Regional operation state recognition method based on power multi-source data fusion

Abstract

This invention relates to a method for identifying regional operation status based on multi-source power data fusion, comprising the following steps: S1: Obtaining multi-source heterogeneous data from relevant business systems of the regional power grid and establishing a unified data object model to obtain a standardized multi-source dataset; S2: Constructing multi-dimensional operation features based on the standardized multi-source dataset; S3: Performing multi-level data fusion based on the multi-dimensional operation features to construct a regional status profile; S4: Obtaining regional operation status identification results based on the regional operation status identification model according to the regional status profile; S5: Performing risk tracing based on the power operation knowledge graph and the regional operation status identification results; S6: Combining the regional operation status identification results and the risk tracing results, performing a comprehensive risk score to obtain the final regional operation status level. This invention effectively improves the accuracy of regional power grid operation status identification.

Description

Technical Field

[0001] This invention relates to the field of power grid status identification, and in particular to a method for identifying regional operation status based on the fusion of multi-source power data. Background Technology

[0002] With the continuous advancement of new power system construction, the operational characteristics of regional power grids are undergoing significant changes. On the one hand, the large-scale integration of new entities such as distributed photovoltaic, wind power, energy storage, and electric vehicle charging facilities is gradually transforming regional power grids from a traditional unidirectional power supply model to a multi-faceted interactive model involving power sources, grids, loads, and storage. On the other hand, user electricity consumption behaviors are becoming more diverse, with increasing load peak-to-valley differences, random fluctuations, and localized clustering. Coupled with factors such as extreme weather, equipment aging, and power outages, the operational status of regional power grids exhibits characteristics of multi-factor coupling, multi-scale changes, and dynamic evolution. Therefore, accurately, timely, and comprehensively identifying the operational status of regional power grids has become a crucial technical issue for ensuring the safe and stable operation of the power grid, improving power supply reliability, and supporting lean operation and maintenance.

[0003] Current regional power grid operation status analysis typically relies on data from multiple business systems, including dispatch automation, distribution automation, electricity consumption information collection, equipment asset management, fault repair, GIS, marketing management, and meteorological monitoring. However, these data differ significantly in their sources, formats, temporal granularity, spatial granularity, and business object coding. For example, the dispatch system focuses on real-time operation data of the main grid and critical equipment; the distribution automation system focuses on feeder, switch, and transformer operation data; the electricity consumption information collection system focuses on user-side load and electricity consumption data; the equipment asset system focuses on equipment ledgers, defects, and maintenance records; the GIS system focuses on spatial location and topological relationships; the fault repair system focuses on power outage events and work order information; and the meteorological system provides information on external environmental changes. Due to the lack of a unified data object model and association mechanism among these systems, it is difficult to accurately correspond equipment, lines, transformer areas, users, regions, events, and time, resulting in prominent data silos and hindering the overall assessment of the regional operation status. Summary of the Invention

[0004] To address the aforementioned problems, the present invention aims to provide a regional operation status identification method based on the fusion of multi-source power data, which effectively improves the accuracy of regional power grid operation status identification.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A regional operation status identification method based on multi-source power data fusion includes the following steps: S1: Obtain multi-source heterogeneous data from regional power grid related business systems, establish a unified data object model, and uniformly encode equipment, lines, transformer areas, users, regions, events, and times from different systems to obtain a standardized multi-source dataset; S2: Based on standardized multi-source datasets, data from different time scales, spatial scales, and business objects are linked into a unified analysis framework, and multi-dimensional operational features are constructed. S3: Based on multi-dimensional operational characteristics, perform multi-level data fusion to construct a regional status profile; S4: Based on the regional status profile and the regional operation status recognition model, obtain the regional operation status recognition results; S5: Based on the power operation knowledge graph, risk tracing is performed according to the regional operation status identification results; S6: Combining the regional operation status identification results and risk tracing results, a comprehensive risk score is performed to obtain the final regional operation status level.

[0006] Furthermore, a unified data object model is established to uniformly encode devices, lines, transformer substations, users, regions, events, and times from different systems, resulting in a standardized multi-source dataset, as detailed below: Objects in regional power grid-related business systems are categorized into standard object types—equipment, lines, distribution areas, users, regions, events, and time—based on power grid business semantics. Through this object abstraction, data from different systems with varying field names, data structures, and business granularities are uniformly converted into standard objects. Based on a unified data object model, devices, lines, transformer substations, users, regions, events, and times in different systems are uniformly coded. When the same device has different numbers in different systems, a mapping method is used with the assistance of device master data verification, GIS topology constraints, and name matching algorithms to generate a unique device identifier. After generating a unique device identifier, it is written back to the standardized data object model, thereby forming a standardized multi-source dataset.

[0007] Furthermore, for situations where the same device has different IDs in different systems, a mapping method is used with the assistance of device master data verification, GIS topology constraints, and name matching algorithms to generate a unique device identifier. Specifically, the device master data is used as a baseline database to extract key attributes and compare them with device records in each business system; the upstream and downstream connections, spatial location relationships, and power supply attribution relationships between devices are verified using GIS topology to determine whether device records in different systems point to the same physical device; for devices with different names, abbreviations, aliases, or historical IDs, a name matching algorithm is used to calculate similarity; based on a comprehensive score of attribute similarity, topology consistency, spatial distance, and name similarity, the device mapping relationship is determined, and a globally unique device identifier is generated for the confirmed matching device.

[0008] Furthermore, based on standardized multi-source datasets, data from different time scales, spatial scales, and business objects are correlated into a unified analysis framework, and multi-dimensional operational features are constructed. Specifically, data with different sampling frequencies and occurrence forms are converted into time-aligned datasets indexed by a unified time window; after unifying the time scale, data from different spatial scales and business objects are mapped to a unified regional analysis unit; multi-dimensional operational features for regional operational status identification are constructed, including at least load features, voltage features, equipment load features, power supply reliability features, new energy impact features, meteorological risk features, and equipment health features.

[0009] Furthermore, the regional status profile includes regional load profile, voltage quality profile, equipment stress profile, power supply reliability profile, new energy impact profile, meteorological risk profile, and equipment health profile. After forming multiple thematic profiles, regional-level fusion is carried out to construct a regional status profile that can comprehensively represent the regional operating status. The regional status profile takes the regional analysis unit as the object and the analysis time window as the index, and unifies the profile results of load, voltage, equipment, reliability, new energy, meteorology, and health to form a structured profile result that includes profile index values, profile level, change trend, anomaly label, and dominant influencing factors.

[0010] Furthermore, based on the regional status profile and the regional operational status recognition model, the regional operational status recognition results are obtained, as follows: Using each region-time window as the basic identification object, the regional load profile, voltage quality profile, equipment stress profile, power supply reliability profile, new energy impact profile, meteorological risk profile, and equipment health profile are converted into a state feature vector that the model can identify. Let the set of regions be denoted as . The time window set is For any region r i In time window tj The regional status profile can be represented as follows: ; in, This represents a regional load profile. This represents a voltage quality profile. This represents a pressure profile of the equipment. This indicates a power supply reliability profile. This depicts the impact profile of new energy sources. This represents a weather risk profile. This represents a device health profile; the indicators from various profiles are concatenated to obtain the region status recognition input vector: ; Where d represents the input feature dimension, x i,j,d Representing region r i In time window t j The d-th image feature value; Normalize the input features: ; in, and Let represent the minimum and maximum values ​​of the k-th feature in the historical samples, respectively, and ε be a very small positive number to prevent the denominator from being zero; thus, the standardized input vector is obtained: ; A regional operational status identification model is established to classify and identify regional operational status. Regional operational status can be divided into four levels: normal, watchful, warning, abnormal, and severe abnormal, denoted as: C = {c1, c2, c3, c4, c5}; Where c1 represents normal status, c2 represents status of concern, c3 represents warning status, c4 represents abnormal status, and c5 represents severe abnormal status; the regional operation status identification model is represented as: ; Where F(·) represents the regional operation status identification model, Θ represents the model parameters, and Y... i,j Representing region r i In time window t j The state recognition results are as follows.

[0011] Furthermore, the regional operational status identification model uses a multi-layer neural network for identification, and its calculation process is expressed as follows: ; in, W represents the output of the l-th hidden layer.(l) and b (l) Let z represent the weight matrix and bias term of the l-th layer, respectively; σ(·) represents the non-linear activation function; L represents the number of hidden layers; z i,j This indicates the unnormalized score of the output layer; The probability of each state category is obtained using the Softmax function: ; Where, p i,j,q Representing region r i In time window t j Belongs to state c of class q q The probability, z i,j,q This represents the score of the output layer corresponding to the q-th state; The final state category is represented as:

[0012] This yields preliminary identification results of the region's operational status; Combining temporal trends and threshold correction to output recognition results, let region r i The model output probability within the most recent M time windows is Then the temporal smoothing probability of the q-th state is: ; Where, α m This represents the time weight of the m-th historical time window; After obtaining the smoothing probability, calculate the regional comprehensive anomaly tendency value: ; Where, β q The outlier weights represent the q-th class of states; The identification results are graded based on the overall abnormal tendency value: ; Where τ1, τ2, τ3, and τ4 are the threshold values ​​for classifying state levels; The final output is the regional operational status identification result, including the region number, time window, status category, probability of each status, comprehensive anomaly tendency value, and main contribution profile. The output result is represented as follows: ; Among them, G i,j This represents the set of images that contribute significantly to the recognition results, calculated based on the contribution of image features: ; Among them, Ω u Let ω represent the feature set corresponding to the u-th type of image.k g represents the importance weight of the k-th feature in the model. i,j,u This represents the contribution of the u-th type of image to the region state recognition result. Furthermore, based on the power operation knowledge graph and the regional operation status identification results, risk tracing is performed, as follows: Using region ID, time window, status category, comprehensive anomaly tendency value, and major contribution profile as inputs for risk tracing, a power operation knowledge graph is constructed or invoked. This knowledge graph expresses the relationships between equipment, lines, transformer substations, users, regions, events, meteorology, new energy sources, and operational indicators within the regional power grid. Let the power operation knowledge graph be: KG=(V,E,A); Where V represents the set of nodes, E represents the set of edges, and A represents the set of node attributes and edge attributes; The node set is represented as: V=VD∪VL∪VT∪VU∪VR∪VE∪VW∪VN∪VI; Wherein, VD represents the set of equipment nodes, VL represents the set of line nodes, VT represents the set of transformer area nodes, VU represents the set of user nodes, VR represents the set of regional nodes, VE represents the set of event nodes, VW represents the set of meteorological nodes, VN represents the set of new energy nodes, and VI represents the set of operational indicator nodes; Edge sets are used to describe electrical connections, spatial affiliations, power supply relationships, event impacts, indicator correlations, and causal transmission relationships between different entities, and are represented as: ; Where ρ represents the relation type, For a set of relation types; for edge e=(v a, ρ,v b Set edge weight w e Used to characterize risk from node v a Propagated to node v b Strength: ; Among them, s e Indicates the strength of electrical or spatial association, d e h represents the electrical, topological, or spatial distance between nodes. e Indicates the frequency of historical risk co-occurrence. f ρ (·) This represents the function for calculating the edge weights corresponding to relation type ρ. Let the output of S4 be:

[0013] If A i,j Exceeding the preset risk trigger threshold τ s Then, initiate knowledge graph risk tracing: ; When δ i,j When =1, the region node v ri Abnormal profile node v g and the region within the time window t j Internally related abnormal equipment, abnormal lines, abnormal transformer areas, abnormal events, and abnormal weather nodes serve as a set of risk anchor points: ; Among them, V abn (r i ,t j ) represents region r i In time window t j The corresponding set of abnormal entity nodes; Abnormal entity nodes can be identified based on the degree of deviation of node operation indicators: ; Among them, Ω v This represents the set of operational metrics associated with node v. This indicates that node v is in time window t. j The k-th standardized operating indicator, μ v,k and σv ,k η represents the historical mean and standard deviation of the indicator, respectively. k This represents the indicator weight, where ε is a very small positive number; If the following conditions are met:

[0014] Then node v is determined to be an abnormal entity node; subsequently, the risk anchor point set S is used. i,j Starting from this point, a path search is performed in the power operation knowledge graph based on electrical topology, power supply path, spatial adjacency, event impact, and causal relationships to obtain a set of candidate risk propagation paths: ; Any candidate path is represented as: ; Where, v0∈S i,j l represents the path length; After obtaining the set of candidate risk propagation paths, the risk propagation intensity and source tracing contribution are calculated for each path. Its path propagation intensity is expressed as:

[0015] Where D(v0,t) j The ) indicates the degree of abnormal deviation of the starting node. Let γ represent the risk propagation weight of the a-th edge in the path, γ represent the path length attenuation coefficient, and l represent the path length. Considering the ability of the path termination node to explain the anomalies in the current region's state, the overall contribution of the path is further calculated: ; in, This indicates the degree of matching between the terminating node and the current state category. This indicates the degree of correlation between the terminating node and the profile of the major contributors; The top K paths, ranked from highest to lowest based on their overall contribution, were selected as the primary risk tracing paths. ; For a candidate risk source node v, the contribution scores of all paths that terminate at v or are key intermediate nodes are summed to obtain the risk source contribution score: ; in, This indicates that node v is on path p. q Importance coefficient in; Ultimately, the node or set of nodes with the highest contribution was selected as the risk source for abnormal regional operation status. .

[0016] Furthermore, combining the regional operational status identification results and risk tracing results, a comprehensive risk score is performed to obtain the final regional operational status level, as follows: A multi-dimensional risk assessment system is constructed, including five core evaluation dimensions: status identification risk component, risk source severity component, risk propagation path component, impact range component, and risk change trend component; the status identification risk component calculates the risk intensity corresponding to the current regional status based on the regional operational status category probability and comprehensive anomaly tendency value output by S4 using a weighted summation method; the risk source severity component assesses the severity of the risk source based on the contribution of the main risk source nodes identified by S5, combined with the risk source type weight and equipment importance coefficient. The risk severity is assessed using several methods: the risk propagation path component calculates the probability of risk propagation and diffusion in the power grid based on the comprehensive contribution and path length of the main risk propagation paths using an exponential decay weighting method; the impact range component assesses the breadth of the risk's impact on the power grid and users based on the number of affected users, the number of important users, the number of affected equipment, and the amount of affected load; the risk change trend component determines whether the risk is showing a continuous upward, stable, or downward trend based on the rate of change of the current risk score relative to the historical time window, and adjusts the comprehensive score accordingly; after completing the calculation of risk components in each dimension, a weighted fusion method is used to generate a comprehensive risk score, and the final regional operation status level is determined based on preset level thresholds.

[0017] The regional operation status identification system based on power multi-source data fusion includes a processor, a memory, and a computer program stored in the memory. When the processor executes the computer program, it specifically performs the steps in the regional operation status identification method based on power multi-source data fusion as described above.

[0018] The present invention has the following beneficial effects: 1. By forming a standardized multi-source dataset, this invention enables accurate matching and fusion of operational data, equipment data, user data, event data, and external environment data from different business systems under a unified object framework, thereby improving the accuracy of regional power grid operation status analysis and data utilization efficiency. 2. This invention can fully reflect the complexity and dynamic changes of the regional power grid operation status. It can identify obvious problems such as local equipment anomalies, voltage overruns, and sudden load increases, as well as potential risks caused by the superposition of factors such as new energy fluctuations, meteorological disturbances, and equipment health deterioration. This improves the comprehensiveness, precision, and intelligence of regional operation status identification. 3. After obtaining the regional operation status identification results, this invention further introduces a power operation knowledge graph for risk tracing, and combines the status identification results and risk tracing results to carry out a comprehensive risk score, and finally obtains the regional operation status level. Through the joint analysis of equipment topology relationships, power supply relationships, spatial affiliation relationships, event impact relationships and causal relationships in the knowledge graph, it is possible to trace the abnormal status at the regional level to specific equipment, lines, transformer areas, events, meteorological factors or new energy fluctuation factors. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0020] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments: refer to Figure 1 In this embodiment, a method for identifying regional operation status based on multi-source power data fusion is provided, including the following steps: S1: Obtain multi-source heterogeneous data from regional power grid related business systems, establish a unified data object model, and uniformly encode equipment, lines, transformer areas, users, regions, events, and times from different systems to obtain a standardized multi-source dataset; S2: Based on standardized multi-source datasets, data from different time scales, spatial scales, and business objects are linked into a unified analysis framework, and multi-dimensional operational features are constructed. S3: Based on multi-dimensional operational characteristics, perform multi-level data fusion to construct a regional status profile; S4: Based on the regional status profile and the regional operation status recognition model, obtain the regional operation status recognition results; S5: Based on the power operation knowledge graph, risk tracing is performed according to the regional operation status identification results; S6: Combining the regional operation status identification results and risk tracing results, a comprehensive risk score is performed to obtain the final regional operation status level.

[0021] Preferably, in this implementation, the multi-source power data includes: bus voltage, line power flow, main transformer load rate, switch status, and protection action information from the dispatch automation system; feeder current, distribution transformer load rate, fault indicator action information, and transformer substation voltage from the distribution automation system; user load curves, power consumption, maximum demand, and outage events from the power consumption information acquisition system; photovoltaic output, wind power output, energy storage charging and discharging power, and electric vehicle charging load from the new energy monitoring system; temperature, humidity, wind speed, rainfall, lightning density, typhoon path, and disaster warnings from the meteorological system; equipment location, line topology, equipment ledger, defect records, maintenance records, and equipment health index from the GIS and equipment asset system; and repair work orders, fault types, outage scope, number of affected users, and repair progress from the fault repair system.

[0022] In this embodiment, a unified data object model is established to uniformly encode devices, lines, transformer substations, users, regions, events, and times from different systems, resulting in a standardized multi-source dataset, as detailed below: The regional power grid-related business systems include dispatch automation systems, distribution automation systems, electricity consumption information collection systems, and GIS. Objects in systems such as the system, equipment asset management system, fault repair system, and new energy monitoring system are categorized into standard object types—equipment, line, distribution area, user, region, event, and time—according to power grid business semantics. Equipment objects represent entities such as substations, main transformers, busbars, switches, distribution transformers, reactive power compensation devices, and new energy grid-connected equipment. Line objects represent transmission lines, distribution feeders, branch lines, and their connections. Distribution area objects represent the distribution transformer's power supply range, area capacity, area load, and user set. User objects represent user ID, user category, power consumption capacity, load curve, and power supply affiliation. Region objects represent administrative regions, power supply zones, grid units, or feeder power supply ranges. Event objects represent operational events such as faults, power outages, alarms, maintenance, emergency repairs, and weather warnings. Time objects represent data acquisition time, event occurrence time, event recovery time, and the analysis time window required for status identification. Through this object abstraction, data from different systems with varying field names, data structures, and business granularities are uniformly converted into standard objects. Based on a unified data object model, unified coding is applied to equipment, lines, transformer substations, users, regions, events, and times across different systems. For equipment objects, a globally unique equipment code is established based on the equipment master data in the equipment asset management system, combined with dispatch naming, distribution network equipment ledgers, GIS spatial locations, and field operation numbers. For lines, a unified line code is generated based on the line name, voltage level, start and end nodes, affiliated substation, GIS line path, and topological connection relationship. For transformer substations, a unified substation code is established based on the transformer code, substation name, substation capacity, power supply range, and user affiliation. For users, a unified user code is established based on the marketing user number, metering point number, data acquisition terminal number, and substation affiliation relationship. For regions, a unified code is generated by combining administrative division codes, power supply area codes, and GIS data. For grid coding and equipment power supply range, a unified area coding system is established. For event objects, event codes are generated based on event type, object, time of occurrence, scope of impact, and processing status, ensuring that events such as faults, power outages, alarms, and maintenance can be accurately linked to the corresponding equipment, lines, transformer substations, users, or areas. For time objects, a unified timestamp and standard time format are adopted, and second-level, minute-level, hour-level, or day-level time window codes are generated according to business analysis needs. For cases where the same equipment has different numbers in different systems, mapping is performed using equipment master data verification, GIS topology constraints, and name matching algorithms to generate a unique equipment identifier. After generating a unique device identifier, it is written back to the standardized data object model, so that the device's operational data, ledger data, spatial data and event data in different systems such as scheduling, power distribution, GIS, assets, data acquisition, and emergency repair can be accurately correlated, thereby forming a unified, reliable and traceable standardized multi-source dataset.

[0023] In this embodiment, for situations where the same device has different numbers in different systems, a mapping method is used with the assistance of device master data verification, GIS topology constraints, and name matching algorithms to generate a unique device identifier. Specifically, the device master data is used as a baseline database to extract key attributes such as device name, voltage level, device type, station line, capacity parameters, commissioning date, and asset number, and these are compared with the device records in each business system. The upstream and downstream connection relationships, spatial location relationships, and power supply attribution relationships between devices are verified using GIS topology to determine whether the device records in different systems point to the same physical device. For devices with different names, abbreviations, aliases, or historical numbers, a name matching algorithm is used to calculate similarity. For example, the device name is standardized, invalid characters are removed, and the voltage level expression is unified. Candidate matching results are obtained by combining edit distance, keyword matching, pinyin matching, or semantic similarity calculation. Based on the comprehensive score of attribute similarity, topology consistency, spatial distance, and name similarity, the device mapping relationship is determined, and a globally unique device identifier is generated for the confirmed matching device.

[0024] In this embodiment, based on standardized multi-source datasets, data from different time scales, spatial scales, and business objects are correlated into a unified analysis framework, and multi-dimensional operational features are constructed, as follows: Because the data sampling periods of regional power grid related business systems differ—for example, dispatch automation data is typically at the second or minute level, distribution automation data is mostly at the minute level, electricity consumption information collection data is mostly at the 15-minute or hour level, meteorological data may be at the minute or hour level, and fault repair and power outage information is usually event-level data—it is necessary to set a unified analysis time window based on the real-time requirements of regional operation status identification. For continuous operation measurement data, such as voltage, current, active power, reactive power, load rate, renewable energy output, and user load, the average, maximum, minimum, standard deviation, and rate of change are calculated within the unified time window. For status data such as switch status, protection status, and communication status, the latest status, longest duration, or abnormal priority status within the window is selected. For event-type data such as faults, power outages, alarms, maintenance, and weather warnings, the number of events, duration, number of affected objects, and highest event level are statistically analyzed. Data with different sampling frequencies and different occurrence forms are converted into time-aligned datasets indexed by a unified time window. After unifying the time scale, data from different spatial scales and business objects are mapped to a unified regional analysis unit. This regional analysis unit can be set as an administrative region, power supply zone, substation supply area, feeder supply area, transformer area, or GIS grid, depending on the application scenario. Specifically, based on unique equipment identifiers, GIS coordinates, line topology relationships, power supply range, and user affiliation, objects such as substations, lines, switches, transformers, transformer areas, users, renewable energy access points, and meteorological grids are associated with the corresponding regional analysis units. For user load data, it is aggregated step-by-step through the power supply relationship of user—metering point—transformer—feeder. For equipment operation data, it is mapped to the corresponding power supply area through the equipment's station / line, upstream / downstream topology relationships, and spatial location. For meteorological data, spatial matching is performed between meteorological monitoring point coordinates or gridded meteorological data and the regional boundary; spatial interpolation is used to generate regional meteorological characteristics when necessary. For fault repair and power outage events, the event is linked to the corresponding region based on the event location, affected equipment, outage range, and affected users. A multi-dimensional operational feature structure is constructed for regional operational status identification. This multi-dimensional operational feature includes at least load characteristics, voltage characteristics, equipment load characteristics, power supply reliability characteristics, renewable energy impact characteristics, meteorological risk characteristics, and equipment health characteristics. Specifically, load characteristics include total regional load, average load, maximum load, load growth rate, load fluctuation rate, peak-to-valley difference, and the proportion of load from important users; voltage characteristics include voltage qualification rate, minimum voltage, maximum voltage, proportion exceeding the upper limit, proportion exceeding the lower limit, voltage deviation, and voltage fluctuation amplitude; equipment load characteristics include main transformer load rate, line load rate, distribution transformer load rate, number of heavily loaded devices, number of overloaded devices, and load rate change trends; power supply reliability characteristics include the number of users experiencing power outages, number of fault work orders, fault duration, number of repeated power outages, and the number of important users affected; renewable energy impact characteristics include the proportion of renewable energy output, output fluctuation rate, number of reverse power flow events, energy storage charging and discharging status, and renewable energy prediction deviation; meteorological risk characteristics include temperature, rainfall, wind speed, lightning density, and typhoon or icing warning levels; and equipment health characteristics include the equipment's years of operation, number of defects, overdue maintenance, historical fault count, and health index.

[0025] In this embodiment, the regional status profile includes a regional load profile, voltage quality profile, equipment stress profile, power supply reliability profile, new energy impact profile, meteorological risk profile, and equipment health profile. Specifically, the regional load profile depicts the regional load level, trend, and load sensitivity based on the total regional load, load growth rate, peak-to-valley difference, load volatility, and the proportion of load from important users. The voltage quality profile depicts the regional voltage stability and power quality level based on the voltage qualification rate, highest voltage, lowest voltage, over-limit ratio, voltage deviation, and fluctuation amplitude. The equipment stress profile depicts the load-bearing capacity of key regional equipment based on the load rates of main transformers, lines, and distribution transformers, the number of heavily loaded equipment, the number of overloaded equipment, and the trend of load rate changes. Pressure; Power supply reliability profile: Based on the number of users experiencing power outages, the number of fault work orders, the duration of faults, the number of repeated power outages, and the number of important users affected, it depicts the continuity of regional power supply and the degree of fault impact; New energy impact profile: Based on the proportion of new energy output, output fluctuation rate, number of reverse power flows, energy storage charging and discharging status, and prediction deviation, it depicts the degree of disturbance of new energy access to regional operation; Meteorological risk profile: Based on temperature, rainfall, wind speed, lightning density, typhoon, icing, or high temperature warning levels, it depicts the impact of the external environment on regional power grid operation; Equipment health profile: Based on the equipment's years of operation, number of defects, overdue maintenance, historical fault count, and health index, it depicts the aging of regional equipment, accumulation of defects, and operational reliability level. The various profiles independently reflect different dimensions of operational status, while also achieving horizontal correlation through the same time window and the same regional object. After forming multiple thematic profiles, they are integrated at the regional level to construct a regional status profile that comprehensively represents the regional operational status. The regional status profile takes the regional analysis unit as the object and the analysis time window as the index, and unifies the profile results of load, voltage, equipment, reliability, new energy, meteorology, and health to form a structured profile result that includes profile index values, profile level, change trend, anomaly labels, and dominant influencing factors. For example, when a region simultaneously experiences rapid load growth, increased proportion of heavily loaded distribution transformers, decreased voltage qualification rate, and enhanced high temperature warnings, the regional status profile can integrate and express this as "increased equipment pressure and decreased voltage quality driven by high load and high temperature"; when a region experiences drastic fluctuations in new energy output, increased reverse power flow frequency, and increased voltage fluctuation amplitude, the regional status profile can express this as "voltage disturbance and reverse power flow risk caused by new energy fluctuations".

[0026] In this embodiment, based on the regional status profile and the regional operational status recognition model, the regional operational status recognition result is obtained, as follows: Using each region-time window as the basic identification object, the regional load profile, voltage quality profile, equipment stress profile, power supply reliability profile, new energy impact profile, meteorological risk profile, and equipment health profile are converted into a state feature vector that the model can identify. Let the set of regions be denoted as . The time window set is For any region r i In time window t j The regional status profile can be represented as follows: ; in, This represents a regional load profile. This represents a voltage quality profile. This represents a pressure profile of the equipment. This indicates a power supply reliability profile. This depicts the impact profile of new energy sources. This represents a weather risk profile. This represents a device health profile; the indicators from various profiles are concatenated to obtain the region status recognition input vector: ; Where d represents the input feature dimension, x i,j,d Representing region r i In time window t j The d-th image feature value; Normalize the input features: ; in, and Let represent the minimum and maximum values ​​of the k-th feature in the historical samples, respectively, and ε be a very small positive number to prevent the denominator from being zero; thus, the standardized input vector is obtained: ; A regional operational status identification model is established to classify and identify regional operational status. Regional operational status can be divided into four levels: normal, watchful, warning, abnormal, and severe abnormal, denoted as: C = {c1, c2, c3, c4, c5}; Where c1 represents normal status, c2 represents status of concern, c3 represents warning status, c4 represents abnormal status, and c5 represents severe abnormal status; the regional operation status identification model is represented as: ; Where F(·) represents the regional operation status identification model, Θ represents the model parameters, and Y... i,j Representing region r i In time window tj The state recognition results are as follows.

[0027] In this embodiment, the regional operation status identification model uses a multi-layer neural network for identification, and its calculation process is expressed as follows: ; in, W represents the output of the l-th hidden layer. (l) and b (l) Let z represent the weight matrix and bias term of the l-th layer, respectively; σ(·) represents the non-linear activation function; L represents the number of hidden layers; z i,j This indicates the unnormalized score of the output layer; The probability of each state category is obtained using the Softmax function: ; Where, p i,j,q Representing region r i In time window t j Belongs to state c of class q q The probability, z i,j,q This represents the score of the output layer corresponding to the q-th state; The final state category is represented as:

[0028] This yields preliminary identification results of the region's operational status; Combining temporal trends and threshold correction to output recognition results, let region r i The model output probability within the most recent M time windows is Then the temporal smoothing probability of the q-th state is: ; Where, α m This represents the time weight of the m-th historical time window; After obtaining the smoothing probability, calculate the regional comprehensive anomaly tendency value: ; Where, β q The outlier weights represent the q-th class of states; The identification results are graded based on the overall abnormal tendency value: ; Where τ1, τ2, τ3, and τ4 are the threshold values ​​for classifying state levels; The final output is the regional operational status identification result, including the region number, time window, status category, probability of each status, comprehensive anomaly tendency value, and main contribution profile. The output result is represented as follows: ; Among them, G i,j This represents the set of images that contribute significantly to the recognition results, calculated based on the contribution of image features: ; Among them, Ω u Let ω represent the feature set corresponding to the u-th type of image. k g represents the importance weight of the k-th feature in the model. i,j,u This represents the contribution of the u-th type of image to the region state recognition result. In this embodiment, risk tracing is performed based on the power operation knowledge graph and the regional operation status identification results, as detailed below: Using region ID, time window, status category, comprehensive anomaly tendency value, and major contribution profile as inputs for risk tracing, a power operation knowledge graph is constructed or invoked. This knowledge graph expresses the relationships between equipment, lines, transformer substations, users, regions, events, meteorology, new energy sources, and operational indicators within the regional power grid. Let the power operation knowledge graph be: KG=(V,E,A); Where V represents the set of nodes, E represents the set of edges, and A represents the set of node attributes and edge attributes; The node set is represented as: V=VD∪VL∪VT∪VU∪VR∪VE∪VW∪VN∪VI; Wherein, VD represents the set of equipment nodes, VL represents the set of line nodes, VT represents the set of transformer area nodes, VU represents the set of user nodes, VR represents the set of regional nodes, VE represents the set of event nodes, VW represents the set of meteorological nodes, VN represents the set of new energy nodes, and VI represents the set of operational indicator nodes; Edge sets are used to describe electrical connections, spatial affiliations, power supply relationships, event impacts, indicator correlations, and causal transmission relationships between different entities, and are represented as: ; Where ρ represents the relation type, This is a set of relation types, including "connected to", "belongs to", "powered by", "affects", "causes", "associated with", "located in", "occurs in", "upstream in", and "downstream in". For edge e=(v a, ρ,v b Set edge weight w e Used to characterize risk from node v a Propagated to node v b Strength: ; Among them, s e Indicates the strength of electrical or spatial association, d e h represents the electrical, topological, or spatial distance between nodes. e Indicates the frequency of historical risk co-occurrence. f ρ (·) This represents the function for calculating the edge weights corresponding to relation type ρ. Let the output of S4 be:

[0029] If A i,j Exceeding the preset risk trigger threshold τ s Then, initiate knowledge graph risk tracing: ; When δ i,j When =1, the region node v ri Abnormal profile node v g and the region within the time window t j Internally related abnormal equipment, abnormal lines, abnormal transformer areas, abnormal events, and abnormal weather nodes serve as a set of risk anchor points: ; Among them, V abn (r i ,t j ) represents region r i In time window t j The corresponding set of abnormal entity nodes; Abnormal entity nodes can be identified based on the degree of deviation of node operation indicators: ; Among them, Ω v This represents the set of operational metrics associated with node v. This indicates that node v is in time window t. j The k-th standardized operating indicator, μ v,k and σv ,k η represents the historical mean and standard deviation of the indicator, respectively. k This represents the indicator weight, where ε is a very small positive number; If the following conditions are met:

[0030] Then node v is determined to be an abnormal entity node; subsequently, the risk anchor point set S is used. i,jStarting from this point, a path search is performed in the power operation knowledge graph based on electrical topology, power supply path, spatial adjacency, event impact, and causal relationships to obtain a set of candidate risk propagation paths: ; Any candidate path is represented as: ; Where, v0∈S i,j l represents the path length; After obtaining the set of candidate risk propagation paths, the risk propagation intensity and source tracing contribution are calculated for each path. Its path propagation intensity is expressed as:

[0031] Where D(v0,t) j The ) indicates the degree of abnormal deviation of the starting node. Let γ represent the risk propagation weight of the a-th edge in the path, γ represent the path length attenuation coefficient, and l represent the path length. Considering the ability of the path termination node to explain the anomalies in the current region's state, the overall contribution of the path is further calculated: ; in, This indicates the degree of matching between the terminating node and the current state category. This indicates the degree of correlation between the terminating node and the profile of the major contributors; In this embodiment, the matching degree can be calculated based on the co-occurrence probability of node anomalies and state categories in historical samples:

[0032] in, Represents node v in historical samples l Abnormal and the region status is The number of times, N(v) l ) represents node v l The total number of anomalies; the correlation degree is expressed as:

[0033] in, Represents node v l The set of associated image types.

[0034] The top K paths, ranked from highest to lowest based on their overall contribution, were selected as the primary risk tracing paths. ; For a candidate risk source node v, the contribution scores of all paths that terminate at v or are key intermediate nodes are summed to obtain the risk source contribution score: ; in, This indicates that node v is on path p. q Importance coefficient in; Ultimately, the node or set of nodes with the highest contribution was selected as the risk source for abnormal regional operation status. .

[0035] In this embodiment, a comprehensive risk score is performed by combining the regional operation status identification results and risk tracing results to obtain the final regional operation status level. Specifically, a multi-dimensional risk assessment system is constructed, including five core evaluation dimensions: status identification risk component, risk source severity component, risk propagation path component, impact range component, and risk change trend component. The status identification risk component calculates the risk intensity corresponding to the current regional status based on the regional operation status category probability and comprehensive anomaly tendency value output by S4 using a weighted summation method. The risk source severity component assesses the severity of the risk source based on the contribution of the main risk source nodes identified by S5, combined with the risk source type weight and equipment importance coefficient. The risk score is calculated using a weighted average method based on the severity of the risk, the risk propagation path component (calculating the probability of risk propagation and diffusion in the power grid using an exponential decay weighting method based on the comprehensive contribution and path length of the main risk propagation paths), the impact range component (assessing the breadth of the risk's impact on the power grid and users based on the number of affected users, the number of important users, the number of affected equipment, and the amount of affected load), and the risk change trend component (determining whether the risk is showing a continuous upward, stable, or downward trend based on the rate of change of the current risk score relative to the historical time window, and adjusting the comprehensive score accordingly). After calculating the risk components of each dimension, a weighted fusion method is used to generate a comprehensive risk score, and the final regional operation status level is determined according to a preset level threshold. The comprehensive risk score is obtained by weighting and summing the components of status identification, risk source, propagation path, impact range, and trend correction. The weight of each component can be set according to business importance, historical experience, and expert knowledge. For example, the status identification component has a higher weight to reflect the core role of model judgment, the risk source component has a lower weight to highlight the importance of cause identification, the impact range component has a higher weight to reflect user concerns about service quality, and the propagation path and trend correction components have relatively lower weights to play an auxiliary adjustment role. After obtaining the comprehensive risk score, the regional operation status is divided into five levels: normal, attention, warning, abnormal, and severe abnormality, based on the preset level thresholds. The normal level corresponds to a low risk score and no obvious abnormality. The attention level corresponds to a slightly increased risk score or local abnormality with limited impact. The warning level corresponds to a medium risk score and a clear risk source but no serious consequences. The abnormal level corresponds to a high risk score and obvious abnormalities such as equipment overload, voltage exceeding limits, and power outage. The severe abnormality level corresponds to a very high risk score and serious situations such as large-scale power outage, impact on important users, failure of key equipment, or extreme weather disasters.

[0036] The regional operation status identification system based on power multi-source data fusion includes a processor, a memory, and a computer program stored in the memory. When the processor executes the computer program, it specifically performs the steps in the regional operation status identification method based on power multi-source data fusion as described above.

[0037] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0038] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0039] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0040] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0041] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for identifying regional operation status based on multi-source power data fusion, characterized in that, Includes the following steps: S1: Obtain multi-source heterogeneous data from regional power grid related business systems, establish a unified data object model, and uniformly encode equipment, lines, transformer areas, users, regions, events, and times from different systems to obtain a standardized multi-source dataset; S2: Based on standardized multi-source datasets, data from different time scales, spatial scales, and business objects are linked into a unified analysis framework, and multi-dimensional operational features are constructed. S3: Based on multi-dimensional operational characteristics, perform multi-level data fusion to construct a regional status profile; S4: Based on the regional status profile and the regional operation status recognition model, obtain the regional operation status recognition results; S5: Based on the power operation knowledge graph, risk tracing is performed according to the regional operation status identification results; S6: Combining the regional operation status identification results and risk tracing results, a comprehensive risk score is performed to obtain the final regional operation status level.

2. The regional operation status identification method based on multi-source power data fusion according to claim 1, characterized in that, The establishment of a unified data object model encodes devices, lines, transformer substations, users, regions, events, and times from different systems in a unified manner, resulting in a standardized multi-source dataset, as detailed below: Objects in regional power grid-related business systems are categorized into standard object types—equipment, lines, distribution areas, users, regions, events, and time—based on power grid business semantics. Through this object abstraction, data from different systems with varying field names, data structures, and business granularities are uniformly converted into standard objects. Based on a unified data object model, devices, lines, transformer substations, users, regions, events, and times in different systems are uniformly coded. When the same device has different numbers in different systems, a mapping method is used with the assistance of device master data verification, GIS topology constraints, and name matching algorithms to generate a unique device identifier. After generating a unique device identifier, it is written back to the standardized data object model, thereby forming a standardized multi-source dataset.

3. The regional operation status identification method based on multi-source power data fusion according to claim 2, characterized in that, For situations where the same device has different IDs in different systems, a mapping method is used with the assistance of device master data verification, GIS topology constraints, and name matching algorithms to generate a unique device identifier. Specifically, the device master data is used as a baseline database to extract key attributes and compare them with device records in each business system; the upstream and downstream connections, spatial location relationships, and power supply attribution relationships between devices are verified using GIS topology to determine whether device records in different systems point to the same physical device; for devices with different names, abbreviations, aliases, or historical IDs, a name matching algorithm is used to calculate similarity; based on a comprehensive score of attribute similarity, topology consistency, spatial distance, and name similarity, the device mapping relationship is determined, and a globally unique device identifier is generated for the confirmed matching device.

4. The regional operation status identification method based on multi-source power data fusion according to claim 1, characterized in that, The standardized multi-source dataset links data from different time scales, spatial scales, and business objects into a unified analysis framework, and constructs multi-dimensional operational features, as detailed below: Data with different sampling frequencies and occurrence formats are converted into time-aligned datasets indexed by a unified time window; after unifying the time scale, data with different spatial scales and different business objects are mapped to a unified regional analysis unit. A multi-dimensional operational feature for regional operational status identification is constructed, which includes at least load features, voltage features, equipment load features, power supply reliability features, new energy impact features, meteorological risk features, and equipment health features.

5. The regional operation status identification method based on multi-source power data fusion according to claim 1, characterized in that, The established regional status profile includes regional load profile, voltage quality profile, equipment stress profile, power supply reliability profile, new energy impact profile, meteorological risk profile, and equipment health profile. After forming multiple thematic profiles, regional-level fusion is carried out to construct a regional status profile that can comprehensively represent the regional operational status. The regional status profile takes the regional analysis unit as the object and the analysis time window as the index, and unifies the profile results of load, voltage, equipment, reliability, new energy, meteorology and health to form a structured profile result that includes profile index values, profile level, change trend, anomaly label and dominant influencing factors.

6. The method for regional operation status identification based on multi-source power data fusion according to claim 1, characterized in that, The process of obtaining the regional operational status identification result based on the regional status profile and the regional operational status identification model is as follows: Using each region-time window as the basic identification object, the regional load profile, voltage quality profile, equipment stress profile, power supply reliability profile, new energy impact profile, meteorological risk profile, and equipment health profile are converted into a state feature vector that the model can identify. Let the set of regions be denoted as . The time window set is For any region r i In time window t j The regional status profile can be represented as follows: ； in, This represents a regional load profile. This represents a voltage quality profile. This represents a pressure profile of the equipment. This indicates a power supply reliability profile. This depicts the impact profile of new energy sources. This represents a weather risk profile. This represents a device health profile; the indicators from various profiles are concatenated to obtain the region status recognition input vector. ； Where d represents the input feature dimension, x i,j,d Representing region r i In time window t j The d-th portrait feature value; Normalize the input features: ; in, and Let represent the minimum and maximum values ​​of the k-th feature in the historical samples, respectively, and ε be a very small positive number to prevent the denominator from being zero; thus, the standardized input vector is obtained: ; A regional operational status identification model is established to classify and identify regional operational status. Regional operational status can be divided into four levels: normal, watchful, warning, abnormal, and severe abnormal, denoted as: C = {c1, c2, c3, c4, c5}; Where c1 represents normal status, c2 represents status of concern, c3 represents warning status, c4 represents abnormal status, and c5 represents severe abnormal status; the regional operation status identification model is represented as: ; Where F(·) represents the regional operation status identification model, Θ represents the model parameters, and Y... i,j Representing region r i In time window t j The state recognition results are as follows.

7. The regional operation status identification method based on multi-source power data fusion according to claim 6, characterized in that, The regional operation status identification model uses a multi-layer neural network for identification, and its calculation process is expressed as follows: ; in, W represents the output of the l-th hidden layer. (l) and b (l) Let z represent the weight matrix and bias term of the l-th layer, respectively; σ(·) represents the non-linear activation function; L represents the number of hidden layers; z i,j This indicates the unnormalized score of the output layer; The probability of each state category is obtained using the Softmax function: ; Where, p i,j,q Representing region r i In time window t j Belongs to state c of class q q The probability, z i,j,q This represents the score of the output layer corresponding to the q-th state; The final state category is represented as: ; This yields preliminary identification results of the region's operational status; Combining temporal trends and threshold correction to output recognition results, let the region r i The model output probability within the most recent M time windows is Then the temporal smoothing probability of the q-th state is: ; Where, α m This represents the time weight of the m-th historical time window; After obtaining the smoothing probability, calculate the regional comprehensive anomaly tendency value: ; Where, β q The outlier weights represent the q-th class of states; The identification results are graded based on the overall abnormal tendency value: ; Where τ1, τ2, τ3, and τ4 are the threshold values ​​for classifying state levels; The final output of the regional operational status identification results includes the region number, time window, status category, probability of each status, comprehensive anomaly tendency value, and main contribution profile. The output results are represented as follows: ; Among them, G i,j This represents the set of images that contribute significantly to the recognition results, calculated based on the contribution of image features: ; Among them, Ω u Let ω represent the feature set corresponding to the u-th type of image. k g represents the importance weight of the k-th feature in the model. i,j,u This represents the contribution of the u-th type of image to the region state recognition result.

8. The method for regional operation status identification based on multi-source power data fusion according to claim 7, characterized in that, The risk tracing based on the power operation knowledge graph and the regional operation status identification results is as follows: Using region ID, time window, status category, comprehensive anomaly tendency value, and major contribution profile as inputs for risk tracing, a power operation knowledge graph is constructed or invoked. This knowledge graph expresses the relationships between equipment, lines, transformer substations, users, regions, events, meteorology, new energy sources, and operational indicators within the regional power grid. Let the power operation knowledge graph be: KG=(V,E,A); Where V represents the set of nodes, E represents the set of edges, and A represents the set of node attributes and edge attributes; The node set is represented as: V=VD∪VL∪VT∪VU∪VR∪VE∪VW∪VN∪VI; Wherein, VD represents the set of equipment nodes, VL represents the set of line nodes, VT represents the set of transformer area nodes, VU represents the set of user nodes, VR represents the set of regional nodes, VE represents the set of event nodes, VW represents the set of meteorological nodes, VN represents the set of new energy nodes, and VI represents the set of operational indicator nodes; Edge sets are used to describe electrical connections, spatial affiliations, power supply relationships, event impacts, indicator correlations, and causal transmission relationships between different entities, and are represented as: ； Where ρ represents the relation type, For a set of relation types; for edge e=(v a, ρ,v b Set edge weight w e Used to characterize risk from node v a Propagated to node v b Strength: ； Among them, s e Indicates the strength of electrical or spatial association, d e h represents the electrical, topological, or spatial distance between nodes. e Indicates the frequency of historical risk co-occurrence. f ρ (·) This represents the function for calculating the edge weights corresponding to relation type ρρ; Let the output of S4 be: ; If A i,j Exceeding the preset risk trigger threshold τ s Then, initiate knowledge graph risk tracing: ； When δ i,j When =1, the region node v ri Abnormal profile node v g and the region within the time window t j Internally related abnormal equipment, abnormal lines, abnormal transformer areas, abnormal events, and abnormal weather nodes serve as a set of risk anchor points: ； Among them, V abn (r i ,t j ) represents region r i In time window t j The corresponding set of abnormal entity nodes; Abnormal entity nodes can be identified based on the degree of deviation of node operation indicators: ； Among them, Ω v This represents the set of operational metrics associated with node v. This indicates that node v is in time window t. j The k-th standardized operating indicator, μ v,k and σv ,k η represents the historical mean and standard deviation of the indicator, respectively. k This represents the indicator weight, where ε is a very small positive number; If the following conditions are met: ; Then node v is determined to be an abnormal entity node; subsequently, the risk anchor point set S is used. i,j Starting from this point, a path search is performed in the power operation knowledge graph based on electrical topology, power supply path, spatial adjacency, event impact, and causal relationships to obtain a set of candidate risk propagation paths: ； Any candidate path is represented as: ； Where, v0∈S i,j l represents the path length; After obtaining the set of candidate risk propagation paths, the risk propagation intensity and source tracing contribution are calculated for each path. Its path propagation intensity is expressed as: ; Where D(v0,t) j The ) indicates the degree of abnormal deviation of the starting node. Let γ represent the risk propagation weight of the a-th edge in the path, γ represent the path length attenuation coefficient, and l represent the path length. Considering the ability of the path termination node to explain the current region's state anomalies, the overall contribution of the path is further calculated: ； in, This indicates the degree of matching between the terminating node and the current state category. This indicates the degree of correlation between the terminating node and the profile of the major contributors; The top K paths, ranked from highest to lowest based on their overall contribution, were selected as the primary risk tracing paths. ； For a candidate risk source node v, the contribution scores of all paths that terminate at v or are key intermediate nodes are summed to obtain the risk source contribution score: ； in, This indicates that node v is on path p. q Importance coefficient in; Ultimately, the node or set of nodes with the highest contribution was selected as the risk source for abnormal regional operation status. 。 9. The method for regional operation status identification based on multi-source power data fusion according to claim 1, characterized in that, The method combines the regional operational status identification results and risk tracing results to conduct a comprehensive risk score and obtain the final regional operational status level, as follows: A multi-dimensional risk assessment system is constructed, including five core evaluation dimensions: status identification risk component, risk source severity component, risk propagation path component, impact range component, and risk change trend component; the status identification risk component is calculated based on the regional operational status category probability and comprehensive abnormality tendency value output by S4, and the risk intensity corresponding to the current regional status is calculated by weighted summation. The severity component of the risk source is based on the contribution of the main risk source nodes identified by S5, combined with the risk source type weight and the equipment importance coefficient, to assess the severity of the risk source. The risk propagation path component is calculated using an exponential decay weighting method based on the comprehensive contribution and path length of the main risk propagation paths; The scope of impact component assesses the breadth of the risk’s impact on the power grid and users based on the number of affected users, the number of important users, the number of affected equipment, and the amount of affected load. The risk change trend component determines whether the risk is continuously rising, stable, or declining based on the rate of change of the current risk score relative to the historical time window, and adjusts the comprehensive score accordingly. After completing the calculation of risk components in each dimension, a weighted fusion method is used to generate a comprehensive risk score, and the final regional operation status level is determined according to the preset level threshold.

10. A regional operation status identification system based on multi-source power data fusion, characterized in that, It includes a processor, a memory, and a computer program stored in the memory. When the processor executes the computer program, it specifically performs the steps in the regional operation status identification method based on power multi-source data fusion as described in any one of claims 1-9.

Images