Multi-protection linkage control method for power distribution system

Abstract

The invention discloses a multi-protection linkage control method for a power distribution system, and relates to the technical field of power distribution control, and the method comprises the steps: obtaining a power distribution system topological structure of a target power distribution system, and carrying out the monitoring data and sequential collection at the same time; performing interaction association identification to obtain a monitoring data interaction topological structure; performing operation stability identification to obtain an overall stability factor; performing abnormal topological node identification to obtain an abnormal topological node set; performing abnormal path identification to obtain an abnormal path set; and performing multi-protection linkage control scheme identification to obtain a target linkage control scheme. According to the invention, the technical problem of insufficient operation stability and reliability caused by incomplete monitoring and inaccurate analysis of the power distribution system in the prior art is solved, and comprehensive monitoring, accurate analysis and intelligent control of the operation state of the target power distribution system are realized. And the accuracy of fault diagnosis of the power distribution system and the stability and reliability of overall operation are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution control, and in particular to a multiple protection linkage control method for a power distribution system. Background Art

[0002] With the continuous growth of electricity demand and the continuous expansion of the scale of power systems, the safe and stable operation of distribution systems faces many challenges. In traditional distribution system technology, the monitoring of system operation status often has problems such as insufficient coverage of monitoring points and asynchronous data collection, which makes it impossible to obtain comprehensive and real-time system operation information. At the same time, in the analysis of monitoring data, there is a lack of effective means of interactive correlation and stability identification, which makes it difficult to accurately judge the operating status of the system, which greatly reduces the accuracy of fault diagnosis, often resulting in misjudgment or missed judgment, thus affecting the timeliness and effectiveness of fault handling. Moreover, when faced with complex fault conditions, traditional technologies lack intelligent linkage control mechanisms, and each protection device acts independently and cannot work together to achieve rapid isolation of faults and stable recovery of the system, which seriously reduces the stability and reliability of the overall operation of the distribution system and makes it difficult to meet the needs of modern society for high-quality power supply.

[0003] The existing technology has technical problems such as incomplete monitoring and inaccurate analysis of the distribution system, which leads to insufficient operational stability and reliability. Summary of the invention

[0004] The present application provides a multiple protection linkage control method for a power distribution system, which is used to solve the technical problems in the prior art of insufficient operation stability and reliability caused by incomplete monitoring and inaccurate analysis of the power distribution system.

[0005] In view of the above problems, the present application provides a multiple protection linkage control method for a power distribution system, the method comprising:

[0006] Acquire the distribution system topology of the target distribution system, extract the monitoring sensors of each topological node in the distribution system topology to collect monitoring data in parallel, and obtain a monitoring data topological structure sequence; perform interactive correlation identification on the monitoring data topological structure sequence in chronological order from front to back, and obtain a monitoring data interactive topological structure; use a stability identifier to identify the system operation stability of the monitoring data interactive topological structure, and obtain an overall stability factor; when the overall stability factor is greater than or equal to a preset stability factor threshold, traverse the monitoring data topological structure sequence to identify abnormal topological nodes, and obtain an abnormal topological node set; combine the distribution system topology and the abnormal topological node set to identify abnormal paths, and obtain an abnormal path set; traverse the abnormal path set to identify multiple protection linkage control schemes, and obtain a target linkage control scheme.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] The distribution system topology of the target distribution system is obtained, and the monitoring sensors of each topological node in the distribution system topology are extracted to collect monitoring data in parallel and obtain a monitoring data topology sequence; interactive association identification is performed in chronological order from front to back to obtain a monitoring data interactive topology structure; system operation stability is identified for the monitoring data interactive topology structure to obtain an overall stability factor; abnormal topology node identification is performed through the monitoring data topology sequence to obtain an abnormal topology node set; abnormal path identification is performed in combination with the distribution system topology and the abnormal topology node set to obtain an abnormal path set; multiple protection linkage control schemes are identified through the abnormal path set to obtain a target linkage control scheme. The technical effect of realizing comprehensive monitoring, precise analysis and intelligent control of the operating status of the target distribution system, improving the accuracy of distribution system fault diagnosis, and the stability and reliability of the overall operation is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0010] Figure 1 A flowchart of a multiple protection linkage control method for a power distribution system provided in an embodiment of the present application;

[0011] Figure 2 A schematic diagram of a flow chart of obtaining an interactive topology structure of monitoring data in a multiple protection linkage control method for a power distribution system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0012] The present application provides a multiple protection linkage control method for a power distribution system, which is used to solve the technical problems in the prior art of insufficient operation stability and reliability caused by incomplete monitoring and inaccurate analysis of the power distribution system.

[0013] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0014] Examples, such as Figure 1 As shown, the present application provides a multiple protection linkage control method for a power distribution system, the method comprising:

[0015] Step S100: acquiring a distribution system topology structure of a target distribution system, extracting monitoring sensors of each topology node in the distribution system topology structure to collect monitoring data in parallel, and obtaining a monitoring data topology structure sequence.

[0016] Specifically, firstly, a professional topology identification and acquisition system is used to connect with various control and communication interfaces of the target distribution system, and a depth-first search algorithm is used to traverse the distribution network, identify and obtain the complete topological structure information of the distribution system, and store it as a structured graph data model, including node information (such as device number, type, location, etc.) and edge information (connection relationship, cable parameters, etc.). Then, for each topological node in the acquired topological structure, according to its pre-configured monitoring sensor type and interface protocol, multi-threaded concurrent acquisition technology is used to send acquisition instructions to each sensor at the same time, triggering it to perform synchronous data acquisition operations on the corresponding electrical parameters (such as voltage, current, power factor, etc.), equipment status parameters (such as switch status, equipment temperature, insulation resistance, etc.) and environmental parameters (such as temperature, humidity, harmful gas concentration, etc.). During the acquisition process, a high-precision time synchronization protocol is used to ensure that the data collected by all sensors have strict time sequence to avoid data inconsistency caused by time deviation. After the collection is completed, the various parameter data collected by each topological node at the same time are encapsulated and integrated, and a monitoring data topology structure sequence is constructed according to the pre-defined topological structure data format. It is stored in the cache area in the form of an ordered array or linked list for rapid data access and processing in subsequent steps, providing an accurate, real-time and complete data foundation for further system analysis and decision-making, ensuring that the operating status and dynamic change trends of the distribution system can be accurately grasped.

[0017] Step S200: performing interactive association identification on the monitoring data topological structure sequence in chronological order to obtain a monitoring data interactive topological structure.

[0018] Specifically, the constructed monitoring data topology sequence is read from the cache and arranged in ascending order according to the timestamp to ensure its order. For two adjacent monitoring data topologies in the sequence, for the same type of parameters of each topological node, the dynamic time warping (DTW) algorithm based on sliding window is used to measure the time series similarity, and the matching distance and path cost are calculated to determine the degree of association between node parameters in adjacent time segments and construct a preliminary association matrix. At the same time, the linear correlation between different node parameters is analyzed in combination with the Pearson correlation coefficient to supplement and improve the association information. Subsequently, the association results obtained in the previous step are used as input, and the graph neural network (GNN) model is used to extract features and update the graph structure. Through information propagation and aggregation operations between nodes, the association relationship between nodes is further optimized. This process is iterated, and the subsequent monitoring data topology structures are gradually included in the analysis scope. The association matrix is ​​continuously updated and improved, and finally a monitoring data interactive topology structure that can fully reflect the complex interactive relationship of monitoring data in the time dimension is constructed, providing strong support for in-depth insight into the operating dynamics and potential change trends of the distribution system.

[0019] Step S300: using a stability identifier to identify the system operation stability of the monitoring data interaction topology structure to obtain an overall stability factor.

[0020] Specifically, the monitoring data interaction topology structure constructed in the previous steps is input into the stability identifier, which is pre-trained based on a deep neural network model. Its training process uses a large number of distribution system monitoring data interaction topology structure samples covering different operating states, as well as the corresponding overall stability factor values ​​annotated by domain experts based on their rich experience and professional knowledge. After the monitoring data interaction topology structure to be evaluated is input, the stability identifier uses its internal multi-layer neuron structure to perform comprehensive and in-depth feature extraction and analysis of key information such as node data features, node association patterns, and overall topological morphology in the structure. For example, it will focus on the parameter fluctuations of key nodes, the coordinated change trends of closely related node groups, and the energy distribution characteristics of the entire topological structure. Based on the features extracted by these deep analyses, the stability identifier uses the mapping relationship between the complex data patterns and stability factors learned during the training process, and through a series of complex matrix operations and nonlinear transformations, it finally calculates the overall stability factor that can accurately quantify the degree of operational stability of the distribution system represented by the monitoring data interaction topology structure. This factor will serve as the key basis for subsequent judgment on whether the distribution system is in normal operation and whether further measures need to be taken, providing important decision-making support for the safe and reliable operation of the distribution system.

[0021] Step S400: When the overall stability factor is greater than or equal to a preset stability factor threshold, the monitoring data topology structure sequence is traversed to identify abnormal topology nodes to obtain an abnormal topology node set.

[0022] Specifically, the overall stability factor output by the stability identifier is first compared with the preset stability factor threshold. Once it is determined that the overall stability factor is greater than or equal to the preset threshold, the identification process of abnormal topological nodes is immediately started. The complete monitoring data topological structure sequence is retrieved from the storage area, and each monitoring data topological structure in the sequence is analyzed in detail in turn according to the established traversal algorithm. For each node in the topological structure, a statistical anomaly detection method is used, combined with the normal operating parameter range and historical data characteristics of the distribution system, to calculate the statistical indicators of each node data, such as mean, standard deviation, skewness and kurtosis. At the same time, the sliding window technology is used to monitor the change trend of node data in time series in real time, and the dynamic stability of node data is judged by comparing the change rate of data in the current window with the historical normal change rate. When the statistical indicators of a node exceed the normal range or its data change trend shows abnormal fluctuations, the node will be marked as a potential abnormal node. After completing the traversal of the entire monitoring data topological structure sequence, all marked potential abnormal nodes are summarized and screened, and misjudged nodes that may be caused by noise or short-term interference are removed, and finally an accurate and reliable abnormal topological node set is formed. This collection will provide key target objects for subsequent fault diagnosis, risk assessment and formulation of targeted maintenance measures, effectively ensuring the stable operation and safety of the distribution system.

[0023] Step S500: abnormal path identification is performed in combination with the power distribution system topology structure and the abnormal topology node set to obtain an abnormal path set.

[0024] Specifically, the distribution system topology structure obtained in the early stage and the determined abnormal topology node set are used as key input information. Based on the node connection relationship and line direction defined by the distribution system topology structure, starting from each node in the abnormal topology node set, the depth-first search algorithm is used to explore the path. During the search process, the lines connecting the abnormal nodes are gradually expanded, and the node and line information passed through are recorded to construct a series of possible path branches. By analyzing the electrical parameter characteristics of these path branches, such as calculating the total current, voltage drop distribution, and power loss on the path, and comparing them with the standard parameter range under normal operating conditions, once the electrical parameters of a path exceed the preset normal range threshold, it is judged as an abnormal path. Continue such search and judgment operations until all potential paths starting from abnormal topology nodes are traversed, and finally all paths that meet the abnormal judgment conditions are summarized and integrated to form a complete abnormal path set. This abnormal path set can clearly present the areas and line directions of the power distribution system where there may be hidden fault hazards or abnormal operation conditions, providing precise target orientation for subsequent troubleshooting, repair and system optimization, greatly improving the efficiency and accuracy of power distribution system operation and maintenance, and ensuring the stability and reliability of power supply.

[0025] Step S600: traverse the abnormal path set to identify multiple protection linkage control schemes and obtain a target linkage control scheme.

[0026] Specifically, the generated abnormal path set is read from the storage area, and a comprehensive and in-depth multiple protection linkage control scheme identification process is started for each abnormal path in the set. Combined with the key information such as the overall architecture, operating parameters and equipment characteristics of the distribution system, the pre-built protection strategy library is called, which stores a large number of diversified protection measures and their combinations for different types of abnormal situations, such as various feasible linkage schemes for various protection means such as overcurrent protection, undervoltage protection, short-circuit protection and reclosing. For each abnormal path, an intelligent matching algorithm is used to select multiple potential protection linkage control schemes adapted to it from the protection strategy library according to the fault characteristics on the path, such as fault type, fault location, and expected fault impact range. Then, by establishing an accurate power system simulation model, these potential schemes are simulated one by one in a virtual environment. During the simulation process, the real-time status of the system, load changes, and various complex working conditions that may occur are considered, and the protection effect of each scheme is quantitatively evaluated. The evaluation indicators include fault isolation speed, power outage range, system recovery time, and the degree of impact on other normal areas. After simulating and evaluating the potential solutions for all abnormal paths, taking into account the advantages and disadvantages of each solution and the overall coordination, the optimization algorithm was used to integrate and optimize the solutions, and finally a set of target linkage control solutions that can achieve fast, accurate and efficient protection under different abnormal conditions was determined. This solution will serve as the core strategy to guide the actual distribution system to deal with various faults in operation, ensuring that when abnormal conditions occur, measures can be taken quickly and effectively to minimize the impact of faults on the distribution system and ensure the safe, reliable and stable operation of power supply.

[0027] In one possible implementation, Figure 2 As shown, step S200 also includes:

[0028] Step S210: interactively associate and identify the first monitoring data topology structure and the second monitoring data topology structure in the monitoring data topology structure sequence in chronological order to obtain a first interactively associated topology structure.

[0029] Step S220: performing interactive association identification on the first interactive association topology structure and the third monitoring data topology structure in the monitoring data topology structure sequence to obtain a second interactive association topology structure.

[0030] Step S230: performing interactive association identification based on the second interactive association topology structure in a time-delayed order until reaching the last monitoring data topology structure, thereby obtaining the monitoring data interactive topology structure.

[0031] Specifically, a sequence of monitoring data topological structures sorted by time is retrieved from the storage medium, and the first and second monitoring data topological structures are selected. For each topological node data, the dynamic time warping (DTW) algorithm is used to construct an n×m cost matrix (where n and m are the number of nodes in the two topological structures, respectively) to calculate the optimal matching path of the node data in the time series, thereby measuring the similarity between nodes; at the same time, the principal component analysis (PCA) algorithm is used to reduce the dimension of the high-dimensional node data and extract the main characteristic components to reduce the amount of calculation and highlight the key information. Based on the similarity and characteristic information between nodes obtained by the above algorithm, an initial association matrix is ​​constructed. Then, the message passing mechanism in the graph neural network (GNN) is used to pass the characteristic information of the node along the edge to the adjacent node, and aggregate and update it to further optimize the association relationship between the nodes, and finally generate the first interactive correlation topological structure that can accurately reflect the internal connection of the monitoring data at two adjacent time points.

[0032] The first interactive correlation topology is extracted, and the third monitoring data topology is called out from the monitoring data topology sequence. For both, the cosine similarity algorithm based on node feature vectors is first used to calculate the similarity between each node in the first interactive correlation topology and the corresponding node in the third monitoring data topology, so as to determine the preliminary correlation between the nodes. Then, the graph convolutional neural network (GCN) algorithm is used, and the feature information and preliminary correlation of these nodes are used as input. Through multi-layer convolution operations, the nodes can aggregate the information of their neighbor nodes and update their own feature representations, so as to more accurately capture the complex correlation between the nodes. In this process, the connection weights between nodes are dynamically adjusted through the attention mechanism according to the change trend and correlation strength of the node data, so as to strengthen important connections and weaken irrelevant connections. After such a series of operations, the updated node relationships and feature information are integrated to construct the second interactive correlation topology, which integrates the correlation features of the monitoring data at three time points, further enriches and refines the internal connection between the data, and provides a more comprehensive and accurate basis for subsequent system analysis and decision-making, which strongly supports the in-depth insight and accurate grasp of the operating status of the distribution system.

[0033] Taking the obtained second interactive correlation topology as the starting basis, strictly following the time sequence from front to back, the subsequent monitoring data topology is gradually introduced to carry out the interactive correlation identification process. For each newly introduced monitoring data topology, a time series association model based on deep learning is used. This model can simultaneously consider the long-term trend, seasonal changes and short-term fluctuation characteristics of node data, and effectively capture the time dependency in sequence data by constructing a multi-layer recurrent neural network (RNN) structure, such as a long short-term memory network (LSTM). During the operation of the model, the hidden state of the interactive correlation topology at the previous moment is used as input, and the node data in the newly introduced monitoring data topology is fused and calculated to update the association weights and feature representations between nodes. At the same time, the attention mechanism is used to dynamically focus on the nodes and connections that have a key indicative role in the changes in the system operation state, highlight important information, and suppress noise interference. As new monitoring data topological structures are continuously included in the analysis, the association relationships are continuously iterated and optimized, and the information richness and accuracy of the entire topological structure are gradually improved until the last monitoring data topological structure in the sequence is processed. Finally, a monitoring data interaction topological structure is generated that comprehensively and accurately reflects the data interaction relationship during the entire monitoring time period. This provides solid and powerful basic data support for subsequent in-depth analysis of the distribution system's operating stability, potential risks, and fault prediction, and greatly enhances the comprehensive control capabilities of the distribution system's operating status.

[0034] In a possible implementation, step S210 further includes:

[0035] Step S211: performing same-node similarity calculation on the first monitoring data topology structure and the second monitoring data topology structure according to a cosine similarity calculation formula to obtain a first interactive similarity topology structure.

[0036] Step S212: performing graph convolution calculation using the first interaction similarity topological structure and the second monitoring data topological structure to obtain the first interaction association topological structure.

[0037] Specifically, the first and second monitoring data topological structures are accurately extracted from the monitoring data topological structure sequence. For each corresponding topological node in the two structures, the various types of monitoring data contained therein, such as voltage amplitude, current intensity, power factor and other parameter information, are deeply analyzed, and these parameters are integrated into feature vectors. Subsequently, the feature vector of a node in the first monitoring data topological structure is similar to the feature vector of the corresponding node in the second monitoring data topological structure in strict accordance with the cosine similarity calculation formula. By calculating the cosine value of the angle between the two vectors in the vector space, an accurate similarity value is obtained, which can accurately quantify the similarity between the two nodes in different monitoring data dimensions. The similarity values ​​of all corresponding nodes are integrated and structured to construct a new first interactive similarity topological structure, in which the connection weight or relationship strength between nodes is represented by the corresponding similarity value, thereby clearly presenting the similarity distribution pattern of the two monitoring data topological structures at the same node level, providing a key basic data framework for the subsequent in-depth mining of the correlation relationship and potential laws between data, and effectively promoting the in-depth analysis and accurate interpretation of the distribution system operation data.

[0038] According to the similarity weights between nodes in the first interactive similarity topology, the neighborhood range and influence weight distribution of each node in the graph convolution calculation are determined. For each node, its own feature vector and the feature vector of the neighboring node are aggregated according to a specific weight distribution rule. In this process, the node similarity information provided by the first interactive similarity topology is fully considered, so that nodes with high correlation contribute greater weights when the features are aggregated. Then, the predefined graph convolution kernel function is used to perform a convolution operation on the aggregated node features, and more advanced and complex association features between nodes are extracted by performing information propagation and feature transformation in the local neighborhood of the node. In the process of multiple iterations of graph convolution calculation, the feature representation of the node and the connection weights between the nodes are continuously updated, and the node connections that are consistent and relevant in the trend of monitoring data changes are gradually strengthened, and accidental similar or irrelevant connections are weakened or eliminated. Finally, after a series of graph convolution calculation steps, the updated node features and connection relationships are integrated to obtain the first interactive correlation topological structure, which effectively integrates the key information in the two monitoring data topological structures, clearly shows the deep interactive correlation between nodes based on data similarity, and provides a more accurate and rich information basis for further analysis of the evolution of the operating status of the distribution system.

[0039] In a possible implementation, step S300 further includes:

[0040] Step S310: Acquire multiple sample monitoring data interaction topological structures, use expert investigation method to identify the overall stability factors of the multiple sample monitoring data interaction topological structures respectively, and obtain multiple sample overall stability factors.

[0041] Step S320: Use the interaction topological structure of the multiple sample monitoring data and the overall stability factors of the multiple samples as training data, perform supervised training on the framework constructed based on the feedforward neural network, learn the one-to-one mapping relationship between the interaction topological structure of the monitoring data and the overall stability factor, until the training converges, and obtain the trained stability identifier.

[0042] Specifically, firstly, through extensive and in-depth data collection of distribution systems in different regions, types and operating periods, a rich and diverse sample monitoring data interaction topology is constructed using the established monitoring data processing process. Then, the stability of the sample monitoring data interaction topology is evaluated based on the expertise and experience of experts. Experts will first check the monitoring data of each node in the topology and the interaction between nodes. For the nodes themselves, key electrical parameters (such as voltage and current) will be checked to see if they are within the normal operating range, and the fluctuation range of these parameters will be evaluated to see if it is within the acceptable range. For example, if the voltage of a node frequently exceeds the upper and lower limits of the normal voltage range, or the current changes dramatically and irregularly, this may mean that the stability of the system near this node is poor. In terms of the interaction between nodes, experts will consider whether the energy transmission is smooth and efficient. For example, analyze whether the power transmission between adjacent nodes is as expected, and whether the association between nodes can maintain sufficient flexibility in the face of external interference or internal faults, so as not to cause the rapid spread of faults. At the same time, experts will also consider the redundancy and fault tolerance of the entire topology. A distribution system topology with good stability should have enough backup paths or redundant equipment so that when some nodes or links have problems, the system can still maintain basic power supply functions and resume normal operation as soon as possible. Experts will combine these factors, compare and weigh the importance of each factor, and combine their previous experience in dealing with similar system problems, and finally give an overall stability factor evaluation value for each sample monitoring data interaction topology. This evaluation value may be a numerical range (for example, 0-100, 0 means extremely unstable and 100 means very stable). These evaluation values ​​constitute the overall stability factors of multiple samples required for subsequent training of stability identifiers.

[0043] The obtained interactive topological structures of multiple sample monitoring data are used as input data, and the corresponding multiple sample overall stability factors are used as output labels to form a complete training data set for supervised learning training based on the framework constructed by the feedforward neural network. The feedforward neural network framework consists of an input layer, several hidden layers and an output layer. The number of neurons in the input layer is set according to the characteristic dimensions of the interactive topological structure of the sample monitoring data, so as to receive and process various node data and connection relationship information in the topological structure. The hidden layer uses different activation functions (such as ReLU function) to increase the nonlinear expression ability of the network. By extracting and transforming the input data layer by layer, the deep-level characteristic patterns and laws in the interactive topological structure of the monitoring data are mined. During the training process, the interactive topological structure of the sample monitoring data is input into the network, and the predicted overall stability factor is obtained through forward propagation calculation. Then, according to the difference between the predicted value and the true sample overall stability factor, the error is calculated using a suitable loss function (such as the mean square error loss function), and the error is back-propagated from the output layer to the input layer through the back-propagation algorithm, and the connection weights between the neurons in each layer of the network are updated, so that the predicted output of the network gradually approaches the true stability factor value. As the training progresses, the network continuously adjusts its own parameters and gradually learns the one-to-one mapping relationship between the complex feature combination in the monitoring data interactive topology and the overall stability factor. When the value of the loss function converges below the preset threshold, or reaches the preset training round, it indicates that the network has been trained. At this time, the obtained stability identifier can accurately and quickly identify the system operation stability of the newly input monitoring data interactive topology structure, and output a reliable overall stability factor, which provides an efficient and intelligent technical means for the real-time monitoring and stability evaluation of the distribution system, and effectively ensures the safe and stable operation of the distribution system.

[0044] In a possible implementation, step S400 further includes:

[0045] Step S410: taking each topological node in the power distribution system topological structure as an index, extracting data from the monitoring data topological structure sequence to obtain a topological node monitoring data sequence set.

[0046] Step S420: traverse the topological node monitoring data sequence set to perform data center sampling, and determine a topological node monitoring data center neighborhood set.

[0047] Step S430: traverse the topological node monitoring data center neighborhood set to perform data fluctuation variance identification and data median identification, and obtain a topological node monitoring data fluctuation variance set and a topological node monitoring data median set.

[0048] Step S440: Acquire a topological node standard data set, perform difference identification on a median set of topological node monitoring data, obtain a topological node monitoring data difference set, perform weighted calculation in combination with the topological node monitoring data fluctuation variance set, and obtain a topological node abnormality factor set.

[0049] Step S450: adding the topological nodes corresponding to the topological node abnormality factors greater than or equal to the preset abnormality factor threshold in the topological node abnormality factor set into the abnormal topological node set.

[0050] Specifically, accurate topological structure information of the power distribution system is obtained, and each topological node is identified. Subsequently, these topological nodes are used as key indexes for retrieval one by one, and the topological structure sequence of the monitoring data is deeply searched. For each topological node, all the corresponding monitoring data are carefully screened out through a rigorously designed data extraction program. These monitoring data contain various electrical parameters and status information of the node at different times, which fully reflect the dynamic change process of its operation. The monitoring data extracted from each topological node are sorted into independent sequence forms, and finally a topological node monitoring data sequence set is successfully constructed. The formation of this set provides a solid and detailed basic data support for further analysis of the operating characteristics of each topological node and discovery of potential abnormalities. It is a crucial starting link in the entire abnormal topological node analysis process, and effectively ensures that the subsequent steps can be processed in depth and effectively based on accurate data.

[0051] Each data sequence in the topological node monitoring data sequence set is visited one by one. These sequences record the operating data information of each topological node at different time points. For each sequence, the mean of all the data in it is calculated, and this mean is used as the data center. Then, with the mean as the core, according to the preset neighborhood range determination rules, with the mean as the center, a certain proportion of the standard deviation range is extended to both sides, or a fixed data interval length is set to select data points within this central neighborhood range. These selected data points together constitute the neighborhood set of the topological node monitoring data center. Through this mean-based central sampling method, abnormal fluctuations and noise interference in the data can be filtered out to a certain extent, focusing on the main distribution area of ​​the data, making the subsequent analysis of the operating status of the topological node more accurate, and providing a more representative and reliable data basis for further identifying potential abnormal situations, thereby improving the efficiency and accuracy of abnormal detection of the entire distribution system.

[0052] A detailed traversal operation is carried out on the neighborhood set of the topological node monitoring data center. For each topological node monitoring data center neighborhood set, the variance calculation formula in statistics is first used to identify the fluctuation variance of the data in the set. By calculating the average of the sum of squares of the difference between each data and the neighborhood mean, the discrete degree of the data in the central neighborhood is accurately quantified to reflect the fluctuation of the data, and then the fluctuation variance set of the topological node monitoring data is obtained. At the same time, the average value of all data in the central neighborhood is calculated and determined as the median value of the data set (neighborhood mean). Through this operation on each topological node, the median value set of the topological node monitoring data set is formed. These sets of variances and central values ​​can characterize the characteristics of the data from different angles. The fluctuation variance shows the stability of the data and the severity of the change, while the central value represents the typical value level of the data, which provides a key quantitative basis for the subsequent in-depth analysis of the operating status of the topological node and the accurate judgment of whether it is abnormal, which helps to more accurately grasp the working conditions of each topological node in the distribution system and improve the ability to discover and diagnose potential problems.

[0053] It is necessary to obtain a set of standard data of key topological nodes. This set is obtained by in-depth analysis, statistics and collation of a large amount of data accumulated by the distribution system under long-term stable and normal operation. It covers the typical data characteristics of each topological node under normal working conditions, such as the standard voltage range, current intensity range, power factor and other standard value ranges of various electrical parameters. They constitute an important benchmark for judging whether the operation status of the topological node is abnormal. Then, the difference recognition work is carried out for the median value set of the topological node monitoring data. Specifically, for each median value of the topological node monitoring data in the set, an algorithm based on difference calculation and normalization processing is adopted. The difference between the median value of the monitoring data of each topological node and the standard value in the corresponding standard data set is calculated element by element. For example, for the voltage parameter, the standard voltage value is subtracted from the monitored voltage concentration value, and the difference of all parameters is squared and summed, and a dimensionless difference value is obtained by dividing by a normalization factor (the factor is determined according to the overall scale of the data and the range of parameter changes). In this way, by performing the same operation on all topological nodes, a set of topological node monitoring data differences is obtained. Each element in the set intuitively reflects the degree to which the median value of the corresponding topological node monitoring data deviates from the standard value. Subsequently, a weighted calculation is performed in combination with the fluctuation variance set of the topological node monitoring data. For each variance value in the fluctuation variance set, the degree of association with the abnormal situation is considered, and combined with the importance of the difference in reflecting the abnormality, appropriate weighting coefficients are determined for the two. These weighting coefficients are determined based on the analysis results of historical fault data, the experience judgment of experts, and the statistical data of a large number of simulation experiments to ensure that the roles of the two in judging the degree of abnormality of the topological node can be accurately balanced. Then, for each topological node, its corresponding difference value is multiplied by the weighting coefficient of the difference, and the fluctuation variance value is multiplied by the weighting coefficient of the fluctuation variance. Through such a weighted summation operation, an abnormal factor that can comprehensively reflect the degree of abnormality of the topological node is obtained. After performing the above weighted calculation on all topological nodes, we finally successfully obtained a set of topological node abnormality factors. Each of these abnormal factors accurately quantifies the degree of abnormality of the corresponding topological node, providing a key and reliable basis for the subsequent accurate identification of abnormal topological nodes, and strongly supporting the stable operation and fault diagnosis of the distribution system.

[0054] The obtained set of abnormal factors of topological nodes is clarified, in which each element represents the quantitative value of the abnormal degree of the corresponding topological node. At the same time, based on the deep mining of the historical operation data of the distribution system, the analysis of a large number of fault cases and the experience judgment of professional and technical personnel, an abnormal factor threshold is pre-set, which serves as the key boundary for judging whether the topological node is abnormal. Then, each abnormal factor in the set of abnormal factors of the topological node is checked and compared one by one. When it is found that the abnormal factor of a topological node is greater than or equal to the preset abnormal factor threshold, the topological node corresponding to the abnormal factor is accurately screened out and added to the set of abnormal topological nodes. The formation of this abnormal topological node set enables operation and maintenance personnel to quickly locate those topological nodes that may have problems, so as to carry out further detection, diagnosis and repair work in a targeted manner, greatly improving the efficiency and accuracy of the operation and maintenance of the distribution system, effectively ensuring the safe and stable operation of the distribution system, reducing the risk of power outages and equipment damage caused by potential faults, and providing solid technical support for the reliability of power supply.

[0055] In a possible implementation, step S420 further includes:

[0056] Step S421: traverse the topology node monitoring data sequence set to perform mean calculation to obtain a topology node monitoring data center set.

[0057] Step S422: Taking the topological node monitoring data center set as the starting point, constructing a topological node initial neighborhood set according to a preset neighborhood radius and combining the monitoring data in the topological node monitoring data sequence set.

[0058] Step S423: performing neighborhood edge diffusion on the initial neighborhood set of the topological node according to a preset neighborhood radius to obtain a first diffused neighborhood set of the topological node.

[0059] Step S424: Determine whether the difference in the amount of neighborhood data between the first diffuse neighborhood set of the topological node and the initial neighborhood set of the topological node is less than or equal to a preset difference. If so, stop center sampling and use the first diffuse neighborhood set of the topological node as the neighborhood set of the topological node monitoring data center.

[0060] Specifically, a comprehensive traversal is carried out on the set of topological node monitoring data sequences. For each topological node monitoring data sequence in the set, its mean is calculated, all monitoring data in the sequence are added one by one, and then divided by the total number of data. The result is the mean of the topological node monitoring data sequence. By performing the same operation on each such sequence, the various means obtained are summarized, and finally a topological node monitoring data center set is successfully constructed. Each element in this set, that is, the mean of the monitoring data of each topological node, represents the central position of its corresponding monitoring data in the overall distribution, which lays a solid foundation for the subsequent determination of the neighborhood range around these central positions, in-depth mining of data features, and accurate identification of potential abnormalities.

[0061] For each element in the topological node monitoring data center set, that is, the mean value of the monitoring data corresponding to each topological node, it is regarded as the starting reference point. A fixed neighborhood radius value has been set scientifically and reasonably in advance based on factors such as the characteristics of the power distribution system, the distribution law of historical data, and actual analysis needs. Subsequently, the topological node monitoring data sequence set is traversed, and for each topological node monitoring data sequence, a distance measurement algorithm (such as the Euclidean distance algorithm, which can be simplified to the absolute value calculation of the numerical difference for one-dimensional data) is used to calculate the distance between each monitoring data in the sequence and the mean value of the monitoring data of the topological node one by one. Those monitoring data whose calculated distance is less than or equal to the preset neighborhood radius are screened out, and these qualified data are summarized to construct an initial neighborhood set for the corresponding topological node.

[0062] The initial neighborhood sets of the constructed topological nodes are clearly defined. These sets define an initial data range for each topological node. Then, according to the pre-set fixed neighborhood radius value (the value is determined based on in-depth research and repeated experiments on the data characteristics, noise level and potential abnormal patterns of the distribution system), the neighborhood edge diffusion operation is performed on the initial neighborhood set of the topological node. For the initial neighborhood set of each topological node, it is expanded outward along the boundary of the data dimension. By applying the distance measurement algorithm again, the monitoring data whose distance to the boundary data points of the initial neighborhood set is less than or equal to the preset neighborhood radius are included. These newly included data points and the data in the original initial neighborhood set together constitute the first diffusion neighborhood set of the topological node. This process aims to further expand the coverage of the data, so as to more comprehensively capture the distribution characteristics and change trends of the monitoring data of the topological nodes, reduce the possibility of missing important information due to the limited initial neighborhood range, and provide a richer and more representative data basis for the subsequent more accurate analysis of the operating status of the topological nodes and the identification of potential abnormal situations, so as to ensure that the accuracy and reliability of the anomaly detection process of the entire distribution system are effectively improved.

[0063] The amount of data in the first diffusion neighborhood set of the topological node and the initial neighborhood set of the topological node are counted respectively, and the difference between the two is calculated. The difference is calculated by counting the elements of the two sets, and then subtracting the amount of data in the initial neighborhood set from the amount of data in the first diffusion neighborhood set. Then, this difference is compared with a preset difference that is set in advance. The preset difference is an empirical threshold determined based on an in-depth understanding of the data characteristics of the distribution system and a large number of experimental simulations. It aims to balance the relationship between the richness of the data and the stability of the data, avoid excessive collection of data that may introduce too much noise, and ensure that there is enough data to accurately reflect the actual operating status of the topological node. If it is found by comparison that the difference in the amount of neighborhood data between the first diffusion neighborhood set of a topological node and the initial neighborhood set is less than or equal to the preset difference, it means that at the current degree of neighborhood expansion, the increase in the amount of data has stabilized, and continuing to expand the neighborhood may not bring more valuable information, but may increase the complexity of data processing and introduce unnecessary interference. At this time, the center sampling process of the topological node will be stopped, and the first diffusion neighborhood set of the topological node will be formally determined as the final neighborhood set of the topological node monitoring data center. This selected central neighborhood set will serve as the basic data source for subsequent in-depth analysis of the operating status of topological nodes, identification of data fluctuation variance, identification of data median values, and calculation of abnormal factors. It will provide reliable data support for the precise positioning of abnormal topological nodes in the distribution system, thereby ensuring that the entire distribution system can operate stably and efficiently.

[0064] In a possible implementation, step S425 further includes:

[0065] Step S4251: If not, the first diffusion neighborhood set of the topological node is subjected to neighborhood edge diffusion according to a preset neighborhood radius until the difference in the amount of neighborhood data between two adjacent diffusions is less than or equal to a preset difference, and the topological node diffusion neighborhood set obtained by the last diffusion is used as the topological node monitoring data center neighborhood set.

[0066] Specifically, when it is judged that the difference in the amount of neighborhood data between the first diffusion neighborhood set of the topological node and the initial neighborhood set of the topological node is greater than the preset difference, it indicates that the current data range has not yet reached the ideal stable state, and the sampling range needs to be further expanded to obtain more representative data. Therefore, the neighborhood edge diffusion operation is performed again on the first diffusion neighborhood set of the topological node according to the preset fixed neighborhood radius value to form a new diffusion neighborhood set. Then, the data volume difference between the new diffusion neighborhood set and the previous diffusion neighborhood set (i.e., the first diffusion neighborhood set or the last newly generated diffusion neighborhood set) is repeatedly calculated and compared with the preset difference. This process will continue to iterate, and each diffusion is intended to include more data that may reflect the actual operating status of the topological node. At the same time, the boundary of data collection is controlled by comparing with the preset difference to avoid unlimited expansion of the data range and the introduction of too much noise or irrelevant data. Until a certain iteration, the difference in the amount of data in the neighborhood set obtained by two adjacent diffusions is less than or equal to the preset difference, which means that the growth of the data volume has slowed down. At this time, the diffuse neighborhood set obtained has reached a relatively ideal balance between data richness and data quality. The diffuse neighborhood set of the topological node obtained by the last diffusion is determined as the final topological node monitoring data center neighborhood set, which serves as the basic data source for subsequent data feature analysis and anomaly judgment, providing solid data support for the accurate identification of abnormal topological nodes in the distribution system, and ensuring the stable operation of the entire distribution system and the accuracy of fault diagnosis.

[0067] In a possible implementation, step S600 further includes:

[0068] Step S610: traverse the abnormal path set to identify the protection control scheme and obtain a protection control scheme identification result set.

[0069] Step S620: Combined with the power distribution system topology, linkage control is performed on the protection control scheme identification result set to obtain the target linkage control scheme.

[0070] Specifically, the abnormal path set is traversed, and for each abnormal path in the set, its fault characteristics are first analyzed comprehensively and carefully, including the location of the fault, the type of fault (such as short circuit fault, open circuit fault, etc.), the changes in electrical parameters around the fault point, and the power equipment information involved in the abnormal path. Then, based on the pre-built large and complete protection control scheme database, which covers a variety of protection control strategies designed for various fault scenarios and different distribution system structures, the protection control scheme that best matches the current abnormal path characteristics is accurately selected from a large number of schemes through an intelligent matching algorithm. These schemes involve the setting of fast tripping action of specific circuit breakers to quickly cut off the fault current and prevent the further expansion of the fault range; they may also include the sensitivity adjustment of related relays to ensure that they can detect fault signals in a timely and accurate manner and trigger corresponding protection actions; they may also include automatic switching instructions for backup power supplies to ensure continuous power supply to critical loads when a fault occurs, as well as optimized control strategies for reactive power compensation devices to maintain system voltage stability. By performing such operations on each abnormal path, the various protection and control schemes obtained are summarized, and finally a set of protection and control scheme identification results is formed, which provides a solid foundation and specific operation guidance for subsequent linkage control, ensuring that the distribution system can implement effective protection and control measures in an orderly manner when facing abnormal situations, reduce the impact of faults on power supply, and maintain the safe and stable operation of the system.

[0071] Fully retrieve and deeply study the topological structure information of the distribution system, which covers key elements such as the connection relationship between each node, the direction and parameters of the line, the distribution of electrical equipment, and the flow of electricity. Then, using these topological structure information as the basic framework, each scheme in the protection and control scheme identification result set is finely integrated and coordinated. Taking into account the position of different abnormal paths in the topological structure and the electrical connection between each other, the execution order and action coordination method of each protection and control scheme are accurately arranged. For example, when a fault in a certain area triggers multiple abnormal paths, the impact range and priority between each path are judged according to the topological structure, and it is reasonably determined which protection actions are executed first to isolate the key fault point, and then the equipment on other related paths is controlled through linkage to transfer load or adjust voltage, so as to ensure that when dealing with abnormal situations, the operating state of the entire distribution system can be smoothly transitioned to avoid new problems or unnecessary expansion of power outages caused by the independent execution of a single scheme. Through such comprehensive and precise linkage control operations, various protection and control schemes are organically combined together, ultimately forming a target linkage control scheme that can comprehensively respond to a variety of abnormal situations and ensure the overall stability and reliability of the distribution system, so that the power supply can be restored to normal in the shortest time or maintained in an acceptable safe operating state.

[0072] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0073] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0074] This specification and the drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A multi-protection linkage control method for a power distribution system, characterized in that: The method comprises: Acquire a distribution system topology structure of a target distribution system, extract monitoring sensors of each topology node in the distribution system topology structure to collect monitoring data in parallel, and obtain a monitoring data topology structure sequence; Performing interactive association identification on the monitoring data topological structure sequence in a time-ordered order to obtain the monitoring data interactive topological structure; Using a stability identifier to identify the system operation stability of the monitoring data interaction topology structure to obtain an overall stability factor; When the overall stability factor is greater than or equal to a preset stability factor threshold, traversing the monitoring data topology structure sequence to identify abnormal topology nodes and obtain an abnormal topology node set; Combining the power distribution system topology structure and the abnormal topology node set to identify abnormal paths, and obtain an abnormal path set; The abnormal path set is traversed to identify multiple protection linkage control schemes to obtain a target linkage control scheme.

2. The multiple protection linkage control method for a power distribution system according to claim 1, characterized in that: Performing interactive association identification on the monitoring data topology structure sequence in a time-ordered order to obtain the monitoring data interactive topology structure, including: Interactively associate and identify the first monitoring data topology structure and the second monitoring data topology structure in the monitoring data topology structure sequence in a time-ordered order to obtain a first interactively associated topology structure; Interactively associate and identify the first interactively associated topological structure with a third monitoring data topological structure in the monitoring data topological structure sequence to obtain a second interactively associated topological structure; In order from front to back in time, interactive association identification is performed based on the second interactive association topology structure until the last monitoring data topology structure is reached, thereby obtaining the monitoring data interactive topology structure.

3. The multiple protection linkage control method for a power distribution system according to claim 2, characterized in that: The first monitoring data topology structure and the second monitoring data topology structure in the monitoring data topology structure sequence are interactively associated and identified in a time-ordered manner to obtain a first interactively associated topology structure, including: Performing same-node similarity calculation on the first monitoring data topology structure and the second monitoring data topology structure according to a cosine similarity calculation formula to obtain a first interactive similarity topology structure; The first interaction similarity topological structure and the second monitoring data topological structure are used to perform graph convolution calculation to obtain the first interaction association topological structure.

4. The multiple protection linkage control method for a power distribution system according to claim 1, characterized in that: The stability identifier is used to identify the system operation stability of the monitoring data interaction topology structure to obtain the overall stability factor, including: Acquire multiple sample monitoring data interactive topological structures, use expert survey method to identify the overall stability factors of the multiple sample monitoring data interactive topological structures respectively, and obtain the overall stability factors of the multiple samples; The interactive topological structures of the multiple sample monitoring data and the overall stability factors of the multiple samples are used as training data, and supervised training is performed on the framework constructed based on the feedforward neural network to learn the one-to-one mapping relationship between the interactive topological structures of the monitoring data and the overall stability factors until the training converges, thereby obtaining the trained stability identifier.

5. The multiple protection linkage control method for a power distribution system according to claim 1, characterized in that: When the overall stability factor is greater than or equal to a preset stability factor threshold, traversing the monitoring data topology structure sequence to identify abnormal topology nodes and obtaining an abnormal topology node set includes: Taking each topological node in the power distribution system topology structure as an index, extracting data from the monitoring data topological structure sequence to obtain a topological node monitoring data sequence set; Traversing the topological node monitoring data sequence set to perform data center sampling, and determining a topological node monitoring data center neighborhood set; Traversing the neighborhood set of the topological node monitoring data center to perform data fluctuation variance identification and data median identification, and obtaining a topological node monitoring data fluctuation variance set and a topological node monitoring data median set; Acquire a topological node standard data set, perform difference identification on a median set of topological node monitoring data, obtain a topological node monitoring data difference set, perform weighted calculation on the topological node monitoring data fluctuation variance set, and obtain a topological node abnormality factor set; The topological nodes corresponding to the topological node abnormality factors greater than or equal to the preset abnormality factor threshold in the topological node abnormality factor set are added into the abnormal topological node set.

6. The multiple protection linkage control method for a power distribution system according to claim 5, characterized in that: Traversing the topological node monitoring data sequence set to perform data center sampling and determine a topological node monitoring data center neighborhood set, including: Traversing the topological node monitoring data sequence set to perform mean calculation to obtain a topological node monitoring data center set; Taking the topological node monitoring data center set as the starting point respectively, constructing the topological node initial neighborhood set according to the preset neighborhood radius and the monitoring data in the topological node monitoring data sequence set; Performing neighborhood edge diffusion on the initial neighborhood set of the topological node according to a preset neighborhood radius to obtain a first diffused neighborhood set of the topological node; Determine whether the difference in the amount of neighborhood data between the first diffuse neighborhood set of the topological node and the initial neighborhood set of the topological node is less than or equal to a preset difference. If so, stop center sampling and use the first diffuse neighborhood set of the topological node as the neighborhood set of the topological node monitoring data center.

7. The multiple protection linkage control method for a power distribution system according to claim 6, characterized in that: include: If not, the first diffusion neighborhood set of the topological node is diffused at the neighborhood edge according to the preset neighborhood radius until the difference in the amount of neighborhood data between two adjacent diffusions is less than or equal to the preset difference, and the topological node diffusion neighborhood set obtained by the last diffusion is used as the topological node monitoring data center neighborhood set.

8. The multiple protection linkage control method for a power distribution system according to claim 1, characterized in that: Traversing the abnormal path set to identify multiple protection linkage control schemes and obtain a target linkage control scheme, including: Traversing the abnormal path set to identify the protection control scheme, and obtaining a protection control scheme identification result set; In combination with the power distribution system topology, linkage control is performed on the protection control scheme identification result set to obtain the target linkage control scheme.

Images