Decision path generation method, device and equipment in power supply scenario, storage medium

Abstract

This application relates to a method, apparatus, device, and storage medium for generating decision paths in power supply scenarios. The method includes: extracting power supply-related data from multiple heterogeneous data sources across various business processes of the power supply system; preprocessing the data; constructing a heterogeneous knowledge graph in the power supply domain based on the preprocessed data; determining the confidence weights of graph edges based on the quantity and quality of data sources; defining a decision range on the graph with power supply anomaly scenario nodes as the starting set and power supply decision target nodes as the ending set; using a breadth-first search algorithm to expand paths hierarchically from the starting point; stopping branch expansion when a path reaches the ending point to generate a candidate power supply decision path set; and finally selecting the optimal power supply decision path and auxiliary power supply decision paths from this set. This method can improve the efficiency and accuracy of power supply decisions, while the auxiliary paths can provide alternative solutions, enhancing the flexibility and robustness of the power supply system.

Description

Technical Field

[0001] This application relates to the field of power grid technology, and in particular to a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for generating decision paths in a power supply scenario. Background Technology

[0002] With the development of power grid technology, decision path mining and evaluation techniques have emerged. However, these techniques rely on a single data source or simple cross-data source queries, failing to effectively integrate diverse data. Furthermore, they often employ static models, using fixed rules or rules of thumb to establish decision paths. However, with the large-scale grid integration of new energy sources, the complexity and uncertainty of the power grid have significantly increased. Consequently, various factors in the power supply scenario change over time, leading to inaccurate decision paths in traditional techniques. Summary of the Invention

[0003] Therefore, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for generating decision paths in power supply scenarios that can respond in real time and adjust flexibly, in order to address the above-mentioned technical problems.

[0004] Firstly, this application provides a decision path generation method for a power supply scenario, including:

[0005] Power supply-related data are extracted from multiple heterogeneous data sources across various business processes within the power supply system. These heterogeneous data sources include power grid operation monitoring data, equipment maintenance records, power load data, and power supply fault handling data.

[0006] Perform abnormal data processing and missing data supplementation on power supply related data; perform word segmentation and stop word removal on text fields in power supply related data; and perform normalization processing on numerical fields in power supply related data.

[0007] Based on the preprocessed power supply-related data, a heterogeneous knowledge graph in the power supply field is constructed according to the pattern of entity, attribute, and relationship. The edges of the heterogeneous knowledge graph are attached with multi-dimensional attribute indicators. The confidence weight of the edges in the heterogeneous knowledge graph is determined based on the number and quality of data sources.

[0008] On the heterogeneous knowledge graph, a starting set and an ending set for power supply decision-making are defined. The starting set includes nodes of power supply anomaly scenarios, and the ending set includes nodes of power supply decision targets. Starting from the starting point in the starting set, a breadth-first search algorithm is used to expand the path on the heterogeneous knowledge graph hierarchically. When the path reaches the ending point in the ending set, the branch expansion stops, resulting in a set of candidate power supply decision paths.

[0009] From the set of candidate power supply decision paths, the optimal power supply decision path and the auxiliary power supply decision path are selected.

[0010] In one embodiment, selecting the optimal power supply decision path and the auxiliary power supply decision path from the set of candidate power supply decision paths includes:

[0011] For the candidate power supply decision paths in the candidate power supply decision path set, the attribute indicators attached to each edge of the candidate power supply decision path are dimensionless; based on the correlation between the attribute indicators and the power supply business, and the historical power supply data related to the attribute indicators, the corresponding weights of the attribute indicators are determined; based on the dimensionless attribute indicators and their corresponding weights, and the confidence weights of each edge of the candidate power supply decision path, the comprehensive score of the candidate power supply decision path is calculated and sorted to obtain the path ranking result;

[0012] For the sorted power supply decision paths, the original data, the corresponding unique identifier of the original data, the data source label, and the data collection time are associated with the power supply-related data for each edge and node, along with the confidence weight of each edge, to construct a complete power supply decision evidence chain; based on the evidence chain integrity assessment standard, the integrity index of the power supply decision evidence chain is calculated; the interpretability of the power supply decision path is verified according to the integrity index to obtain the evidence chain verification result, and the power supply decision paths with insufficient evidence or confidence weights below the preset threshold are marked with risks;

[0013] Based on the path ranking results and evidence chain verification results, the optimal power supply decision path and the auxiliary power supply decision path are selected.

[0014] In one embodiment, the method further includes:

[0015] It continuously receives incremental data from various business processes of the power supply system. Based on the incremental data, it updates attribute indicators, their corresponding weights, the confidence weights of edges in the heterogeneous knowledge graph, and the comprehensive scores of candidate power supply decision paths. It also retains the corresponding update records so that it can be used to restore the evidence and conclusions of power supply decisions at any time.

[0016] In one embodiment, the method further includes:

[0017] Based on the comprehensive scores, attribute indicators, confidence weights and completeness indices of the optimal power supply decision path and the auxiliary power supply decision path, a power supply decision result dataset containing a complete chain of evidence is generated. The power supply decision result dataset and corresponding power supply decision optimization suggestions are output in the form of a graphical network view or a tabular list.

[0018] In one embodiment, abnormal data processing and missing data supplementation are performed on power supply-related data, including:

[0019] The sliding statistical method is used to calculate the center value and fluctuation range of power supply related data within a preset time period. Power supply related data outside the fluctuation range are marked as abnormal, the abnormal data are deleted, and the center value is used to replace and mark them.

[0020] Missing numerical power supply related data will be filled in with the median of the same type of data. Missing categorized power supply related data will be placed and labeled with preset fields. Missing timestamp-type power supply related data will be backfilled with data collection time and labeled.

[0021] In one embodiment, the entity nodes in the heterogeneous knowledge graph include power grid equipment, operation and maintenance personnel, power supply lines, substations, and user load center types in the power supply field; the semantic relationships represented by the edges in the heterogeneous knowledge graph include equipment operation and maintenance, line connection, power supply service, and fault association types; the multidimensional attributes of the edges in the heterogeneous knowledge graph include equipment operating years, line load rate, power supply reliability level, fault handling time, and operation and maintenance cost.

[0022] Secondly, this application also provides a decision path generation device for a power supply scenario, comprising:

[0023] The extraction module is used to extract power supply-related data from multiple heterogeneous data sources in various business processes of the power supply system; the heterogeneous data sources include power grid operation monitoring data, equipment maintenance record data, power load data, and power supply fault handling data;

[0024] The processing module is used to process abnormal data and supplement missing data in power supply related data, perform word segmentation and stop word removal on text fields in power supply related data, and normalize numerical fields in power supply related data.

[0025] The module is used to construct a heterogeneous knowledge graph in the power supply field based on preprocessed power supply-related data and in accordance with the pattern of entities, attributes, and relationships. The edges of the heterogeneous knowledge graph are attached with multi-dimensional attribute indicators. The confidence weight of the edges in the heterogeneous knowledge graph is determined based on the number and quality of data sources.

[0026] The candidate module is used to set the starting point set and the ending point set for power supply decisions on a heterogeneous knowledge graph. The starting point set includes power supply anomaly scenario nodes, and the ending point set includes power supply decision target nodes. Starting from the starting point in the starting point set, a breadth-first search algorithm is used to expand the path on the heterogeneous knowledge graph hierarchically, and the branch expansion stops when the path reaches the ending point in the ending point set, thus obtaining a candidate power supply decision path set.

[0027] The filtering module is used to filter out the optimal power supply decision path and the auxiliary power supply decision path from the set of candidate power supply decision paths.

[0028] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0029] Power supply-related data are extracted from multiple heterogeneous data sources across various business processes within the power supply system. These heterogeneous data sources include power grid operation monitoring data, equipment maintenance records, power load data, and power supply fault handling data.

[0030] Perform abnormal data processing and missing data supplementation on power supply related data; perform word segmentation and stop word removal on text fields in power supply related data; and perform normalization processing on numerical fields in power supply related data.

[0031] Based on the preprocessed power supply-related data, a heterogeneous knowledge graph in the power supply field is constructed according to the pattern of entity, attribute, and relationship. The edges of the heterogeneous knowledge graph are attached with multi-dimensional attribute indicators. The confidence weight of the edges in the heterogeneous knowledge graph is determined based on the number and quality of data sources.

[0032] On the heterogeneous knowledge graph, a starting set and an ending set for power supply decision-making are defined. The starting set includes nodes of power supply anomaly scenarios, and the ending set includes nodes of power supply decision targets. Starting from the starting point in the starting set, a breadth-first search algorithm is used to expand the path on the heterogeneous knowledge graph hierarchically. When the path reaches the ending point in the ending set, the branch expansion stops, resulting in a set of candidate power supply decision paths.

[0033] From the set of candidate power supply decision paths, the optimal power supply decision path and the auxiliary power supply decision path are selected.

[0034] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0035] Power supply-related data are extracted from multiple heterogeneous data sources across various business processes within the power supply system. These heterogeneous data sources include power grid operation monitoring data, equipment maintenance records, power load data, and power supply fault handling data.

[0036] Perform abnormal data processing and missing data supplementation on power supply related data; perform word segmentation and stop word removal on text fields in power supply related data; and perform normalization processing on numerical fields in power supply related data.

[0037] Based on the preprocessed power supply-related data, a heterogeneous knowledge graph in the power supply field is constructed according to the pattern of entity, attribute, and relationship. The edges of the heterogeneous knowledge graph are attached with multi-dimensional attribute indicators. The confidence weight of the edges in the heterogeneous knowledge graph is determined based on the number and quality of data sources.

[0038] On the heterogeneous knowledge graph, a starting set and an ending set for power supply decision-making are defined. The starting set includes nodes of power supply anomaly scenarios, and the ending set includes nodes of power supply decision targets. Starting from the starting point in the starting set, a breadth-first search algorithm is used to expand the path on the heterogeneous knowledge graph hierarchically. When the path reaches the ending point in the ending set, the branch expansion stops, resulting in a set of candidate power supply decision paths.

[0039] From the set of candidate power supply decision paths, the optimal power supply decision path and the auxiliary power supply decision path are selected.

[0040] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0041] Power supply-related data are extracted from multiple heterogeneous data sources across various business processes within the power supply system. These heterogeneous data sources include power grid operation monitoring data, equipment maintenance records, power load data, and power supply fault handling data.

[0042] Perform abnormal data processing and missing data supplementation on power supply related data; perform word segmentation and stop word removal on text fields in power supply related data; and perform normalization processing on numerical fields in power supply related data.

[0043] Based on the preprocessed power supply-related data, a heterogeneous knowledge graph in the power supply field is constructed according to the pattern of entity, attribute, and relationship. The edges of the heterogeneous knowledge graph are attached with multi-dimensional attribute indicators. The confidence weight of the edges in the heterogeneous knowledge graph is determined based on the number and quality of data sources.

[0044] On the heterogeneous knowledge graph, a starting set and an ending set for power supply decision-making are defined. The starting set includes nodes of power supply anomaly scenarios, and the ending set includes nodes of power supply decision targets. Starting from the starting point in the starting set, a breadth-first search algorithm is used to expand the path on the heterogeneous knowledge graph hierarchically. When the path reaches the ending point in the ending set, the branch expansion stops, resulting in a set of candidate power supply decision paths.

[0045] From the set of candidate power supply decision paths, the optimal power supply decision path and the auxiliary power supply decision path are selected.

[0046] The decision path generation method, device, computer equipment, computer-readable storage medium, and computer program product in the above power supply scenario first extract power supply-related data from multiple heterogeneous data sources in various business links of the power supply system, such as power grid operation monitoring, equipment maintenance records, power load, and power supply fault handling. Secondly, the data undergoes preprocessing operations such as anomaly handling, missing word supplementation, text segmentation to remove stop words, and numerical normalization to ensure data quality and consistency, providing comprehensive and reliable data support for decision path generation. Subsequently, a heterogeneous knowledge graph of the power supply domain with multi-dimensional attribute indicators is constructed based on the preprocessed data according to the entity-attribute-relationship pattern. The confidence weight of the graph edges is determined according to the number and quality of data sources. Then, a decision range is set on the graph with the power supply anomaly scenario nodes as the starting set and the power supply decision target nodes as the ending set. A breadth-first search algorithm is used to expand the path hierarchically from the starting point. When the path reaches the ending point, the branch expansion stops to generate a set of candidate power supply decision paths. Finally, the optimal power supply decision path and auxiliary power supply decision path are selected from this set to improve the efficiency and accuracy of power supply decision. At the same time, the auxiliary path can provide alternative solutions for power supply anomaly handling, enhancing the flexibility and robustness of the power supply system in dealing with faults. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is an application environment diagram of the decision path generation method in a power supply scenario of one embodiment;

[0049] Figure 2 This is a flowchart illustrating a decision path generation method in a power supply scenario, as shown in one embodiment.

[0050] Figure 3 This is a flowchart illustrating the decision path generation method in a power supply scenario, as shown in another embodiment.

[0051] Figure 4 This is a structural block diagram of a decision path generation device in a power supply scenario according to one embodiment;

[0052] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0054] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0055] The decision path generation method for power supply scenarios provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated on server 104 or placed on the cloud or other network servers. First, power supply-related data is extracted from multiple heterogeneous data sources across various business processes of the power supply system, such as power grid operation monitoring, equipment maintenance records, power load, and power supply fault handling. Then, preprocessing operations are performed on the data, including anomaly handling, missing word filling, text segmentation and stop word removal, and numerical normalization. Subsequently, a heterogeneous knowledge graph of the power supply domain with edges bearing multi-dimensional attribute indicators is constructed based on the preprocessed data according to an entity-attribute-relationship model. The confidence weight of the graph edges is determined based on the number and quality of data sources. Next, a decision range is set on the graph, with power supply anomaly scenario nodes as the starting point set and power supply decision target nodes as the ending point set. A breadth-first search algorithm is used to expand the path hierarchically from the starting point. When the path reaches the ending point, branch expansion stops to generate a candidate power supply decision path set. Finally, the optimal power supply decision path and auxiliary power supply decision path are selected from this set.

[0056] Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, drones, low-altitude aircraft, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, and projection equipment. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted displays. Head-mounted displays can be virtual reality (VR) devices, augmented reality (AR) devices, and smart glasses. Server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0057] In one exemplary embodiment, such as Figure 2 As shown, a decision path generation method for power supply scenarios is provided, which is then applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps 202 to 210. Wherein:

[0058] Step 202: Extract power supply-related data from multiple heterogeneous data sources in various business segments of the power supply system; heterogeneous data sources include power grid operation monitoring data, equipment maintenance record data, power load data, and power supply fault handling data;

[0059] The term "power supply system business links" refers to the core business scenarios throughout the entire power supply system process, encompassing all business links such as power grid operation monitoring, power supply equipment maintenance, power load monitoring and control, and troubleshooting and handling of power supply faults. "Heterogeneous data sources" refers to the collection of various data sources within the power supply system that differ in storage format, data type, business affiliation, and management entity. "Power grid operation monitoring data" refers to indicator-type data collected through monitoring terminals, dispatching systems, etc., reflecting the real-time / historical operating status of the power grid and power supply equipment, such as voltage, current, power, and equipment operating status. "Equipment maintenance record data" refers to ledger-type record data generated during the maintenance process of power supply equipment throughout its entire lifecycle, including information such as maintenance time, maintenance object, maintenance personnel, fault cause, handling measures, and maintenance results. "Power load data" refers to statistical data collected through load monitoring systems, reflecting the power consumption load status of different time periods and regions, including information such as regional load values, peak and valley loads, load change trends, and load characteristics. Power supply failure handling data refers to the business data generated throughout the entire process from the occurrence of a power supply failure to its completion, including information such as failure type, occurrence time / location, scope of impact, handling process, handling results, and loss assessment.

[0060] Specifically, during the data extraction process, each extracted data point can be assigned a unique identifier, and data source tags and data collection times can be added. The extracted data can then be stored hierarchically.

[0061] Step 204: Perform abnormal data processing and missing data supplementation on power supply related data; perform word segmentation and stop word removal on text fields in power supply related data; and perform normalization processing on numerical fields in power supply related data.

[0062] Among these, abnormal data processing refers to the use of methods such as data verification and outlier detection to identify abnormal data in power supply-related data that does not conform to business logic, exceeds reasonable limits, or contains errors, and then processing the abnormal data through methods such as removal, correction, and interpolation to ensure data accuracy. Missing data completion refers to the use of methods such as statistical analysis, interpolation, nearest neighbor filling, and business rule derivation to fill in null values ​​and missing items in power supply-related data to ensure data integrity. Text fields refer to non-numerical data fields in power supply-related data presented in text form, such as the fault cause and handling measures in equipment maintenance records, and the fault type description in power supply fault handling data.

[0063] Step 206: Based on the preprocessed power supply-related data, construct a heterogeneous knowledge graph in the power supply field according to the pattern of entity, attribute, and relationship. The edges of the heterogeneous knowledge graph are attached with multi-dimensional attribute indicators. Determine the confidence weight of the edges in the heterogeneous knowledge graph based on the number and quality of data sources.

[0064] In this context, "entities" refer to core research objects with independent semantic and practical business significance within the power supply field. They are the core nodes of the power supply knowledge graph, such as specific objects related to power supply operations, including organizations, personnel, power grid equipment, fault events, and maintenance projects. "Attributes" refer to various types of information describing the characteristics, states, or parameters of each entity in the power supply knowledge graph. They are concrete representations of entities, such as the model, installation location, and operating parameters of power grid equipment, and the occurrence time and type of fault events. "Relationships" refer to the logical or semantic connections between different entities in the power supply knowledge graph. They are the links connecting the entity nodes, such as the connection between equipment and maintenance records, the connection between fault events and handling personnel, and the connection between power grid equipment and operation monitoring data.

[0065] Step 208: On the heterogeneous knowledge graph, set the starting point set and the ending point set for power supply domain decision-making. The starting point set includes power supply anomaly scenario nodes, and the ending point set includes power supply decision target nodes. Starting from the starting point in the starting point set, use the breadth-first search algorithm to expand the path on the heterogeneous knowledge graph hierarchically, and stop branch expansion when the path reaches the ending point in the ending point set to obtain a candidate power supply decision path set.

[0066] The starting point set refers to the set of entity nodes in the heterogeneous knowledge graph of the power supply field, which serves as the starting point for path search in order to generate power supply decision-making paths. The power supply anomaly scenario nodes refer to the entity nodes in the heterogeneous knowledge graph of the power supply field that represent various abnormal operating scenarios such as operational anomalies, equipment failures, and load anomalies in the power supply system; these are the triggering nodes for power supply decisions. The ending point set refers to the set of entity nodes in the heterogeneous knowledge graph of the power supply field, which serves as the ending point for path search in order to generate power supply decision-making paths.

[0067] Step 210: Select the optimal power supply decision path and the auxiliary power supply decision path from the set of candidate power supply decision paths.

[0068] Among them, the candidate power supply decision path set refers to the overall set of all valid power supply decision paths generated by graph search algorithms in the heterogeneous knowledge graph of the power supply field, from the starting set composed of power supply anomaly scenario nodes to the ending set composed of power supply decision target nodes. It is the basic object for screening the optimal and auxiliary paths. The optimal power supply decision path refers to the power supply decision path selected from the candidate power supply decision path set that has the best comprehensive performance under preset evaluation indicators and best meets the actual business needs of power supply. It is the core execution path for handling power supply anomaly scenarios.

[0069] The decision path generation method in the aforementioned power supply scenario first extracts power supply-related data from multiple heterogeneous data sources across various business processes of the power supply system, including power grid operation monitoring, equipment maintenance records, power load, and power supply fault handling. Secondly, preprocessing operations are performed on the data, including anomaly handling, missing data supplementation, text segmentation and stop word removal, and numerical normalization, ensuring data quality and consistency and providing comprehensive and reliable data support for decision path generation. Subsequently, a heterogeneous knowledge graph of the power supply domain with edges bearing multi-dimensional attribute indicators is constructed based on the preprocessed data according to an entity-attribute-relationship model. The confidence weights of the graph edges are determined based on the quantity and quality of the data sources. Then, a decision range is defined on the graph, with the power supply anomaly scenario nodes as the starting point set and the power supply decision target nodes as the ending point set. A breadth-first search algorithm is used to expand the path hierarchically from the starting point. When the path reaches the ending point, branch expansion stops to generate a candidate power supply decision path set. Finally, the optimal power supply decision path and auxiliary power supply decision paths are selected from this set, improving the efficiency and accuracy of power supply decisions. Simultaneously, the auxiliary paths can provide alternative solutions for power supply anomaly handling, enhancing the flexibility and robustness of the power supply system in responding to faults.

[0070] In one embodiment, selecting the optimal power supply decision path and the auxiliary power supply decision path from the set of candidate power supply decision paths includes:

[0071] For the candidate power supply decision paths in the candidate power supply decision path set, the attribute indicators attached to each edge of the candidate power supply decision path are dimensionless; based on the correlation between the attribute indicators and the power supply business, and the historical power supply data related to the attribute indicators, the corresponding weights of the attribute indicators are determined; based on the dimensionless attribute indicators and their corresponding weights, and the confidence weights of each edge of the candidate power supply decision path, the comprehensive score of the candidate power supply decision path is calculated and sorted to obtain the path ranking result;

[0072] For the sorted power supply decision paths, the original data, the corresponding unique identifier of the original data, the data source label, and the data collection time are associated with the power supply-related data for each edge and node, along with the confidence weight of each edge, to construct a complete power supply decision evidence chain; based on the evidence chain integrity assessment standard, the integrity index of the power supply decision evidence chain is calculated; the interpretability of the power supply decision path is verified according to the integrity index to obtain the evidence chain verification result, and the power supply decision paths with insufficient evidence or confidence weights below the preset threshold are marked with risks;

[0073] Based on the path ranking results and evidence chain verification results, the optimal power supply decision path and the auxiliary power supply decision path are selected.

[0074] Specifically, the candidate path set is as follows: For any path in the set Let the number of its edges be . The system extracts the same set of parameters for each edge on the path. The item attribute index is used, and its direction is unified to obtain the edge in the first position. Dimensionless values ​​of the index .

[0075] The weights are calculated using a simplified method, which involves manually setting initial weights and normalizing them so that the sum of all weights is 1. The initial weights are set for each indicator. A preliminary weight is manually set. These weights can be assigned values ​​based on the importance of the business, expert experience, or historical data. Furthermore, to ensure that the sum of the weights of all indicators is 1, the initially set weights need to be adjusted. Normalization is performed using the following formula:

[0076]

[0077] in, This indicates the initial weight settings. This represents the normalized weights, ensuring that the sum of all weights is 1. This results in a set of indicator weights. .

[0078] Based on this, the path The overall score is defined as follows:

[0079]

[0080] in, Indicates the number of edges in the path. Indicates the road The edge at the 1st Normalized values ​​for each indicator. Indicates the first The system assigns weights to each indicator. From high to low The paths are sorted, and the one with the highest score is taken as the main path, while the remaining paths are used as auxiliary paths for selection and comparative analysis.

[0081] The system establishes a chain of evidence for each edge and node of the main path and candidate paths, completely linking "which table was used, which field, what the value was, what the source system was, what the collection time was, and what the UID was." The entities and attributes in each path correspond to the fields, table names, source systems, and timestamps in the original data records, constructed in the following format:

[0082]

[0083] in, This represents a set of evidence chain data. For field names, For field values, For timestamps, This is the data source. And it will undergo integrity checks. Integrity check formula:

[0084]

[0085] in, This represents the index of the completeness of the chain of evidence. This indicates the actual pieces of evidence found. Evidence items required to represent the path.

[0086] when At this point, the path is considered interpretable and can proceed with further analysis and decision-making. At that time, the path will also be analyzed and decided upon, but risks will be marked as "insufficient evidence" and a risk warning will be issued.

[0087] Finally, by combining the path ranking results and the evidence chain verification results, the optimal power supply decision path and the auxiliary power supply decision path are selected from the candidate set.

[0088] In the above embodiments, by dimensionless processing of the attribute indicators of the path edges and determining the weights by combining business relevance and historical data, and calculating the comprehensive score by combining the confidence weight of the edges, the quantitative evaluation of candidate paths is made more in line with the actual power supply business, ensuring the objectivity and accuracy of path ranking. By associating original data and traceability identifiers with each edge and node of the path and constructing an evidence chain, combined with the integrity index to verify interpretability and label risky paths, the full-link evidence traceability of power supply decision paths is realized, improving the interpretability and credibility of the paths and avoiding the decision-making risks of insufficient evidence and low confidence paths. Finally, the optimal and auxiliary paths are selected by combining the score ranking and evidence chain verification results, which not only ensures the superiority of the selected optimal path in terms of comprehensive performance, but also ensures that the auxiliary paths meet the requirements of sufficient evidence and qualified confidence, effectively improving the scientificity, reliability and traceability of power supply decisions, while making the selected decision paths more suitable for the actual business handling needs of the power supply system.

[0089] In one embodiment, the method further includes:

[0090] It continuously receives incremental data from various business processes of the power supply system. Based on the incremental data, it updates attribute indicators, their corresponding weights, the confidence weights of edges in the heterogeneous knowledge graph, and the comprehensive scores of candidate power supply decision paths. It also retains the corresponding update records so that it can be used to restore the evidence and conclusions of power supply decisions at any time.

[0091] Specifically, it continuously receives incremental events from various business systems, merges multiple versions of the same object according to event time, and maintains a summary frequency of minutes. When key data changes, it automatically recalculates indicator values, weights, and path scores. To support auditing, all version changes retain validity periods and operation type tags, allowing evidence and conclusions from any point in time to be restored at any time.

[0092] In the above embodiments, by continuously receiving and utilizing incremental data from the power supply system to conduct multi-dimensional dynamic updates, the attribute indicators, various weights, and path comprehensive scores are always synchronized with the actual business status of the power supply system, ensuring the timeliness and adaptability of the power supply decision path and avoiding decision deviations caused by data lag. At the same time, the update records are completely retained, realizing the reproducibility of power supply decision evidence and conclusions at any time, further improving the full-link traceability system of power supply decisions, and enhancing the auditability and reliability of the decision-making process.

[0093] In one embodiment, the method further includes:

[0094] Based on the comprehensive scores, attribute indicators, confidence weights and completeness indices of the optimal power supply decision path and the auxiliary power supply decision path, a power supply decision result dataset containing a complete chain of evidence is generated. The power supply decision result dataset and corresponding power supply decision optimization suggestions are output in the form of a graphical network view or a tabular list.

[0095] Specifically, the decision path output includes a list of main and candidate paths, the total score and breakdown of each indicator for each path, the average relationship confidence level, the evidence integrity index, and a complete evidence chain. It provides multiple formats: a graphical network view, a tabular list, and a standardized interface return structure. The interface return includes the order of path nodes, the unique numbers involved, the source and time of each step, and the final values ​​and contributions of each indicator, ensuring that external systems or reports can directly reference it without further processing.

[0096] In the above embodiments, by integrating core information such as the comprehensive score and attribute indicators of the path with a complete chain of evidence to generate a standardized decision result dataset, the information presentation of the power supply decision results is more systematic and comprehensive. The output format of graphical network view or tabular list can be adapted to the viewing and usage needs of different power supply business scenarios, presenting the core decision information intuitively and clearly, and improving the readability and usability of the decision results. The power supply decision optimization suggestions output simultaneously can provide targeted guidance for the handling of power supply anomalies and subsequent system operation management, so that the power supply decision results have both immediate handling value and long-term optimization reference value, further enhancing the practical application value of power supply decisions.

[0097] In one embodiment, abnormal data processing and missing data supplementation are performed on power supply-related data, including:

[0098] The sliding statistical method is used to calculate the center value and fluctuation range of power supply related data within a preset time period. Power supply related data outside the fluctuation range are marked as abnormal, the abnormal data are deleted, and the center value is used to replace and mark them.

[0099] Missing numerical power supply related data will be filled in with the median of the same type of data. Missing categorized power supply related data will be placed and labeled with preset fields. Missing timestamp-type power supply related data will be backfilled with data collection time and labeled.

[0100] Specifically, data preprocessing mainly involves eliminating outliers, normalizing the data, and recording text. Outliers include missing data and anomalous data values. First, anomaly detection is performed on numerical fields. A moving average method is used to calculate the median value and fluctuation range over a period of time. Records with significant deviations are marked as outliers and deleted according to rules, replaced by the median. Outlier detection is performed using the moving average method:

[0101]

[0102] in, For the original data, The sample mean. is the standard deviation. When this happens, the value is considered an abnormal deletion and replaced with the median, and then marked.

[0103] For missing data, numerical fields are padded with the median of similar data; categorical fields are marked as "unknown" and annotated; missing timestamps are filled with the data arrival time, but must be labeled for display in the output. Text fields are uniformly segmented, stop words are removed, and keyword vectors are generated, preserving both the original text and the processed representation for easier subsequent similarity calculations and topic classification. Furthermore, all numerical fields are uniformly processed using the minimum-maximum normalization formula for comparative analysis. The minimum-maximum normalization formula is:

[0104]

[0105] in, This represents data after normalization by maximum and minimum values, and its value range is usually between [0,1]. This represents the i-th sample value in the original data to be normalized. This represents the minimum value among the original data of the same type. This represents the maximum value among the original data of the same type.

[0106] In the above embodiments, the sliding statistical method, combined with the central value and fluctuation range, is used to determine and process abnormal data. This method can accurately identify outliers in power supply time series data. By replacing and labeling with the central value, data errors are eliminated while data integrity is preserved, adapting to the dynamic change characteristics of power supply system data. For different types of missing data, a strategy of classification completion and labeling is adopted. For numerical data, the median is used to complete the data, ensuring the statistical rationality of quantitative data. For categorical and timestamp data, preset field placeholders and collection time are used to fill in and label the data. This not only completes the data structure but also distinguishes between the original data and the completed data, avoiding interference from the completed data to subsequent analysis. Overall, this improves the accuracy, completeness, and standardization of power supply-related data.

[0107] In one embodiment, the entity nodes in the heterogeneous knowledge graph include power grid equipment, operation and maintenance personnel, power supply lines, substations, and user load center types in the power supply field; the semantic relationships represented by the edges in the heterogeneous knowledge graph include equipment operation and maintenance, line connection, power supply service, and fault association types; the multidimensional attributes of the edges in the heterogeneous knowledge graph include equipment operating years, line load rate, power supply reliability level, fault handling time, and operation and maintenance cost.

[0108] Specifically, a heterogeneous knowledge graph is constructed using an entity-attribute-relationship model. Entity nodes in the graph include power grid equipment, maintenance personnel, power lines, substations, and user load center types within the power supply field. Edges represent semantic relationships between entities, including equipment maintenance, line connections, power supply services, and fault association types. Using Neo4j as the graph database backend, the system automatically identifies data and imports it into the graph. Each edge is accompanied by multi-dimensional attributes, including equipment operating years, line load rate, power supply reliability level, fault handling time, and maintenance costs.

[0109] In the above embodiments, by clearly defining the specific types of entities, edge relationships, and edge attributes in the heterogeneous knowledge graph of the power supply field, the graph can accurately and comprehensively map the core business objects and business logic associations of the power supply system. Five types of entity nodes cover the core elements of the entire power supply chain from facilities to personnel and from the power supply end to the load end. Four types of semantic relationships clearly represent the core business associations between entities. Five types of multi-dimensional attributes quantify the relationship characteristics from multiple dimensions such as equipment, lines, services, faults, and costs. This provides a quantitative basis that fits the actual power supply business for the subsequent indicator evaluation and weight calculation of path search. This makes the constructed heterogeneous knowledge graph more suitable for the decision path generation needs of the power supply scenario, and improves the semantic expression capability and practical application value of the graph for power supply business.

[0110] In one embodiment, such as Figure 3 As shown, this is a decision path generation method in a power supply scenario according to a specific embodiment, including:

[0111] S1: Data Extraction and Structured Processing

[0112] Structured and semi-structured data are extracted from multiple heterogeneous data sources within the power supply system using interfaces or data integration tools. These heterogeneous data sources include, but are not limited to, power grid operation monitoring data, equipment maintenance records, power load data, and power supply fault handling data. During data extraction, each record is assigned a unique identifier (UID) for subsequent graph construction and evidence backtracking. A hash function is used to ensure that the generated UIDs are unique and do not conflict.

[0113] (1)

[0114] Duplicate records are identified and merged based on their UIDs. All records are accompanied by raw data such as a "data source tag" and arrival time for subsequent evidence retrieval. Data is stored hierarchically: the original layer preserves the original data and metadata, while the graph loading layer is split and stored according to "node tables and relationship tables" for use in subsequent knowledge graph construction.

[0115] S2: Data Preprocessing

[0116] Data preprocessing primarily involves eliminating outliers, normalizing data, and recording text. Outliers include missing data and anomalous data values. First, anomaly detection is performed on numerical fields. A moving average method is used to calculate the median value and fluctuation range over a period of time. Records showing significant deviations are marked as outliers and deleted according to rules, replaced by the median. Outlier detection is achieved using the moving average method.

[0117] (2)

[0118] in, For the original data, The sample mean. is the standard deviation. When this happens, the value is considered an abnormal deletion and replaced with the median, and then marked.

[0119] For missing data, numerical fields are padded with the median of similar data; categorical fields are marked as "unknown" and annotated; missing timestamps are filled with the data arrival time, but must be labeled for display in the output. Text fields are uniformly segmented, stop words are removed, and keyword vectors are generated, preserving both the original text and the processed representation for easier subsequent similarity calculations and topic classification. Furthermore, all numerical fields are uniformly processed using the minimum-maximum normalization formula for comparative analysis. The minimum-maximum normalization formula is:

[0120] (3)

[0121] in, This represents data after normalization by maximum and minimum values, and its value range is usually between [0,1]. This represents the i-th sample value in the original data to be normalized. This represents the minimum value among the original data of the same type. This represents the maximum value among the original data of the same type.

[0122] S3: Knowledge Graph Construction

[0123] Based on the obtained structured data, a heterogeneous knowledge graph is constructed according to the entity-attribute-relationship model. Entity nodes in the graph include power grid equipment, maintenance personnel, power lines, substations, and user load center types in the power supply field. Edges represent semantic relationships between entities, including equipment maintenance, line connections, power supply services, and fault association types. Using Neo4j as the graph database backend, the system automatically identifies the data and imports it into the graph. Each edge is accompanied by multi-dimensional attributes, including equipment operating years, line load rate, power supply reliability level, fault handling time, and maintenance costs. To enhance the weighted expressive power of the graph structure, a confidence weight mechanism is introduced. The confidence score for each edge is defined as:

[0124] (4)

[0125] in, This represents a relation edge. This indicates the number of independent data sources that support the relationship. This represents a weighted average of source quality, composed of factors such as source authority, information freshness, and consistency. This parameter represents the trade-off between the number of sources and the quality of sources (default 0.5). Indicates a smoothing term, avoiding Excessive magnification at very small values ​​(default 2).

[0126] The formula for calculating the weighted average of source quality is as follows:

[0127] (5)

[0128] in, This indicates the source's authority rating, which is obtained from the corresponding institutionalized rating table. Indicates freshness. This indicates the trade-off parameter between the two (default 0.7).

[0129] The formula for calculating freshness is:

[0130] (6)

[0131] in, This indicates the time elapsed since then, expressed in days. This represents the time decay constant (default 180 days).

[0132] S4: Multi-indicator decision path generation and scoring

[0133] On the constructed knowledge graph, the starting set is defined as the evaluation target S (power supply anomaly scenario nodes, such as power supply fault points, load exceeding nodes, etc.), and the ending set is defined as the decision support object T (power supply decision target nodes, such as fault handling plan nodes, load control plan nodes, etc.), with a maximum path length of no more than 5 edges. A layer-wise breadth-first search (BFS) is used for path mining: initially, each starting point is considered a path of length 0. Subsequently, the path expands layer by layer, allowing only one edge to extend outward along the allowed relation type at each step. If a new path generated by expansion contains duplicate nodes (forming a cycle), it will be added to the prohibited entity sequence and discarded immediately. Once a new path reaches any ending point, it is added to the result set. The branch is no longer expanded. The search stops when all candidate paths have reached their maximum length or the candidate set is exhausted, resulting in a set of candidate paths. .

[0134] For any path in the set Let the number of its edges be . The system extracts the same set of parameters for each edge on the path. The item attribute index is used, and its direction is unified to obtain the edge in the first position. Dimensionless values ​​of the index .

[0135] The weights are calculated using a simplified method, which involves manually setting initial weights and normalizing them so that the sum of all weights is 1. The initial weights are set for each indicator. A preliminary weight is manually set. These weights can be assigned values ​​based on the importance of the business, expert experience, or historical data. Furthermore, to ensure that the sum of the weights of all indicators is 1, the initially set weights need to be adjusted. Normalization is performed using the following formula:

[0136] (7)

[0137] in, This indicates the initial weight settings. This represents the normalized weights, ensuring that the sum of all weights is 1. This results in a set of indicator weights. .

[0138] Based on this, the path The overall score is defined as follows:

[0139] (8)

[0140] in, Indicates the number of edges in the path. Indicates the road The edge at the 1st Normalized values ​​for each indicator. Indicates the first The system assigns weights to each indicator. From high to low The paths are sorted, and the one with the highest score is taken as the main path, while the remaining paths are used as auxiliary paths for selection and comparative analysis.

[0141] S5: Evidence Chain Tracing and Integrity Verification

[0142] For each edge and candidate path, an evidence chain is established, linking together "which table was used, which field, what the value was, what the source system was, what the collection time was, and what the UID was." The entities and attributes in each path correspond to the fields, table names, source systems, and timestamps in the original data records, constructed in the following format:

[0143] (9)

[0144] in, This represents a set of evidence chain data. For field names, For field values, For timestamps, This is the data source. And it will undergo integrity checks. Integrity check formula:

[0145] (10)

[0146] in, This represents the index of the completeness of the chain of evidence. This indicates the actual pieces of evidence found. Evidence items required to represent the path.

[0147] when At this point, the path is considered interpretable and can proceed with further analysis and decision-making. At that time, the path will also be analyzed and decided upon, but risks will be marked as "insufficient evidence" and a risk warning will be issued.

[0148] S6: Online update and recalculation mechanism

[0149] To ensure conclusions are updated promptly as data changes, we continuously receive incremental events from various business systems, merging multiple versions of the same object according to event time, maintaining a summary frequency of minutes. When key data changes, we automatically recalculate indicator values, weights, and path scores. To support auditing, all version changes retain validity periods and operation type tags, allowing for the restoration of evidence and conclusions at any point in time.

[0150] S7: Output Results

[0151] The decision path output includes a list of main and candidate paths, the total score and breakdown of each indicator for each path, average relationship confidence, evidence integrity index, and a complete evidence chain. It provides multiple formats: a graphical network view, a tabular list, and a standardized interface return structure. The interface return includes the order of path nodes, unique identifiers involved, the source and time of each step, and the final values ​​and contributions of each indicator, ensuring that external systems or reports can directly reference it without further processing.

[0152] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0153] Based on the same inventive concept, this application also provides a decision path generation device for implementing the decision path generation method in the power supply scenario described above. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the decision path generation device in the power supply scenario provided below can be found in the limitations of the decision path generation method in the power supply scenario described above, and will not be repeated here.

[0154] In one exemplary embodiment, such as Figure 4 As shown, a decision path generation device for a power supply scenario is provided, including: an extraction module 402, a processing module 404, a construction module 406, a candidate module 408, and a filtering module 410, wherein:

[0155] The extraction module 402 is used to extract power supply-related data from multiple heterogeneous data sources in various business links of the power supply system; the heterogeneous data sources include power grid operation monitoring data, equipment maintenance record data, power load data and power supply fault handling data;

[0156] The processing module 404 is used to process abnormal data and supplement missing data in power supply related data, perform word segmentation and stop word removal on text fields in power supply related data, and perform normalization on numerical fields in power supply related data.

[0157] Module 406 is used to construct a heterogeneous knowledge graph in the power supply field based on preprocessed power supply-related data and in accordance with the pattern of entities, attributes, and relationships. The edges of the heterogeneous knowledge graph are attached with multi-dimensional attribute indicators. The confidence weight of the edges in the heterogeneous knowledge graph is determined based on the number and quality of data sources.

[0158] Candidate module 408 is used to set the starting set and the ending set of power supply domain decision-making on the heterogeneous knowledge graph. The starting set includes power supply anomaly scenario nodes, and the ending set includes power supply decision target nodes. Starting from the starting point in the starting set, a breadth-first search algorithm is used to expand the path on the heterogeneous knowledge graph hierarchically, and the branch expansion stops when the path reaches the ending point in the ending set, thus obtaining a candidate power supply decision path set.

[0159] The filtering module 410 is used to filter out the optimal power supply decision path and the auxiliary power supply decision path from the set of candidate power supply decision paths.

[0160] In one embodiment, the filtering module 410 is further configured to:

[0161] For the candidate power supply decision paths in the candidate power supply decision path set, the attribute indicators attached to each edge of the candidate power supply decision path are dimensionless; based on the correlation between the attribute indicators and the power supply business, and the historical power supply data related to the attribute indicators, the corresponding weights of the attribute indicators are determined; based on the dimensionless attribute indicators and their corresponding weights, and the confidence weights of each edge of the candidate power supply decision path, the comprehensive score of the candidate power supply decision path is calculated and sorted to obtain the path ranking result;

[0162] For the sorted power supply decision paths, the original data, the corresponding unique identifier of the original data, the data source label, and the data collection time are associated with the power supply-related data for each edge and node, along with the confidence weight of each edge, to construct a complete power supply decision evidence chain; based on the evidence chain integrity assessment standard, the integrity index of the power supply decision evidence chain is calculated; the interpretability of the power supply decision path is verified according to the integrity index to obtain the evidence chain verification result, and the power supply decision paths with insufficient evidence or confidence weights below the preset threshold are marked with risks;

[0163] Based on the path ranking results and evidence chain verification results, the optimal power supply decision path and the auxiliary power supply decision path are selected.

[0164] In one embodiment, the method further includes:

[0165] It continuously receives incremental data from various business processes of the power supply system. Based on the incremental data, it updates attribute indicators, their corresponding weights, the confidence weights of edges in the heterogeneous knowledge graph, and the comprehensive scores of candidate power supply decision paths. It also retains the corresponding update records so that it can be used to restore the evidence and conclusions of power supply decisions at any time.

[0166] In one embodiment, the method further includes:

[0167] Based on the comprehensive scores, attribute indicators, confidence weights and completeness indices of the optimal power supply decision path and the auxiliary power supply decision path, a power supply decision result dataset containing a complete chain of evidence is generated. The power supply decision result dataset and corresponding power supply decision optimization suggestions are output in the form of a graphical network view or a tabular list.

[0168] In one embodiment, the processing module 404 is further configured to:

[0169] The sliding statistical method is used to calculate the center value and fluctuation range of power supply related data within a preset time period. Power supply related data outside the fluctuation range are marked as abnormal, the abnormal data are deleted, and the center value is used to replace and mark them.

[0170] Missing numerical power supply related data will be filled in with the median of the same type of data. Missing categorized power supply related data will be placed and labeled with preset fields. Missing timestamp-type power supply related data will be backfilled with data collection time and labeled.

[0171] In one embodiment, the construction module 406 includes:

[0172] The entity nodes in the heterogeneous knowledge graph include power grid equipment, operation and maintenance personnel, power supply lines, substations, and user load center types in the power supply field; the semantic relationships represented by the edges in the heterogeneous knowledge graph include equipment operation and maintenance, line connection, power supply service, and fault association types; the multidimensional attributes of the edges in the heterogeneous knowledge graph include equipment operating years, line load rate, power supply reliability level, fault handling time, and operation and maintenance cost.

[0173] Each module in the decision path generation device under the aforementioned power supply scenario can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0174] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 5As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When executed by the processor, the computer program implements a decision path generation method under a power supply scenario. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0175] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0176] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0177] In one embodiment, selecting the optimal power supply decision path and the auxiliary power supply decision path from the set of candidate power supply decision paths includes:

[0178] For the candidate power supply decision paths in the candidate power supply decision path set, the attribute indicators attached to each edge of the candidate power supply decision path are dimensionless; based on the correlation between the attribute indicators and the power supply business, and the historical power supply data related to the attribute indicators, the corresponding weights of the attribute indicators are determined; based on the dimensionless attribute indicators and their corresponding weights, and the confidence weights of each edge of the candidate power supply decision path, the comprehensive score of the candidate power supply decision path is calculated and sorted to obtain the path ranking result;

[0179] For the sorted power supply decision paths, the original data, the corresponding unique identifier of the original data, the data source label, and the data collection time are associated with the power supply-related data for each edge and node, along with the confidence weight of each edge, to construct a complete power supply decision evidence chain; based on the evidence chain integrity assessment standard, the integrity index of the power supply decision evidence chain is calculated; the interpretability of the power supply decision path is verified according to the integrity index to obtain the evidence chain verification result, and the power supply decision paths with insufficient evidence or confidence weights below the preset threshold are marked with risks;

[0180] Based on the path ranking results and evidence chain verification results, the optimal power supply decision path and the auxiliary power supply decision path are selected.

[0181] In one embodiment, the method further includes:

[0182] It continuously receives incremental data from various business processes of the power supply system. Based on the incremental data, it updates attribute indicators, their corresponding weights, the confidence weights of edges in the heterogeneous knowledge graph, and the comprehensive scores of candidate power supply decision paths. It also retains the corresponding update records so that it can be used to restore the evidence and conclusions of power supply decisions at any time.

[0183] In one embodiment, the method further includes:

[0184] Based on the comprehensive scores, attribute indicators, confidence weights and completeness indices of the optimal power supply decision path and the auxiliary power supply decision path, a power supply decision result dataset containing a complete chain of evidence is generated. The power supply decision result dataset and corresponding power supply decision optimization suggestions are output in the form of a graphical network view or a tabular list.

[0185] In one embodiment, abnormal data processing and missing data supplementation are performed on power supply-related data, including:

[0186] The sliding statistical method is used to calculate the center value and fluctuation range of power supply related data within a preset time period. Power supply related data outside the fluctuation range are marked as abnormal, the abnormal data are deleted, and the center value is used to replace and mark them.

[0187] Missing numerical power supply related data will be filled in with the median of the same type of data. Missing categorized power supply related data will be placed and labeled with preset fields. Missing timestamp-type power supply related data will be backfilled with data collection time and labeled.

[0188] In one embodiment, the entity nodes in the heterogeneous knowledge graph include power grid equipment, operation and maintenance personnel, power supply lines, substations, and user load center types in the power supply field; the semantic relationships represented by the edges in the heterogeneous knowledge graph include equipment operation and maintenance, line connection, power supply service, and fault association types; the multidimensional attributes of the edges in the heterogeneous knowledge graph include equipment operating years, line load rate, power supply reliability level, fault handling time, and operation and maintenance cost.

[0189] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0190] In one embodiment, selecting the optimal power supply decision path and the auxiliary power supply decision path from the set of candidate power supply decision paths includes:

[0191] For the candidate power supply decision paths in the candidate power supply decision path set, the attribute indicators attached to each edge of the candidate power supply decision path are dimensionless; based on the correlation between the attribute indicators and the power supply business, and the historical power supply data related to the attribute indicators, the corresponding weights of the attribute indicators are determined; based on the dimensionless attribute indicators and their corresponding weights, and the confidence weights of each edge of the candidate power supply decision path, the comprehensive score of the candidate power supply decision path is calculated and sorted to obtain the path ranking result;

[0192] For the sorted power supply decision paths, the original data, the corresponding unique identifier of the original data, the data source label, and the data collection time are associated with the power supply-related data for each edge and node, along with the confidence weight of each edge, to construct a complete power supply decision evidence chain; based on the evidence chain integrity assessment standard, the integrity index of the power supply decision evidence chain is calculated; the interpretability of the power supply decision path is verified according to the integrity index to obtain the evidence chain verification result, and the power supply decision paths with insufficient evidence or confidence weights below the preset threshold are marked with risks;

[0193] Based on the path ranking results and evidence chain verification results, the optimal power supply decision path and the auxiliary power supply decision path are selected.

[0194] In one embodiment, the method further includes:

[0195] It continuously receives incremental data from various business processes of the power supply system. Based on the incremental data, it updates attribute indicators, their corresponding weights, the confidence weights of edges in the heterogeneous knowledge graph, and the comprehensive scores of candidate power supply decision paths. It also retains the corresponding update records so that it can be used to restore the evidence and conclusions of power supply decisions at any time.

[0196] In one embodiment, the method further includes:

[0197] Based on the comprehensive scores, attribute indicators, confidence weights and completeness indices of the optimal power supply decision path and the auxiliary power supply decision path, a power supply decision result dataset containing a complete chain of evidence is generated. The power supply decision result dataset and corresponding power supply decision optimization suggestions are output in the form of a graphical network view or a tabular list.

[0198] In one embodiment, abnormal data processing and missing data supplementation are performed on power supply-related data, including:

[0199] The sliding statistical method is used to calculate the center value and fluctuation range of power supply related data within a preset time period. Power supply related data outside the fluctuation range are marked as abnormal, the abnormal data are deleted, and the center value is used to replace and mark them.

[0200] Missing numerical power supply related data will be filled in with the median of the same type of data. Missing categorized power supply related data will be placed and labeled with preset fields. Missing timestamp-type power supply related data will be backfilled with data collection time and labeled.

[0201] In one embodiment, the entity nodes in the heterogeneous knowledge graph include power grid equipment, operation and maintenance personnel, power supply lines, substations, and user load center types in the power supply field; the semantic relationships represented by the edges in the heterogeneous knowledge graph include equipment operation and maintenance, line connection, power supply service, and fault association types; the multidimensional attributes of the edges in the heterogeneous knowledge graph include equipment operating years, line load rate, power supply reliability level, fault handling time, and operation and maintenance cost.

[0202] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0203] In one embodiment, selecting the optimal power supply decision path and the auxiliary power supply decision path from the set of candidate power supply decision paths includes:

[0204] For the candidate power supply decision paths in the candidate power supply decision path set, the attribute indicators attached to each edge of the candidate power supply decision path are dimensionless; based on the correlation between the attribute indicators and the power supply business, and the historical power supply data related to the attribute indicators, the corresponding weights of the attribute indicators are determined; based on the dimensionless attribute indicators and their corresponding weights, and the confidence weights of each edge of the candidate power supply decision path, the comprehensive score of the candidate power supply decision path is calculated and sorted to obtain the path ranking result;

[0205] For the sorted power supply decision paths, the original data, the corresponding unique identifier of the original data, the data source label, and the data collection time are associated with the power supply-related data for each edge and node, along with the confidence weight of each edge, to construct a complete power supply decision evidence chain; based on the evidence chain integrity assessment standard, the integrity index of the power supply decision evidence chain is calculated; the interpretability of the power supply decision path is verified according to the integrity index to obtain the evidence chain verification result, and the power supply decision paths with insufficient evidence or confidence weights below the preset threshold are marked with risks;

[0206] Based on the path ranking results and evidence chain verification results, the optimal power supply decision path and the auxiliary power supply decision path are selected.

[0207] In one embodiment, the method further includes:

[0208] It continuously receives incremental data from various business processes of the power supply system. Based on the incremental data, it updates attribute indicators, their corresponding weights, the confidence weights of edges in the heterogeneous knowledge graph, and the comprehensive scores of candidate power supply decision paths. It also retains the corresponding update records so that it can be used to restore the evidence and conclusions of power supply decisions at any time.

[0209] In one embodiment, the method further includes:

[0210] Based on the comprehensive scores, attribute indicators, confidence weights and completeness indices of the optimal power supply decision path and the auxiliary power supply decision path, a power supply decision result dataset containing a complete chain of evidence is generated. The power supply decision result dataset and corresponding power supply decision optimization suggestions are output in the form of a graphical network view or a tabular list.

[0211] In one embodiment, abnormal data processing and missing data supplementation are performed on power supply-related data, including:

[0212] The sliding statistical method is used to calculate the center value and fluctuation range of power supply related data within a preset time period. Power supply related data outside the fluctuation range are marked as abnormal, the abnormal data are deleted, and the center value is used to replace and mark them.

[0213] Missing numerical power supply related data will be filled in with the median of the same type of data. Missing categorized power supply related data will be placed and labeled with preset fields. Missing timestamp-type power supply related data will be backfilled with data collection time and labeled.

[0214] In one embodiment, the entity nodes in the heterogeneous knowledge graph include power grid equipment, operation and maintenance personnel, power supply lines, substations, and user load center types in the power supply field; the semantic relationships represented by the edges in the heterogeneous knowledge graph include equipment operation and maintenance, line connection, power supply service, and fault association types; the multidimensional attributes of the edges in the heterogeneous knowledge graph include equipment operating years, line load rate, power supply reliability level, fault handling time, and operation and maintenance cost.

[0215] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0216] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0217] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0218] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A decision path generation method in a power supply scenario, characterized in that, The method includes: Power supply-related data are extracted from multiple heterogeneous data sources across various business processes of the power supply system; these heterogeneous data sources include power grid operation monitoring data, equipment maintenance record data, power load data, and power supply fault handling data. The power supply-related data is processed for abnormal data and missing data is supplemented. The text fields in the power supply-related data are processed for word segmentation and stop word removal. The numerical fields in the power supply-related data are processed for normalization. Based on the preprocessed power supply-related data, a heterogeneous knowledge graph in the power supply field is constructed according to the pattern of entity, attribute, and relationship. The edges of the heterogeneous knowledge graph are attached with multi-dimensional attribute indicators. The confidence weight of the edges in the heterogeneous knowledge graph is determined based on the number and quality of data sources. On the heterogeneous knowledge graph, a starting set and an ending set for power supply decisions are set. The starting set includes power supply anomaly scenario nodes, and the ending set includes power supply decision target nodes. Starting from the starting point in the starting set, a breadth-first search algorithm is used to expand the path on the heterogeneous knowledge graph hierarchically. When the path reaches the ending point in the ending set, the branch expansion stops, resulting in a candidate power supply decision path set. From the set of candidate power supply decision paths, the optimal power supply decision path and the auxiliary power supply decision path are selected.

2. The method according to claim 1, characterized in that, The step of selecting the optimal power supply decision path and the auxiliary power supply decision path from the candidate power supply decision path set includes: For the candidate power supply decision paths in the candidate power supply decision path set, the attribute indicators attached to each edge of the candidate power supply decision path are dimensionless; based on the correlation between the attribute indicators and power supply services, and the historical power supply data related to the attribute indicators, the corresponding weights of the attribute indicators are determined; based on the dimensionless attribute indicators and their corresponding weights, and the confidence weights of each edge of the candidate power supply decision path, the comprehensive score of the candidate power supply decision path is calculated and sorted to obtain the path ranking result; For the sorted power supply decision path, the original data, the corresponding unique identifier of the original data, the data source label, and the data collection time in the power supply related data are associated with each edge and node, and the confidence weight of each edge is attached to construct a complete power supply decision evidence chain; based on the evidence chain integrity evaluation standard, the integrity index of the power supply decision evidence chain is calculated; the interpretability of the power supply decision path is verified according to the integrity index to obtain the evidence chain verification result, and the power supply decision path with insufficient evidence or confidence weight below the preset threshold is marked with risk; Based on the path sorting results and the evidence chain verification results, the optimal power supply decision path and the auxiliary power supply decision path are selected.

3. The method according to claim 1, characterized in that, The method further includes: The system continuously receives incremental data from various business processes of the power supply system. Based on the incremental data, it updates the attribute indicators, the corresponding weights of the attribute indicators, the confidence weights of the edges in the heterogeneous knowledge graph, and the comprehensive score of the candidate power supply decision paths. It also retains the corresponding update records so that the power supply decision evidence and conclusions can be restored at any time.

4. The method according to claim 1, characterized in that, The method further includes: Based on the comprehensive scores of the optimal power supply decision path and the auxiliary power supply decision path, the attribute indicators, the confidence weights and completeness indices of each side, a power supply decision result dataset containing a complete chain of evidence is generated. The power supply decision result dataset and corresponding power supply decision optimization suggestions are output in the form of a graphical network view or a tabular list.

5. The method according to claim 1, characterized in that, The process of handling abnormal data and supplementing missing data in the power supply-related data includes: The sliding statistical method is used to calculate the center value and fluctuation range of power supply related data within a preset time period. Power supply related data outside the fluctuation range are marked as abnormal, abnormal data are deleted, and the center value is used to replace and label them. Missing numerical power supply related data will be filled in with the median of the same type of data. Missing categorized power supply related data will be placed and labeled with preset fields. Missing timestamp-type power supply related data will be backfilled with data collection time and labeled.

6. The method according to claim 1, characterized in that, The entity nodes in the heterogeneous knowledge graph include power grid equipment, operation and maintenance personnel, power supply lines, substations, and user load center types in the power supply field; the semantic relationships represented by the edges in the heterogeneous knowledge graph include equipment operation and maintenance, line connection, power supply service, and fault association types; the multidimensional attributes of the edges in the heterogeneous knowledge graph include equipment operating years, line load rate, power supply reliability level, fault handling time, and operation and maintenance cost.

7. A decision path generation device for a power supply scenario, characterized in that, The device includes: The extraction module is used to extract power supply-related data from multiple heterogeneous data sources in various business processes of the power supply system; the heterogeneous data sources include power grid operation monitoring data, equipment maintenance record data, power load data, and power supply fault handling data. The processing module is used to process abnormal data and supplement missing data in the power supply related data, perform word segmentation and stop word removal on the text fields in the power supply related data, and perform normalization on the numerical fields in the power supply related data. The construction module is used to construct a heterogeneous knowledge graph in the power supply field based on preprocessed power supply-related data and in accordance with the pattern of entities, attributes, and relationships. The edges of the heterogeneous knowledge graph are attached with multi-dimensional attribute indicators. The confidence weight of the edges in the heterogeneous knowledge graph is determined based on the number and quality of data sources. The candidate module is used to set a starting point set and an ending point set for power supply domain decisions on the heterogeneous knowledge graph. The starting point set includes power supply anomaly scenario nodes, and the ending point set includes power supply decision target nodes. Starting from the starting point in the starting point set, a breadth-first search algorithm is used to expand the path on the heterogeneous knowledge graph hierarchically, and the branch expansion stops when the path reaches the ending point in the ending point set, thus obtaining a candidate power supply decision path set. The filtering module is used to filter out the optimal power supply decision path and the auxiliary power supply decision path from the set of candidate power supply decision paths.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images