Fire-fighting detection method and system based on electric power data analysis

Abstract

The invention provides a fire protection detection method and system based on electric power data analysis, and the method comprises the steps: obtaining a to-be-analyzed electric power equipment set and a historical abnormal monitoring data set of each to-be-analyzed electric power equipment in the to-be-analyzed electric power equipment set, and determining one or more abnormal operation events in the historical abnormal monitoring data set; the method comprises the following steps: generating a power analysis graph by taking to-be-analyzed power equipment and an abnormal operation event as graph vertexes according to a historical abnormal monitoring data set, carrying out representation information mining on the power analysis graph to obtain power equipment representation vectors of the to-be-analyzed power equipment, grouping the to-be-analyzed power equipment according to the power equipment representation vectors, and carrying out analysis on the to-be-analyzed power equipment. Obtaining one or more to-be-determined power equipment groups, and determining a target power equipment group in the to-be-determined power equipment groups; according to the method and the device, the accuracy of the determined power equipment group can be improved, the equipment of the group is helped to be emphatically monitored in fire-fighting monitoring, and the pertinence of fire-fighting monitoring is higher.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a fire detection method and system based on power data analysis. Background Art

[0002] In modern society, the stable operation of the power system is crucial to the normal operation of various fields. As the core component of the power system, the operating status of power equipment directly affects the reliability and safety of power supply. However, during the long-term operation of power equipment, due to various factors such as equipment aging, environmental changes, improper operation, etc., abnormal operating conditions may occur. These abnormal operating conditions will not only affect the normal service life of power equipment, but may also cause safety accidents, especially fire accidents, posing a serious threat to people's lives and property safety.

[0003] Existing fire detection methods mainly rely on traditional monitoring methods, such as smoke sensors, temperature sensors, etc. These sensors can only issue an alarm after a fire occurs or is about to occur, and cannot effectively monitor and warn potential abnormalities of power equipment in advance. Moreover, traditional monitoring methods can often only monitor a single device, making it difficult to conduct a comprehensive and systematic analysis and evaluation of power equipment in the entire power system.

[0004] Some fire detection methods based on data analysis usually simply count and analyze the operating data of power equipment, ignoring the correlation between power equipment and the mutual influence between abnormal operation events. This monotonous analysis method can easily lead to misjudgment of abnormal status of power equipment and fail to accurately determine the power equipment groups that need to be monitored, thereby reducing the pertinence and effectiveness of fire detection. At the same time, due to the lack of comprehensive consideration of the structural characteristics of the abnormal operation graph of power equipment, existing methods often need to process a large amount of redundant data during the data analysis process, which increases the cost and difficulty of data analysis. Summary of the invention

[0005] In view of this, the present application provides a fire detection method and system based on power data analysis.

[0006] The technical solution of this application is implemented as follows: On the one hand, the present application provides a fire detection method based on power data analysis, which is applied to a computer system, and the method includes: obtaining a set of power equipment to be analyzed and a historical abnormal monitoring data set of each power equipment to be analyzed in the set of power equipment to be analyzed; determining one or more abnormal operation events in the historical abnormal monitoring data set, and the abnormal operation event is the abnormal classification corresponding to the abnormal state of the power equipment to be analyzed; based on the historical abnormal monitoring data set, using the power equipment to be analyzed and the abnormal operation event as a graph vertices to generate a power analysis graph, and the power analysis graph is used to indicate the triggering connection between the power equipment to be analyzed and the abnormal operation event; performing characterization information mining on the power analysis graph to obtain a power equipment characterization vector of the power equipment to be analyzed; based on the power equipment characterization vector, grouping the power equipment to be analyzed to obtain one or more pending power equipment groups, and determining a target power equipment group in the pending power equipment group, so as to perform fire detection on the power equipment in the target power equipment group.

[0007] On the other hand, the present application provides a computer system, including a memory and a processor, wherein the memory stores a computer program executable on the processor, and the processor implements the steps in the above method when executing the program.

[0008] After obtaining the set of power equipment to be analyzed and the historical abnormal monitoring data set of each power equipment to be analyzed in the set of power equipment to be analyzed, the present invention determines one or more abnormal operation events in the historical abnormal monitoring data set, and generates a power analysis map based on the historical abnormal monitoring data set, using the power equipment to be analyzed and the abnormal operation events as graph vertices, and then performs characterization information mining on the power analysis map to obtain the power equipment characterization vector of the power equipment to be analyzed, and groups the power equipment to be analyzed according to the power equipment characterization vector to obtain one or more pending power equipment groups, and determines the target power equipment group in the pending power equipment group. Because the present application can determine one or more abnormal operation events in the historical abnormal monitoring data set, and generate a power analysis map based on the abnormal operation events, the map structure features of abnormal operation and the features of the graph vertices can be combined, and the error of determining the target power equipment group based on monotonous connections can be prevented based on the power analysis map. In other words, the present application can improve the accuracy of the determined power equipment group, help focus on the equipment of the group in fire monitoring, make fire monitoring more targeted, and reduce the cost of data analysis.

[0009] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and are not intended to limit the technical solutions of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic diagram of the implementation flow of a fire detection method based on power data analysis provided in an embodiment of the present application.

[0011] Figure 2 A hardware entity diagram of a computer system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0012] The embodiment of the present application provides a fire detection method based on power data analysis, which can be executed by a processor of a computer system, wherein the computer system can refer to a device with data processing capabilities such as a server, a laptop, a tablet computer, a desktop computer, a mobile device, etc.

[0013] Figure 1 A schematic diagram of the implementation flow of a fire detection method based on power data analysis provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the method includes: Step 100: Obtain a set of power equipment to be analyzed and a historical abnormal monitoring data set of each power equipment to be analyzed in the set of power equipment to be analyzed.

[0014] In step 100, a set of power equipment to be analyzed and a historical abnormal monitoring data set of each power equipment to be analyzed in the set are obtained. The set of power equipment to be analyzed is a combination of a series of power equipment to be analyzed in depth. These power equipment can be different types of equipment in the same substation, such as transformers, circuit breakers, capacitors, etc., or they can be power equipment distributed in different areas but with similar functions or operating characteristics.

[0015] The historical abnormal monitoring data set is a data set that records the abnormal conditions of each power equipment to be analyzed in the past period of time. This information usually includes the time when the abnormality occurred, the type of abnormality, and the operating parameters of the power equipment when the abnormality occurred. For example, for a transformer, the historical abnormal monitoring data set may contain multiple records of overtemperature abnormalities, each of which will record in detail the specific time when the overtemperature occurred, the oil temperature of the transformer at that time, the winding temperature, the load current and other parameters. It can be filtered from the asset database of the power system. The asset database stores the basic information of all power equipment owned by the power company, including the type, location, and time of commissioning of the equipment. According to the purpose of the analysis, the set of power equipment to be analyzed can be determined by setting filtering conditions. If the operation of power equipment in a specific area is analyzed, all power equipment in the area can be filtered out according to the geographical location information of the equipment; if the abnormal conditions of a certain type of power equipment are analyzed, all power equipment of that type can be filtered out according to the type information of the equipment.

[0016] When obtaining the historical abnormal monitoring data set of each power equipment to be analyzed, data can be extracted from the monitoring system of the power equipment. The power equipment is equipped with various sensors for real-time monitoring of the operating status of the equipment. These sensors transmit the monitored data to the monitoring system for storage. By establishing a data connection with the monitoring system, the historical abnormal monitoring data of the corresponding power equipment can be extracted from the monitoring system according to the identification information of the equipment.

[0017] Step 200: One or more abnormal operation events are determined in the historical abnormal monitoring data set, where the abnormal operation event is an abnormal classification corresponding to an abnormal state of the power equipment to be analyzed.

[0018] In step 200, one or more abnormal operation events are determined in the historical abnormal monitoring data set. The abnormal operation event is the abnormal classification corresponding to the abnormal state of the power equipment to be analyzed, which can help extract key abnormal information from the massive historical abnormal monitoring data. The historical abnormal monitoring data set contains detailed records of abnormal conditions of the power equipment to be analyzed in the past period of time. These records may involve various operating parameters, time information, etc. of the equipment.

[0019] In order to determine abnormal operation events from these data, a data classification algorithm can be used to classify historical abnormal monitoring data according to the characteristics and types of the abnormalities. For the abnormal data of the transformer, it can be classified according to different types such as temperature abnormality and current abnormality; for the abnormal data of the circuit breaker, it can be classified according to abnormal opening and closing time, abnormal contact wear, etc. Through this classification, similar abnormal data can be classified into one category, so as to determine different abnormal operation events. Machine learning algorithms can also be used to identify abnormal operation events. By learning a large amount of historical abnormal monitoring data, an abnormal recognition model can be established, which can automatically identify new abnormal data and classify it into corresponding abnormal operation events. The support vector machine (SVM) algorithm can be used to train an SVM model that can accurately identify abnormal operation events by taking historical abnormal monitoring data as training samples.

[0020] In practical applications, the severity of the abnormality can be determined by setting a threshold. For the abnormal oil temperature of the transformer, if the oil temperature exceeds a certain percentage of the normal operating range, it is considered a serious abnormality; if the oil temperature only slightly exceeds the normal range, it is considered a slight abnormality. According to the classification of abnormal operation events, corresponding early warning strategies can be formulated to promptly notify relevant personnel to handle them. It is also possible to perform correlation analysis on abnormal operation events to find out the potential connection between different abnormal operation events. In a power system, the temperature abnormality of the transformer may be related to the abnormal load current. Through the correlation analysis of these abnormal operation events, we can deeply understand the operating status of the power equipment and discover potential safety hazards. The association rule mining algorithm is used to analyze the historical abnormal monitoring data to find the association rules between different abnormal operation events. Through the Apriori algorithm, the combination of frequently occurring abnormal events can be mined from the historical abnormal monitoring data, so as to find the correlation between different abnormal operation events.

[0021] Step 300: Based on the historical abnormal monitoring data set, the power equipment to be analyzed and the abnormal operation event are used as graph vertices to generate a power analysis graph, which is used to indicate the triggering connection between the power equipment to be analyzed and the abnormal operation event.

[0022] In step 300, based on the historical abnormal monitoring data set, the power equipment to be analyzed and the abnormal operation event are used as graph vertices to generate a power analysis graph, which is used to indicate the triggering connection between the power equipment to be analyzed and the abnormal operation event, which is crucial for in-depth understanding of the operation status and potential risks of the power equipment. The historical abnormal monitoring data set contains detailed records of abnormal conditions that occurred in the past operation of the power equipment to be analyzed, and these records cover various operating parameters of the equipment, the time when the abnormality occurred, and other information.

[0023] To clarify the triggering relationship between the power equipment to be analyzed and the abnormal operation event, it is necessary to extract relevant information from the historical abnormal monitoring data set. For example, count the number, time, and frequency of each power equipment triggering a specific abnormal operation event. For a transformer, you can count the number of times it triggered an over-temperature anomaly in the past year, the specific time of each over-temperature anomaly, and the frequency of over-temperature anomalies. By analyzing these data, you can understand the degree of correlation between power equipment and abnormal operation events.

[0024] In order to more accurately determine the triggering connection, data analysis algorithms can be used, such as association rule mining algorithms, such as the Apriori algorithm, to mine frequent association rules between power equipment and abnormal operation events from historical abnormal monitoring data. This algorithm calculates the frequency of item sets in the data set to find association rules with high support and confidence. Support represents the frequency of an item set appearing in the data set, and confidence represents the probability of another item set appearing under the condition that one item set appears. Suppose the support of power equipment A triggering abnormal operation event B is ,in represents the number of times that power equipment A triggers abnormal operation event B, N represents the total number of records; the confidence level is , where Count(A) represents the number of abnormal records of power equipment A. The abnormal classification information of abnormal operation events can also be considered. Different types of abnormal operation events may have different effects on power equipment. For transformers, the severity and potential risks of over-temperature anomalies and over-current anomalies may be different. The triggering relationship can be weighted according to the abnormal classification information to more accurately reflect the relationship between power equipment and abnormal operation events. After determining the triggering relationship, the power equipment to be analyzed and the abnormal operation event are used as graph vertices to generate a power analysis graph. Graph vertices are basic elements in the graph structure, which represent power equipment and abnormal operation events. Graph databases can be used to store and manage power analysis graphs. Graph databases can efficiently process graph structure data and support fast graph query and analysis operations. When generating power analysis graphs, edges can be added between graph vertices according to the triggering relationship. The attributes of the edges can include information such as the number of triggers, time, and frequency. For the two graph vertices of transformer and over-temperature anomaly, if the transformer triggered the over-temperature anomaly 10 times in the past year, an edge can be added between the two graph vertices, and the attribute of the edge can be set to the number of triggers 10 times.

[0025] Step 400: mining the characterization information of the power analysis graph to obtain a power equipment characterization vector of the power equipment to be analyzed.

[0026] In step 400, the power analysis map is mined for characterization information to obtain the power equipment characterization vector of the power equipment to be analyzed. This step is an important step to extract key information from the power analysis map to perform a more accurate analysis of the power equipment. The power analysis map is generated based on the historical abnormal monitoring data set, which contains the power equipment to be analyzed and the abnormal operation events as graph vertices, as well as the triggering relationship between them. For example, in the power system of a factory, the power analysis map may show the association between multiple motors and abnormal operation events such as overload anomalies and overheating anomalies.

[0027] To mine the representation information of the power analysis graph, it is necessary to assign a representation vector to each vertex in the graph to represent the characteristic information of the vertex. These representation vectors can be constructed based on the attribute information of the graph vertices, such as the type and rated power of the power equipment, the type and severity of abnormal operation events, etc. For a transformer, its graph vertex representation vector may contain information such as the capacity, voltage level, and number of historical faults of the transformer; for the abnormal operation event of overtemperature anomaly, its graph vertex representation vector may contain information such as the frequency and duration of the abnormal occurrence. In order to make the representation vectors of different graph vertices comparable, these graph vertex representation vectors need to be projected into a unified representation domain to obtain the basic graph vertex representation vector. Since the original representation vectors of different graph vertices may have different dimensions and scales, direct comparison and analysis will be difficult. The linear transformation method can be used to project the graph vertex representation vector to the graph vertex representation vector corresponding to the preset unified projection standard dimension by determining a transformation matrix. Assume that the preset unified projection standard dimension is m, the actual vector dimension of the graph vertex representation vector is n, the transformation matrix is ​​T, and the graph vertex representation vector is X. Then the projected basic graph vertex representation vector Y=T×X, where T is an m×n matrix.

[0028] After obtaining the basic graph vertex characterization vectors, these vectors need to be integrated to obtain the power equipment characterization vectors of the power equipment to be analyzed. The operating status of the power equipment is not only related to its own characteristics, but also to the abnormal operating events around it and the status of other related power equipment. It is necessary to consider the connection relationship between the graph vertices and integrate the characterization vectors of the target graph vertex (i.e., the graph vertex corresponding to the power equipment to be analyzed) and its connected graph vertices (adjacent graph vertices). The graph neural network (GNN) method can be used for vector integration. Graph neural networks can effectively process graph structure data and aggregate the information of the target graph vertex and its connected graph vertices through a message passing mechanism. The target graph vertex can receive information from its connected graph vertices and update its own characterization vector based on this information. Different weights can be given to the information transmission of the connected graph vertices according to the attributes of the edges between the graph vertices, such as the number of triggers and time, so as to more accurately reflect the degree of association between them.

[0029] The power equipment characterization vector can comprehensively reflect the operating status and characteristics of the power equipment. By analyzing the power equipment characterization vector, we can find the similarities and differences between the power equipment, which provides a basis for the subsequent power equipment grouping. If the characterization vectors of two power equipment are highly similar, it means that they are similar in terms of operating status and abnormal triggering conditions, and may belong to the same type of equipment group.

[0030] The power equipment characterization vector can also be used for fault prediction and risk assessment. By establishing a fault prediction model and taking the power equipment characterization vector as input, the probability and type of power equipment failure can be predicted. Machine learning algorithms, such as support vector machines and random forests, can be used to train a large amount of historical data to obtain an accurate fault prediction model. The potential risks of power equipment can also be evaluated based on the power equipment characterization vector to provide a reference for fire detection. For power equipment with higher risks, early warnings can be issued in a timely manner to remind relevant personnel to conduct inspections and maintenance to prevent safety accidents such as fires.

[0031] Step 500: Group the power equipment to be analyzed according to the power equipment characterization vector to obtain one or more pending power equipment groups, and determine a target power equipment group in the pending power equipment groups so as to perform fire detection on the power equipment in the target power equipment group.

[0032] In step 500, the power equipment to be analyzed is grouped according to the power equipment characterization vector to obtain one or more pending power equipment groups, and a target power equipment group is determined in the pending power equipment group, so as to perform fire detection on the power equipment in the target power equipment group. The power equipment characterization vector is obtained after characterization information mining of the power analysis graph, and it comprehensively reflects the operating status and characteristics of the power equipment.

[0033] The power equipment commonality metric value between the power equipment to be analyzed in the power equipment set to be analyzed is determined by calculating the power equipment characterization vector. The power equipment commonality metric value is used to measure the similarity between different power equipment. The higher the similarity, the closer these power equipment are in terms of operating status, abnormal triggering conditions, etc. The cosine similarity algorithm can be used to calculate the power equipment commonality metric value. Suppose the characterization vectors of two power equipment are A and B, then their cosine similarity is ,in is the dot product of vectors A and B, and are the moduli of vectors A and B, respectively. The closer the cosine similarity value is to 1, the more similar the characterization vectors of the two power devices are, that is, the higher their commonality measure value is.

[0034] According to the calculated commonality metric values ​​of the power equipment, the power equipment to be analyzed is used as the graph vertex to generate an interconnection graph. The interconnection graph is an undirected graph, in which each graph vertex represents a power equipment, and the edges between the vertices represent the commonality metric values ​​between the power equipment. The weight of the edge can be set to the commonality metric value of the power equipment. The larger the weight, the higher the similarity between the two power equipment. In the power system of an industrial park, the interconnection graph can intuitively show the similarity relationship between different power equipment.

[0035] Next, the graph vertices in the interconnected graph are grouped to obtain one or more pending power equipment groups. The purpose of grouping is to group similar power equipment into the same group for more targeted analysis and management. Community discovery algorithms, such as the Louvain algorithm, can be used to group the interconnected graph. The Louvain algorithm is a community discovery algorithm based on modularity optimization. It maximizes the grouping effectiveness of the entire graph by continuously moving graph vertices to different communities. Grouping effectiveness Q is an indicator to measure the quality of community division in the graph, and its calculation formula is: , where m is the total number of edges in the graph, is the weight of the edge between vertices i and j in the graph, and are the degrees of vertices i and j respectively, and are the communities to which vertices i and j belong respectively, is an indicator function, when 1 if the value is 0, otherwise it is 0.

[0036] After obtaining the pending power equipment groups, the target power equipment groups are determined in these groups. The target power equipment group is a group composed of power equipment that is more likely to have fire hazards. The pending power equipment group can be evaluated and screened based on historical abnormal data of the power equipment, operating environment and other factors. Once the target power equipment group is determined, fire detection can be performed on the power equipment in the group. Fire detection can include real-time monitoring of parameters such as temperature, current, and voltage of the power equipment, and checking the insulation performance and grounding conditions of the power equipment. Relevant parameters can be obtained through sensors installed on the power equipment, and these parameters are compared with preset safety thresholds. Once it is found that the parameters exceed the threshold, an early warning signal is immediately issued to remind relevant personnel to handle it. In the above manner, the power equipment to be analyzed can be effectively grouped according to the power equipment characterization vector, the target power equipment group can be determined, and fire detection can be performed on it, which is helpful to improve the safety and reliability of the power system, timely discover and eliminate potential fire hazards, and ensure the normal operation of the power equipment and the safety of life and property of personnel.

[0037] In one implementation, step 200, determining one or more abnormal operation events in the historical abnormal monitoring data set includes: Step 210: extracting abnormal category data of the power equipment to be analyzed from the historical abnormal monitoring data set; Step 220: Determine in the abnormal category data one or more abnormal operation events corresponding to the power equipment to be analyzed.

[0038] In step 210, the abnormal category data of the power equipment to be analyzed is extracted from the historical abnormal monitoring data set. The historical abnormal monitoring data set contains detailed records of abnormal situations that occurred in the power equipment to be analyzed during past operations. These records cover various operating parameters of the equipment, the time when the abnormality occurred, and other information.

[0039] In order to extract abnormal category data from historical abnormal monitoring data sets, data screening and classification methods can be used. The historical abnormal monitoring data can be screened according to the preset abnormal type labels to extract data that meets the specific abnormal type. For over-temperature anomalies, a temperature threshold can be set. When the monitored device temperature exceeds the threshold, it is marked as over-temperature abnormal data. Data classification algorithms, such as decision tree algorithms, can also be used to classify historical abnormal monitoring data and distinguish different types of abnormal data. The decision tree algorithm can effectively identify different types of abnormal data by constructing a decision tree model and classifying according to the characteristics of the data.

[0040] After the abnormal category data is extracted, in step 220, one or more abnormal operation events corresponding to the power equipment to be analyzed are determined in the abnormal category data. The abnormal operation event is the abnormal classification corresponding to the abnormal state of the power equipment to be analyzed, which is a further abstraction and summary of the abnormal situation. For the over-temperature anomaly of the transformer, the abnormal operation event can be divided into different levels of abnormal classification such as mild over-temperature, moderate over-temperature, and severe over-temperature.

[0041] To determine abnormal operation events in abnormal category data, rule matching and machine learning methods can be used. Rule matching refers to matching abnormal category data according to preset rules to determine the corresponding abnormal operation events. For the over-temperature anomaly of the transformer, different temperature intervals can be set. When the monitored temperature falls in different intervals, it is classified as over-temperature abnormal operation events of different levels. The machine learning method is to establish an abnormal operation event recognition model by learning a large amount of historical abnormal category data. The model can automatically identify the abnormal operation events corresponding to the new abnormal category data. The support vector machine (SVM) algorithm can be used to train an SVM model that can accurately identify abnormal operation events using historical abnormal category data as training samples. In practical applications, the severity of the anomaly can be determined by setting a threshold. For the over-current anomaly of the transformer, if the current exceeds a certain proportion of the rated current, it is considered a serious anomaly; if the current only slightly exceeds the rated current, it is considered a minor anomaly. According to the classification of abnormal operation events, corresponding early warning strategies can be formulated to notify relevant personnel in time to handle them. It is also possible to perform correlation analysis on abnormal operation events to find out the potential connection between different abnormal operation events. In a power system, transformer overtemperature anomalies may be related to overcurrent anomalies. By analyzing the correlation between these abnormal operating events, we can gain an in-depth understanding of the operating status of power equipment and discover potential safety hazards.

[0042] The association rule mining algorithm can be used to analyze the abnormal category data and find the association rules between different abnormal operation events. Through the Apriori algorithm, the frequently occurring abnormal event combinations can be mined from the abnormal category data, thereby discovering the association relationship between different abnormal operation events.

[0043] In one implementation, step 220, determining in the abnormal category data one or more abnormal operation events corresponding to the power equipment to be analyzed, includes: Step 221: extracting the abnormal occurrence time and abnormal electrical nodes of the power equipment to be analyzed from the abnormal category data; Step 222: Determine one or more abnormal operation events corresponding to the power equipment to be analyzed based on the abnormal occurrence time and the abnormal electrical node.

[0044] In step 221, the abnormal occurrence time and abnormal electrical node of the power equipment to be analyzed are extracted from the abnormal category data. The abnormal category data is data related to the abnormal type extracted from the historical abnormal monitoring data set, which contains various information when the power equipment to be analyzed is abnormal. Taking the power system of a large factory as an example, the abnormal category data may cover different abnormal situation records of equipment such as transformers and motors. The abnormal occurrence time refers to the specific moment when the power equipment has an abnormal state, which can reflect the time law and trend of the abnormal occurrence. For the over-temperature anomaly of the transformer, the abnormal occurrence time can be accurate to the specific year, month, day, hour, minute, and second. By analyzing this time information, it can be found whether the over-temperature anomaly occurs frequently in a specific time period, such as in the high temperature period in summer or the high-load operation period of the equipment. The abnormal electrical node refers to the specific electrical part of the power equipment where the abnormality occurs, which can help locate the specific location of the abnormality. In a complex power network, a transformer may have multiple electrical nodes, such as high-voltage side windings, low-voltage side windings, etc. The abnormal electrical node can clearly indicate which winding has an abnormality.

[0045] In order to extract the abnormal occurrence time and abnormal electrical nodes from the abnormal category data, data parsing and filtering methods can be used. According to the storage format of the abnormal category data, the time information and electrical node information can be extracted using the corresponding parsing algorithm. If the abnormal category data is stored in a table, the column where the abnormal occurrence time and abnormal electrical node are located can be located by column name or data position, and then these data can be extracted. You can also set filtering conditions to only extract the abnormal occurrence time and abnormal electrical node data related to the power equipment to be analyzed to improve the accuracy and pertinence of the data.

[0046] After the abnormal occurrence time and abnormal electrical node are extracted, in step 222, one or more abnormal operation events corresponding to the power equipment to be analyzed are determined based on this information. Abnormal operation events are the classification and summary of abnormal states of power equipment, which can help to understand the abnormal conditions of power equipment more clearly. For transformers, abnormal operation events may include over-temperature abnormality, over-current abnormality, short-circuit abnormality, etc.

[0047] When determining abnormal operation events based on the abnormal occurrence time and abnormal electrical nodes, rule matching and machine learning methods can be used. Rule matching refers to matching the abnormal occurrence time and abnormal electrical nodes according to preset rules to determine the corresponding abnormal operation events. For the over-temperature anomaly of the transformer, if the abnormal occurrence time is in the high temperature period in summer, and the abnormal electrical node is the high-voltage side winding of the transformer, it can be classified as an abnormal operation event of the high-voltage side winding over-temperature caused by the high temperature in summer according to the preset rules. The machine learning method is to establish an abnormal operation event recognition model by learning a large amount of historical abnormal data. The model can automatically identify the corresponding abnormal operation events based on the input abnormal occurrence time and abnormal electrical node information. A neural network algorithm can be used to take the abnormal occurrence time and abnormal electrical node in the historical abnormal data as input and the abnormal operation event as output to train a neural network model that can accurately identify abnormal operation events. When determining abnormal operation events, factors such as the frequency and duration of abnormal occurrence can also be considered. If an abnormal electrical node frequently has abnormalities in a short period of time, it means that the node may have serious potential faults, and its corresponding abnormal operation event can be marked as a high-risk event. The duration of the abnormality can be calculated based on the time when the abnormality occurred. If the duration of the abnormality is long, the risk level of the abnormal operation event will also increase. It is also possible to perform correlation analysis on abnormal operation events, combining the time of abnormal occurrence and abnormal electrical nodes to find the potential connection between different abnormal operation events. In a power system, the over-temperature anomaly and over-current anomaly of the transformer may occur at the same time, and the abnormal electrical nodes are close. Through the correlation analysis of these abnormal operation events, we can gain a deep understanding of the operating status of the power equipment and discover potential safety hazards.

[0048] In one implementation, step 300, based on the historical abnormal monitoring data set, uses the power equipment to be analyzed and the abnormal operation events as graph vertices to generate a power analysis graph, including: Step 310: determining triggering statistical information of the power equipment to be analyzed and the abnormal operation event in the historical abnormal monitoring data set; Step 320: determining the triggering relationship between the power equipment to be analyzed and the abnormal operation event according to the abnormal classification information and triggering statistical information of the abnormal operation event; Step 330: Based on the triggering connection, the power equipment to be analyzed and the abnormal operation events are used as graph vertices to generate a power analysis graph.

[0049] In step 310, the triggering statistical information of the power equipment to be analyzed and the abnormal operation event is determined in the historical abnormal monitoring data set. The historical abnormal monitoring data set contains detailed records of abnormal conditions that occurred in the past operation of the power equipment to be analyzed. These records cover various operating parameters of the equipment, the time when the abnormality occurred, and other information. Taking an electric power system as an example, the historical abnormal monitoring data set may contain voltage and current abnormal records of equipment such as transformers, distribution cabinets, and elevator control cabinets, as well as the specific time when these abnormalities occurred. Triggering statistical information refers to the relevant statistical data of abnormal operation events triggered by the power equipment to be analyzed, such as the number of triggers, time, frequency, etc. This information can reflect the degree of correlation between the power equipment and the abnormal operation event. For a transformer, the triggering statistical information may include the number of times it triggered over-temperature anomalies in the past year, the specific time of each over-temperature anomaly, and the frequency of over-temperature anomalies.

[0050] After the trigger statistical information is determined, in step 320, the trigger connection between the power equipment to be analyzed and the abnormal operation event is determined based on the abnormal classification information and trigger statistical information of the abnormal operation event. Abnormal classification information refers to information for classifying abnormal operation events, and different types of abnormal operation events may have different effects on power equipment. For transformers, the severity and potential risks of over-temperature anomalies and over-current anomalies may be different. The trigger connection refers to the correlation between the power equipment to be analyzed and the abnormal operation event, which can reflect the possibility and intensity of the power equipment triggering the abnormal operation event. A weighted calculation method can be used to determine the trigger connection based on the abnormal classification information and trigger statistical information. For different types of abnormal operation events, different weights can be set according to their severity and potential risks. For over-temperature anomalies, since it may cause transformer insulation aging or even cause fire, a higher weight can be set; for minor voltage fluctuation anomalies, a lower weight can be set. The trigger statistical information and the corresponding weights can be weighted calculated to obtain the trigger connection value between the power equipment to be analyzed and the abnormal operation event. Assume that the abnormal operation event The weight of , the power equipment to be analyzed is D j Trigger abnormal operation event E i The number of triggers is n ij , then the power equipment D j With abnormal operation event E i The trigger contact value C ij =w i ×n ij .

[0051] We can also consider factors such as trigger time and frequency to make a more accurate assessment of the trigger connection. If the power equipment frequently triggers an abnormal operation event in a short period of time, it means that the trigger connection between the equipment and the abnormal operation event is strong. We can introduce a time decay factor to attenuate the trigger events with a long trigger time to highlight the impact of recent trigger events. Let the time decay factor be , For power equipment Trigger abnormal operation event E i time, the current time is t0, then the trigger contact value after time decay is considered .

[0052] After the trigger connection is determined, in step 330, according to the trigger connection, the power equipment to be analyzed and the abnormal operation event are used as graph vertices to generate a power analysis graph. Graph vertices are basic elements in the graph structure, which represent the power equipment to be analyzed and the abnormal operation event. The power analysis graph is a graph structure that can intuitively display the trigger connection between the power equipment to be analyzed and the abnormal operation event. A graph database can be used to store and manage the power analysis graph. A graph database is a database specifically used to process graph structure data, which can efficiently store and query the information of graph vertices and edges. The power equipment to be analyzed and the abnormal operation event can be stored in the graph database as graph vertices, and edges can be added between the graph vertices according to the trigger connection. The attribute of the edge can be set to the trigger connection value. The larger the weight of the edge, the stronger the trigger connection between the power equipment to be analyzed and the abnormal operation event. The power analysis graph can be dynamically updated. As new historical abnormal monitoring data is continuously generated, the information of the graph vertices and edges in the graph can be updated in real time to reflect the latest changes in the operating status of the power equipment. New data can be periodically extracted from the historical anomaly monitoring data set, trigger statistics and trigger connections can be recalculated, and then the power analysis map can be updated.

[0053] Risk assessment can be performed based on the power analysis map. For each power device to be analyzed, the risk score of the device can be calculated based on the triggering connection between it and the abnormal operation event. The higher the risk score, the greater the possibility of failure of the device.

[0054] Through the above methods, the trigger statistics can be accurately determined from the historical abnormal monitoring data set, the trigger connection can be determined based on the abnormal classification information and the trigger statistics, and the power analysis map can be generated based on the trigger connection. The power analysis map can intuitively show the relationship between the power equipment to be analyzed and the abnormal operation event, which provides an important basis for subsequent steps such as characterization information mining, power equipment grouping, and fire detection, which helps to improve the safety and reliability of the power system and reduce potential fire risks.

[0055] In one implementation, step 400, performing characterization information mining on the power analysis graph to obtain a power equipment characterization vector of the power equipment to be analyzed, includes: Step 410: extracting a graph vertex representation vector of each graph vertex in the power analysis graph; Step 420: Project the graph vertex representation vector to a unified representation domain to obtain a basic graph vertex representation vector; Step 430: Integrate the basic graph vertex characterization vectors to obtain the power equipment characterization vector of the power equipment to be analyzed.

[0056] In step 410, a vertex characterization vector of each vertex is extracted from the power analysis map. The power analysis map is generated based on a historical abnormal monitoring data set, which includes the power equipment to be analyzed and abnormal operation events as graph vertices, as well as the triggering relationship between them. The vertex characterization vector is a vector used to represent the characteristics of the graph vertex, which can comprehensively reflect the various attribute information of the power equipment or abnormal operation events represented by the vertex. Taking the power analysis map of a city's power supply network as an example, for the vertex representing the transformer, its vertex characterization vector may include information such as the transformer's capacity, voltage level, operating years, and number of historical faults; for the vertex representing the over-temperature anomaly, its vertex characterization vector may include information such as the frequency, duration, and impact range of the anomaly.

[0057] In order to extract the graph vertex representation vector in the power analysis graph, it can be constructed based on the attribute information of the graph vertex. If the power analysis graph is stored in a graph database, the attribute data of each graph vertex can be obtained by querying the graph database. These attribute data can be converted into vector form according to the pre-defined feature dimensions. For the graph vertices of the transformer, the attribute data such as capacity, voltage level, operating years, and number of historical faults can be arranged in a certain order to form a vector. Feature engineering methods can also be used to process and transform the attribute data to improve the quality of the graph vertex representation vector. For some continuous attribute data, normalization can be performed and mapped to the interval [0, 1] to eliminate the dimensional differences between different attribute data.

[0058] After extracting the graph vertex representation vector of each graph vertex, in step 420, the graph vertex representation vector is projected to a unified representation domain to obtain a basic graph vertex representation vector. Since the original representation vectors of different graph vertices may have different dimensions and scales, direct comparison and analysis may be difficult. A unified representation domain refers to a vector space with the same dimension and scale. After the graph vertex representation vectors are projected to the space, they can be compared and analyzed under the same standard. A linear transformation method can be used to project the graph vertex representation vector to the graph vertex representation vector corresponding to the preset unified projection standard dimension by determining a transformation matrix. Assuming the preset unified projection standard dimension is m, the actual vector dimension of the graph vertex representation vector is n, the transformation matrix is ​​T, and the graph vertex representation vector is X, then the projected basic graph vertex representation vector Y=T×X, where T is an m×n matrix. The transformation matrix T can be determined by methods such as singular value decomposition (SVD). Singular value decomposition is a matrix decomposition method that can decompose a matrix into the product of three matrices. By processing these matrices, a suitable transformation matrix T can be obtained.

[0059] After projecting the graph vertex characterization vectors to a unified characterization domain, the characterization vectors of different graph vertices are comparable, and subsequent analysis and processing can be performed more conveniently. By calculating the similarity between the basic graph vertex characterization vectors, the similarity relationship between the graph vertices can be found, such as which power equipment is more similar in operating characteristics, and which abnormal operating events have similar characteristics. After obtaining the basic graph vertex characterization vector, in step 430, the basic graph vertex characterization vector is integrated to obtain the power equipment characterization vector of the power equipment to be analyzed. The operating status of the power equipment is not only related to its own characteristics, but also to the abnormal operating events around it and the status of other related power equipment. It is necessary to consider the connection relationship between the graph vertices, and integrate the characterization vectors of the target graph vertex (that is, the graph vertex corresponding to the power equipment to be analyzed) and its connected graph vertices (adjacent graph vertices).

[0060] Vector integration can be performed using the graph neural network (GNN) method. A graph neural network is a neural network specifically designed to process graph structured data, which can effectively capture the dependencies between graph vertices. Through the message passing mechanism, the graph neural network can aggregate the information of the target graph vertex and its connected graph vertices. The target graph vertex can receive information from its connected graph vertices and update its own representation vector based on this information. Different weights can be assigned to the information transmission of connected graph vertices based on the attributes of the edges between graph vertices, such as the number of triggers, time, etc., to more accurately reflect the degree of association between them.

[0061] The attention mechanism can also be used to enhance the effect of vector integration. The attention mechanism can automatically assign different attention weights to different connection graph vertices, so that the target graph vertex can pay more attention to the connection graph vertices with stronger correlation with itself. Suppose the target graph vertex is , whose connection graph vertices are , attention weight It can be calculated by the following formula: ,in Represents the target graph vertex and connect graph vertices The similarity between Represents the target graph vertex v t The attention mechanism can be used to more flexibly integrate the basic graph vertex representation vectors and improve the quality of the power equipment representation vectors.

[0062] The power equipment characterization vector can comprehensively reflect the operating status and characteristics of the power equipment. By analyzing the power equipment characterization vector, we can find the similarities and differences between the power equipment, which provides a basis for the subsequent power equipment grouping. If the characterization vectors of two power equipment are highly similar, it means that they are similar in terms of operating status and abnormal triggering conditions, and may belong to the same type of equipment group.

[0063] The power equipment characterization vector can be used for fault prediction and risk assessment. By establishing a fault prediction model, the power equipment characterization vector is used as input to predict the probability and type of power equipment failure. Machine learning algorithms, such as support vector machines and random forests, can be used to train a large amount of historical data to obtain an accurate fault prediction model. Let the power equipment characterization vector be x and the fault prediction model be f(x), then the predicted fault probability P=f(x).

[0064] The potential risk of power equipment can also be evaluated based on the power equipment characterization vector to provide a reference for fire detection. For power equipment with higher risks, early warnings can be issued in time to remind relevant personnel to conduct inspections and maintenance to prevent safety accidents such as fires. The risk score of power equipment can be calculated based on certain features in the power equipment characterization vector, such as the frequency and severity of abnormalities. Suppose the power equipment characterization vector is x and the risk assessment function is g(x), then the risk score of the power equipment is R=g(x).

[0065] Through the above method, the representation information of the power analysis graph can be effectively mined, and the extraction of graph vertex representation vectors, projection to a unified representation domain, and integration of basic graph vertex representation vectors are completed in sequence, and finally the power equipment representation vector of the power equipment to be analyzed is obtained.

[0066] In one implementation, step 420, projecting the graph vertex representation vectors to a unified representation domain to obtain the basic graph vertex representation vectors, includes: Step 421: Obtaining a preset unified projection standard dimension and an actual vector dimension of each graph vertex representation vector; Step 422: Determine the transformation matrix corresponding to the graph vertex representation vector according to the preset unified projection standard dimension and the actual vector dimension; Step 423: According to the transformation matrix, the graph vertex representation vector is projected to the graph vertex representation vector corresponding to the preset unified projection standard dimension to obtain the basic graph vertex representation vector.

[0067] In step 421, the preset unified projection standard dimension and the actual vector dimension of each graph vertex representation vector are obtained. The preset unified projection standard dimension is a pre-set fixed dimension, which is used to unify different graph vertex representation vectors into the vector space of this dimension for subsequent comparison and analysis. The actual vector dimension of each graph vertex representation vector refers to the dimension of each graph vertex representation vector before the projection operation is performed. Taking an electric power analysis graph containing a variety of power equipment and abnormal operation events as an example, for the graph vertex representing the transformer, its graph vertex representation vector may contain information such as the capacity, voltage level, operating years, and historical fault number of the transformer, and the actual vector dimension may be 4; and for the graph vertex representing the overcurrent anomaly, its graph vertex representation vector may contain information such as the frequency, duration, and impact range of the anomaly, and the actual vector dimension may be 3.

[0068] In order to obtain the preset unified projection standard dimension, it can be set according to the specific analysis requirements and subsequent processing requirements. If a specific machine learning algorithm is required for analysis later, and the algorithm has certain requirements on the dimension of the input vector, the preset unified projection standard dimension will be set to the dimension required by the algorithm. You can also select a suitable preset unified projection standard dimension through experiments and optimization methods to improve the effect of subsequent analysis.

[0069] While obtaining the preset unified projection standard dimension, obtain the actual vector dimension of each graph vertex representation vector. You can directly obtain its dimension information by querying the data structure of the graph vertex representation vector. If the graph vertex representation vector is stored in the form of an array, you can determine its actual vector dimension by obtaining the length of the array.

[0070] After obtaining the preset unified projection standard dimension and the actual vector dimension of each graph vertex representation vector, in step 422, the transformation matrix corresponding to the graph vertex representation vector is determined based on this information. The transformation matrix is ​​a matrix used to project the graph vertex representation vector from its actual vector dimension to the preset unified projection standard dimension. Suppose the preset unified projection standard dimension is m, the actual vector dimension of the graph vertex representation vector is n, and the transformation matrix is ​​T, then T is an m×n matrix. The singular value decomposition (SVD) method can be used to determine the transformation matrix T. Singular value decomposition is a matrix decomposition method that can decompose a matrix A into the product of three matrices, namely , where U is an m×m orthogonal matrix, is an m×n diagonal matrix, V T is an n×n orthogonal matrix. The matrix composed of the graph vertex representation vectors can be subjected to singular value decomposition, and then, according to the preset unified projection standard dimension and the actual vector dimension, the appropriate part is selected from the decomposed matrix to form the transformation matrix T. In addition to singular value decomposition, the principal component analysis (PCA) method can also be used to determine the transformation matrix.

[0071] After the transformation matrix T is determined, in step 423, the graph vertex representation vector is projected to the graph vertex representation vector corresponding to the preset unified projection standard dimension according to the transformation matrix to obtain the basic graph vertex representation vector. Let the graph vertex representation vector be X and the transformation matrix be T, then the projected basic graph vertex representation vector Y can be calculated by the formula Y=T×X.

[0072] When performing the projection operation, the same processing is performed on each graph vertex representation vector. For each graph vertex in the power analysis graph, the corresponding graph vertex representation vector X is matrix multiplied with the transformation matrix T to obtain the projected basic graph vertex representation vector Y. In this way, all graph vertex representation vectors can be unified into a vector space with a preset unified projection standard dimension. The original representation vectors of different graph vertices may have different dimensions and scales, and direct comparison and analysis will be difficult. By projecting to a unified representation domain, the differences in dimensions and scales can be eliminated, so that the representation vectors of different graph vertices can be compared and analyzed under the same standard. This helps to more accurately discover the similarity and differences between graph vertices, and provide a more reliable data basis for subsequent power equipment grouping, fault prediction and other analyses. After obtaining the basic graph vertex representation vectors, these vectors can be further analyzed and processed. The similarity between the basic graph vertex representation vectors can be calculated to find the similarity between power equipment or abnormal operation events. Commonly used similarity calculation methods include cosine similarity, Euclidean distance, etc. The basic graph vertex representation vectors can also be used for cluster analysis to classify similar graph vertices into the same category. Commonly used clustering algorithms include K-means clustering, hierarchical clustering, etc. Through cluster analysis, power equipment or abnormal operation events can be classified for more targeted management and processing.

[0073] In addition, the base graph vertex representation vector can be used as input to build a machine learning model for fault prediction and risk assessment. For the fault prediction of power equipment, machine learning algorithms such as support vector machines and neural networks can be used to train a model that can accurately predict power equipment failures by using the base graph vertex representation vector as input features. Let the base graph vertex representation vector be Y and the fault prediction model be f(Y), then the predicted fault probability P=f(Y).

[0074] The potential risk of power equipment can be evaluated based on the base graph vertex characterization vector. For power equipment with higher risks, early warnings can be issued in time to remind relevant personnel to conduct inspections and maintenance to prevent safety accidents such as fires. The risk score of power equipment can be calculated based on certain features in the base graph vertex characterization vector, such as the frequency and severity of abnormalities. Let the base graph vertex characterization vector be Y and the risk assessment function be g(Y), then the risk score of the power equipment is R=g(Y).

[0075] Through the above method, the transformation matrix can be determined according to the preset unified projection standard dimension and the actual vector dimension of the graph vertex representation vector, and the graph vertex representation vector is projected to a unified representation domain to obtain the basic graph vertex representation vector.

[0076] In one implementation, step 430, integrating the base graph vertex characterization vectors to obtain the power equipment characterization vector of the power equipment to be analyzed, includes: Step 431: determining the graph vertex corresponding to the power equipment to be analyzed in the graph vertices, obtaining the target graph vertex, and sampling one or more connection graph vertices of the target graph vertex in the power analysis graph; Step 432: Determine the target graph vertex representation vector corresponding to the target graph vertex and the connection graph vertex representation vector corresponding to the connection graph vertex in the base graph vertex representation vector; Step 433: Integrate the target graph vertex characterization vector and the connection graph vertex characterization vector to obtain the power equipment characterization vector of the power equipment to be analyzed.

[0077] In step 431, the graph vertex corresponding to the power equipment to be analyzed is determined in the graph vertices, the target graph vertex is obtained, and one or more connection graph vertices of the target graph vertex are sampled in the power analysis spectrum. In the power analysis spectrum, each graph vertex represents a power equipment or an abnormal operation event, and the graph vertex corresponding to the power equipment to be analyzed is the target graph vertex. Taking the power analysis spectrum of the power system of a large industrial park as an example, if the operating status of a transformer is to be analyzed, then the graph vertex representing this transformer is the target graph vertex. The connection graph vertex is the graph vertex that has a connection relationship with the target graph vertex in the spectrum, and they may represent abnormal operation events related to the power equipment or other related power equipment. For example, if the transformer has an over-temperature anomaly, then the graph vertex representing the over-temperature anomaly is the connection graph vertex of the target graph vertex.

[0078] In order to determine the target graph vertex, it is necessary to search in the power analysis graph according to the identification information of the power equipment to be analyzed. If the power analysis graph is stored in a graph database, the corresponding graph vertex can be located according to the identification information such as the equipment number and name through a query statement. After determining the target graph vertex, it is necessary to sample one or more of its connected graph vertices. The purpose of sampling is to reduce the amount of calculation and improve processing efficiency while ensuring the integrity of information. Sampling can be performed using methods such as random sampling and stratified sampling. Random sampling refers to randomly selecting a certain number of graph vertices from all connected graph vertices of the target graph vertex; stratified sampling is to stratify according to certain characteristics of the connected graph vertices, and then select a certain number of graph vertices from each layer.

[0079] After obtaining the target graph vertex and the connection graph vertex, in step 432, the target graph vertex representation vector corresponding to the target graph vertex and the connection graph vertex representation vector corresponding to the connection graph vertex are determined in the base graph vertex representation vector. The base graph vertex representation vector is a vector obtained by projecting the graph vertex representation vector to a unified representation domain. They have the same dimension, which is convenient for comparison and analysis. The target graph vertex representation vector and the connection graph vertex representation vector represent the characteristic information of the target graph vertex and the connection graph vertex, respectively. For example, for a transformer represented by a target graph vertex, its target graph vertex representation vector may contain characteristic information such as the capacity, voltage level, and operating life of the transformer; for an over-temperature anomaly represented by a connection graph vertex, its connection graph vertex representation vector may contain characteristic information such as the frequency and duration of the anomaly.

[0080] The corresponding vector can be found in the basic graph vertex representation vector set according to the identification information of the graph vertex. If the basic graph vertex representation vector is stored in the form of an array or a list, each element corresponds to a representation vector of a graph vertex, and the corresponding element can be found according to the index or identification information of the graph vertex.

[0081] After determining the target graph vertex characterization vector and the connection graph vertex characterization vector, in step 433, these vectors are integrated to obtain the power equipment characterization vector of the power equipment to be analyzed. The operating status of the power equipment is not only related to its own characteristics, but also to abnormal operating events around it and the status of other related power equipment. Therefore, integrating the target graph vertex characterization vector and the connection graph vertex characterization vector can more comprehensively reflect the operating status of the power equipment, such as using a weighted summation method for vector integration, or using an attention mechanism for vector integration. The attention mechanism automatically assigns different attention weights to different connection graph vertices, so that the target graph vertex can pay more attention to the connection graph vertices that are more relevant to itself. Suppose the target graph vertex is , whose connection graph vertices are , attention weight It can be calculated by the following formula: ,in Represents the target graph vertex representation vector and the connection graph vertex representation vector Then, the integrated power equipment characterization vector V can be obtained by the formula Calculated.

[0082] The power equipment characterization vector can comprehensively reflect the operating status and characteristics of the power equipment. By analyzing the power equipment characterization vector, we can find the similarities and differences between the power equipment, which provides a basis for the subsequent grouping of power equipment. If the characterization vectors of two power equipment have a high similarity, it means that they are similar in terms of operating status and abnormal triggering conditions, and may belong to the same type of equipment group. The similarity between the power equipment characterization vectors can be calculated using methods such as cosine similarity and Euclidean distance.

[0083] Through the above method, it is possible to determine the target graph vertex and the connection graph vertex in the graph vertices, obtain the corresponding characterization vectors, and integrate them to finally obtain the power equipment characterization vector of the power equipment to be analyzed.

[0084] In one implementation, in step 431, sampling one or more connection graph vertices of a target graph vertex in the power analysis graph includes: Step 4311: determining the number of graph vertices of the surrounding graph vertices adjacent to the target graph vertex in the power analysis graph; Step 4312: Determine the sampling strategy of the connection graph vertices of the target graph vertex according to the number of graph vertices. For example, according to the number of graph vertices, determine the sampling strategy of the connection graph vertices of the target graph vertex, including: comparing the number of graph vertices with the reference value, if the number of graph vertices is greater than the reference value, determine that the sampling strategy is a non-repetitive sampling strategy; if the number of graph vertices is less than or equal to the reference value, determine that the sampling strategy is a reusable sampling strategy; if the sampling strategy is a non-repetitive sampling strategy, directly sample the connection graph vertices corresponding to the reference value from the surrounding graph vertices, and if the sampling strategy is a reusable sampling strategy, sample the connection graph vertices corresponding to the reference value from the surrounding graph vertices using the reusable sampling strategy.

[0085] Step 4313: According to the sampling strategy, one or more connection graph vertices are sampled from the surrounding graph vertices.

[0086] In step 4311, the number of graph vertices of the peripheral graph vertices adjacent to the target graph vertex is determined in the power analysis graph. In the power analysis graph, the target graph vertex represents the power equipment to be analyzed, and the peripheral graph vertices are graph vertices that are directly connected to the target graph vertex. These peripheral graph vertices may represent abnormal operation events related to the power equipment or other related power equipment. In order to determine the number of graph vertices of the peripheral graph vertices, it can be achieved by querying the graph structure information of the power analysis graph. If the power analysis graph is stored in a graph database, the query language provided by the graph database can be used to find the graph vertices directly connected to the target graph vertex according to its identification information, and count the number of these graph vertices.

[0087] After determining the number of graph vertices of the surrounding graph vertices, in step 4312, a sampling strategy for connecting graph vertices of the target graph vertex is determined based on the number of graph vertices. The sampling strategy refers to the method used when selecting connecting graph vertices from the surrounding graph vertices. Different numbers of graph vertices may require different sampling strategies to ensure the validity and representativeness of the sampling results. The number of graph vertices is usually compared with a reference value, which is a pre-set threshold used to determine which sampling strategy to use.

[0088] If the number of graph vertices is greater than the reference value, it means that there are many surrounding graph vertices. At this time, the sampling strategy can be determined as a non-repetitive sampling strategy. The non-repetitive sampling strategy means that during the sampling process, each surrounding graph vertex can only be selected once, which can avoid repeated selection of the same graph vertex and ensure the randomness and diversity of sampling. For example, if the reference value is set to 10, if the number of surrounding graph vertices of the target graph vertex is 15, the non-repetitive sampling strategy can be used to randomly select a certain number of graph vertices from these 15 surrounding graph vertices as connected graph vertices.

[0089] If the number of graph vertices is less than or equal to the reference value, it means that the number of surrounding graph vertices is small. At this time, the sampling strategy can be determined to be a reusable sampling strategy. The reusable sampling strategy means that during the sampling process, the same surrounding graph vertex can be selected multiple times, which can ensure that a sufficient number of connected graph vertices can be extracted to meet the needs of subsequent analysis. For example, if the reference value is set to 10, if the number of surrounding graph vertices of the target graph vertex is 5, a reusable sampling strategy can be used to extract 10 graph vertices from these 5 surrounding graph vertices as connected graph vertices, and some of the graph vertices may be extracted repeatedly.

[0090] After the sampling strategy is determined, in step 4313, one or more connected graph vertices are sampled from the surrounding graph vertices according to the sampling strategy. If the sampling strategy is a non-repetitive sampling strategy, a random number generator can be used to assign a random number to each surrounding graph vertex, and then the surrounding graph vertices are sorted according to the size of the random number, and a certain number of graph vertices with the highest sorting are selected as connected graph vertices. Let the set of surrounding graph vertices be , for each graph vertex v i Generate a random number r i , sort the graph vertices from small to large according to random numbers. If k connected graph vertices need to be extracted, the first k graph vertices after sorting are selected.

[0091] If the sampling strategy is a reusable sampling strategy, we can continuously generate random numbers and select graph vertices from the surrounding graph vertex set according to the index corresponding to the random number until a sufficient number of connected graph vertices are extracted. Let the surrounding graph vertex set be , each time generate a random integer j in the range [1, n], select the vertex v j As the connected graph vertices, repeat this process k times to obtain k connected graph vertices.

[0092] The representation vectors of the sampled connection graph vertices contain information about abnormal operation events or other related power equipment related to the target graph vertices. Integrating them with the representation vectors of the target graph vertices can more comprehensively reflect the operating status of the power equipment represented by the target graph vertices. After obtaining the connection graph vertices, vector integration is performed based on the representation vectors of these connection graph vertices and the representation vectors of the target graph vertices to generate the power equipment representation vector. The power equipment representation vector can comprehensively reflect the operating status and characteristics of the power equipment. By analyzing the power equipment representation vector, the similarities and differences between the power equipment can be found, providing a basis for the subsequent grouping of power equipment. If the representation vectors of two power equipment have a high similarity, it means that they have similarities in terms of operating status and abnormal triggering conditions, and may belong to the same type of equipment group. The similarity between the power equipment representation vectors can be calculated using methods such as cosine similarity and Euclidean distance.

[0093] In addition, the characterization vector of power equipment can also be used for fault prediction and risk assessment. By establishing a fault prediction model and taking the characterization vector of power equipment as input, the probability and type of fault of power equipment can be predicted. Machine learning algorithms, such as support vector machines and random forests, can be used to train a large amount of historical data to obtain an accurate fault prediction model. Let the characterization vector of power equipment be V and the fault prediction model be f(V), then the predicted fault probability P=f(V). The potential risk of power equipment can also be evaluated based on the characterization vector of power equipment to provide a reference for fire detection. For power equipment with higher risks, early warnings can be issued in time to remind relevant personnel to conduct inspections and maintenance to prevent the occurrence of safety accidents such as fires. The risk score of power equipment can be calculated based on certain features in the characterization vector of power equipment, such as the frequency and severity of abnormalities. Let the characterization vector of power equipment be V and the risk assessment function be g(V), then the risk score of power equipment is R=g(V).

[0094] In one implementation, step 433, integrating the target graph vertex characterization vector and the connection graph vertex characterization vector to obtain the power equipment characterization vector of the power equipment to be analyzed, includes: Step 4331: determining the number of graph vertex hops between the connection graph vertex and the target graph vertex in the power analysis graph; Step 4332: performing jump grouping on the connection graph vertices according to the number of graph vertex hops, obtaining a graph vertex group corresponding to each connection graph vertex, and determining a target connection graph vertex among the connection graph vertices according to the graph vertex grouping; Step 4333: Integrate the connection graph vertex characterization vector corresponding to the target connection graph vertex with the target graph vertex characterization vector to obtain the power equipment characterization vector of the power equipment to be analyzed.

[0095] In step 4331, the number of graph vertex hops between the connection graph vertex and the target graph vertex is determined in the power analysis graph. The number of graph vertex hops refers to the minimum number of edges passed from one graph vertex to another graph vertex in the power analysis graph, which reflects the distance and degree of association between two graph vertices. Taking a power analysis graph as an example, if the target graph vertex represents a transformer of a substation, and the connection graph vertex represents an over-temperature abnormal event related to the transformer, if only one edge passes from the transformer graph vertex to the over-temperature abnormal event graph vertex, then the number of graph vertex hops is 1; if two edges are required to reach it, then the number of graph vertex hops is 2.

[0096] In order to determine the number of graph vertex hops between the connection graph vertex and the target graph vertex, a breadth-first search (BFS) algorithm can be used. The breadth-first search algorithm is an algorithm used to traverse or search a tree or graph. It starts from the starting vertex and visits the vertices in the graph layer by layer. Taking the target graph vertex as the starting vertex, the breadth-first search algorithm is used to traverse the power analysis graph, and the number of layers when each connection graph vertex is visited is recorded. This number of layers is the number of graph vertex hops between the connection graph vertex and the target graph vertex.

[0097] After the number of graph vertex jumps is determined, in step 4332, the connection graph vertices are jump-grouped according to the number of graph vertex jumps, the graph vertex group corresponding to each connection graph vertex is obtained, and the target connection graph vertex is determined in the connection graph vertices according to the graph vertex grouping. Jump grouping refers to dividing the connection graph vertices into different groups according to the number of graph vertex jumps, which is similar to dividing into layers. The target graph vertex is in the first group (first layer), and the one with a jump number of 2 is in the second group (second layer), and so on. Through jump grouping, the distance relationship between the connection graph vertex and the target graph vertex can be more clearly understood, providing a basis for subsequent vector integration.

[0098] A hash table or array can be used to store the graph vertex grouping information of each connected graph vertex. For each connected graph vertex, it is assigned to the corresponding group according to its graph vertex hop count. Let the set of connected graph vertices be , the set of graph vertex hop counts is , create an array G, the index of the array represents the group number, and each element is a set used to store the connected graph vertices in the group. For each connected graph vertex v ci , add it to G[h i ]middle.

[0099] After completing the jump grouping, the target connected graph vertices are determined in the connected graph vertices according to the graph vertex grouping. The target connected graph vertices are those connected graph vertices that play an important role in the integration of the target graph vertex representation vector. The target connected graph vertices can be determined based on factors such as the level of the grouping and the importance of the connected graph vertices. For groups with lower levels (smaller hops), the connected graph vertices therein are more closely associated with the target graph vertices, and the connected graph vertices in these groups can be preferentially determined as the target connected graph vertices. The connected graph vertices can also be sorted according to the importance of the features of the connected graph vertices, such as the frequency and severity of abnormalities, and the connected graph vertices with the highest rankings are selected as the target connected graph vertices.

[0100] After the target connection graph vertex is determined, in step 4333, the connection graph vertex characterization vector corresponding to the target connection graph vertex is integrated with the target graph vertex characterization vector to obtain the power equipment characterization vector of the power equipment to be analyzed. The power equipment characterization vector can comprehensively reflect the operating status and characteristics of the power equipment. Integrating the information of the target connection graph vertex into the target graph vertex characterization vector can make the power equipment characterization vector more comprehensively reflect the actual situation of the power equipment. The weighted summation method can be used for vector integration, or the attention mechanism can be used for vector integration.

[0101] In one implementation, step 4333, integrating the connection graph vertex characterization vector corresponding to the target connection graph vertex with the target graph vertex characterization vector to obtain the power equipment characterization vector of the power equipment to be analyzed, includes: Step 43331: Integrate the connection graph vertex representation vector corresponding to the target connection graph vertex with the target graph vertex representation vector to obtain the integrated graph vertex representation vector corresponding to the current graph vertex group; Step 43332: if the level of the current graph vertex grouping is at the maximum number of layers of the preset graph vertex grouping, the integrated graph vertex representation vector is used as the target graph vertex representation vector, and the execution is jumped to the step of determining the target connected graph vertex in the connected graph vertices according to the graph vertex grouping, until the current graph vertex grouping reaches the maximum number of layers of the preset graph vertex grouping, and the target integrated graph vertex representation vector of the power equipment to be analyzed is obtained; Step 43333: perform a normalization operation on the target integrated graph vertex representation vector to obtain a compressed graph vertex representation vector, and use the compressed graph vertex representation vector as the power equipment representation vector of the power equipment to be analyzed.

[0102] In step 43331, the connection graph vertex characterization vector corresponding to the target connection graph vertex is integrated with the target graph vertex characterization vector to obtain the integrated graph vertex characterization vector corresponding to the current graph vertex group. In the power analysis graph, the target graph vertex represents the power equipment to be analyzed, and the target connection graph vertex is an important connection graph vertex that is closely related to the target graph vertex and is determined after grouping according to the number of graph vertex hops. Each target connection graph vertex has its corresponding connection graph vertex characterization vector, which contains the characteristic information of the abnormal operation event or related power equipment represented by the connection graph vertex.

[0103] Take the power analysis graph of a large factory's power system as an example. If the target graph vertex represents a key transformer, the target connection graph vertex may represent the transformer's over-temperature anomaly, over-current anomaly and other events. The target graph vertex representation vector may contain information such as the transformer's rated power and operating time, while the target connection graph vertex representation vector contains information such as the frequency of over-temperature anomaly and the current peak of over-current anomaly. The weighted summation method can be used for vector integration.

[0104] After obtaining the integrated graph vertex representation vector corresponding to the current graph vertex grouping, in step 43332, determine whether the level of the current graph vertex grouping is within the preset maximum number of levels of the graph vertex grouping. The preset maximum number of levels of the graph vertex grouping is a preset threshold used to control the depth of vector integration. If the level of the current graph vertex grouping is within the preset maximum number of levels of the graph vertex grouping, it means that the maximum depth of vector integration has been reached. At this time, the integrated graph vertex representation vector is used as the target graph vertex representation vector. At the same time, jump to execute the step of determining the target connected graph vertex in the connected graph vertices based on the graph vertex grouping until the current graph vertex grouping reaches the preset maximum number of levels of the graph vertex grouping, and obtain the target integrated graph vertex representation vector of the power equipment to be analyzed.

[0105] Continuing with the factory transformer as an example, assume that the maximum number of layers of the preset graph vertex grouping is 3. After completing the vector integration of the first layer (the layer where the target graph vertex is located) and the second layer of the target connection graph vertex, the integrated graph vertex representation vector corresponding to the current graph vertex grouping (second layer) is obtained. At this time, check whether the second layer is the maximum number of layers of the preset graph vertex grouping. If not, continue to determine the target connection graph vertex of the third layer, and integrate its connection graph vertex representation vector with the current integrated graph vertex representation vector to obtain a new integrated graph vertex representation vector. This cycle continues until the maximum number of layers of the preset graph vertex grouping is reached, and the target integrated graph vertex representation vector is obtained.

[0106] In step 43333, the target integrated graph vertex representation vector is standardized to obtain a compressed graph vertex representation vector, which is used as the power equipment representation vector of the power equipment to be analyzed. The standardization operation is to eliminate the dimensional differences of different features in the vector, so that each feature in the vector is comparable, and it also helps the subsequent machine learning algorithm to run more stably. The Z-score standardization method can be used. Let the target integrated graph vertex representation vector be , where the jth feature is , the mean of this feature is , the standard deviation is , then the jth feature after standardization The formula can be Calculated. The normalized vector It is the compressed graph vertex representation vector.

[0107] Through the above method, the connection graph vertex representation vector corresponding to the target connection graph vertex can be integrated with the target graph vertex representation vector. After multiple iterations until the preset maximum number of graph vertex grouping layers is reached, the target integrated graph vertex representation vector is obtained, and then it is standardized to finally obtain the power equipment representation vector of the power equipment to be analyzed.

[0108] In one implementation, step 500, based on the power device characterization vector, the power devices to be analyzed are grouped to obtain one or more pending power device groups, including: Step 510: determining the power equipment commonality metric value between the power equipment to be analyzed in the set of power equipment to be analyzed according to the power equipment characterization vector; Step 520: Generate an interconnection graph by taking the power equipment to be analyzed as a graph vertex according to the commonality metric value of the power equipment; Step 530: Group the graph vertices in the interconnection graph to obtain one or more pending power device groups.

[0109] In step 510, based on the power equipment characterization vector, the power equipment commonality measurement value between the power equipment to be analyzed in the power equipment set to be analyzed is determined. The power equipment characterization vector is obtained by mining the characterization information of the power analysis graph, and it comprehensively reflects the operating status and characteristics of the power equipment. The power equipment commonality measurement value is used to measure the similarity between different power equipment. The higher the similarity, the closer these power equipment are in terms of operating status, abnormal triggering conditions, etc.

[0110] Taking a power system as an example, the set of power equipment to be analyzed may include multiple transformers, distribution cabinets, motors, etc. Each power equipment has its corresponding power equipment characterization vector, which contains various characteristic information of the equipment, such as power, voltage, operating time, abnormal frequency, etc. The cosine similarity algorithm, Euclidean distance, Manhattan distance and other methods can be used to calculate the commonality measurement value of power equipment.

[0111] After determining the commonality metric value of the power equipment, in step 520, based on the commonality metric value of the power equipment, the power equipment to be analyzed is used as a graph vertex to generate an interconnection graph. The interconnection graph is an undirected graph, in which each graph vertex represents a power equipment, and the edges between the vertices represent the commonality metric value between the power equipment. The weight of the edge can be set to the commonality metric value of the power equipment. The larger the weight, the higher the similarity between the two power equipment. A graph database can be used to store and manage the interconnection graph. The graph database can efficiently process graph structure data and support fast graph query and analysis operations.

[0112] In step 530, the graph vertices in the interconnected graph are grouped to obtain one or more pending power equipment groups. The purpose of grouping is to divide similar power equipment into the same group for more targeted analysis and management. A community discovery algorithm, such as the Louvain algorithm, can be used to group the interconnected graph, by continuously moving the graph vertices to different communities to maximize the grouping effectiveness of the entire graph, that is, the modularity. Please refer to the above introduction for the formula, which will not be repeated here. When using the Louvain algorithm for grouping, each graph vertex is first assigned to a separate community, and then the current grouping effectiveness is calculated. Next, traverse each graph vertex, try to move it to an adjacent community, and calculate the grouping effectiveness after the move. If the grouping effectiveness after the move increases, move the vertex to the new community. This process is repeated until the grouping effectiveness no longer increases. After completing a round of vertex movement, each community is regarded as a new vertex, the interconnected graph is reconstructed, and the above process is repeated until the grouping effectiveness reaches the maximum value. Through the Louvain algorithm, the graph vertices in the interconnected graph can be divided into different communities, each of which corresponds to a pending power equipment group. The power equipment in these pending power equipment groups have high similarity, and they may have common characteristics in terms of operating status, abnormal triggering conditions, etc. For example, in the power system of a commercial building, all transformers may be divided into one pending power equipment group, and all motors may be divided into another pending power equipment group.

[0113] For each pending power equipment group, unified monitoring and analysis can be performed to predict the failure probability and risk level of the power equipment in the group. Targeted maintenance plans and emergency plans can be formulated based on the common characteristics of the power equipment in the group. If the power equipment in a pending power equipment group often has overtemperature anomalies, the temperature monitoring of the group equipment can be strengthened, and heat dissipation equipment and maintenance tools can be prepared in advance.

[0114] The pending power equipment group can be dynamically updated. With the continuous generation of new historical abnormal monitoring data, the operating status of the power equipment will change, and the similarity relationship between the power equipment will also change accordingly. It is necessary to regularly recalculate the commonality metric values ​​of the power equipment, update the interconnection map, and re-group the operation to ensure that the pending power equipment group can accurately reflect the latest status of the power equipment. The pending power equipment group can also be used for knowledge discovery and decision support. By analyzing the common characteristics and abnormal patterns of the power equipment in the pending power equipment group, potential fault hazards and operating laws can be discovered. Based on these findings, decision-making basis can be provided for the optimization and upgrading of the power system, such as reasonably adjusting the operating parameters of the equipment, planning the replacement and expansion of the equipment, etc. The information of the pending power equipment group can be combined with fire detection. For some high-risk pending power equipment groups, fire monitoring of the power equipment in the group can be strengthened, and more fire equipment such as temperature sensors and smoke alarms can be installed to timely discover and deal with potential fire hazards. Different levels of fire warning strategies can be formulated according to the failure probability and risk level of the power equipment in the group to improve the pertinence and effectiveness of fire detection.

[0115] Through the above method, the commonality measurement value of the power equipment can be determined according to the power equipment characterization vector, an interconnection map can be generated according to the commonality measurement value, and the map vertices in the interconnection map can be grouped to obtain one or more pending power equipment groups.

[0116] In one implementation, step 530, grouping the graph vertices in the interconnection graph to obtain one or more pending power device groups, includes: Step 531: group each graph vertex in the interconnected graph into a corresponding dense graph vertex group, and determine the basic grouping validity of the dense graph vertex group; Step 532: Determine the adjacent graph vertices corresponding to each graph vertex in the interconnected graph, and group the graph vertices into target dense graph vertex groups corresponding to the adjacent graph vertices; Step 533: Determine the current grouping validity of the target tight graph vertex group, and determine the power equipment to be analyzed corresponding to one or more target tight graph vertex groups in the set of power equipment to be analyzed based on the basic grouping validity and the current grouping validity, and obtain the pending power equipment group.

[0117] In step 531, each graph vertex in the interconnected graph is grouped into a corresponding close graph vertex group, and the basic grouping validity of the close graph vertex group is determined. A close graph vertex group can be understood as a node community, which is a collection of vertices in the interconnected graph. The connections between these vertices are relatively close, representing power equipment with similar characteristics or relationships.

[0118] In order to group the graph vertices into the corresponding dense graph vertex groups, a clustering algorithm such as a spectral clustering algorithm can be used. The spectral clustering algorithm is based on the spectral theory of the graph. By performing eigendecomposition on the Laplacian matrix of the graph, the vertices are mapped to a low-dimensional space, and then clustered in the low-dimensional space. First, the Laplacian matrix L=DA of the interconnected graph is calculated, where D is the degree matrix and A is the adjacency matrix. Then, the Laplacian matrix is ​​eigendecomposed to obtain the eigenvector. The eigenvectors corresponding to the first k smallest non-zero eigenvalues ​​are selected to form a matrix U, and the row vector of each vertex in U is used as its representation in the low-dimensional space. Finally, the K-means clustering algorithm is used to cluster these low-dimensional vectors to divide the vertices into different dense graph vertex groups. The concept of modularity is needed to determine the basic grouping validity of dense graph vertex groups. Modularity is an indicator to measure the quality of community division in the graph. Its calculation formula refers to the introduction of the aforementioned step 500. The closer the value of modularity is to 1, the more reasonable the grouping is, the tighter the connection within the dense graph vertex group is, and the sparser the connection between the groups is. According to the current grouping situation, the corresponding values ​​are substituted into the grouping validity formula for calculation to obtain the basic grouping validity of the compact graph vertex group.

[0119] After completing the collection of graph vertices and determining the validity of the basic grouping, in step 532, the adjacent graph vertices corresponding to each graph vertex are determined in the interconnected graph, and the graph vertices are collected into the target tight graph vertex group corresponding to the adjacent graph vertices. Adjacent graph vertices refer to other graph vertices directly connected to the current graph vertex. In the power supply network interconnection graph, if a graph vertex represents a transformer, then the graph vertices corresponding to other power equipment directly connected to the transformer in the circuit are its adjacent graph vertices.

[0120] The adjacent graph vertex corresponding to each graph vertex can be determined by querying the adjacency matrix of the interconnected graph. For each graph vertex i, check the i-th row of the adjacency matrix A. If , it means that vertex j is an adjacent graph vertex of vertex i. Then, try to group graph vertex i into the target dense graph vertex group corresponding to its adjacent graph vertex j. During the grouping process, it is necessary to recalculate the grouping validity to evaluate whether the overall grouping validity is improved after moving vertex i to the new group.

[0121] In step 533, the current group validity of the target tight graph vertex group is determined, and based on the basic group validity and the current group validity, one or more target tight graph vertex groups corresponding to the power equipment to be analyzed are determined in the set of power equipment to be analyzed to obtain the pending power equipment group. After the graph vertex is moved to the target tight graph vertex group corresponding to the adjacent graph vertex, the current group validity of the target tight graph vertex group is calculated again using the group validity formula.

[0122] By comparing the validity of the basic grouping with the validity of the current grouping, we can determine whether the grouping is optimized. If the validity of the current grouping is greater than the validity of the basic grouping, it means that after moving the graph vertex to the new group, the grouping is more reasonable, and this move will be retained; if the validity of the current grouping is less than the validity of the basic grouping, it means that this move has reduced the quality of the grouping, and the graph vertex will be restored to the original group.

[0123] In the process of continuously moving graph vertices to the target dense graph vertex groups corresponding to adjacent graph vertices and comparing the grouping effectiveness, multiple iterations will be performed until the grouping effectiveness is no longer improved. In this process, the composition of the target dense graph vertex groups will be continuously updated, making the connections between graph vertices in each group more dense and the connections between groups more sparse.

[0124] When the effectiveness of the grouping reaches a stable state, the power equipment to be analyzed corresponding to one or more target tight graph vertex groups are determined in the set of power equipment to be analyzed, and the pending power equipment group is obtained. The power equipment represented by the graph vertices in each target tight graph vertex group has a high similarity, and they may have common characteristics in terms of operating status, abnormal triggering conditions, etc. In the power supply network, a target tight graph vertex group may contain power equipment of the same type, the same area, or with similar failure modes.

[0125] The pending power equipment group can be used for fault prediction and risk assessment. By analyzing the common characteristics and historical fault data of the power equipment in the group, a fault prediction model can be established to predict the probability and type of failure of the power equipment in the group. For a pending power equipment group that often has overcurrent anomalies, it can be predicted that other power equipment in the group may also have overcurrent failures, and timely warnings can be issued. The information of the pending power equipment group can also be combined with fire detection. For some high-risk pending power equipment groups, fire monitoring of the power equipment in the group can be strengthened, and more fire-fighting equipment such as temperature sensors and smoke alarms can be installed to timely discover and deal with potential fire hazards. Different levels of fire warning strategies can be formulated according to the failure probability and risk level of the power equipment in the group to improve the pertinence and effectiveness of fire detection.

[0126] In one implementation, step 533, based on the validity of the basic grouping and the validity of the current grouping, determines the power devices to be analyzed corresponding to one or more target tight graph vertex groups in the set of power devices to be analyzed, and obtains the pending power device group, including: Step 5331: determine the error between the basic grouping validity and the current grouping validity of the target dense graph vertex group, and obtain the grouping validity error; Step 5332: iterating the graph vertices in the target dense graph vertex group according to the grouping validity error to obtain the iterated dense graph vertex group, and performing group strength verification on the iterated dense graph vertex group; Step 5333: If the iterative compact graph vertex group meets the strength requirement, the iterative compact graph vertex group is used as a graph vertex to generate an interconnection graph, and the current interconnection graph is obtained. According to the current interconnection graph, one or more compact graph vertex groups after iteration are determined in the set of power equipment to be analyzed to obtain the power equipment group to be determined. For example, if all the graph vertex groups are grouped into the target compact graph vertex group, and the grouping validity reaches the maximum value, it is determined that the iterative compact graph vertex group meets the strength requirement. Otherwise, jump to the step of grouping the graph vertex groups into the target compact graph vertex groups corresponding to the adjacent graph vertex groups until the iterative compact graph vertex group meets the strength requirement. When the iterative compact graph vertex group meets the strength requirement, the method of using the iterative compact graph vertex group as a graph vertex to generate an interconnection graph is, for example, to merge all the graph vertices in the iterative compact graph vertex group into one graph vertex, that is, each iterative compact graph vertex group is regarded as one graph vertex. Then, the edges between the compact graph vertex groups remain unchanged, and the current interconnection graph is generated.

[0127] In step 5331, the error between the basic group validity and the current group validity of the target tight graph vertex group is determined to obtain the group validity error. The basic group validity is the group validity value calculated using the group validity formula (refer to the above content) after the graph vertices are initially grouped into tight graph vertex groups, and the current group validity is the group validity value calculated again using the same group validity formula after the graph vertices are moved to the target tight graph vertex group corresponding to the adjacent graph vertex.

[0128] Taking a power supply network interconnection graph as an example, during the initial grouping, the graph vertices representing the power equipment in different substations are grouped into different compact graph vertex groups, and the basic grouping validity Q1 is calculated. Then, try to move some graph vertices to the target compact graph vertex groups corresponding to their adjacent graph vertices, calculate the grouping validity again, and get the current grouping validity Q2. Grouping validity error The formula can be The grouping validity error reflects the degree of change in grouping validity after grouping adjustment. The larger the error, the greater the impact of grouping adjustment on the overall grouping quality.

[0129] After obtaining the group validity error, in step 5332, the graph vertices in the target tight graph vertex group are iterated according to the group validity error to obtain the iterated tight graph vertex group, and the group strength of the iterated tight graph vertex group is verified. It is determined whether it is necessary to continue to move and group the graph vertices according to the group validity error. If the group validity error is large, it means that there is still room for group optimization. The graph vertices will continue to be moved to the target tight graph vertex group corresponding to the adjacent graph vertex, and the group validity will be recalculated until the group validity error is less than a preset threshold.

[0130] During each iteration, traverse each graph vertex in the target tight graph vertex group, try to move it to different groups, and compare the grouping effectiveness before and after the move. If the grouping effectiveness increases after the move, keep this move; otherwise, restore the graph vertex to the original group. Through continuous iteration, the grouping can be gradually optimized so that the connection between the graph vertices in each tight graph vertex group is tighter and the connection between groups is sparser. After obtaining the tight graph vertex group after iteration, the group strength is verified. The group strength verification is to ensure that the tight graph vertex group after iteration has sufficient stability and rationality. Verification is performed by checking the average connection strength between the graph vertices in the group, the size of the group and other indicators. If the average connection strength between the graph vertices in the group is low, it means that the relationship between the vertices in the group is not tight enough, and the group may need further adjustment; if the size of the group is too large or too small, it may not meet the actual grouping requirements.

[0131] In step 5333, it is determined whether the dense graph vertex group after iteration meets the strength requirement. If it meets the strength requirement, the dense graph vertex group after iteration is used as the graph vertex to generate an interconnection graph, and the current interconnection graph is obtained. According to the current interconnection graph, the power equipment to be analyzed corresponding to one or more dense graph vertex groups after iteration is determined in the set of power equipment to be analyzed, and the pending power equipment group is obtained. It can be determined whether the dense graph vertex group after iteration meets the strength requirement according to multiple conditions. If all the graph vertices are collected into the target dense graph vertex group, and the group validity reaches the maximum value, that is, the group validity no longer increases, it can be determined that the dense graph vertex group after iteration meets the strength requirement.

[0132] When the tight graph vertex group after iteration meets the strength requirement, the tight graph vertex group after iteration is used as the graph vertex to generate the interconnection graph. The specific method is to merge all the graph vertex in the tight graph vertex group after iteration into one graph vertex, that is, each tight graph vertex group after iteration is regarded as a graph vertex, and then, the edges between the tight graph vertex groups remain unchanged to generate the current interconnection graph. In the interconnection graph of the power supply network, if after iteration, several tight graph vertex groups representing some power equipment in different substations meet the strength requirements, these groups will be merged into one graph vertex, and the weights of the edges between these new graph vertices are determined according to the connection between the original groups.

[0133] According to the current interconnection graph, the power equipment to be analyzed corresponding to one or more iterative close graph vertex groups are determined in the power equipment set to be analyzed to obtain the pending power equipment group. The corresponding power equipment can be searched in the power equipment set to be analyzed according to the information of the iterative close graph vertex group represented by each graph vertex in the current interconnection graph. If a graph vertex represents an iterative close graph vertex group containing multiple transformers, these transformers will be found in the power equipment set to be analyzed and regarded as a pending power equipment group.

[0134] These pending power equipment groups are of great significance for the management and maintenance of power equipment. A personalized monitoring and maintenance strategy can be formulated for each pending power equipment group. For a pending power equipment group containing multiple old transformers, the temperature, oil quality and other parameters of the transformers in the group can be strengthened to formulate maintenance and replacement plans in advance.

[0135] The pending power equipment group can also be used for fault prediction and risk assessment. By analyzing the common characteristics and historical fault data of the power equipment in the group, a fault prediction model can be established to predict the probability and type of faults of the power equipment in the group. For a pending power equipment group that often has overcurrent anomalies, it can be predicted that other power equipment in the group may also have overcurrent faults, and timely warnings can be issued.

[0136] In one implementation, in step 5333, according to the current interconnection graph, one or more power devices to be analyzed corresponding to the vertex groups of the iterative dense graph are determined in the set of power devices to be analyzed, and the pending power device group is obtained, including: Step 53331: determine the current interconnection graph as the interconnection graph, and jump to the step of grouping each graph vertex in the interconnection graph into a corresponding dense graph vertex group until the current interconnection graph stops changing, thereby obtaining a target interconnection graph; Step 53332: Determine in the target interconnected graph the group label of the current dense graph vertex group corresponding to each graph vertex; Step 53333: According to the group label, one or more power devices to be analyzed in the current tight graph vertex group are determined in the power device set to be analyzed to obtain a pending power device group. For example, if the current interconnection graph stops changing, it means that the tight graph vertex group in the current interconnection graph has stabilized. At this time, the group label of the current tight graph vertex group corresponding to each graph vertex can be determined in the obtained target interconnection graph. The group label here can be the graph vertex label of the graph vertex included in the current tight graph vertex group. According to the graph vertex label, the power devices to be analyzed included in the current tight graph vertex group can be determined in the power device set to be analyzed, and the power devices to be analyzed in the current tight graph vertex group can be used as a pending power device group, thereby obtaining a pending power device group corresponding to each current tight graph vertex group in the target interconnection graph.

[0137] In step 53331, the current interconnection graph is determined as the interconnection graph, and the step of grouping each graph vertex in the interconnection graph into the corresponding dense graph vertex group is executed until the current interconnection graph stops changing, and the target interconnection graph is obtained. The current interconnection graph is a new interconnection graph generated by using the dense graph vertex group after iteration as the graph vertex when the dense graph vertex group after iteration meets the strength requirement. In this new interconnection graph, each graph vertex represents a dense graph vertex group after iteration, and the edges between vertices represent the connection relationship between the groups.

[0138] In the process of repeated grouping operations, the grouping effectiveness is continuously calculated and the changes in grouping effectiveness after each grouping adjustment are compared. If after multiple adjustments, the grouping effectiveness no longer increases, and the grouping of the graph vertices no longer changes, that is, the current interconnection graph stops changing, then the target interconnection graph is obtained. The target interconnection graph is a graph that reaches a stable state after multiple optimizations, which can more accurately reflect the similarity relationship and group structure between power equipment.

[0139] After obtaining the target interconnection graph, in step 53332, the group label of the current tight graph vertex group corresponding to each graph vertex is determined in the target interconnection graph. The group label is information used to identify each tight graph vertex group, and it can be the graph vertex label of the graph vertex contained in the current tight graph vertex group. In the target interconnection graph of the power system of the industrial park, a graph vertex represents a tight graph vertex group composed of multiple transformers, and the group label of the group can be a set of numbers of these transformers. By traversing each graph vertex in the target interconnection graph, the graph vertex labels of the graph vertices in the tight graph vertex group represented by it can be obtained, and these labels can be combined as the group label of the group.

[0140] In step 53333, one or more power devices to be analyzed in the current dense graph vertex group are determined in the power device set to be analyzed according to the group label, and the pending power device group is obtained. The power device set to be analyzed is a series of power devices that need to be analyzed initially.

[0141] Through the above method, it is possible to obtain the target interconnection map through multiple iterations and optimizations based on the current interconnection map, determine the group label, and find the corresponding pending power equipment group in the set of power equipment to be analyzed.

[0142] Figure 2 A hardware entity diagram of a computer system provided in an embodiment of the present application is shown in FIG. Figure 2 As shown, the hardware entity of the computer system 1000 includes: a processor 1001 and a memory 1002, wherein the memory 1002 stores a computer program that can be run on the processor 1001, and the processor 1001 implements the steps in the method of any of the above embodiments when executing the program.

[0143] The memory 1002 stores computer programs that can be run on the processor. The memory 1002 is configured to store instructions and applications executable by the processor 1001. It can also cache data to be processed or processed by the processor 1001 and various modules in the computer system 1000 (for example, image data, audio data, voice communication data, and video communication data). This can be achieved through flash memory (FLASH) or random access memory (Random Access Memory, RAM).

[0144] When the processor 1001 executes the program, the steps of any of the above-mentioned fire detection methods based on power data analysis are implemented. The processor 1001 generally controls the overall operation of the computer system 1000.

[0145] The above is only an implementation method of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A fire detection method based on power data analysis, characterized in that: Applied to a computer system, the method comprises: Acquire a set of power equipment to be analyzed and a historical abnormal monitoring data set of each power equipment to be analyzed in the set of power equipment to be analyzed; Determine one or more abnormal operation events in the historical abnormal monitoring data set, wherein the abnormal operation event is an abnormal classification corresponding to an abnormal state of the power equipment to be analyzed; According to the historical abnormal monitoring data set, a power analysis graph is generated by taking the power equipment to be analyzed and the abnormal operation event as graph vertices, wherein the power analysis graph is used to indicate the triggering connection between the power equipment to be analyzed and the abnormal operation event; Performing characterization information mining on the power analysis graph to obtain a power equipment characterization vector of the power equipment to be analyzed; The power equipment to be analyzed is grouped according to the power equipment characterization vector to obtain one or more pending power equipment groups, and a target power equipment group is determined in the pending power equipment groups so as to perform fire detection on the power equipment in the target power equipment group.

2. The fire detection method based on power data analysis according to claim 1 is characterized in that: Determining one or more abnormal operation events in the historical abnormal monitoring data set includes: Extracting the abnormality category data of the power equipment to be analyzed from the historical abnormality monitoring data set; Extracting the abnormal occurrence time and abnormal electrical nodes of the power equipment to be analyzed from the abnormal category data; Determine one or more abnormal operation events corresponding to the power equipment to be analyzed according to the abnormal occurrence time and the abnormal electrical node; The step of generating a power analysis graph by taking the power equipment to be analyzed and the abnormal operation events as graph vertices according to the historical abnormal monitoring data set includes: Determining in the historical abnormal monitoring data set to obtain triggering statistical information of the power equipment to be analyzed and the abnormal operation event; Determining a triggering relationship between the power equipment to be analyzed and the abnormal operation event according to the abnormal classification information of the abnormal operation event and the triggering statistical information; According to the triggering connection, the power equipment to be analyzed and the abnormal operation events are used as graph vertices to generate a power analysis graph.

3. The fire detection method based on power data analysis according to claim 1 or 2, characterized in that: The performing characterization information mining on the power analysis graph to obtain the power equipment characterization vector of the power equipment to be analyzed includes: Extracting a graph vertex characterization vector of each graph vertex from the power analysis graph; Projecting the graph vertex representation vectors to a unified representation domain to obtain basic graph vertex representation vectors; The basic graph vertex characterization vectors are integrated to obtain the power equipment characterization vector of the power equipment to be analyzed.

4. The fire detection method based on power data analysis according to claim 3 is characterized in that: The step of projecting the graph vertex representation vector to a unified representation domain to obtain a basic graph vertex representation vector includes: Obtaining a preset unified projection standard dimension and an actual vector dimension of each of the graph vertex representation vectors; Determine the transformation matrix corresponding to the graph vertex representation vector according to the preset unified projection standard dimension and the actual vector dimension; According to the transformation matrix, the graph vertex representation vector is projected onto the graph vertex representation vector corresponding to the preset unified projection standard dimension to obtain a basic graph vertex representation vector; The step of integrating the base graph vertex characterization vectors to obtain the power equipment characterization vector of the power equipment to be analyzed includes: Determine the graph vertex corresponding to the power equipment to be analyzed in the graph vertices to obtain a target graph vertex, and sample one or more connection graph vertices of the target graph vertex in the power analysis graph; Determine the target graph vertex representation vector corresponding to the target graph vertex and the connection graph vertex representation vector corresponding to the connection graph vertex in the base graph vertex representation vector; The target graph vertex characterization vector and the connection graph vertex characterization vector are integrated to obtain the power equipment characterization vector of the power equipment to be analyzed.

5. The fire detection method based on power data analysis according to claim 4 is characterized in that: The step of sampling one or more connection graph vertices of the target graph vertex in the power analysis graph includes: Determining the number of graph vertices of the surrounding graph vertices adjacent to the target graph vertex in the power analysis graph; Determine, according to the number of graph vertices, a sampling strategy for connection graph vertices of the target graph vertices; According to the sampling strategy, one or more connection graph vertices are sampled from the peripheral graph vertices; Wherein, the step of determining a sampling strategy for connection graph vertices of the target graph vertices according to the number of graph vertices includes: Compare the number of graph vertices with a reference value. If the number of graph vertices is greater than the reference value, determine that the sampling strategy is a non-repetitive sampling strategy; if the number of graph vertices is less than or equal to the reference value, determine that the sampling strategy is a reusable sampling strategy; if the sampling strategy is a non-repetitive sampling strategy, directly sample the surrounding graph vertices to obtain the connected graph vertices corresponding to the reference value; if the sampling strategy is a reusable sampling strategy, sample the surrounding graph vertices using the reusable sampling strategy to obtain the connected graph vertices corresponding to the reference value; The step of integrating the target graph vertex characterization vector and the connection graph vertex characterization vector to obtain the power equipment characterization vector of the power equipment to be analyzed includes: Determining the number of graph vertex hops between the connection graph vertex and the target graph vertex in the power analysis graph; According to the number of graph vertex jumps, the connection graph vertices are jumped grouped to obtain a graph vertex group corresponding to each of the connection graph vertices, and according to the graph vertex group, a target connection graph vertex is determined among the connection graph vertices; The connection graph vertex characterization vector corresponding to the target connection graph vertex is integrated with the target graph vertex characterization vector to obtain the power equipment characterization vector of the power equipment to be analyzed.

6. The fire detection method based on power data analysis according to claim 5 is characterized in that: The step of integrating the connection graph vertex characterization vector corresponding to the target connection graph vertex with the target graph vertex characterization vector to obtain the power equipment characterization vector of the power equipment to be analyzed includes: Integrate the connection graph vertex representation vector corresponding to the target connection graph vertex with the target graph vertex representation vector to obtain an integrated graph vertex representation vector corresponding to the current graph vertex group; If the level of the current graph vertex grouping is at the maximum number of layers of the preset graph vertex grouping, the integrated graph vertex characterization vector is used as the target graph vertex characterization vector, and the step of determining the target connected graph vertex in the connected graph vertices according to the graph vertex grouping is jumped to be executed until the current graph vertex grouping reaches the maximum number of layers of the preset graph vertex grouping, thereby obtaining the target integrated graph vertex characterization vector of the power equipment to be analyzed; A normalization operation is performed on the target integrated graph vertex representation vector to obtain a compressed graph vertex representation vector, and the compressed graph vertex representation vector is used as the power equipment representation vector of the power equipment to be analyzed.

7. The fire detection method based on power data analysis according to claim 1 or 2, characterized in that: The step of grouping the power equipment to be analyzed according to the power equipment characterization vector to obtain one or more pending power equipment groups includes: Determining, based on the power equipment characterization vector, a power equipment commonality metric value between the power equipment to be analyzed in the set of power equipment to be analyzed; According to the commonality metric value of the power equipment, the power equipment to be analyzed is used as a graph vertex to generate an interconnection graph; grouping each graph vertex in the interconnected graph into a corresponding dense graph vertex group, and determining a base grouping validity of the dense graph vertex group; Determine adjacent graph vertices corresponding to each graph vertex in the interconnected graph, and group the graph vertices into target dense graph vertex groups corresponding to the adjacent graph vertices; Determine the current grouping validity of the target tight graph vertex group, and determine the power equipment to be analyzed corresponding to one or more target tight graph vertex groups in the set of power equipment to be analyzed based on the basic grouping validity and the current grouping validity, to obtain the pending power equipment group.

8. The fire detection method based on power data analysis according to claim 7 is characterized in that: The step of determining the power devices to be analyzed corresponding to one or more target dense graph vertex groups in the power device set to be analyzed based on the validity of the basic group and the validity of the current group, and obtaining the pending power device group, comprises: Determine the error between the basic grouping validity and the current grouping validity of the target dense graph vertex group to obtain a grouping validity error; Iterating the graph vertices in the target dense graph vertex group according to the grouping validity error to obtain the iterated dense graph vertex group, and performing group strength verification on the iterated dense graph vertex group; If the iterated tight graph vertex group meets the strength requirement, the iterated tight graph vertex group is used as the graph vertex to generate an interconnection graph to obtain the current interconnection graph, and based on the current interconnection graph, one or more power devices to be analyzed corresponding to the iterated tight graph vertex group are determined in the set of power devices to be analyzed to obtain the pending power equipment group.

9. The fire detection method based on power data analysis according to claim 8, characterized in that: The step of determining, based on the current interconnection graph, one or more power devices to be analyzed corresponding to the vertex groups of the iterative dense graph from the set of power devices to be analyzed, and obtaining the pending power device group, comprises: Determine the current interconnection graph as the interconnection graph, and jump to the step of grouping each graph vertex in the interconnection graph into a corresponding dense graph vertex group until the current interconnection graph stops changing, thereby obtaining a target interconnection graph; Determine in the target interconnected graph the group label of the current dense graph vertex group corresponding to each graph vertex; According to the group label, one or more power devices to be analyzed within the current dense graph vertex group are determined in the set of power devices to be analyzed to obtain the pending power device group.

10. A computer system comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, wherein: When the processor executes the program, the steps in the method according to any one of claims 1 to 9 are implemented.

Images