于预设的图神经网络对所述综合拓扑向量进行编码，并对所述行波特征和编码后的综合拓扑向量进行节点特征融合，生成初始节点特征，根据所述初始节点特征，结合预先引入的注意力机制，通过所述图神经网络更新所述初始节点特征，生成输入节点特征，根据所述输入节点特征，结合所述行波特征，通过所述图神经网络中的随机森林模块进行故障定位，确定故障位置。The third unit is used to encode the comprehensive topology vector based on the preset graph neural network, and perform node feature fusion on the traveling wave features and the encoded comprehensive topology vector to generate initial node features. According to the initial Node features, combined with the attention mechanism introduced in advance, update the initial node features through the graph neural network, and generate input node features. According to the input node features, combined with the traveling wave features, through the graph neural network The random forest module performs fault location and determines the fault location.
本发明实施例的第三方面，A third aspect of the embodiment of the present invention,
提供一种电子设备，包括：An electronic device is provided, including:
处理器；processor;
用于存储处理器可执行指令的存储器；Memory used to store instructions executable by the processor;
其中，所述处理器被配置为调用所述存储器存储的指令，以执行前述所述的方法。Wherein, the processor is configured to call instructions stored in the memory to execute the aforementioned method.
本发明实施例的第四方面，The fourth aspect of the embodiment of the present invention,
提供一种计算机可读存储介质，其上存储有计算机程序指令，所述计算机程序指令被处理器执行时实现前述所述的方法。A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.
本发明中，通过行波传感器获取故障电磁波的时间戳和行波传播速度，结合故障定位算法，能够精确定位故障发生的区域，提高了故障定位的准确性，利用电网拓扑数据提取结构特征，包括电网中节点的最短路径和聚类系数有助于更好地捕捉电力系统的结构信息，提供更全面的系统认知，利用图神经网络对行波特征和编码后的综合拓扑向量进行节点特征融合，生成初始节点特征提高了模型对节点状态的综合把握，为后续的故障定位提供更准确的输入，引入注意力机制对初始节点特征进行更新，根据不同节点的重要性进行加权有助于提高模型对电网中不同节点的关注程度，更有针对性地进行故障定位，综上，本发明实现了高效、准确的电力系统故障定位并实现了用户和系统间的数据交互。In the present invention, the time stamp and traveling wave propagation speed of the fault electromagnetic wave are obtained through the traveling wave sensor, and combined with the fault location algorithm, the area where the fault occurs can be accurately located, which improves the accuracy of fault location. The power grid topology data is used to extract structural features, including The shortest paths and clustering coefficients of nodes in the power grid help to better capture the structural information of the power system and provide a more comprehensive system understanding. The graph neural network is used to fuse traveling wave features and encoded comprehensive topology vectors for node features. , generating initial node features improves the model's comprehensive grasp of the node status, and provides more accurate input for subsequent fault location. Introducing an attention mechanism to update the initial node features, and weighting according to the importance of different nodes helps improve the model The degree of attention to different nodes in the power grid can be used to perform more targeted fault location. In summary, the present invention realizes efficient and accurate power system fault location and realizes data interaction between users and systems.
附图说明Description of the drawings
图1为本发明实施例人机终端画面数据交互方法的流程示意图；Figure 1 is a schematic flow chart of a human-machine terminal screen data interaction method according to an embodiment of the present invention;
图2为本发明实施例人机终端画面数据交互系统的结构示意图。Figure 2 is a schematic structural diagram of a human-machine terminal screen data interaction system according to an embodiment of the present invention.
具体实施方式Detailed ways
为使本发明实施例的目的、技术方案和优点更加清楚，下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments These are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.
下面以具体地实施例对本发明的技术方案进行详细说明。下面这几个具体的实施例可以相互结合，对于相同或相似的概念或过程可能在某些实施例不再赘述。The technical solution of the present invention will be described in detail below with specific examples. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.
图1为本发明实施例人机终端画面数据交互方法的流程示意图，如图1所示，所述方法包括：Figure 1 is a schematic flow chart of a human-machine terminal screen data interaction method according to an embodiment of the present invention. As shown in Figure 1, the method includes:
S1.安装行波传感器，根据所述行波传感器获取故障电磁波，确定所述故障电磁波对应的时间戳和所述故障电磁波的行波传播速度，根据所述时间戳和所述行波传播速度，通过故障定位算法确定故障区域；S1. Install a traveling wave sensor, obtain the fault electromagnetic wave according to the traveling wave sensor, determine the timestamp corresponding to the fault electromagnetic wave and the traveling wave propagation speed of the fault electromagnetic wave, according to the time stamp and the traveling wave propagation speed, Determine the fault area through fault location algorithm;
所述行波传感器是一种用于检测电力系统中的故障的传感器，当电力系统中发生故障时，故障点处会产生电磁波，行波传感器可以探测这些电磁波并提供关于故障的信息，所述故障电磁波是在电力系统中发生故障时产生的电磁辐射，所述行波传播速度是故障电磁波在电力系统中传播的速度，所述故障定位算法用于根据行波传感器测量到的电磁波信息，确定电力系统中的故障位置。The traveling wave sensor is a sensor used to detect faults in the power system. When a fault occurs in the power system, electromagnetic waves will be generated at the fault point. The traveling wave sensor can detect these electromagnetic waves and provide information about the fault. The fault electromagnetic wave is the electromagnetic radiation generated when a fault occurs in the power system. The traveling wave propagation speed is the speed at which the fault electromagnetic wave propagates in the power system. The fault location algorithm is used to determine based on the electromagnetic wave information measured by the traveling wave sensor. Location of faults in power systems.
在一种可选的实施方式中，In an alternative implementation,
所述安装行波传感器，根据所述行波传感器获取故障电磁波，确定所述故障电磁波对应的时间戳和所述故障电磁波的行波传播速度，根据所述时间戳和所述行波传播速度，通过故障定位算法确定故障区域包括：The traveling wave sensor is installed, and the fault electromagnetic wave is acquired according to the traveling wave sensor, and the time stamp corresponding to the fault electromagnetic wave and the traveling wave propagation speed of the fault electromagnetic wave are determined. According to the time stamp and the traveling wave propagation speed, Determining the fault area through fault location algorithm includes:
在电力线路中安装行波传感器，根据所述行波传感器获取因故障引起的故障电磁波，并根据所述行波传感器获取所述故障电磁波的时间为行波检测数据添加时间戳，确定所述电力线路中的线路长度和导线类型，确定所述故障电磁波的行波传播速度；Install a traveling wave sensor in the power line, acquire the fault electromagnetic wave caused by the fault according to the traveling wave sensor, and add a time stamp to the traveling wave detection data according to the time when the fault electromagnetic wave is acquired by the traveling wave sensor, and determine the power The line length and conductor type in the line determine the traveling wave propagation speed of the fault electromagnetic wave;
根据所述时间戳和所述行波传播速度，判断故障发生区域中行波传感器的数量，若有两台及以上行波传感器，则通过比较所述故障电磁波的波形在不同行波传感器的到达时间，确定故障区域，若只有一台行波传感器，则通过反射行波的信息确定故障区域。According to the timestamp and the traveling wave propagation speed, the number of traveling wave sensors in the fault occurrence area is determined. If there are two or more traveling wave sensors, the arrival time of the waveform of the fault electromagnetic wave at different traveling wave sensors is compared. , determine the fault area. If there is only one traveling wave sensor, the fault area is determined by reflecting the traveling wave information.
安装行波传感器并确保其能够准确捕捉故障电磁波，通过行波传感器获取故障电磁波，并根据所述故障电磁波到达所述行波传感器的时间为每个数据点添加时间戳，根据电力线路的实际情况，测量线路的长度和确定导线的类型，使用已知的线路长度和导线类型，计算故障电磁波在电力线路中的行波传播速度；Install the traveling wave sensor and ensure that it can accurately capture the fault electromagnetic wave, obtain the fault electromagnetic wave through the traveling wave sensor, and add a timestamp to each data point according to the time when the fault electromagnetic wave reaches the traveling wave sensor, according to the actual situation of the power line , measure the length of the line and determine the type of conductor, use the known line length and conductor type to calculate the traveling wave propagation speed of the fault electromagnetic wave in the power line;
根据传感器的布局和电力线路的特性，判断故障发生区域内需要安装或已经安装的行波传感器的数量和位置，如果有两台及以上行波传感器，通过比较故障电磁波在不同传感器上的到达时间，使用时差定位法计算故障发生的位置，如果只有一台行波传感器，通过对反射波形的分析和处理确定故障区域。According to the layout of the sensor and the characteristics of the power line, determine the number and location of traveling wave sensors that need to be installed or have been installed in the fault area. If there are two or more traveling wave sensors, compare the arrival time of the fault electromagnetic wave on different sensors. , use the time difference positioning method to calculate the location of the fault. If there is only one traveling wave sensor, the fault area is determined by analyzing and processing the reflected waveform.
本实施例中，实时监测每个行波传感器捕获到的故障电磁波数据，使得系统能够迅速响应故障事件，提高电力系统的可靠性，利用人机终端的图形界面，将电力线路的布局、传感器位置、故障定位结果等以图形化的方式展示，有助于用户直观地理解故障发生的位置，减少用户的认知负担，用户可以与系统进行交互式分析，如点击某个传感器点来查看该点的详细信息，包括故障电磁波波形、时间戳，使用户能够更深入地了解故障定位的过程，综上，本实施例可以提高用户对电力系统状态的实时监测和分析能力，实现对故障的迅速定位和处理。In this embodiment, the fault electromagnetic wave data captured by each traveling wave sensor is monitored in real time, so that the system can quickly respond to fault events and improve the reliability of the power system. The graphical interface of the human-machine terminal is used to combine the layout of the power line and the location of the sensor. , fault location results, etc. are displayed in a graphical manner, which helps users intuitively understand the location of the fault and reduces the user's cognitive load. Users can conduct interactive analysis with the system, such as clicking on a sensor point to view the point. Detailed information, including fault electromagnetic waveforms and timestamps, enables users to have a deeper understanding of the fault location process. In summary, this embodiment can improve users' real-time monitoring and analysis capabilities of the power system status and achieve rapid fault location. and processing.
S2.根据所述故障区域获取所述故障电磁波的行波信号数据，同时获取电网拓扑数据，提取所述行波信号数据中的行波特征和所述电网拓扑数据中的拓扑特征，根据所述拓扑特征确定结构特征，识别电网拓扑图中节点最短路径，根据所述节点最短路径结合预先引入的聚类系数进行特征融合，得到综合拓扑向量；S2. Acquire the traveling wave signal data of the fault electromagnetic wave according to the fault area, obtain the power grid topology data at the same time, extract the traveling wave characteristics in the traveling wave signal data and the topological features in the power grid topology data, according to the Topological features determine structural features, identify the shortest paths of nodes in the power grid topology diagram, and perform feature fusion based on the shortest paths of nodes combined with pre-introduced clustering coefficients to obtain a comprehensive topology vector;
所述电网拓扑数据是指描述电力系统中各元件（如发电机、变电站、输电线路等）之间连接关系的信息，包括节点、支路、线路的连接方式和拓扑结构，以及它们之间的电气连接关系，所述行波特征是指在电力系统中传播的电磁波的性质和特点，所述拓扑特征描述了电力系统的结构连接关系，通常涉及节点和支路之间的连接关系，所述结构特征在电力系统中通常指描述元件（例如变电站、发电机等）之间的物理结构、连接方式和电气参数等特性，所述聚类系数是一种网络分析中的度量，用于描述图中节点之间的紧密程度，在电力系统中，节点的聚类系数可以表示节点与其邻居节点之间的连接密度，所述综合拓扑向量是一种综合考虑电力系统拓扑结构的特征向量，通常包括节点度、支路的连接方式、聚类系数等多个拓扑特征的组合。The power grid topology data refers to information describing the connection relationships between various components in the power system (such as generators, substations, transmission lines, etc.), including the connection methods and topological structures of nodes, branches, and lines, as well as the connections between them. Electrical connection relationship, the traveling wave characteristics refer to the properties and characteristics of electromagnetic waves propagating in the power system, the topological characteristics describe the structural connection relationship of the power system, usually involving the connection relationship between nodes and branches, the Structural characteristics in power systems usually refer to the physical structure, connection methods, electrical parameters and other characteristics between components (such as substations, generators, etc.). The clustering coefficient is a measure in network analysis and is used to describe graphs. The degree of closeness between nodes in the power system. In the power system, the clustering coefficient of a node can represent the connection density between a node and its neighbor nodes. The comprehensive topology vector is a characteristic vector that comprehensively considers the topology of the power system, and usually includes A combination of multiple topological features such as node degree, branch connection mode, and clustering coefficient.
在一种可选的实施方式中，In an alternative implementation,
所述根据所述故障区域获取所述故障电磁波的行波信号数据，同时获取电网拓扑数据，提取所述行波信号数据中的行波特征和所述电网拓扑数据中的拓扑特征包括：Obtaining the traveling wave signal data of the fault electromagnetic wave according to the fault area and acquiring the power grid topology data at the same time, and extracting the traveling wave characteristics in the traveling wave signal data and the topological features in the power grid topology data include:
根据所述故障区域，获取所述故障电磁波的幅值、频率、能量分布和相位角，记为行波数据信号，获取电网拓扑数据，包括节点信息、连接关系和线路参数；According to the fault area, the amplitude, frequency, energy distribution and phase angle of the fault electromagnetic wave are obtained, recorded as traveling wave data signals, and the power grid topology data is obtained, including node information, connection relationships and line parameters;
对所述行波数据信号进行预处理并提取所述行波数据信号中的幅值、能量和持续时间，并通过傅里叶变换分析所述行波数据信号的频率成分，并将提取得到的数据组合成行波特征；Preprocess the traveling wave data signal and extract the amplitude, energy and duration of the traveling wave data signal, analyze the frequency component of the traveling wave data signal through Fourier transform, and extract the obtained Data are combined into traveling wave features;
识别电网中的每个节点及节点对应的属性，根据节点的连接关系，确定连接类型和连接参数，将所述连接关系、所述连接类型和所述连接参数组合成拓扑特征。Identify each node in the power grid and the corresponding attributes of the node, determine the connection type and connection parameters according to the connection relationship of the nodes, and combine the connection relationship, the connection type and the connection parameters into topological features.
根据已确定的故障区域，使用传感器获取故障电磁波的信号，包括波形、电压、电流等信息，收集电网拓扑数据，包括节点信息（节点编号、类型、坐标等）、连接关系（支路连接、节点连接关系）和线路参数（阻抗、长度等）；According to the identified fault area, sensors are used to obtain fault electromagnetic wave signals, including waveform, voltage, current and other information, and the power grid topology data is collected, including node information (node number, type, coordinates, etc.), connection relationships (branch connections, nodes, etc.) connection relationship) and line parameters (impedance, length, etc.);
对故障电磁波的行波数据信号进行预处理，包括滤波、降噪、去趋势，从行波数据信号中提取幅值、能量、持续时间等特征，使用傅里叶变换分析频率成分，得到频率、频谱分布等信息，将提取得到的各种特征组合成行波特征向量，包括幅值、频率、能量分布、相位角等多个特征参数；Preprocess the traveling wave data signal of the fault electromagnetic wave, including filtering, noise reduction, and detrending. Features such as amplitude, energy, and duration are extracted from the traveling wave data signal. Fourier transform is used to analyze the frequency components to obtain the frequency, Spectrum distribution and other information, the various extracted features are combined into traveling wave feature vectors, including amplitude, frequency, energy distribution, phase angle and other characteristic parameters;
根据电网拓扑数据，识别每个节点及其对应的属性信息，根据节点之间的连接关系，确定连接的类型（支路、变压器、开关等）和连接参数（阻抗、导纳等），将所得到的连接关系、连接类型和连接参数组合成拓扑特征向量，其中，所述拓扑特征向量描述了电力系统的拓扑结构。According to the power grid topology data, identify each node and its corresponding attribute information, determine the connection type (branch, transformer, switch, etc.) and connection parameters (impedance, admittance, etc.) based on the connection relationship between nodes, and combine all The obtained connection relationships, connection types and connection parameters are combined into a topological feature vector, where the topological feature vector describes the topological structure of the power system.
本实施例中，节点之间的连接关系、支路和线路参数等信息通过图形界面呈现，让用户能够清晰地了解电力系统的拓扑结构，通过傅里叶变换对行波数据信号进行频谱分析，将频率成分可视化呈现在人机终端上，用户可以直观地了解电磁波信号的频谱分布情况，幅值、频率、能量分布和相位角等故障特征以图形化方式展示在人机终端画面上，用户可以清晰地看到这些特征随时间的变化趋势，有助于快速识别和定位故障，综上，本实施例提高了系统的可操作性和用户体验，使用户可以在图形化界面上方便地获取、分析和理解电力系统的状态和故障信息。In this embodiment, information such as connection relationships between nodes, branch and line parameters are presented through a graphical interface, allowing users to clearly understand the topology of the power system, and perform spectrum analysis on traveling wave data signals through Fourier transform. The frequency components are visualized on the human-machine terminal. Users can intuitively understand the spectrum distribution of electromagnetic wave signals. Fault characteristics such as amplitude, frequency, energy distribution and phase angle are graphically displayed on the human-machine terminal screen. Users can Clearly seeing the changing trends of these features over time helps to quickly identify and locate faults. In summary, this embodiment improves the operability and user experience of the system, allowing users to easily obtain, Analyze and understand power system status and fault information.
在一种可选的实施方式中，In an alternative implementation,
所述根据所述拓扑特征确定结构特征，识别电网拓扑图中节点最短路径，根据所述节点最短路径结合预先引入的聚类系数进行特征融合，得到综合拓扑向量包括：Determining structural features based on the topological features, identifying the shortest paths of nodes in the power grid topology diagram, performing feature fusion based on the shortest paths of nodes combined with pre-introduced clustering coefficients, and obtaining a comprehensive topology vector includes:
根据所述拓扑特征，对于电网中的每个节点，计算与当前节点直接相连的节点数量，记为当前节点的发散度，根据所述发散度，提取与当前节点直接相连的节点信息，并将与当前节点直接相连的节点记为当前节点的邻居节点；According to the topological characteristics, for each node in the power grid, the number of nodes directly connected to the current node is calculated and recorded as the divergence of the current node. According to the divergence, the node information directly connected to the current node is extracted and Nodes directly connected to the current node are recorded as neighbor nodes of the current node;
根据所述邻居节点，确定所述邻居节点与所述当前节点间的实际连接数，并计算所述邻居节点与所述当前节点间的聚类系数，遍历所述电网，计算所述电网中每个节点的聚类系数和电网的平均聚类系数，并通过预设的局部聚类算法计算当前节点的局部聚类系数，根据所述局部聚类系数将所述发散度，所述邻居节点的特征信息和节点间的路径信息进行融合，得到综合拓扑向量。According to the neighbor node, determine the actual number of connections between the neighbor node and the current node, calculate the clustering coefficient between the neighbor node and the current node, traverse the power grid, and calculate each node in the power grid. clustering coefficient of the nodes and the average clustering coefficient of the power grid, and calculate the local clustering coefficient of the current node through the preset local clustering algorithm. According to the local clustering coefficient, the divergence, the neighbor node's Feature information and path information between nodes are fused to obtain a comprehensive topology vector.
所述发散度指与该节点相连的边的数量，表示了节点的连接程度，是节点的一个基本拓扑特征，所述邻居节点是指与该节点直接连接的其他节点，所述平均聚类系数是图中所有节点的局部聚类系数的平均值，衡量了一个节点的邻居之间相互连接的程度，所述局部聚类系数是指该节点的邻居之间实际连接数与可能的最大连接数之比，反映了节点周围子图的紧密程度。The divergence refers to the number of edges connected to the node, which represents the degree of connection of the node and is a basic topological feature of the node. The neighbor nodes refer to other nodes directly connected to the node. The average clustering coefficient It is the average of the local clustering coefficients of all nodes in the graph, which measures the degree of interconnection between the neighbors of a node. The local clustering coefficient refers to the actual number of connections between the neighbors of the node and the maximum possible number of connections. The ratio reflects the closeness of the subgraph around the node.
对于电网中的每个节点，计算与当前节点直接相连的节点数量，记为当前节点的发散度，根据发散度，提取与当前节点直接相连的节点信息，将其记为当前节点的邻居节点；For each node in the power grid, calculate the number of nodes directly connected to the current node and record it as the divergence of the current node. Based on the divergence, extract the node information directly connected to the current node and record it as the neighbor node of the current node;
遍历每个节点的邻居节点，确定邻居节点与当前节点间的实际连接数，通过计算邻居节点相互连接的边数与可能的最大连接数的比例确定邻居节点与当前节点间的聚类系数，遍历整个电网，计算每个节点的聚类系数，将这些聚类系数求平均，得到电网的平均聚类系数，使用预设的局部聚类算法，计算每个节点的局部聚类系数，其中，所述局部聚类系数是一个与节点直接相连的邻居节点的连接密度相关的指标，根据局部聚类系数将发散度、邻居节点的特征信息和节点间的路径信息进行融合，得到每个节点的综合拓扑向量。Traverse the neighbor nodes of each node to determine the actual number of connections between the neighbor node and the current node. Determine the clustering coefficient between the neighbor node and the current node by calculating the ratio of the number of edges connecting the neighbor nodes to the maximum possible number of connections. Traverse For the entire power grid, calculate the clustering coefficient of each node, average these clustering coefficients, and obtain the average clustering coefficient of the power grid. Use the preset local clustering algorithm to calculate the local clustering coefficient of each node, where The local clustering coefficient is an index related to the connection density of neighbor nodes directly connected to a node. According to the local clustering coefficient, the divergence, the characteristic information of neighbor nodes and the path information between nodes are integrated to obtain the comprehensive information of each node. Topological vector.
本实施例中，针对每个节点，人机终端展示其直接相连的邻居节点信息，用户可以快速了解节点的周围连接情况，对每个节点的邻居节点进行遍历，计算实际连接数和聚类系数，将计算结果以图形化方式呈现在人机终端上，用户能够看到电网中每个节点的聚类情况，将发散度、邻居节点的特征信息和节点间的路径信息融合，形成综合拓扑向量，并在人机终端上以图形方式展示使用户能够全面了解节点的拓扑特征，综上，本实施例使用户能够方便地监控、分析和理解电网的拓扑特征，实现对网络结构的实时可视化和交互式分析，提高了用户对电力系统拓扑特性的感知和管理效率。In this embodiment, for each node, the human-machine terminal displays the information of its directly connected neighbor nodes. The user can quickly understand the surrounding connections of the node, traverse the neighbor nodes of each node, and calculate the actual number of connections and clustering coefficients. , the calculation results are presented graphically on the human-computer terminal. The user can see the clustering situation of each node in the power grid, and integrate the divergence, characteristic information of neighbor nodes and path information between nodes to form a comprehensive topology vector. , and graphically displayed on the human-machine terminal so that the user can fully understand the topological characteristics of the node. In summary, this embodiment enables the user to easily monitor, analyze and understand the topological characteristics of the power grid, and realize real-time visualization and analysis of the network structure. Interactive analysis improves users' perception and management efficiency of power system topological characteristics.
在一种可选的实施方式中，In an alternative implementation,
所述通过预设的局部聚类算法计算当前节点的局部聚类系数如下公式所示：The local clustering coefficient of the current node is calculated through the preset local clustering algorithm as shown in the following formula:
； ;
其中，Cwi表示节点i的局部聚类系数，△ijk表示节点i，j和k之间的三角关系，wij表示边ij的权重，wik表示边ik的权重，wjk表示边jk的权重，fi表示节点i的发散度。Among them,Cwi represents the local clustering coefficient of node i,△ijk represents the triangular relationship between nodesi , j and k,wij represents the weight of edge ij,wik represents the weight of edge ik,wjk represents the weight of edge jk The weight,fi represents the divergence of node i.
本函数中，通过在人机终端上实时应用上述局部聚类系数计算公式，系统可以动态计算每个节点的局部聚类系数，实现了对电网拓扑动态特性的实时监测，考虑了边的权重使得局部聚类系数的计算更为精准，能够更好地反映节点与邻居节点之间的实际连接强度，通过综合考虑三角关系，可以更全面地评估节点的局部连接紧密性，有助于理解节点在网络中的角色，将发散度引入局部聚类系数的计算中，使得计算结果不仅考虑了连接的紧密性，还考虑了节点在整个网络中的多样性，综上，本函数使用户可以深入了解电网中每个节点的局部连接特性，从而更好地进行网络拓扑分析和故障诊断。In this function, by applying the above local clustering coefficient calculation formula in real time on the human-machine terminal, the system can dynamically calculate the local clustering coefficient of each node, realizing real-time monitoring of the dynamic characteristics of the power grid topology, taking into account the weight of the edges so that The calculation of the local clustering coefficient is more accurate and can better reflect the actual connection strength between the node and its neighbor nodes. By comprehensively considering the triangular relationship, the local connection tightness of the node can be more comprehensively evaluated, which helps to understand the location of the node. Role in the network, the divergence is introduced into the calculation of the local clustering coefficient, so that the calculation results not only consider the tightness of the connection, but also the diversity of the nodes in the entire network. In summary, this function allows users to have an in-depth understanding Local connection characteristics of each node in the power grid to better conduct network topology analysis and fault diagnosis.
S3.基于预设的图神经网络对所述综合拓扑向量进行编码，并对所述行波特征和编码后的综合拓扑向量进行节点特征融合，生成初始节点特征，根据所述初始节点特征，结合预先引入的注意力机制，通过所述图神经网络更新所述初始节点特征，生成输入节点特征，根据所述输入节点特征，结合所述行波特征，通过所述图神经网络中的随机森林模块进行故障定位，确定故障位置。S3. Encode the comprehensive topology vector based on the preset graph neural network, and perform node feature fusion on the traveling wave features and the coded comprehensive topology vector to generate initial node features. Based on the initial node features, combine The attention mechanism introduced in advance updates the initial node features through the graph neural network and generates input node features. Based on the input node features and combined with the traveling wave features, the random forest module in the graph neural network is used. Carry out fault location and determine the fault location.
所述图神经网络是一类用于处理图数据的神经网络，目标是学习图结构中节点之间的复杂关系，从而能够对图中节点的特征进行有效的表示学习，所述初始节点特征是指图中每个节点最初具有的特征，所述注意力机制是一种模型学习不同部分之间关注度的方法，在图神经网络中，注意力机制可以用于调整节点之间信息传递的权重，使得网络更加关注重要的节点，所述输入节点特征是指图神经网络中每一轮迭代时，节点更新所依据的特征，所述随机森林是一种集成学习方法，通常由多个决策树组成。The graph neural network is a type of neural network used to process graph data. The goal is to learn the complex relationships between nodes in the graph structure, so as to be able to effectively represent the characteristics of the nodes in the graph. The initial node characteristics are Refers to the initial characteristics of each node in the graph. The attention mechanism is a method for the model to learn the degree of attention between different parts. In the graph neural network, the attention mechanism can be used to adjust the weight of information transfer between nodes. , making the network pay more attention to important nodes. The input node characteristics refer to the characteristics based on node updates in each iteration of the graph neural network. The random forest is an integrated learning method, usually consisting of multiple decision trees. composition.
在一种可选的实施方式中，In an alternative implementation,
所述基于预设的图神经网络对所述综合拓扑向量进行编码，并对所述行波特征和编码后的综合拓扑向量进行节点特征融合，生成初始节点特征，根据所述初始节点特征，结合预先引入的注意力机制，通过所述图神经网络更新所述初始节点特征，生成输入节点特征包括：The preset-based graph neural network encodes the comprehensive topology vector, and performs node feature fusion on the traveling wave features and the encoded comprehensive topology vector to generate initial node features. According to the initial node features, the combined The attention mechanism introduced in advance updates the initial node features through the graph neural network and generates input node features including:
获取所述综合拓扑向量，根据所述综合拓扑向量和所述行波特征初始化节点特征并构建当前节点的邻接矩阵并确定当前节点的邻居节点，根据所述邻居节点更新当前节点，记为初始节点特征，对于每个初始节点特征，收集所述邻居节点的节点信息，并根据预先引入的多头注意力机制计算每个邻居节点对当前节点的重要性，即注意力系数；Obtain the comprehensive topology vector, initialize the node characteristics according to the comprehensive topology vector and the traveling wave characteristics and construct the adjacency matrix of the current node and determine the neighbor nodes of the current node, update the current node according to the neighbor nodes, and record it as the initial node Features, for each initial node feature, collect the node information of the neighbor nodes, and calculate the importance of each neighbor node to the current node according to the pre-introduced multi-head attention mechanism, that is, the attention coefficient;
将每个邻居节点的节点信息进行合并，结合每个邻居节点对应的注意力系数通过激活函数加权计算邻居节点的综合特征，根据所述综合特征，更新当前节点的特征表示，生成输入节点特征。The node information of each neighbor node is merged, and the comprehensive features of the neighbor nodes are calculated by weighting the activation function in combination with the attention coefficient corresponding to each neighbor node. Based on the comprehensive features, the feature representation of the current node is updated to generate input node features.
从之前的步骤中获取计算得到的综合拓扑向量和行波特征，使用综合拓扑向量和行波特征初始化每个节点的特征表示，构建当前节点的邻接矩阵，确定当前节点的邻居节点，对于每个初始节点特征，收集其邻居节点的节点信息，包括邻居节点的当前特征表示、行波特征，引入预先定义的多头注意力机制，通过计算注意力得分、应用softmax函数确定每个邻居节点对当前节点的重要性，即注意力系数；Obtain the calculated comprehensive topology vector and traveling wave characteristics from the previous steps, use the comprehensive topology vector and traveling wave characteristics to initialize the feature representation of each node, construct the adjacency matrix of the current node, and determine the neighbor nodes of the current node. For each Initial node features collect node information of its neighbor nodes, including the current feature representation and traveling wave features of neighbor nodes, introduce a predefined multi-head attention mechanism, and determine the impact of each neighbor node on the current node by calculating the attention score and applying the softmax function. The importance of , that is, the attention coefficient;
将每个邻居节点的节点信息进行合并，结合每个邻居节点对应的注意力系数，通过激活函数加权计算邻居节点的综合特征，根据综合特征，通过将综合特征与当前节点的原始特征表示进行融合，例如使用残差连接，更新当前节点的特征表示，生成输入节点特征。Merge the node information of each neighbor node, combine it with the attention coefficient corresponding to each neighbor node, and calculate the comprehensive features of the neighbor nodes through activation function weighting. According to the comprehensive features, fuse the comprehensive features with the original feature representation of the current node. , for example, using residual connection to update the feature representation of the current node and generate input node features.
本实施例中，在人机终端上通过图形化展示多头注意力机制的运作，包括计算注意力系数的过程，用户可以通过交互方式了解每个邻居节点对当前节点的重要性，以及注意力系数的变化情况，通过图形界面呈现邻居节点信息的收集和综合特征的计算过程，用户可以直观地了解邻居节点信息是如何被整合到当前节点的特征表示中的，以及每个邻居节点对当前节点的贡献程度，综上，本实施例有助于用户更全面地理解图结构中的信息传递和节点表示学习过程。In this embodiment, the operation of the multi-head attention mechanism is graphically displayed on the human-machine terminal, including the process of calculating the attention coefficient. The user can interactively understand the importance of each neighbor node to the current node and the attention coefficient. The changes of the neighbor node information and the calculation process of comprehensive features are presented through the graphical interface. The user can intuitively understand how the neighbor node information is integrated into the feature representation of the current node, and the impact of each neighbor node on the current node. Degree of Contribution,To sum up, this embodiment helps users to have a more,comprehensive understanding of the information transfer and node,representation learning process in the graph structure.
在一种可选的实施方式中，In an alternative implementation,
所述根据所述输入节点特征，结合所述行波特征，通过所述图神经网络中的随机森林模块进行故障定位，确定故障位置包括：According to the input node characteristics, combined with the traveling wave characteristics, fault location is performed through the random forest module in the graph neural network, and the determination of the fault location includes:
根据所述行波特征初始化所述图神经网络中的随机森林模块，设置所述随机森林模块的超参数，即树的数量和深度，将所述输入节点特征作为输入信息添加至所述随机森林模块中；Initialize the random forest module in the graph neural network according to the traveling wave characteristics, set the hyperparameters of the random forest module, that is, the number and depth of trees, and add the input node characteristics as input information to the random forest in module;
对于所述随机森林模块中的每个树，随机生成训练集，根据所述训练集交叉验证不同树的数量下所述随机森林模块的性能，选择性能最佳时树的数量，对于每个树，增加最大深度，并根据所述训练集设置所述树中每个节点进行拆分的最小样本数，同时设置叶节点的最小样本数，在调整超参数的过程中观察所述随机森林模块的性能，当性能不再提升时，根据对应的超参数调整所述随机森林模块中树的深度和树的数量；For each tree in the random forest module, a training set is randomly generated. According to the training set cross-validation of the performance of the random forest module under different numbers of trees, the number of trees with the best performance is selected. For each tree , increase the maximum depth, and set the minimum number of samples for each node in the tree to be split according to the training set. At the same time, set the minimum number of samples for leaf nodes, and observe the performance of the random forest module during the process of adjusting the hyperparameters. Performance, when the performance no longer improves, adjust the depth of the trees and the number of trees in the random forest module according to the corresponding hyperparameters;
所述随机森林模块将所述输入节点特征发送至模块中的每个树中，所述树生成每个节点对应的预测故障概率，结合节点对应的行波信息判断所述预测故障概率的置信度，若所述置信度大于预设的置信度阈值，则认为当前节点存在故障，获取故障位置并返回。The random forest module sends the input node characteristics to each tree in the module. The tree generates the predicted failure probability corresponding to each node, and determines the confidence of the predicted failure probability based on the traveling wave information corresponding to the node. , if the confidence is greater than the preset confidence threshold, it is considered that the current node has a fault, the fault location is obtained and returned.
所述超参数是指在模型训练之前设置的参数，所述置信度是指模型对于某个预测结果的信心程度或置信水平，通常，置信度是以概率的形式表示，所述置信度阈值是一个在二分类或多分类问题中的阈值，用于决定模型输出的预测结果是否被接受，如果模型输出的置信度大于等于阈值，则将其判定为正类别；反之，则判定为负类别。The hyperparameters refer to parameters set before model training. The confidence refers to the degree of confidence or confidence level of the model for a certain prediction result. Usually, the confidence is expressed in the form of probability, and the confidence threshold is A threshold in a binary or multi-classification problem is used to determine whether the prediction result of the model output is accepted. If the confidence of the model output is greater than or equal to the threshold, it will be judged as a positive category; otherwise, it will be judged as a negative category.
使用行波特征作为输入，初始化图神经网络中的随机森林模块。设置随机森林模块的超参数，包括树的数量和深度，将输入节点特征作为信息添加到随机森林模块中；Initialize the random forest module in the graph neural network using traveling wave features as input. Set the hyperparameters of the random forest module, including the number and depth of trees, and add input node features as information to the random forest module;
对于随机森林模块中的每个树，随机生成训练集，并在不同树的数量下进行交叉验证，以选择性能最佳的树的数量，对于每个树，逐步增加最大深度，并根据训练集设置树中每个节点进行拆分的最小样本数，同时设置叶节点的最小样本数，在调整超参数的过程中观察随机森林模块的性能，当性能不再提升时，根据对应的超参数调整随机森林模块中树的深度和树的数量；For each tree in the random forest module, a training set is randomly generated and cross-validated under different numbers of trees to select the number of trees with the best performance. For each tree, the maximum depth is gradually increased and based on the training set Set the minimum number of samples for each node in the tree to split, and set the minimum number of samples for leaf nodes. Observe the performance of the random forest module during the process of adjusting the hyperparameters. When the performance no longer improves, adjust according to the corresponding hyperparameters. The depth of trees and the number of trees in the random forest module;
随机森林模块将输入节点特征发送至每个树中，每个树生成每个节点对应的预测故障概率，结合节点对应的行波信息，判断预测故障概率的置信度。如果置信度大于预设的置信度阈值，则认为当前节点存在故障，当判断存在故障时，根据随机森林模块中的树的预测结果，获取故障位置，并将结果返回。The random forest module sends the input node features to each tree, and each tree generates the predicted failure probability corresponding to each node, and combines the traveling wave information corresponding to the node to determine the confidence of the predicted failure probability. If the confidence is greater than the preset confidence threshold, the current node is considered to have a fault. When it is judged that there is a fault, the fault location is obtained based on the prediction results of the tree in the random forest module, and the result is returned.
综上，本实施例在界面上实时显示随机森林模块的性能指标，例如准确度、精确度、召回率等，随着超参数的调整，用户可以即时观察到性能的变化，帮助其做出更明智的调参决策，通过图表或曲线展示超参数调优的过程，显示不同树的数量和深度下模型性能的变化趋势，有助于用户更好地理解超参数选择对模型的影响，在界面上实时显示每个节点的预测故障概率以及相应的置信度，通过热图、图表等方式展示，帮助用户直观地了解哪些节点可能存在故障以及置信度的大小，综上，本实施例提高了用户对系统运行状态和故障预测过程的理解，同时也增加了用户对系统的操作和调参的参与感。In summary, this embodiment displays the performance indicators of the random forest module in real time on the interface, such as accuracy, precision, recall, etc. As the hyperparameters are adjusted, the user can immediately observe the changes in performance to help them make better decisions. Wise parameter tuning decisions, showing the process of hyperparameter tuning through charts or curves, showing the changing trend of model performance under different number and depth of trees, helping users better understand the impact of hyperparameter selection on the model, in the interface The predicted failure probability of each node and the corresponding confidence level are displayed in real time on the computer, and are displayed through heat maps, charts, etc., to help users intuitively understand which nodes may have faults and the level of confidence. In summary, this embodiment improves user The understanding of the system's operating status and fault prediction process also increases the user's sense of participation in the system's operation and parameter adjustment.
图2为本发明实施例人机终端画面数据交互系统的结构示意图，如图2所示，所述系统包括：Figure 2 is a schematic structural diagram of a human-machine terminal screen data interaction system according to an embodiment of the present invention. As shown in Figure 2, the system includes:
第一单元，用于安装行波传感器，根据所述行波传感器获取故障电磁波，确定所述故障电磁波对应的时间戳和所述故障电磁波的行波传播速度，根据所述时间戳和所述行波传播速度，通过故障定位算法确定故障区域；The first unit is used to install a traveling wave sensor, obtain a fault electromagnetic wave according to the traveling wave sensor, determine the time stamp corresponding to the fault electromagnetic wave and the traveling wave propagation speed of the fault electromagnetic wave, and determine the time stamp corresponding to the fault electromagnetic wave and the traveling wave propagation speed of the fault electromagnetic wave. The wave propagation speed determines the fault area through the fault location algorithm;
第二单元，用于根据所述故障区域获取所述故障电磁波的行波信号数据，同时获取电网拓扑数据，提取所述行波信号数据中的行波特征和所述电网拓扑数据中的拓扑特征，根据所述拓扑特征确定结构特征，识别电网拓扑图中节点最短路径，根据所述节点最短路径结合预先引入的聚类系数进行特征融合，得到综合拓扑向量；The second unit is used to obtain the traveling wave signal data of the fault electromagnetic wave according to the fault area, obtain the power grid topology data at the same time, and extract the traveling wave characteristics in the traveling wave signal data and the topological features in the power grid topology data. , determine the structural features according to the topological features, identify the shortest paths of nodes in the power grid topology diagram, perform feature fusion based on the shortest paths of nodes combined with the pre-introduced clustering coefficients, and obtain a comprehensive topology vector;
第三单元，用于基于预设的图神经网络对所述综合拓扑向量进行编码，并对所述行波特征和编码后的综合拓扑向量进行节点特征融合，生成初始节点特征，根据所述初始节点特征，结合预先引入的注意力机制，通过所述图神经网络更新所述初始节点特征，生成输入节点特征，根据所述输入节点特征，结合所述行波特征，通过所述图神经网络中的随机森林模块进行故障定位，确定故障位置。The third unit is used to encode the comprehensive topology vector based on the preset graph neural network, and perform node feature fusion on the traveling wave features and the encoded comprehensive topology vector to generate initial node features. According to the initial Node features, combined with the attention mechanism introduced in advance, update the initial node features through the graph neural network, and generate input node features. According to the input node features, combined with the traveling wave features, through the graph neural network The random forest module performs fault location and determines the fault location.
本发明实施例的第三方面，A third aspect of the embodiment of the present invention,
提供一种电子设备，包括：An electronic device is provided, including:
处理器；processor;
用于存储处理器可执行指令的存储器；Memory used to store instructions executable by the processor;
其中，所述处理器被配置为调用所述存储器存储的指令，以执行前述所述的方法。Wherein, the processor is configured to call instructions stored in the memory to execute the aforementioned method.
本发明实施例的第四方面，The fourth aspect of the embodiment of the present invention,
提供一种计算机可读存储介质，其上存储有计算机程序指令，所述计算机程序指令被处理器执行时实现前述所述的方法。A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.
本发明可以是方法、装置、系统和/或计算机程序产品。计算机程序产品可以包括计算机可读存储介质，其上载有用于执行本发明的各个方面的计算机可读程序指令。The invention may be a method, apparatus, system and/or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions thereon for performing various aspects of the invention.
最后应说明的是：以上各实施例仅用以说明本发明的技术方案，而非对其限制；尽管参照前述各实施例对本发明进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分或者全部技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本发明各实施例技术方案的范围。Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention, but not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: The technical solutions described in the foregoing embodiments can still be modified, or some or all of the technical features can be equivalently replaced; and these modifications or substitutions do not deviate from the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention. scope.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311770009.6ACN117454315B (en) | 2023-12-21 | 2023-12-21 | Man-machine terminal picture data interaction method and system |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311770009.6ACN117454315B (en) | 2023-12-21 | 2023-12-21 | Man-machine terminal picture data interaction method and system |
| Publication Number | Publication Date |
|---|---|
| CN117454315Atrue CN117454315A (en) | 2024-01-26 |
| CN117454315B CN117454315B (en) | 2024-05-28 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311770009.6AActiveCN117454315B (en) | 2023-12-21 | 2023-12-21 | Man-machine terminal picture data interaction method and system |
| Country | Link |
|---|---|
| CN (1) | CN117454315B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118473910A (en)* | 2024-07-08 | 2024-08-09 | 鄂尔多斯市泛胜数据技术有限公司 | Power Internet of Things fault detection method and system based on edge-cloud collaboration |
| CN119004278A (en)* | 2024-07-19 | 2024-11-22 | 国网河南省电力公司郑州供电公司 | Power grid fault on-line identification system based on big data |
| CN119510976A (en)* | 2024-11-18 | 2025-02-25 | 云南电网有限责任公司电力科学研究院 | A distribution line fault location method based on power frequency and traveling wave data |
| CN119669663A (en)* | 2024-11-15 | 2025-03-21 | 中国科学院计算技术研究所 | Extrapolation model for multivariate time sequence analysis and training method thereof |
| CN119986261A (en)* | 2025-04-17 | 2025-05-13 | 北京智芯微电子科技有限公司 | Transmission network fault location method, device, equipment, storage medium and chip |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6430150B1 (en)* | 1996-02-14 | 2002-08-06 | Fujitsu Limited | Communication node, restoration method and communication network |
| WO2017196821A1 (en)* | 2016-05-09 | 2017-11-16 | Strong Force Iot Portfolio 2016, Llc | Methods and systems for the industrial internet of things |
| WO2019232595A1 (en)* | 2018-06-07 | 2019-12-12 | Federation University Australia | A method of estimating the location of a fault on an electrical distribution network and an associated system |
| US20210156899A1 (en)* | 2019-11-25 | 2021-05-27 | Shandong University Of Technology | Method for fault location to multi-terminal traveling wave in direct current distribution line |
| CN113835000A (en)* | 2021-09-23 | 2021-12-24 | 南方电网科学研究院有限责任公司 | A distribution network fault location method, device, terminal and storage medium |
| CN113945799A (en)* | 2021-10-15 | 2022-01-18 | 广东电网有限责任公司惠州供电局 | Power line network fault positioning method and device, electronic equipment and storage medium |
| US20220050134A1 (en)* | 2020-08-13 | 2022-02-17 | Siemens Aktiengesellschaft | Method and device for identifying the location of a fault on a line of an electrical power supply network |
| CN114636894A (en)* | 2022-03-09 | 2022-06-17 | 四川思极科技有限公司 | Power distribution network topology change identification method for optimizing traveling wave positioning |
| CN114675134A (en)* | 2022-04-18 | 2022-06-28 | 四川思极科技有限公司 | Power distribution network fault positioning method and system based on traveling wave space-time matrix |
| CN114839473A (en)* | 2022-03-31 | 2022-08-02 | 广西电网有限责任公司电力科学研究院 | Active power distribution network fault positioning method based on model migration diagram self-encoder network |
| CN115730261A (en)* | 2022-11-21 | 2023-03-03 | 山东博鸿电气股份有限公司 | Power distribution network line fault detection method and system based on traveling waves |
| CN115954865A (en)* | 2022-12-09 | 2023-04-11 | 广西电网有限责任公司 | A low-voltage topology analysis method and system for distribution network based on artificial intelligence |
| CN116243110A (en)* | 2023-04-14 | 2023-06-09 | 长沙理工大学 | A primary distribution network fault identification and location method and related equipment |
| CN116822334A (en)* | 2023-05-30 | 2023-09-29 | 贵州黔驰信息股份有限公司 | A visual power grid model fault response method and system |
| CN116973694A (en)* | 2023-09-22 | 2023-10-31 | 国网浙江宁波市鄞州区供电有限公司 | Power distribution network fault diagnosis optimization method and system |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6430150B1 (en)* | 1996-02-14 | 2002-08-06 | Fujitsu Limited | Communication node, restoration method and communication network |
| WO2017196821A1 (en)* | 2016-05-09 | 2017-11-16 | Strong Force Iot Portfolio 2016, Llc | Methods and systems for the industrial internet of things |
| WO2019232595A1 (en)* | 2018-06-07 | 2019-12-12 | Federation University Australia | A method of estimating the location of a fault on an electrical distribution network and an associated system |
| US20210156899A1 (en)* | 2019-11-25 | 2021-05-27 | Shandong University Of Technology | Method for fault location to multi-terminal traveling wave in direct current distribution line |
| US20220050134A1 (en)* | 2020-08-13 | 2022-02-17 | Siemens Aktiengesellschaft | Method and device for identifying the location of a fault on a line of an electrical power supply network |
| CN113835000A (en)* | 2021-09-23 | 2021-12-24 | 南方电网科学研究院有限责任公司 | A distribution network fault location method, device, terminal and storage medium |
| CN113945799A (en)* | 2021-10-15 | 2022-01-18 | 广东电网有限责任公司惠州供电局 | Power line network fault positioning method and device, electronic equipment and storage medium |
| CN114636894A (en)* | 2022-03-09 | 2022-06-17 | 四川思极科技有限公司 | Power distribution network topology change identification method for optimizing traveling wave positioning |
| CN114839473A (en)* | 2022-03-31 | 2022-08-02 | 广西电网有限责任公司电力科学研究院 | Active power distribution network fault positioning method based on model migration diagram self-encoder network |
| CN114675134A (en)* | 2022-04-18 | 2022-06-28 | 四川思极科技有限公司 | Power distribution network fault positioning method and system based on traveling wave space-time matrix |
| CN115730261A (en)* | 2022-11-21 | 2023-03-03 | 山东博鸿电气股份有限公司 | Power distribution network line fault detection method and system based on traveling waves |
| CN115954865A (en)* | 2022-12-09 | 2023-04-11 | 广西电网有限责任公司 | A low-voltage topology analysis method and system for distribution network based on artificial intelligence |
| CN116243110A (en)* | 2023-04-14 | 2023-06-09 | 长沙理工大学 | A primary distribution network fault identification and location method and related equipment |
| CN116822334A (en)* | 2023-05-30 | 2023-09-29 | 贵州黔驰信息股份有限公司 | A visual power grid model fault response method and system |
| CN116973694A (en)* | 2023-09-22 | 2023-10-31 | 国网浙江宁波市鄞州区供电有限公司 | Power distribution network fault diagnosis optimization method and system |
| Title |
|---|
| SHIHAO ZHANG 等: "Application of Transient Traveling Wave Technology in Fault Location and Condition Monitoring of Collecting Line in Wind Farm", 《2023 IEEE 3RD INTERNATIONAL CONFERENCE ON POWER, ELECTRONICS AND COMPUTER APPLICATIONS (ICPECA)》, 29 March 2023 (2023-03-29)* |
| WEIHUA SUN 等: "Fault location method for multi-segment hybrid line based on traveling wave time difference", 《2022 4TH INTERNATIONAL CONFERENCE ON ELECTRICAL ENGINEERING AND CONTROL TECHNOLOGIES》, 7 February 2023 (2023-02-07)* |
| 张翼等: "广域行波信息与图注意力网络相结合的 输电网故障定位", 《仪器仪表学报》, 30 June 2022 (2022-06-30)* |
| 曲广龙;杨洪耕;吴晓清;周辉;: "考虑电网拓扑结构的行波故障定位方法", 电力系统及其自动化学报, no. 06, 15 December 2013 (2013-12-15)* |
| 陈日成;金涛;: "面向复杂信息系统的多源异构数据融合技术", 中国测试, no. 07, 31 July 2020 (2020-07-31)* |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118473910A (en)* | 2024-07-08 | 2024-08-09 | 鄂尔多斯市泛胜数据技术有限公司 | Power Internet of Things fault detection method and system based on edge-cloud collaboration |
| CN119004278A (en)* | 2024-07-19 | 2024-11-22 | 国网河南省电力公司郑州供电公司 | Power grid fault on-line identification system based on big data |
| CN119669663A (en)* | 2024-11-15 | 2025-03-21 | 中国科学院计算技术研究所 | Extrapolation model for multivariate time sequence analysis and training method thereof |
| CN119510976A (en)* | 2024-11-18 | 2025-02-25 | 云南电网有限责任公司电力科学研究院 | A distribution line fault location method based on power frequency and traveling wave data |
| CN119986261A (en)* | 2025-04-17 | 2025-05-13 | 北京智芯微电子科技有限公司 | Transmission network fault location method, device, equipment, storage medium and chip |
| Publication number | Publication date |
|---|---|
| CN117454315B (en) | 2024-05-28 |
| Publication | Publication Date | Title |
|---|---|---|
| CN117454315B (en) | Man-machine terminal picture data interaction method and system | |
| CN116505665B (en) | Fault monitoring method and system for power grid distribution line | |
| CN119881542B (en) | Intelligent positioning and diagnosing method for optical fiber composite faults of power distribution network based on hybrid sensing | |
| CN110232524A (en) | Social networks cheats the construction method of model, antifraud method and apparatus | |
| CN119510969B (en) | Double-end fault positioning method based on signal enhancement | |
| CN110347717A (en) | A kind of big data storage method based on urban electric power equipment monitoring | |
| CN110412368A (en) | On-line monitoring method and system of power equipment based on voiceprint recognition | |
| CN118818215A (en) | Fault diagnosis method and system based on FTU of distribution network | |
| CN118568640A (en) | Intelligent monitoring and diagnosing method and device for high-voltage cable grounding system | |
| CN118885944A (en) | Multimodal fault diagnosis and early warning method | |
| CN117725536A (en) | Cable joint diagnosis and abnormality early warning method and device based on multi-parameter fusion analysis | |
| CN119577556B (en) | Power grid fault online identification system and method based on big data | |
| CN118445703A (en) | Vibration signal characteristic identification method | |
| CN115079052A (en) | A transformer fault diagnosis method and system | |
| CN118761016A (en) | Cable status monitoring method, system and storage medium based on deep reinforcement learning | |
| Oliveira et al. | Traveling wave fault location for AC transmission lines: an approach based on the application of EMD and Teager energy operator | |
| CN120354328B (en) | AI-based multi-source data anomaly perception analysis method for Internet of things | |
| CN120355409B (en) | A multimodal data fusion intelligent inspection method for hydropower plants | |
| CN119335317B (en) | A method for evaluating the severity of power system faults based on optical CT measurements | |
| CN118381683B (en) | Distributed monitoring method and device for industrial control network attack event | |
| CN119719952B (en) | A wiring monitoring method and system based on digital twin | |
| CN120596358B (en) | A method for realizing the logic breakpoint debugging function of industrial configuration control system | |
| CN115423221B (en) | Facility operation trend prediction method | |
| CN120389938A (en) | Communication equipment fault prediction method and system based on big data and optical fiber sensors | |
| CN119814133A (en) | Online monitoring method for multi-core optical cable faults based on optical fiber sensing technology |
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |