Power outage event-driven distribution network state analysis decision system and method

Abstract

This invention discloses a power outage event-driven distribution network status analysis and decision-making system and method, relating to the field of distribution network operation management and reliability analysis technology. The system includes: a data access module: obtaining intermediate data packets based on system data from different systems; a data cleaning module: correcting intermediate data packets and mounting them to the power grid topology model to obtain a power outage data sequence; a data association module: obtaining association results based on the power outage data sequence; an event generation module: obtaining a power outage event pool based on the association results; a data calculation module: analyzing the power outage event pool to obtain analysis data; a data analysis module: obtaining analysis results based on the analysis data; and a decision support module: obtaining decision schemes based on the analysis results, early warning information, and investment benefit evaluation models. This invention addresses the problems of existing distribution network power outage event analysis systems, which struggle to effectively integrate multi-source heterogeneous power outage data, and where the identification of power outage events and the actual operating status of equipment relies on human experience, resulting in low efficiency and a high risk of misjudgment or omission.

Description

Technical Field

[0001] This invention relates to the field of distribution network operation management and reliability analysis technology, specifically to a distribution network status analysis and decision-making system and method driven by power outage events. Background Technology

[0002] With the continuous expansion of power distribution networks and the increasing demands from users for power supply reliability, the traditional operation and management model, which relies heavily on manual labor and experience, faces severe challenges. Currently, the perception and analysis of power outage events in power distribution networks mainly suffer from the following prominent problems:

[0003] First, power outage data is scattered and heterogeneous, forming data silos. In actual operation and maintenance, power outage-related information is dispersed across multiple independent systems, such as dispatch automation systems, fault indicators, electricity consumption information collection systems, and 95598 customer service work orders. These data are inconsistent in terms of format, recording time, and data quality, lacking a unified data governance and integration mechanism. This makes it difficult to form a complete and accurate panoramic view of power outage events, creating obstacles for subsequent analysis and decision-making.

[0004] Secondly, the identification, judgment, and in-depth analysis of power outage events still heavily rely on human experience, with insufficient automation and intelligence. Current methods require maintenance personnel to sift through massive amounts of mixed alarm and event information, relying on experience to filter, judge, and summarize power outage events and the actual operating status and level of power grid equipment. This process is cumbersome, inefficient, and prone to misjudgments or omissions due to information overload or differences in experience, affecting the precision of maintenance and repair, the speed of fault repair response, and the timeliness of power restoration. Summary of the Invention

[0005] To address the shortcomings of existing power outage event analysis systems, such as the inability to effectively integrate multi-source heterogeneous outage data and the reliance on manual experience for identifying outage events and the actual operating status of grid equipment, which leads to low efficiency and susceptibility to misjudgments or omissions, this invention provides a power outage event-driven distribution network status analysis and decision-making system. The system includes:

[0006] Data access module: used to obtain intermediate data packets based on system data from different target systems, and to obtain a cache database based on the intermediate data packets;

[0007] Data cleaning module: used to correct the intermediate data packets, attach the corrected intermediate data packets to the power grid topology model, and obtain the power outage data sequence;

[0008] Data association module: used to obtain several target signals based on the power outage data sequence, associate the target signals, and obtain association results;

[0009] Event generation module: used to obtain power outage events based on the correlation results, and to obtain a power outage event pool based on the power outage events;

[0010] Data calculation module: used to analyze the power outage event pool to obtain analysis data;

[0011] Data analysis module: used to analyze the data and obtain analysis results;

[0012] Decision support module: used to obtain early warning information based on a pre-trained health assessment model and third-party data; to construct an investment benefit assessment model; and to obtain decision-making solutions based on the analysis results, the early warning information, and the investment benefit assessment model.

[0013] This system accesses raw outage-related data from multiple heterogeneous source systems in the distribution network, standardizes it to obtain intermediate data packets, and then maps all data onto a unified distribution network topology model (including substations, lines, transformers, and user hierarchical relationships) according to a time base, forming a standardized outage data sequence with spatiotemporal correlation. Through unified data governance and fusion, it effectively integrates scattered multi-source outage data, breaks down data silos, achieves integrated governance, and solves the problems of inconsistent data formats, time scales, and topologies, laying a high-quality data foundation for advanced applications. Utilizing a data association module, relevant target signals are correlated, and preliminary outage event records are automatically generated based on the association results, thereby improving the automation and accuracy of event identification. This enables automatic, rapid, and accurate assessment of outage events, significantly reducing manual intervention and improving the efficiency and reliability of event handling. A high-confidence outage event database is automatically generated and updated based on the association results. Based on this database, power supply reliability indicators are calculated and analyzed in real time and from multiple dimensions. Equipment risk prediction and investment benefit simulation are performed based on real-time indicators, and early warning and decision support information is output.

[0014] Furthermore, the data access module is specifically used for:

[0015] A data source registry and data adapters are constructed. The data source registry contains registration data for different target systems. The registration data includes source system identifier, interface type, connection parameters, and data characteristics. Each data adapter corresponds to a different target system.

[0016] Based on the data source registry and the data adapter, obtain the system data for each of the target systems;

[0017] The intermediate data packet is obtained based on the system data, the source system identifier corresponding to the system data, and the original reception timestamp of the system data;

[0018] Configure metadata for each intermediate data packet, the metadata including a unique serial number, a data confidence identifier, and a topology association key;

[0019] The cache database is obtained based on the intermediate data packet.

[0020] By configuring diverse communication interface adapters, raw power outage-related data can be reliably, efficiently, and automatically collected from multiple heterogeneous external source systems. This enables unified, reliable, and traceable access to multi-source, heterogeneous, and asynchronous data, unifying the data entry point, breaking down data silos, and providing the possibility for subsequent integrated analysis. It also provides a stable, rich data stream with preliminary contextual information for subsequent in-depth governance and integration, demonstrating a targeted solution capability for the complex data environment of the distribution network and ensuring the robustness and scalability of the system. The data is written to a highly available cache database (such as Redis) to achieve data buffering and decoupling, ensuring that the raw data is not lost and does not affect the performance of the source system when facing instantaneous data surges or temporary congestion of downstream processing modules.

[0021] Furthermore, the data cleaning module is specifically used for:

[0022] Construct a data parser for each of the target systems, with each data parser corresponding one-to-one with a source system identifier;

[0023] The data parser parses the intermediate data packet to obtain field-value pairs, verifies the compliance of the field-value pairs, and obtains the quality label of the intermediate data packet.

[0024] Based on the quality label, a problem data packet is obtained. The topology association key of the problem data packet is obtained to obtain the problem device identifier. If the problem device identifier does not exist in the power grid topology model or the problem device identifier conflicts with an existing device identifier in the power grid topology model, the problem device identifier is associated and mapped based on the device alias library and historical ledger change records to obtain the corrected device identifier of the problem data packet. Based on the corrected device identifier, the problem device identifier of the problem data packet is corrected to obtain the corrected intermediate data packet.

[0025] The corrected intermediate data packet is attached to the power grid topology model, and a topology coordinate chain of the corrected intermediate data packet is generated based on the power grid topology model and the topology association key. A standard data object is obtained based on the corrected intermediate data packet and the topology coordinate chain.

[0026] According to the time sequence and topological relationship, the standard data objects are stored in the time series database and the graph database respectively, and the power outage data sequence is obtained based on the time series database and the graph database.

[0027] The system performs in-depth processing on the raw data units provided by the data access module, which contain preliminary metadata, and strictly aligns them to spatiotemporal and topological benchmarks. This achieves the transformation from raw, heterogeneous, and messy data to clean, consistent, and high-quality information with clear spatiotemporal and topological details. Through a topology-forced attachment mechanism, each piece of data (such as a signal or a user's repair request) is precisely attached to a unified power grid equipment node (such as a substation-feeder-transformer-user), and a complete topological location coding chain is generated. Finally, a standardized data stream with a unified time benchmark, unified equipment identification, and clear topological relationships is output. This fundamentally solves the core pain point of data not being able to be correlated in a unified space, laying an indispensable data foundation for building a precise panoramic view of power outage events.

[0028] Furthermore, the data association module is specifically used for:

[0029] Construct a multi-source evidence chain rule base, time association windows for different signal types, and topological association radii for different signal types. The multi-source evidence chain rule base includes ideal evidence chain templates for different power outage scenarios.

[0030] Preset basic weights for different signal sources and signal types; based on the basic weights, obtain several seed signals from the target signal within the preset weight range; based on the time association window and the topological association radius, obtain association signals associated with the seed signals to obtain several association clusters.

[0031] Starting from the seed signal, the network topology is traversed downstream of the seed signal to obtain the ideal affected nodes, the actual affected nodes are obtained based on the associated clusters, and the affected ratio is obtained based on the ideal affected nodes and the actual affected nodes.

[0032] The association clusters are matched with the multi-source evidence chain rule base to obtain the evidence completeness value;

[0033] The overall confidence level of the associated cluster is obtained based on the evidence completeness value and the affected proportion;

[0034] The association results are obtained based on the comprehensive confidence level and the preset confidence threshold, and a power outage event profile is generated based on the association clusters.

[0035] Using high-weighted signals (such as circuit breaker tripping) as seeds, and based on predefined spatiotemporal correlation windows and power grid topology connections, the system automatically searches for and captures all other temporally and spatially related signals (such as fault indicator actions and user power outages), forming a correlation cluster. Subsequently, the completeness and rationality of this signal cluster are evaluated using a multi-source evidence chain rule base, and a comprehensive confidence calculation model is used to determine its likelihood of representing a real power outage event. This automatically aggregates discrete and chaotic signals into a structured preliminary event profile representing the same power outage event. Through a four-step closed-loop process of seed triggering, topology traversal, evidence chain matching, and confidence assessment, a qualitative leap is achieved from passively receiving signals to actively and accurately constructing power outage events. This enables automatic, rapid, and accurate identification of real power outage events from massive amounts of alarm information, significantly improving the efficiency and accuracy of analysis, reducing reliance on human intervention and human error, and effectively solving the pain points of manual labor, low efficiency, and error susceptibility in the background technology. It transforms complex discrete signals into highly reliable structured event knowledge that can be directly used for indicator calculation and decision-making.

[0036] Furthermore, the event generation module is specifically used for:

[0037] Based on the association results, several first events are generated, and the event characteristics of the first events are obtained. The event characteristics include event type, event outage start time, event level, and event impact range.

[0038] Acquire the first recovery signal of the power supply equipment and the second recovery signal of the user equipment in the first event, obtain the reliability value of the first event based on the first recovery signal and the second recovery signal, and correct the power outage end time of the first event based on the reliability value;

[0039] The impact range of the event and the geographic information system layer are overlaid to obtain the overlay result, and the contradiction value is obtained based on the overlay result;

[0040] The overall confidence level of the first event is adjusted based on the reliability value and the contradiction value to obtain the adjusted confidence level;

[0041] The power outage event is obtained based on the corrected confidence level and the preset confidence threshold, and a power outage event pool is obtained based on the power outage event.

[0042] The system automatically adjudicates the associated results, determining the event type, level, start and end times, and refined impact scope. A multi-layered closed-loop verification mechanism is introduced: the generated event is reverse-verified with subsequent power restoration signals (process integrity verification), and reliable restoration signals are used to check and correct the outage end time, making outage duration calculations more accurate. It also verifies the event with external data such as geographic information (logical rationality verification), dynamically calibrating the event confidence level based on the verification results, ultimately generating highly reliable standard outage event records suitable for statistics and decision-making. This constructs a dynamic quality loop of generation-verification-correction, moving beyond simple archiving of associated results to proactive, reverse, and cross-validation with restoration signals and geographic information, significantly improving the accuracy and authority of event records. This mechanism fundamentally solves the problems of error-prone manual judgment and the disconnect between event records and actual conditions in the background technology, ensuring a solid and reliable data foundation for subsequent reliability index calculations and decisions.

[0043] Furthermore, the data calculation module is specifically used for:

[0044] Construct multi-dimensional indicators, and obtain affected indicators based on the multi-dimensional indicators and the power outage event pool. The multi-dimensional indicators include time, administrative region, voltage level, power supply unit, feeder and even distribution transformer.

[0045] Based on the affected users of the power outage event, generate user-power outage event atomic pairs;

[0046] Based on the affected indicators, the user-outage event atomic pairs, and the hierarchical relationship of the power grid topology model, the system triggers the data from bottom to top along the built-in topology relationship chain of the power grid topology model, synchronously updating the indicator values ​​of the affected indicators corresponding to the higher level, thereby obtaining the analysis data.

[0047] Abandoning the traditional offline, batch processing statistical model, it achieves second-level perception and accurate calculation based on real-time updates of a verified power outage event database, using streaming computing and dynamic aggregation technology.

[0048] Furthermore, the data analysis module is specifically used for:

[0049] The analytical data is analyzed based on standard reliability indicators and extended management indicators to obtain indicator analysis results. The standard reliability indicators include the system average power outage time, the system average power outage frequency, the user average power outage time, and the average power supply reliability rate. The extended management indicators include the average duration of fault power outages, the average duration of planned power outages, the frequency of fault power outages, and the number of users experiencing repeated power outages.

[0050] Based on the analysis results of the indicators, abnormal indicators are obtained, and the administrative regions corresponding to the abnormal indicators are obtained to obtain abnormal regions. Based on the abnormal regions and the preset management paths, management responsibility units are obtained, and based on the management responsibility units and the topology paths of the power grid topology model, problem location chains are obtained.

[0051] Obtain several related events by acquiring power outage events related to the problem localization chain, obtain the event causes of the related events, and obtain root cause analysis results based on the event causes;

[0052] The analysis results are obtained based on the indicator analysis results, the problem localization chain, and the root cause analysis results.

[0053] In-depth analysis, top-down problem identification, and bottom-up root cause tracing, solves the problem of unclear reasons behind inflated indicators, and transforms abstract numbers into precise insights that can guide action.

[0054] Furthermore, the decision support module is specifically used for:

[0055] Based on the analysis results, a list of weak links in reliability is obtained, and based on the early warning information, a list of high-risk equipment is obtained.

[0056] Construct an investment solution library and a market price library, wherein the investment solution library contains different types of solutions and quantitative mapping rules corresponding to each solution;

[0057] The decision-making scheme is obtained based on the list of weak links in reliability, the list of high-risk equipment warnings, the market price database, the investment scheme database, and the investment benefit evaluation model.

[0058] By establishing a digital mapping rule for problems, solutions, and benefits, the effects of engineering projects can be quantified in advance as reliability and financial indicators, thereby maximizing the allocation of limited investment resources.

[0059] Further, the first event is generated based on the association result and the event generation rules, wherein the event generation rules include:

[0060] Event type determination rules: If the associated cluster contains a switch protection trip signal of a protection device, and the switch protection trip signal does not match the pre-scheduled power outage plan, it is determined to be a fault power outage; if the switch change signal in the associated cluster matches the work order issued in the production management system in both time and equipment, it is determined to be a planned power outage; if the associated cluster contains a remote control trip signal or a manual trip signal, but there is no corresponding work order, it is determined to be a temporary power outage.

[0061] Event severity rating rules: The severity rating is determined by defining a range of severity levels based on a threshold number of affected users.

[0062] Key event parameter extraction rules: Obtain the switch protection action time as the power outage start time; starting from the switch that triggered the event, traverse downstream along the power grid topology to obtain the power outage impact range.

[0063] This invention also provides a power outage event-driven distribution network status analysis and decision-making method, the method comprising:

[0064] A cache database is obtained based on system data from different target systems;

[0065] Process the cached database to obtain the power outage data sequence;

[0066] Based on the power outage data sequence, several target signals are obtained, and the target signals are correlated to obtain the correlation result;

[0067] Based on the correlation results, power outage events are obtained, and based on the power outage events, a power outage event pool is obtained;

[0068] Analyze the power outage event pool to obtain analytical data;

[0069] The analysis results are obtained by analyzing the data.

[0070] Early warning information is obtained based on a pre-trained health assessment model and third-party data; an investment benefit assessment model is constructed, and a decision-making scheme is obtained based on the analysis results, the early warning information, and the investment benefit assessment model.

[0071] The principle and effect of this method are similar to those of this system, and therefore, no further details will be provided for this method.

[0072] One or more technical solutions provided by this invention have at least the following technical effects or advantages:

[0073] This system accesses raw outage-related data from multiple heterogeneous source systems in the distribution network, standardizes it to obtain intermediate data packets, and then maps all data onto a unified distribution network topology model according to a time base, forming a standardized outage data sequence with spatiotemporal correlation. Through unified data governance and fusion, it effectively integrates scattered multi-source outage data, breaks down data silos, achieves integrated governance, and solves the problems of inconsistent data formats, time scales, and topologies, laying a high-quality data foundation for advanced applications. Utilizing a data association module, relevant target signals are correlated, and preliminary outage event records are automatically generated based on the association results, thereby improving the automation and accuracy of event identification. This enables automatic, rapid, and accurate assessment of outage events, significantly reducing manual intervention and improving the efficiency and reliability of event handling. A high-confidence outage event database is automatically generated and updated based on the association results. Based on this database, power supply reliability indicators are calculated and analyzed in real time and from multiple dimensions. Equipment risk prediction and investment benefit simulation are performed based on real-time indicators, outputting early warning and decision support information. Attached Figure Description

[0074] The accompanying drawings, which are provided to further illustrate embodiments of the invention and constitute a part of this invention, are not intended to limit the scope of the invention.

[0075] Figure 1 This is a flowchart illustrating the power outage event-driven distribution network status analysis and decision-making system of this invention. Detailed Implementation

[0076] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, where there is no conflict, the embodiments of the present invention and the features thereof can be combined with each other.

[0077] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0078] Example 1

[0079] refer to Figure 1 This embodiment provides a power outage event-driven distribution network status analysis and decision-making system, the system comprising:

[0080] Data access module: used to obtain intermediate data packets based on system data from different target systems, and to obtain a cache database based on the intermediate data packets;

[0081] Data cleaning module: used to correct the intermediate data packets, attach the corrected intermediate data packets to the power grid topology model, and obtain the power outage data sequence;

[0082] Data association module: used to obtain several target signals based on the power outage data sequence, associate the target signals, and obtain association results;

[0083] Event generation module: used to obtain power outage events based on the correlation results, and to obtain a power outage event pool based on the power outage events;

[0084] Data calculation module: used to analyze the power outage event pool to obtain analysis data;

[0085] Data analysis module: used to analyze the data and obtain analysis results;

[0086] Decision support module: used to obtain early warning information based on pre-trained health assessment models and third-party data (such as meteorological data and geographic information systems); to construct investment benefit assessment models (for cost-benefit analysis, which can be obtained through pre-training using artificial intelligence or machine learning techniques); and to obtain decision-making solutions based on the analysis results, the early warning information, and the investment benefit assessment models.

[0087] Specifically, the data access module is used for:

[0088] A data source registry and data adapters are constructed. The data source registry contains registration data for different target systems. The registration data includes source system identifiers, interface types, connection parameters, and data characteristics. Each data adapter corresponds to a different target system. In this example, the target systems may include a dispatch automation system (SCADA / DMS), a fault indicator master station system, an electricity information acquisition system (AMI), a production management system (PMS), and a customer service system (such as 95598). For each type of data source, its registration data is defined in the registry.

[0089] Source system identifier: The system from which the data originates;

[0090] Interface types: such as direct database connection (JDBC / ODBC), message queue (Kafka, RabbitMQ), web service (RESTful API / SOAP), file service (FTP / SFTP) or dedicated communication protocols (such as 101 / 104 protocols).

[0091] Connection parameters may include network address, port, access credentials, heartbeat detection frequency, etc.

[0092] Data characteristics: Indicates the data content that the source can provide, such as switch position change and protection action signals, line telemetry over-limit alarms, user meter undervoltage events, fault indicator flip-up and reset signals, planned power outage work orders and fault repair work orders, etc.

[0093] Based on the data source registry and the data adapter, the system data of each target system is obtained; in this example, each data adapter can be dedicated to one type or class of data sources, and different data acquisition strategies are adopted according to the characteristics of the data source.

[0094] For high real-time, event-driven data sources (such as dispatch automation systems and fault indicator systems): the adapter adopts an active listening / subscription mode. For example, a real-time communication channel is established with the SCADA system (e.g., based on the IEC 104 protocol) to continuously listen for real-time alarm events such as communication changes, total fault signals, and protection actions actively sent by the system. Message queue access follows the same principle;

[0095] For data sources that are updated in batches and on a regular schedule (such as electricity consumption information collection systems and production management systems): the adapter adopts a timed polling mode. According to a preset period (such as every 5 minutes), it retrieves the data that has been added or changed since the last polling through API or by querying the database, such as the zero value records of voltage and current for batch users, and newly generated planned power outage plans, etc.

[0096] For hybrid data sources (such as customer service systems): a combination of event notifications and scheduled completion is adopted. On the one hand, event notifications for new work orders are subscribed to; on the other hand, periodic polling is used to ensure that work order status updates (such as order acceptance and archiving) are synchronized.

[0097] Based on the system data, the source system identifier corresponding to the system data, and the original reception timestamp of the system data, the intermediate data packets are obtained; the metadata of each intermediate data packet is configured, and key metadata is attached to form an enhanced data unit. The metadata includes a unique serial number (for end-to-end tracking), a data confidence identifier (preliminarily marked according to the reliability level of the data source itself (e.g., high for SCADA signals, medium for some field terminal signals)) and a topology association key (based on the device number carried in the data packet (e.g., switch ID, transformer ID, user meter ID), the distribution network topology model service is quickly queried to obtain the globally unique code of the device in the unified topology and its line, substation, and other information, and this topology association key is injected into the data).

[0098] The cache database (such as Redis) is obtained based on the intermediate data packet.

[0099] Specifically, the data cleaning module is used for:

[0100] Construct a data parser for each target system, with each data parser corresponding to a source system identifier. Invoke the corresponding data parser based on the source system identifier.

[0101] The data parser parses the intermediate data packets to obtain field-value pairs. For example, it parses the device ID, action time, and alarm type from SCADA alarm messages; and parses the user address, complaint time, and fault description from 95598 work orders. It verifies the compliance of the field-value pairs to obtain the quality label of the intermediate data packets. It then performs compliance checks on the parsed fields, including:

[0102] Non-empty check: Check if key fields (such as device ID, timestamp) are missing;

[0103] Format compliance check: Whether the timestamp format meets expectations, whether the device ID conforms to the coding specifications, etc.;

[0104] Logical rationality check: such as whether the voltage value is within the physically possible range, and whether the timestamp is significantly ahead of the current system time.

[0105] Label each data record with a quality tag (e.g., complete, missing key fields, time anomalies, out-of-bounds values).

[0106] Based on the quality tags, problematic data packets are obtained, and data with quality issues is repaired and enhanced.

[0107] Time zone and precision unification: Convert all timestamps to standard UTC time or local unified time zone time, and unify the precision to the millisecond level;

[0108] Error Time Correction: For obviously erroneous timestamps (such as future times or excessively distant past times), a multi-source alignment strategy is used for inference. For example, if the timestamp of a fault indicator action signal is seriously erroneous, it can be reasonably corrected based on the times of other signals on the same line, the times of related switch changes, or the concentrated time window of power outage events for similar users.

[0109] Fuzzy matching and association: For non-standard or descriptive identifiers (such as user addresses in work orders: XX Road XX No.), fuzzy matching is performed through the address standardization engine and geographic information database to associate them with standard distribution transformers or user metering point numbers.

[0110] Topology-based verification: Obtain the problem device identifier by acquiring the topology association key of the problem data packet. If the problem device identifier does not exist in the power grid topology model or conflicts with an existing device identifier in the power grid topology model, then the problem device identifier is associated and mapped based on the device alias library and historical ledger change records to obtain the corrected device identifier of the problem data packet. Based on the corrected device identifier, the problem device identifier of the problem data packet is corrected to obtain the corrected intermediate data packet. Alternatively, the topology association key can be used to verify whether the device identifier exists in the current power grid topology model. If it does not exist or conflicts, a query is initiated, and the device alias library or historical ledger change records are used for mapping to ensure the uniqueness and accuracy of the identifier.

[0111] Key information completion: Automatically fills in missing key information based on data source type and context. For example, for a trip signal that only has a switch number, it automatically fills in the name of the substation and feeder from the topology model.

[0112] The corrected intermediate data packets are then attached to the power grid topology model, forcibly and precisely linking each data entry to a unified distribution network topology model. This not only associates the data with the device itself but also determines the upstream power supply path and downstream power supply range it affects. For example, if a distribution transformer fails, not only should the transformer be tagged, but it should also be automatically associated with a list of all users affected by the power outage.

[0113] Based on the power grid topology model and the topology association key, a topology coordinate chain is generated for the corrected intermediate data packet. A complete topology hierarchical encoding chain from substation-feeder-section switch-distribution transformer-user is generated for each data packet. This encoding chain is the fundamental basis for subsequent judgment of the scope of event impact and association of signals from different sources.

[0114] Based on the corrected intermediate data packet and the topology coordinate chain, a standard data object is obtained; the processed data is converted into a standard data object defined internally by the system, which may be required to include the following core fields: globally unique ID, authoritative timestamp, standardized device / user topology coordinate chain, event type / status value, data quality score, and original data traceability ID.

[0115] The standard data objects are stored in a time-series database and a graph database according to time sequence and topological relationships, respectively. The power outage data sequence is obtained based on the time-series database and the graph database. The generated standard data objects are written into the time-series database and the graph database according to time sequence and topological relationships. The time-series database supports efficient time-series queries, while the graph database accurately stores the topological relationships between data.

[0116] Specifically, the data association module is used for:

[0117] A multi-source evidence chain rule base, time association windows for different signal types, and topological association radii for different signal types are constructed. Dynamic time association windows and topological association radii are defined for each signal type (e.g., switch tripping, fault indicator action, user voltage loss group). For example, a substation outgoing switch tripping signal triggers a search for related signals from all downstream devices within the next 5 minutes. The time window can be adaptively adjusted according to voltage level and regional characteristics. The multi-source evidence chain rule base contains ideal evidence chain templates for different power outage scenarios; for example, a typical ideal evidence chain for a line fault power outage is: switch protection tripping signal - upstream fault indicator flip-up signal - downstream multiple distribution transformer associated user group batch voltage loss signals (impact range evidence) - possible secondary reclosing failure signals (process evidence). The rules in this multi-source evidence chain rule base define the causal, temporal, and spatial logical relationships between signals.

[0118] Preset basic weights for different signal sources and signal types. For example, SCADA switch position change signals have the highest weight (e.g., 0.9), power distribution automation terminal fault indication signals have the next highest weight (e.g., 0.7), while individual user meter undervoltage signals have a lower weight (e.g., 0.4) because they may be indoor faults. The weight values ​​can also be dynamically learned and updated based on historical accuracy.

[0119] Based on the aforementioned basic weights, several seed signals are obtained from the target signal within a preset weight range, such as using a high-weight signal (e.g., a circuit breaker tripping) as the seed signal; based on the time association window and the topological association radius, associated signals associated with the seed signals are obtained, resulting in several association clusters; for example, based on an existing trained spatiotemporal association window model, all other signals that are related in time and topology are automatically retrieved and captured to form an association cluster.

[0120] Starting with the seed signal, the network topology is traversed downstream of the seed signal to obtain ideally affected nodes. Based on the association clusters, actual affected nodes are obtained, and the affected proportion is calculated based on the ideal and actual affected nodes. The association clusters are matched with the multi-source evidence chain rule base to obtain an evidence completeness value. For example, starting from the device where the seed signal is located, a breadth-first traversal is performed downstream along the distribution network topology to automatically identify all distribution transformers and user nodes that should theoretically be affected by a power outage. Subsequently, based on the association clusters, the actual received user voltage loss, complaint work orders, and other signals are matched with this theoretical impact range to calculate the actual affected proportion, which serves as a key indicator of the event's authenticity and scope completeness.

[0121] Different weighting coefficients are assigned to obtain the overall confidence level of the association cluster based on the evidence completeness value and the affected proportion;

[0122] The association results are obtained based on the comprehensive confidence level and a preset confidence threshold. A power outage event profile is generated based on the association clusters. For association results with a confidence level exceeding a set threshold (e.g., 0.85), a structured power outage event profile record is generated based on the association results. This profile includes at least:

[0123] Unique identifier and core information of the event: estimated power outage start time, recovery time (when a recovery signal is received), and power outage type (fault, planned, defect).

[0124] Event impact topology map: including affected feeder segments, transformer list, and user list;

[0125] A complete list of event-related signals: all original signals and their contribution weights on the related events are arranged in chronological order, forming a complete chain of evidence to support post-event tracing.

[0126] Event assessment confidence level and key judgment criteria: Record the confidence score of this assessment, as well as the one or several key judgment criteria (such as the high concentration of switch tripping and the loss of pressure signals of 80% of downstream users within 2 minutes).

[0127] Specifically, the event generation module is used for:

[0128] Based on the association results, several first events are generated, and the event characteristics of the first events are obtained. The event characteristics include event type, event outage start time, event level, and event impact range. The event type, level, and key parameters are automatically determined according to the characteristics of the association cluster.

[0129] Event type determination: For example, if the associated cluster contains a clear dispatch operation ticket signal, it is determined to be a planned power outage; if the main evidence is the operation of the switch protection, it is determined to be a fault power outage; if the signal is sparse and there is no clear electrical action, it is determined to be a defect alarm.

[0130] Event outage start time: Accurately estimate the outage start time. A multi-signal timestamp collaborative estimation method can be used, which does not simply take the first signal time, but combines signal type weights: priority is given to the accurate time of protection action or switch tripping; if none, the mode of the fault indicator action time is used; and then correction is made by the earliest concentrated time window start of the user's voltage loss signal.

[0131] Event level: Reflects the severity and urgency of the power outage event. The actual affected nodes can be obtained from the associated clusters, i.e., the actual scope of impact.

[0132] Event impact scope: The theoretical impact scope derived from correlation analysis. For example, by checking whether user meters within the theoretical impact scope have a small load during the power outage period (indicating possible transfer from other power sources), or by comparing the consistency of user signals under the same distribution transformer, the actual list of affected equipment and users can be accurately located.

[0133] The system acquires the first recovery signal from the power supply equipment and the second recovery signal from the user equipment in the first event. Based on these signals, it obtains a reliability value for the first event and corrects the power outage end time based on this reliability value. It also monitors the closing and voltage recovery signals of power points (such as switches) associated with the event, as well as the voltage recovery signals of affected users. These recovery signals are used to verify the authenticity and completeness of the power outage event. For example, if no recovery signal is captured for a power outage event, but user complaint work orders within its affected area have been closed, the system will issue an event pending closure alarm, and the reliability value will decrease. The first reliable recovery signal is used to verify and correct the power outage end time, making the power outage duration calculation more accurate.

[0134] The event's impact area and the geographic information system (GIS) layer are overlaid to obtain an overlay result, and a contradiction value is obtained based on the overlay result. The event's impact area is overlaid with the GIS layer for analysis to verify whether there is a contradiction (e.g., the user's marked address affected by the event is not within the power supply area of ​​the power outage line).

[0135] Different weighting coefficients are assigned, and the overall confidence level of the first event is corrected based on the reliability value and the contradiction value to obtain the corrected confidence level;

[0136] The power outage event is obtained based on the corrected confidence level and the preset confidence threshold, and a power outage event pool is obtained based on the power outage event. For events that pass final verification, a complete and tamper-proof standard power outage event record is generated and written into the core verified power outage event library. Each record may contain:

[0137] The event has a unique ID (associated with the power grid topology object and timestamp hash), four-dimensional spatiotemporal topology coordinates (time (start, end), space (complete topology coding chain of feeders, segments, distribution transformers, and users)), event classification and tags (type, cause, technical classification, whether it has been restored), a full evidence chain summary (IDs of all associated source signals), verification process records (results of each re-verification and confidence adjustment logs), and ready fields for derived indicators (pre-calculated number of affected users, number of devices, and power outage duration (for efficient reading by the indicator calculation layer)).

[0138] Specifically, the data calculation module is used for:

[0139] Construct multi-dimensional indicators, and obtain affected indicators based on the multi-dimensional indicators and the power outage event pool. Only incremental calculations are performed on the indicators affected by the event, without the need to scan all historical data, thus ensuring real-time performance. The multi-dimensional indicators include time, administrative region, voltage level, power supply unit, feeder and even distribution transformer.

[0140] Based on the affected users of the power outage event, user-power outage event atomic pairs are generated; each affected user within the scope of the event's influence is analyzed, and the duration of the current power outage is calculated, thereby generating the finest-grained user-power outage event atomic pairs, i.e., atomic data, which may include subject identifier: the unique code of the affected user; event association: the standardized power outage event ID that caused this power outage; quantification value: the duration of the power outage experienced by the user due to this event (usually in minutes); topology affiliation: the identifier of the power supply unit (such as distribution transformer, feeder, power supply station, etc.) to which the user belongs; and event attribute: power outage type (fault / planned, etc.).

[0141] Based on the affected indicators, the user-outage event atomic pairs, and the hierarchical relationship of the power grid topology model (e.g., from bottom to top: user, distribution transformer, feeder, substation, power supply station, and branch company), the analysis data is obtained by triggering the data from bottom to top along the built-in topology relationship chain of the power grid topology model, synchronously updating the indicator values ​​of the affected indicators corresponding to the upper level. For example, after the atomic data is generated, it can be checked which indicators are affected. For instance, if a distribution transformer outage event triggers the update of the distribution transformer indicator, the event will automatically trigger the synchronous update of the indicators of all its upper-level units along the topology relationship chain (distribution transformer-feeder-power supply station-branch company).

[0142] Specifically, the data analysis module is used for:

[0143] The analysis data is analyzed based on standard reliability indicators and extended management indicators, such as whether the indicator values ​​exceed the normal range or preset thresholds, to obtain indicator analysis results. The standard reliability indicators include the system average power outage time, the system average power outage frequency, the user average power outage time, and the average power supply reliability rate. The extended management indicators include the average duration of fault power outages, the average duration of planned power outages, the frequency of fault power outages, and the number of users experiencing repeated power outages.

[0144] Based on the analysis results of the indicators, abnormal indicators are obtained, and the administrative regions corresponding to the abnormal indicators are obtained to obtain abnormal regions. Based on the abnormal regions and the preset management paths (such as administrative regions - power supply branches - power supply stations - power supply grids / teams, following the drilling path of management responsibilities, used to lock the management responsibility list), management responsibility units are obtained. Based on the management responsibility units and the topology path of the power grid topology model (the path used to locate physical fault points), the problem location chain is obtained. For example, after drilling to the power supply grid, it is possible to further drill down along the physical power grid to the feeder group - specific feeder - feeder section / transformer area.

[0145] By obtaining power outage events related to the problem location chain, several related events can be obtained. After locating a specific feeder or transformer area, a one-click query can be performed to find all related power outage events that caused the deterioration of its indicators, obtain the cause of the related events, and obtain the root cause analysis results based on the cause of the events. For example, the related power outage events can be aggregated and analyzed to count the proportion of various causes (such as external damage, equipment aging, weather disasters, and user failures) and form a root cause analysis pie chart.

[0146] The analysis results are obtained based on the indicator analysis results, the problem localization chain, and the root cause analysis results.

[0147] Specifically, the decision support module is used for:

[0148] Based on the analysis results, a list of weak links in reliability was obtained. For example, the 10kV XX line had a SAIDI of 200 minutes in the past year, ranking fifth from the bottom in the entire network. The main reason was repeated failures caused by insulation aging.

[0149] Based on the aforementioned warning information, a list of high-risk equipment is obtained; for example: XX distribution transformer, health index 55 points, risk level red, predicted failure probability of 40% in the next six months, mainly due to heavy load and aging equipment.

[0150] Construct an investment solution library and a market price library. The investment solution library contains different types of solutions and quantitative mapping rules corresponding to each solution. The market price library contains the current market price of the equipment bill of materials (BOM).

[0151] Based on the list of weak reliability links, the list of high-risk equipment warnings, the market price database, the investment plan database, and the investment benefit assessment model, the total investment cost (CAPEX) of each plan is automatically estimated and can be allocated over the entire life cycle to obtain the decision-making plan.

[0152] The first event is generated based on the association result and the event generation rules, wherein the event generation rules include:

[0153] Event type determination rules: If the associated cluster contains a switch protection trip signal of a protection device, and the switch protection trip signal does not match the pre-scheduled power outage plan, it is determined to be a fault power outage; if the switch change signal in the associated cluster matches the work order issued in the production management system in both time and equipment, it is determined to be a planned power outage; if the associated cluster contains a remote control trip signal or a manual trip signal, but there is no corresponding work order, it is determined to be a temporary power outage.

[0154] Event level assessment rules: The level range is defined based on the threshold number of affected users to obtain the event level; in this embodiment, the event level assessment rules may also include: if the affected area includes first-level important users (such as hospitals or governments) or pre-set sensitive users, the event level is automatically upgraded by one level; if the expected or actual power outage time exceeds the set threshold (such as 4 hours), the level is upgraded by one level; if the fault occurs during power supply protection for major events or during nighttime peak hours, the level is upgraded by one level.

[0155] Key event parameter extraction rules: Obtain the switch protection action time as the power outage start time; starting from the switch that triggered the event, traverse downstream along the power grid topology to obtain the power outage impact range.

[0156] Example 2

[0157] Based on Embodiment 1, in this embodiment, the data access module is further used to process system data:

[0158] For data from sources such as SCADA and protection devices, and data based on coding standards:

[0159] The data adapter has a built-in signal encoding-semantic mapping rule base, which is used to automatically and accurately map the original numerical or character encodings in the device or system that have no business meaning into rich semantic descriptions and decision attributes with clear power business management intentions.

[0160] The system extracts signal codes from the data, uses the parsed signal code field as a key to query the rule base, and directly maps it to predefined business semantic tags. For example, signal code 0x11A is mapped as follows:

[0161] {Event type: Protection action; Specific intent: Overcurrent trip; Inferred fault distance: Near zone; Severity level: High}.

[0162] The adapter not only parses the action, but also, based on its specific encoding, determines whether it is an overcurrent stage I action (indicating a serious fault at the near end) or an overload alarm (indicating a risk that has not yet tripped). This predictive information about the distance / severity of the fault is extracted as high-value semantic tags.

[0163] For text data such as customer service work orders and inspection records:

[0164] The adapter integrates an existing lightweight domain-specific natural language processing model:

[0165] The model performs rapid text analysis, including keyword extraction, named entity recognition (identifying device names and addresses), and sentiment / urgency classification.

[0166] As described in the work order: "There was a sudden power outage at No. XX, XX Road. An elderly person in the house needs oxygen!" The model extracted the following:

[0167] {Keywords: sudden power outage; inferred event type: power failure; user emotion: emergency; specific needs mentioned: medical equipment}.

[0168] For text work orders in the customer service system, the adapter performs real-time keyword extraction and initial sentiment assessment (e.g., extracting tags for repeated power outages and strong complaints from texts such as "There's a power outage again!" and "It's been out for a long time!").

[0169] This technology transforms unstructured raw data into structured features with preliminary business semantics at the entry point, providing directly usable high-dimensional features for subsequent associations and greatly reducing the computational complexity and latency of upper-layer modules.

[0170] The intermediate data packet is obtained based on the extracted business semantic tags, the data extracted by the natural language processing model, the source system identifier corresponding to the system data, and the original receiving timestamp of the system data.

[0171] In this example, the specific steps for extracting signal codes from system data may include:

[0172] For raw binary streams that may contain noise, bit errors, missing bytes, or redundancy:

[0173] A fault-tolerant parsing algorithm guided by a state machine is adopted. This algorithm pre-defines a parsing state machine for the communication protocol, and the states include finding the frame header, verifying the length, extracting the encoded fields, and verification.

[0174] During the frame header search stage, fuzzy matching is employed. For example, 1-2 bits of error tolerance are allowed in the frame header flag bits to accommodate channel interference.

[0175] When entering the state of extracting encoded fields, the algorithm not only reads bytes at fixed positions but also initiates a local verification mechanism. For example, it calculates the parity check or sum of the field and compares it with the check bits in the message. If they do not match, it triggers local error correction logic: it attempts to flip the suspicious bits (e.g., using Hamming code error correction) or performs interpolation prediction based on the preceding and following message sequences.

[0176] This results in a clean candidate encoded byte segment that has been corrected to a high confidence level.

[0177] When interfacing with legacy equipment or manufacturer-specific protocols, and in the absence of complete protocol documentation, unsupervised learning and pattern mining are employed to infer the encoding structure.

[0178] A large number of unlabeled historical communication packets from the same type of device are acquired, and cluster analysis is performed on the packet stream to identify which byte segments are relatively fixed (such as address fields) and which are frequently changing. Byte segments that are highly variable but have discrete (non-continuous) values ​​are initially identified as potential signal encoding fields.

[0179] The values ​​of the changing byte segments are correlated with other observable events occurring at the same time (such as switch position records and voltage surges). If the occurrence of a byte segment value of 0x0A is always closely accompanied by a switch tripping event, it can be strongly inferred that 0x0A is the signal code for the tripping event.

[0180] The inferred byte segment location, value, and inference statement are stored as a rule in an adaptive signal coding knowledge base.

[0181] For private protocol messages, the most likely signal encoding and its inferred semantics are matched from the knowledge base.

[0182] The state machine-guided fault-tolerant parsing algorithm combines deterministic finite state machines with various fault-tolerant and error-correcting strategies, and may include the following:

[0183] Deterministic finite state machines define the standard process and expected structure for packet parsing. Each state represents a stage of parsing (such as finding the start-of-frame character, reading the length field, extracting the data body, and verifying the checksum). For example, in the state of reading the length field, if the length value read is absurd (such as negative or excessively large), the state machine can immediately jump to the error handling state instead of continuing to parse blindly.

[0184] Fault tolerance and error correction strategy layer: adopts fuzzy matching / approximate matching algorithm, and allows a certain degree of bit error (such as Hamming distance ≤ 2) between the target byte sequence and the predefined frame header when searching for the frame start symbol, thus solving the problem of frame header distortion caused by channel interference.

[0185] Forward error correction and verification technology: integrated into the verification state of the state machine. When verification fails, it not only reports the error but also triggers the error correction process. For example, if Hamming codes are used, single bit errors can be automatically located and corrected; if stronger error-correcting codes are used, multiple errors or burst errors may be corrected.

[0186] Context-based adaptive parsing: When ambiguity or validation failure occurs in the parsing of a field, the algorithm references the historical context of the current communication session (such as device address, value of the previous normal message) or domain knowledge (such as the reasonable value range of a certain type of signal) to make reasonable inferences and interpolations for the current field. This can be seen as a simple form of Bayesian inference or constraint solving.

[0187] Example 3

[0188] Based on the above embodiments, this implementation also provides a power outage event-driven distribution network status analysis and decision-making method, the method comprising:

[0189] A cache database is obtained based on system data from different target systems;

[0190] Process the cached database to obtain the power outage data sequence;

[0191] Based on the power outage data sequence, several target signals are obtained, and the target signals are correlated to obtain the correlation result;

[0192] Based on the correlation results, power outage events are obtained, and based on the power outage events, a power outage event pool is obtained;

[0193] Analyze the power outage event pool to obtain analytical data;

[0194] The analysis results are obtained by analyzing the data.

[0195] Early warning information is obtained based on a pre-trained health assessment model and third-party data; an investment benefit assessment model is constructed, and a decision-making scheme is obtained based on the analysis results, the early warning information, and the investment benefit assessment model.

[0196] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0197] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A power outage event driven distribution network state analysis decision system, characterized in that, The system includes: Data access module: used to obtain intermediate data packets based on system data from different target systems, and to obtain a cache database based on the intermediate data packets; Data cleaning module: used to correct the intermediate data packets, attach the corrected intermediate data packets to the power grid topology model, and obtain the power outage data sequence; Data association module: used to obtain several target signals based on the power outage data sequence, associate the target signals, and obtain association results; Event generation module: used to obtain power outage events based on the correlation results, and to obtain a power outage event pool based on the power outage events; Data calculation module: used to analyze the power outage event pool to obtain analysis data; Data analysis module: used to analyze the data and obtain analysis results; Decision support module: used to obtain early warning information based on a pre-trained health assessment model and third-party data; to construct an investment benefit assessment model; and to obtain decision-making solutions based on the analysis results, the early warning information, and the investment benefit assessment model.

2. The de-energized event driven distribution state analysis decision system of claim 1, wherein, The data access module is specifically used for: A data source registry and data adapters are constructed. The data source registry contains registration data for different target systems. The registration data includes source system identifier, interface type, connection parameters, and data characteristics. Each data adapter corresponds to a different target system. Based on the data source registry and the data adapter, obtain the system data for each of the target systems; The intermediate data packet is obtained based on the system data, the source system identifier corresponding to the system data, and the original reception timestamp of the system data; Configure metadata for each intermediate data packet, the metadata including a unique serial number, a data confidence identifier, and a topology association key; The cache database is obtained based on the intermediate data packet.

3. The de-energized event driven distribution state analysis decision system of claim 2, wherein, The data cleaning module is specifically used for: Construct a data parser for each of the target systems, with each data parser corresponding one-to-one with a source system identifier; The data parser parses the intermediate data packet to obtain field-value pairs, verifies the compliance of the field-value pairs, and obtains the quality label of the intermediate data packet. Based on the quality label, a problem data packet is obtained. The topology association key of the problem data packet is obtained to obtain the problem device identifier. If the problem device identifier does not exist in the power grid topology model or the problem device identifier conflicts with an existing device identifier in the power grid topology model, the problem device identifier is associated and mapped based on the device alias library and historical ledger change records to obtain the corrected device identifier of the problem data packet. Based on the corrected device identifier, the problem device identifier of the problem data packet is corrected to obtain the corrected intermediate data packet. The corrected intermediate data packet is attached to the power grid topology model, and a topology coordinate chain of the corrected intermediate data packet is generated based on the power grid topology model and the topology association key. A standard data object is obtained based on the corrected intermediate data packet and the topology coordinate chain. According to the time sequence and topological relationship, the standard data objects are stored in the time series database and the graph database respectively, and the power outage data sequence is obtained based on the time series database and the graph database.

4. The de-energized event driven distribution state analysis decision system of claim 1, wherein, The data association module is specifically used for: Construct a multi-source evidence chain rule base, time association windows for different signal types, and topological association radii for different signal types. The multi-source evidence chain rule base includes ideal evidence chain templates for different power outage scenarios. Preset basic weights for different signal sources and different signal types, and obtain several seed signals based on the target signal within the preset weight range based on the basic weights; Based on the time association window and the topological association radius, the association signal associated with the seed signal is obtained, and several association clusters are obtained; Starting from the seed signal, the network topology is traversed downstream of the seed signal to obtain the ideal affected nodes, the actual affected nodes are obtained based on the associated clusters, and the affected ratio is obtained based on the ideal affected nodes and the actual affected nodes. The association clusters are matched with the multi-source evidence chain rule base to obtain the evidence completeness value; The overall confidence level of the associated cluster is obtained based on the evidence completeness value and the affected proportion; The association results are obtained based on the comprehensive confidence level and the preset confidence threshold, and a power outage event profile is generated based on the association clusters.

5. The de-energized event driven distribution state analysis decision system of claim 4, wherein, The event generation module is specifically used for: Based on the association results, several first events are generated, and the event characteristics of the first events are obtained. The event characteristics include event type, event outage start time, event level, and event impact range. Acquire the first recovery signal of the power supply equipment and the second recovery signal of the user equipment in the first event, obtain the reliability value of the first event based on the first recovery signal and the second recovery signal, and correct the power outage end time of the first event based on the reliability value; The impact range of the event and the geographic information system layer are overlaid to obtain the overlay result, and the contradiction value is obtained based on the overlay result; The overall confidence level of the first event is adjusted based on the reliability value and the contradiction value to obtain the adjusted confidence level; The power outage event is obtained based on the corrected confidence level and the preset confidence threshold, and a power outage event pool is obtained based on the power outage event.

6. The de-energized event driven distribution state analysis decision system of claim 1, wherein, The data calculation module is specifically used for: Construct multi-dimensional indicators, and obtain affected indicators based on the multi-dimensional indicators and the power outage event pool. The multi-dimensional indicators include time, administrative region, voltage level, power supply unit, feeder and even distribution transformer. Based on the affected users of the power outage event, generate user-power outage event atomic pairs; Based on the affected indicators, the user-outage event atomic pairs, and the hierarchical relationship of the power grid topology model, the system triggers the data from bottom to top along the built-in topology relationship chain of the power grid topology model, synchronously updating the indicator values ​​of the affected indicators corresponding to the higher level, thereby obtaining the analysis data.

7. The de-energized event driven distribution state analysis decision system of claim 6, wherein, The data analysis module is specifically used for: The analytical data is analyzed based on standard reliability indicators and extended management indicators to obtain indicator analysis results. The standard reliability indicators include the system average power outage time, the system average power outage frequency, the user average power outage time, and the average power supply reliability rate. The extended management indicators include the average duration of fault power outages, the average duration of planned power outages, the frequency of fault power outages, and the number of users experiencing repeated power outages. Based on the analysis results of the indicators, abnormal indicators are obtained, and the administrative regions corresponding to the abnormal indicators are obtained to obtain abnormal regions. Based on the abnormal regions and the preset management paths, management responsibility units are obtained, and based on the management responsibility units and the topology paths of the power grid topology model, problem location chains are obtained. Obtain several related events by acquiring power outage events related to the problem localization chain, obtain the event causes of the related events, and obtain root cause analysis results based on the event causes; The analysis results are obtained based on the indicator analysis results, the problem localization chain, and the root cause analysis results.

8. The de-energized event driven distribution state analysis decision system of claim 7, wherein, The decision support module is specifically used for: Based on the analysis results, a list of weak links in reliability is obtained, and based on the early warning information, a list of high-risk equipment is obtained. Construct an investment solution library and a market price library, wherein the investment solution library contains different types of solutions and quantitative mapping rules corresponding to each solution; The decision-making scheme is obtained based on the list of weak links in reliability, the list of high-risk equipment warnings, the market price database, the investment scheme database, and the investment benefit evaluation model.

9. The de-energized event driven distribution state analysis decision system of claim 5, wherein, The first event is generated based on the association result and the event generation rules, the event generation rules including: Event type determination rules: If the associated cluster contains a switch protection trip signal of a protection device, and the switch protection trip signal does not match the pre-scheduled power outage plan, it is determined to be a fault power outage; if the switch change signal in the associated cluster matches the work order issued in the production management system in both time and equipment, it is determined to be a planned power outage; if the associated cluster contains a remote control trip signal or a manual trip signal, but there is no corresponding work order, it is determined to be a temporary power outage. Event severity rating rules: The severity rating is determined by defining a range of severity levels based on a threshold number of affected users. Key event parameter extraction rules: Obtain the switch protection action time as the power outage start time; starting from the switch that triggered the event, traverse downstream along the power grid topology to obtain the power outage impact range.

10. A power outage event driven distribution network state analysis decision method, characterized in that, The method includes: A cache database is obtained based on system data from different target systems; Process the cached database to obtain the power outage data sequence; Based on the power outage data sequence, several target signals are obtained, and the target signals are correlated to obtain the correlation result; Based on the correlation results, power outage events are obtained, and based on the power outage events, a power outage event pool is obtained; Analyze the power outage event pool to obtain analytical data; The analysis results are obtained by analyzing the data. Early warning information is obtained based on a pre-trained health assessment model and third-party data; an investment benefit assessment model is constructed, and a decision-making scheme is obtained based on the analysis results, the early warning information, and the investment benefit assessment model.

Images