A fault intelligent disposal method and system of a smart grid dispatching control system

Abstract

The present application relates to the technical field of smart grid dispatching and fault handling, and discloses a fault intelligent handling method and system of a smart grid dispatching control system. The method comprises the following steps: introducing asynchronous multi-source monitoring data streams into a distributed message queue buffer to cut peaks; aligning and splicing the data in time sequence to form real-time data records with complete time and space; identifying abnormal features from the records, extracting the features through a fault feature template library to generate a preliminary fault feature vector set; mapping the vector set to a physical power grid in combination with power grid geographic information and topological relations, dynamically calculating the potential impact domain boundary of the fault and generating a description file; screening the equipment state data within the impact domain according to the description file, and iteratively deducing the data in combination with a historical case library to generate a specific operation instruction sequence. The method can achieve accurate and rapid identification of faults and dynamic and accurate judgment of the impact range, and improves the handling efficiency and reliability.

Description

Technical Field

[0001] This invention relates to the field of smart grid dispatching and fault handling technology, specifically to a smart grid dispatching and control system fault intelligent handling method and system. Background Technology

[0002] Currently, fault monitoring in smart grids primarily relies on sensors deployed at various levels, generating massive, asynchronous, multi-source data streams. Conventional fault handling systems typically trigger alarms using over-limit alarms based on fixed thresholds or simple signal logic judgments. These alarm messages are often isolated and fragmented, lacking the ability to deeply integrate and systematically identify patterns in multi-source, multi-dimensional anomaly information. Faced with a large number of alarms emerging simultaneously, operators struggle to quickly identify the true root cause and type of fault, easily falling into information overload and resulting in low efficiency in initial fault location.

[0003] In the fault impact analysis phase, existing methods mostly derive the impact range based on the logical topology of the power grid, typically using substations, voltage levels, or electrical islands as static boundaries. This method fails to adequately consider the actual geographical distribution of equipment, the attenuation effect of electrical distance, and the continuous physical spatial impact of power flow transfer paths after a fault. The defined impact range often deviates from the actual dynamic area that the fault might affect; either the range is too broad, introducing a large amount of irrelevant equipment status interference analysis, or the range is insufficient, omitting key risk points, thus lacking accurate spatial basis for generating subsequent response strategies. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for intelligent fault handling in a smart grid dispatch and control system, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a method for intelligent fault handling in a smart grid dispatching and control system, the method comprising:

[0006] Receive asynchronous multi-source monitoring data streams from different monitoring levels of the power grid, and import the asynchronous multi-source monitoring data streams into a distributed message queue for caching and traffic shaping.

[0007] For the original monitoring data entries cached in the distributed message queue, based on the unique code of the power grid equipment and the data generation time, time sequence alignment and data splicing operations are performed to construct a real-time data record with complete spatiotemporal attributes.

[0008] From the real-time data records, feature data representing abnormal power grid states are identified and separated. Based on a pre-set fault feature template library, pattern matching and feature extraction are performed on the feature data to generate a preliminary fault feature vector set.

[0009] By combining the power grid geographic information layer with the topological connection relationship, the preliminary fault feature vector set is mapped to the physical power grid structure, and the potential impact domain boundary of the fault is calculated based on the mapping result, generating a fault impact domain description file;

[0010] Based on the spatial range defined by the fault impact domain description file, all operating equipment status data within the spatial range are selected from the real-time data records. Combined with the historical fault case library, multiple rounds of iterative handling strategy deduction are performed to finally generate a fault handling plan containing a specific sequence of operation instructions.

[0011] Preferably, the method involves receiving asynchronous multi-source monitoring data streams from different monitoring levels of the power grid and importing these asynchronous multi-source monitoring data streams into a distributed message queue for caching and traffic shaping, including:

[0012] Listen to and receive real-time data messages from the power grid dispatching master station, substation automation system, distribution automation terminal and distributed energy monitoring unit;

[0013] The received real-time data packets are parsed according to the protocol, the communication packet header is stripped, and the payload containing device code, timestamp, measurement value and data quality identifier is extracted;

[0014] The payload is written into message topic channels of different priorities according to its source monitoring level and data urgency label. The distributed message queue dynamically adjusts memory allocation according to the consumption rate of each topic channel.

[0015] Each payload written to the distributed message queue is appended with a globally unique streaming identifier, and the time it enters the queue is recorded to form a raw data buffer stream with a time sequence mark.

[0016] Preferably, for the original monitoring data entries cached in the distributed message queue, based on the unique code of the power grid equipment and the data generation time, time-series alignment and data concatenation operations are performed to construct a real-time data record with complete spatiotemporal attributes, including:

[0017] The raw data cache stream with time-series tags is consumed sequentially from the designated topic channel of the distributed message queue;

[0018] Using the unique code of the power grid equipment as the primary key, multiple data entries from different monitoring levels targeting the same power grid equipment are aggregated within a preset time alignment window;

[0019] The aggregated data entries are sorted in ascending order according to their data generation time, and similar measurement data with time deviations within the tolerance range are processed using a weighted fusion algorithm based on data quality identification to form a unique state snapshot of the power grid equipment within the time window.

[0020] The unique state snapshot of each power grid device is associated and bound with its corresponding geographical coordinates and topology node information to generate a real-time data record containing time, space, device attributes, and operating status.

[0021] Preferably, feature data characterizing abnormal power grid states are identified and separated from the real-time data records. Based on a pre-set fault feature template library, pattern matching and feature extraction are performed on the feature data to generate a preliminary fault feature vector set, including:

[0022] Iterate through the various operational status measurements in the real-time data records, compare them with the normal operating threshold range of the corresponding equipment or line, and mark all data points that exceed the limit and their exceeding the limit.

[0023] Load predefined fault feature templates from the historical database. The fault feature templates define the abnormal change patterns of multiple related state variables in the time dimension and the correlation propagation patterns in the spatial dimension under various typical faults.

[0024] The set of marked out-of-limit data points is analyzed by sliding window on the time axis to capture abrupt changes, step jumps, and oscillation trends of state variables, and similarity calculation is performed with the time patterns defined in the fault feature template.

[0025] The set of data points that exceed the limit and their positional relationship in the power grid topology that meet the similarity conditions are matched with the spatial association pattern defined in the fault feature template. If both the spatiotemporal patterns are successfully matched, the corresponding feature code, fault type identifier and confidence weight are extracted from the template and encapsulated into a fault feature unit.

[0026] All successfully matched fault feature units are summarized, and each unit corresponds to a multi-dimensional preliminary fault feature vector. All vectors constitute the preliminary fault feature vector set.

[0027] Preferably, by combining the power grid geographic information layer with the topological connectivity, the preliminary fault feature vector set is mapped to the physical power grid structure, and the potential impact domain boundary of the fault is calculated based on the mapping result, generating a fault impact domain description file, including:

[0028] Obtain a digital geographic information layer of the power grid that includes the geographic coordinates and electrical connections of transmission lines, substations, distribution transformers, and load points;

[0029] The power grid equipment identifier associated with each fault feature vector in the preliminary fault feature vector set is located in the power grid digital geographic information layer to determine the precise geographical coordinates of the fault occurrence.

[0030] Centered on the precise geographical coordinates, and based on the fault type identifier and confidence weight contained in the preliminary fault feature vector set, the corresponding electrical distance attenuation coefficient and load importance weight coefficient are selected from the preset impact propagation model.

[0031] On the digital geographic information layer of the power grid, a breadth-first search is performed along the topological connection edge with the fault point as the center. The influence intensity is calculated in combination with the electrical distance attenuation coefficient. The search stops when the influence intensity attenuates to below the threshold or when an electrical disconnection point is encountered. All nodes and edges traversed constitute the potential influence range.

[0032] The coordinates of the outer polygon boundary of the potential impact range, the list of affected devices within the range, and the estimated impact level of each device are structured and organized, and output as the fault impact domain description file.

[0033] Preferably, based on the spatial range defined by the fault impact domain description file, all operating equipment status data within the spatial range are filtered from the real-time data records. Combined with a historical fault case library, multiple rounds of iterative handling strategy deduction are performed to ultimately generate a fault handling plan containing a specific sequence of operation instructions, including:

[0034] Parse the fault impact domain description file and extract the bounding polygon boundary coordinates of the potential impact range recorded therein;

[0035] Using the boundary coordinates of the circumscribed polygon as a spatial filtering condition, spatial queries are performed on the real-time data records to quickly retrieve the current operating status data of all power grid equipment whose geographical location falls within the polygon range, forming a dataset of the status of affected equipment.

[0036] From the historical fault case database, using the current fault type, the main affected equipment type, and the current power grid operation mode as combined query conditions, historical similar cases are retrieved, and the handling measures taken in the historical similar cases and the changes in the power grid status after their implementation are extracted.

[0037] Using the affected equipment status dataset as the initial power grid state and the handling measures in the historical similar cases as the candidate operation set, each candidate operation is simulated and executed in the power grid power flow calculation model and the security constraint model.

[0038] During the simulation, the power flow distribution, node voltage, and line load rate of the power grid are evaluated in real time to determine whether they meet the safety operation constraints, and the simulation operation sequence and its resulting final state are recorded.

[0039] From all simulated operation sequences that meet safety constraints, the optimal sequence is selected by evaluating and ranking them based on a comprehensive objective function that maximizes the range of power restoration, minimizes the number of operation steps, and minimizes overall risk.

[0040] Each simulated operation in the optimal sequence is converted into a specific remote control instruction recognizable by the scheduling and control system, along with the preconditions and verification steps for executing the remote control instruction. These are then arranged in sequence to form the final fault handling plan.

[0041] Preferably, using the affected equipment status dataset as the initial power grid state and the handling measures in the historical similar cases as the candidate operation set, each candidate operation is simulated and executed in the power grid power flow calculation model and the security constraint model, including:

[0042] The switching status, active power output, reactive power output, and load value of each device in the affected equipment status dataset are injected into the power grid power flow calculation model to calculate the initial power flow distribution within the current fault influence domain.

[0043] The measures extracted from the historical similar cases are classified into switch operation type, output adjustment type, and load control type. Each type of measure is described as a rule for changing the corresponding equipment parameters or states in the power grid model.

[0044] In the simulation environment, the change rules are applied to the power flow calculation model in the order of the operation in the candidate operation set. After each rule is applied, the power flow calculation is performed again to obtain the new power grid state.

[0045] After each power flow calculation, the calculation results are compared and verified with the line current carrying limit, transformer capacity limit, and voltage upper and lower limits defined in the security constraint model.

[0046] Record all operation sequences in which the state variables obtained by the power flow calculation model always satisfy all constraints in the security constraint model during the simulation execution process, as well as the final power grid state after each sequence is executed, as a set of feasible disposal strategies.

[0047] Preferably, from all simulated operation sequences that meet safety constraints, the optimal sequence is selected by evaluating and ranking them based on a comprehensive objective function that maximizes the range of power restoration, minimizes the number of operation steps, and minimizes overall risk. This includes:

[0048] For each operation sequence in the set of feasible handling strategies, the total number of load points and the total load capacity that can be restored from the power outage state within the range described in the fault impact domain description file after the operation sequence is completed are counted as indicators of the power restoration range.

[0049] The number of specific operation instructions contained in each operation sequence is counted as an indicator of the number of operation steps.

[0050] Based on the data such as line load rate and voltage deviation recorded by the power grid power flow calculation model during the simulation execution of each operation sequence, combined with equipment reliability parameters, the probability and severity of secondary risks such as equipment overload and cascading failures that may be triggered during the execution of the operation sequence are calculated, and the overall risk assessment index is obtained after weighted summation.

[0051] The power restoration range index, the number of operation steps index, and the overall risk assessment index are normalized respectively, and different weight coefficients are assigned to each index according to the scheduling priority strategy.

[0052] Calculate the weighted comprehensive score for each operation sequence, sort all operation sequences from highest to lowest score, and select the operation sequence with the highest score as the optimal sequence.

[0053] Preferably, each simulated operation in the optimal sequence is converted into a specific remote control instruction recognizable by the scheduling and control system, along with the preconditions and verification steps for executing the remote control instruction, and arranged sequentially to form the final fault handling plan, including:

[0054] Each simulated operation in the optimal sequence is mapped to a standard remote control and remote adjustment instruction template of the scheduling and control system, and the specific equipment number, target status value or adjustment value is filled in to generate a machine-readable raw control instruction.

[0055] For each of the original control commands, analyze the real-time operating conditions of the power grid that must be met when the command is actually executed, and convert them into a series of logical judgment statements to form a list of preconditions for the execution of the command.

[0056] For each of the original control commands, a feedback verification step is designed after the command is issued. The feedback verification step includes checking whether the equipment status has changed to the target status and whether the relevant electrical measurement values ​​have entered the expected range within a specified time.

[0057] The original control command, its corresponding precondition list, and the feedback verification steps are combined into a complete step unit according to the execution order in the optimal sequence.

[0058] Assemble all the steps in sequence, and add a fault overview, scope of impact, and overall handling objectives at the beginning, and add a description of the expected power grid state after handling is completed at the end. Finally, package them to generate the fault handling plan file.

[0059] Preferably, the present invention also includes a fault intelligent handling system for a smart grid dispatch and control system, the system including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, it implements the steps of the fault intelligent handling method for a smart grid dispatch and control system as described above.

[0060] Compared with the prior art, the beneficial effects of the present invention are:

[0061] By introducing a pre-built fault feature template library, pattern matching and feature extraction are performed on abnormal features separated from real-time data, transforming unstructured alarm data streams into structured preliminary fault feature vector sets. This method achieves a leap from simple threshold alarms to complex pattern recognition, directly outputting standardized features representing specific fault modes. This provides directly processable and semantically clear input for subsequent intelligent diagnosis and strategy reasoning, overcoming the shortcomings of traditional methods such as messy alarm information and difficulty in quickly identifying the root cause.

[0062] By combining the power grid geographic information layer with topological connectivity, preliminary fault characteristics are mapped to the physical space, and the potential impact domain boundary of the fault is dynamically calculated. The impact domain described in the generated document is a dynamic and continuous range dually defined in electrical connectivity and geographic space. It accurately defines the actual possible propagation range of the fault in the physical network, making the operation of filtering relevant equipment status from massive amounts of data more targeted, improving the accuracy of impact range assessment, and avoiding the range distortion problem caused by purely logical topology analysis. Attached Figure Description

[0063] Figure 1 This is a schematic diagram illustrating the working principle of the intelligent fault handling method of the smart grid dispatch and control system described in this invention.

[0064] Figure 2 A flowchart for data reception and message queue processing;

[0065] Figure 3 This is a flowchart for fault feature identification and extraction;

[0066] Figure 4 Heatmap of safety constraint verification pass rate;

[0067] Figure 5 A graph showing the relationship between matching degree and treatment effect of historical cases. Detailed Implementation

[0068] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0069] Please see Figure 1 This invention provides a method for intelligent fault handling in a smart grid dispatch and control system. The method includes: receiving asynchronous multi-source monitoring data streams from different monitoring levels of the power grid, importing the asynchronous multi-source monitoring data streams into a distributed message queue for caching and traffic peak shaping, performing time-series alignment and data concatenation operations on the original monitoring data entries cached in the distributed message queue based on the unique code of the power grid equipment and the data generation time to construct real-time data records with complete spatiotemporal attributes, identifying and separating feature data representing abnormal states of the power grid from the real-time data records, performing pattern matching and feature extraction on the feature data based on a pre-set fault feature template library to generate a preliminary fault feature vector set, mapping the preliminary fault feature vector set to the physical power grid structure by combining the power grid geographic information layer and topological connection relationship, calculating the potential impact domain boundary of the fault based on the mapping result, generating a fault impact domain description file, filtering all operating equipment status data within the spatial range defined by the fault impact domain description file from the real-time data records, performing multi-round iterative fault handling strategy deduction by combining a historical fault case library, and finally generating a fault handling scheme containing a specific sequence of operation instructions.

[0070] In one embodiment of the present invention, see [reference] Figure 2The system listens for and receives real-time data packets from the power grid dispatch master station, substation automation system, distribution automation terminal, and distributed energy monitoring unit. It performs protocol parsing on the received real-time data packets, strips the communication headers, and extracts the payload containing device codes, timestamps, measurement values, and data quality identifiers. The payloads are then written to message topic channels of different priorities according to their source monitoring level and data urgency labels. The distributed message queue dynamically adjusts memory allocation based on the consumption rate of each topic channel. A globally unique streaming processing identifier is appended to each payload written to the distributed message queue, and the time it enters the queue is recorded, forming a raw data buffer stream with time-series tags. From the distributed message... The queue sequentially consumes raw data cache streams with time-series tags in the designated topic channel. Using the unique code of the power grid equipment as the primary key, multiple data entries from different monitoring levels targeting the same power grid equipment are aggregated within a preset time alignment window. The aggregated data entries are sorted in ascending order according to their data generation time. For similar measurement data with time deviations within the tolerance range, a weighted fusion algorithm based on data quality identifiers is used to process them, forming a unique state snapshot of the power grid equipment within the time window. The unique state snapshot of each power grid equipment is associated and bound with its corresponding geographical coordinate information and the information of its topology node, generating a real-time data record containing time, space, equipment attributes, and operating status.

[0071] In practice, the system monitors and receives real-time data packets from the power grid dispatching master station, substation automation system, distribution automation terminal, and distributed energy monitoring unit. These packets may follow different communication protocols such as IEC60870-5-104, IEC61850, or specific vendor proprietary protocols. The system parses the received real-time data packets, strips the communication headers, and extracts the payload containing equipment codes, timestamps, measurement values, and data quality identifiers. The equipment codes follow the unified asset coding rules of the power grid company, the timestamps are accurate to milliseconds, the measurement values ​​include electrical parameters such as voltage, current, and power, and the data quality identifiers are used to mark the validity of the data, whether it has been set, or whether it comes from bad data. In some embodiments, payloads are written to message topic channels of different priorities according to their source monitoring level and data urgency label. For example, grid operation status data from the dispatch master station is written to a high-priority topic, power generation data from the distributed energy monitoring unit is written to a medium-priority topic, and regular periodic measurement data is written to a low-priority topic. The distributed message queue dynamically adjusts memory allocation according to the consumption rate of each topic channel, with high-consumption-rate topic channels receiving more memory cache space to avoid data backlog. In a specific implementation, each payload written to the distributed message queue is appended with a globally unique streaming processing identifier, and its entry time into the queue is recorded, forming a raw data cache stream with a time sequence label. The streaming processing identifier is generated by combining the data center identifier, timestamp, and sequence number, and the entry time into the queue is used to delay subsequent monitoring data processing.

[0072] It can be understood that the raw data buffer stream with time-series tags is consumed sequentially from a designated topic channel of a distributed message queue. The consumption process follows a first-in, first-out (FIFO) principle, using the unique code of the power grid equipment as the primary key. Within a preset time alignment window, multiple data entries from different monitoring levels targeting the same power grid equipment are aggregated. For example, a transformer may simultaneously have power flow data from the dispatch master station and oil temperature data from the substation automation system within the same time window. The width of the time alignment window is configured based on the maximum communication latency fluctuation of the data source. The aggregated data entries are sorted in ascending order according to their data generation time. For similar measurement data with time deviations within the tolerance range, a weighted fusion algorithm based on data quality identifiers is used to process them, forming a unique state snapshot of the power grid equipment within the time window. The weighted fusion algorithm calculates a weighted average based on the respective data quality identifiers when multiple values ​​of the same physical quantity exist at close times. In specific implementation, the calculation of weighted fusion can be expressed as:

[0073] in: Represents the fused measurement value. Representing the One original measurement value, Representative based on the first Weighting factors calculated from data quality identifiers The number of similar measurement data entries participating in the fusion. Weighting factor. The assignment rules are as follows: data quality marked as "valid" has a weight of 1.0, marked as "suspicious" has a weight of 0.5, and marked as "invalid" has a weight of 0.0. A unique state snapshot of each power grid device is associated and bound to its corresponding geographic coordinates and topology node information. The geographic coordinates are obtained from the power grid resource management system, and the topology node information is obtained from the power grid model file, generating a real-time data record containing time, space, device attributes, and operating status. In some embodiments, the generated real-time data record is persistently stored in structured JSON or Avro format, and a time and space index is established for subsequent rapid retrieval. Optionally, for devices that fail to match any geographic coordinates, the spatial field in their generated real-time data record will be marked as empty, triggering an alarm to notify model maintenance personnel.

[0074] In one embodiment of the present invention, see [reference] Figure 3The system iterates through the various operational status measurements in the real-time data records, comparing them with the normal operating threshold ranges of the corresponding equipment or lines. It marks all data points exceeding the limits and their magnitudes. A predefined fault feature template is loaded from the historical database. This template defines the abnormal change patterns of multiple related state variables under various typical faults in the time dimension and the correlation propagation patterns in the spatial dimension. The marked set of exceeding-limit data points is analyzed using a sliding window on the time axis to capture abrupt changes, steps, and oscillation trends of the state variables. Similarity calculations are performed with the time patterns defined in the fault feature template. The set of exceeding-limit data points that meet the similarity conditions and their positional relationships in the power grid topology are matched with the spatial correlation patterns defined in the fault feature template. If both the temporal and spatial patterns match successfully, the corresponding feature codes, fault type identifiers, and confidence weights are extracted from the template and encapsulated into a fault feature unit. All successfully matched fault feature units are summarized, with each unit corresponding to a multi-dimensional preliminary fault feature vector. All vectors constitute the preliminary fault feature unit. Fault Feature Vector Set: Obtain a digital geographic information layer of the power grid containing the geographic coordinates and electrical connections of transmission lines, substations, distribution transformers, and load points. Locate the power grid equipment identifiers associated with each fault feature vector in the preliminary fault feature vector set within the digital geographic information layer of the power grid to determine the precise geographic coordinates of the fault occurrence. Using the precise geographic coordinates as the center, select the corresponding electrical distance attenuation coefficient and load importance weight coefficient from the pre-set impact propagation model based on the fault type identifier and confidence weight contained in the preliminary fault feature vector set. Perform a breadth-first search along the topological connection edges on the digital geographic information layer of the power grid, centered on the fault point. Calculate the impact intensity using the electrical distance attenuation coefficient. Stop the search when the impact intensity attenuates below a threshold or when an electrical disconnection point is encountered. All nodes and edges traversed constitute the potential impact range. Structure the outer polygon boundary coordinates of the potential impact range, the list of affected equipment within the range, and the estimated impact level of each equipment, and output as a fault impact domain description file.

[0075] In practical implementation, the system iterates through the various operational status measurements in the real-time data records and compares them with the normal operating threshold range of the corresponding equipment or line. The normal operating threshold range is predefined by the equipment technical parameters and power grid operation procedures. All data points exceeding the limit and their exceeding magnitude are marked. For example, if the current measurement value of a 220kV transmission line is 1500A, while its stable operating upper limit is 1200A, then this data point is marked as exceeding the limit, with an exceeding magnitude of 300A. Predefined fault feature templates are loaded from the historical database. The fault feature templates define the abnormal change patterns of multiple related state quantities in the time dimension and the correlation propagation patterns in the spatial dimension under various typical faults. For example, the fault feature template for "single-phase grounding fault of transmission line" may include the time pattern of sudden changes in current of adjacent lines centered on the fault point, the time pattern of a specific phase reduction in bus voltage, and the spatial correlation pattern of fault current propagation between specific electrical connection nodes. The set of marked out-of-limit data points is subjected to sliding window analysis on the time axis to capture abrupt changes, steps, and oscillation trends in state variables. Similarity is calculated between these captured data points and the time patterns defined in the fault feature template. The length of the sliding window matches the time scale of fault development. In some embodiments, the time pattern similarity calculation can be expressed as:

[0076]

[0077] in: This represents the similarity over time, and its value range is [0,1]. Represents the first [number] within the sliding window The characteristic values ​​of state variables observed at each sampling time (such as rate of change, fluctuation amplitude). This represents the expected change in characteristic values ​​at the corresponding time point as defined in the fault characteristic template. This represents the total number of sampling points within the sliding window. Similarity The closer the similarity score is to 1, the better the observed temporal variation pattern matches the template. This can be understood as matching the set of data points that meet the similarity criteria and their positional relationships within the power grid topology with the spatial association patterns defined in the fault feature template. These spatial association patterns define the expected propagation path and attenuation relationship of the faulty electrical quantity within the power grid topology. If both spatiotemporal patterns match successfully, the corresponding feature codes, fault type identifiers, and confidence weights are extracted from the template and encapsulated into a fault feature unit. All successfully matched fault feature units are then aggregated, with each unit corresponding to a multi-dimensional preliminary fault feature vector. The dimensions of the preliminary fault feature vector include feature codes, faulty equipment identifiers, fault occurrence time, fault type, confidence weights, and a list of associated abnormal data points. All vectors constitute the preliminary fault feature vector set.

[0078] In practice, a digital geographic information layer of the power grid, containing the geographic coordinates and electrical connections of transmission lines, substations, distribution transformers, and load points, is acquired. This layer is typically stored in a vector graphic file format with spatial attributes. The power grid equipment identifiers associated with each fault feature vector in the preliminary fault feature vector set are located within the digital geographic information layer of the power grid to determine the precise geographic coordinates of the fault occurrence. For example, this can be achieved by associating the equipment identifier with the unique code of the equipment element in the layer and reading the center point coordinates of the corresponding element. Using the precise geographic coordinates as the center, and based on the fault type identifier and confidence weight contained in the preliminary fault feature vector set, the corresponding electrical distance attenuation coefficient and load importance weight coefficient are selected from a pre-set impact propagation model. For example, for the "bus short circuit" fault type, the electrical distance attenuation coefficient is set relatively large, resulting in faster impact propagation; for high-confidence faults, the impact range threshold used in the calculation may be more stringent. In some embodiments, a breadth-first search is performed on the digital geographic information layer of the power grid, centered on the fault point and along the topological connection edges. The impact intensity is calculated using an electrical distance attenuation coefficient. The search stops when the impact intensity attenuates below a threshold or when an electrical disconnection point is encountered. All nodes and edges traversed constitute the potential impact range. The calculation of the impact intensity comprehensively considers electrical distance, network topology, and fault type. Optionally, during the breadth-first search, the impact intensity is attenuated once based on a preset electrical distance attenuation coefficient after each topological node or line is traversed, and compared with a preset impact intensity threshold. It can be understood that the bounding polygon boundary coordinates of the potential impact range, the list of affected devices within the range, and the estimated impact level of each device are structured. The bounding polygon boundary coordinates are obtained by calculating the minimum bounding rectangle of all device coordinates within the potential impact range. The estimated impact level is determined comprehensively based on the electrical distance of the device from the fault point and the fault type. The output is a fault impact domain description file, containing structured fields such as geographic boundaries, device list, and impact level.

[0079] In one embodiment of the present invention, a fault impact domain description file is parsed, and the bounding polygon boundary coordinates of the potential impact range recorded therein are extracted. Using the bounding polygon boundary coordinates as spatial filtering conditions, a spatial query is performed on the real-time data records to quickly retrieve the current operating status data of all power grid equipment whose geographical locations fall within the polygon range, forming an affected equipment status dataset. Historical similar cases are retrieved from a historical fault case database using a combination of current fault type, main affected equipment type, and current power grid operating mode as query conditions. The handling measures taken in the historical similar cases and the resulting changes in the power grid status are extracted. The affected equipment status dataset is used as the initial power grid status, and the handling measures taken in the historical similar cases are used as the basis for further analysis. The proposed measures are used as a set of candidate operations. Each candidate operation is simulated and executed in the power flow calculation model and the safety constraint model. During the simulation, the power flow distribution, node voltage, and line load rate of the power grid are evaluated in real time to determine whether they meet the safety operation constraints. The simulated operation sequence and its resulting final state are recorded. From all simulated operation sequences that meet the safety constraints, the sequences are evaluated and ranked according to the comprehensive objective function of maximizing the power restoration range, minimizing the operation steps, and minimizing the overall risk. The optimal sequence is selected, and each simulated operation in the optimal sequence is converted into a specific remote control instruction that the dispatch control system can recognize. The preconditions and verification steps for executing the remote control instruction are attached and arranged in sequence to form the final fault handling plan.

[0080] In practical implementation, the fault impact domain description file is parsed to extract the bounding polygon boundary coordinates of the potential impact range recorded within it. These coordinates are stored as a series of latitude and longitude pairs; for example, a rectangular impact domain consists of the coordinates of four points. Using the bounding polygon boundary coordinates as spatial filtering conditions, spatial queries are performed on the real-time data records. Based on the indexing function of the geospatial database, the spatial query quickly retrieves the current operating status data of all power grid equipment whose geographical location falls within the polygon range, forming an affected equipment status dataset. This dataset is a structured data collection containing equipment identifiers, real-time measurement values, and timestamps. Historical similar cases are retrieved from the historical fault case database using a combination of query conditions: current fault type, main affected equipment type, and current power grid operating mode. For example, if the current fault type is "busbar undervoltage," the main affected equipment types include "110kV bus tie switch" and "110kV outgoing line switch," and the current power grid operating mode is "single busbar segmented operation," the historical fault case database stores records of past fault events and their handling logs. It is understandable that extracting the handling measures taken in similar historical cases and the resulting changes in the power grid state after their execution is crucial. These handling measures may include specific operational instructions such as "opening the incoming switch of the faulty bus" or "closing the bus tie switch." The power grid state change results record the voltage and power flow recovery status of key measuring points after the operation. Using the affected equipment state dataset as the initial power grid state and the handling measures from similar historical cases as the candidate operation set, each candidate operation is simulated and executed in the power grid power flow calculation model and the security constraint model.

[0081] In practical implementation, during the simulation execution, the power flow distribution, node voltages, and line load rates of the power grid are evaluated in real time to determine whether they meet the safety operation constraints. These constraints include line current not exceeding the long-term allowable current carrying capacity and node voltages maintained within ±10% of the rated voltage. The simulation operation sequence and its resulting final state are recorded. From all simulation operation sequences that meet the safety constraints, an evaluation and ranking are performed based on a comprehensive objective function that maximizes the restored power supply range, minimizes the operation steps, and minimizes overall risk, selecting the optimal sequence. In some embodiments, the comprehensive objective function can be expressed as:

[0082]

[0083] in: Representing the The overall evaluation value of each operation sequence Representing the The power supply load capacity is restored after the execution of each operation sequence. This represents the maximum value of the power supply load capacity to be restored among all feasible operation sequences. Representing the The number of operation instructions contained in an operation sequence. This represents the maximum number of operation instructions among all feasible operation sequences. Representing the The overall risk assessment value of each operation sequence, This represents the maximum value of the overall risk assessment across all feasible operational sequences. , , These are the weighting coefficients for three indicators: the range of power restoration, the number of operational steps, and the overall risk, and they must satisfy... It is understandable that calculating each operation sequence... Value, according to Sort all operation sequences by value from high to low, and select... The operation sequence with the highest value is selected as the optimal sequence. Each simulated operation in the optimal sequence is converted into a specific remote control instruction recognizable by the dispatching and control system, along with the preconditions and verification steps for executing the remote control instruction. These are then arranged sequentially to form the final fault handling plan. In some embodiments, the fault handling plan is presented in the form of a standardized electronic instruction ticket, including fields such as operation number, operation content, target equipment, target status, preconditions, verification steps, and expected results. Optionally, after the fault handling plan is generated, it is pushed to the dispatcher's console for manual review and confirmation, or executed directly via the automatic control system interface.

[0084] In one embodiment of the present invention, the switching status, active power output, reactive power output, and load value of each device in the affected equipment status dataset are injected into the power grid flow calculation model to calculate the initial power flow distribution within the current fault influence domain. The handling measures extracted from similar historical cases are classified into switching operation type, power output adjustment type, and load control type. Each type of measure is described as a rule for changing the corresponding equipment parameters or status in the power grid model. In the simulation environment, the change rules are applied to the power grid flow calculation model sequentially according to the operation order in the candidate operation set. After each rule is applied, the power flow calculation is re-performed to obtain a new power grid state. After each re-performance of the power flow calculation, the calculation results are compared and verified with the line current carrying limit, transformer capacity limit, and voltage upper and lower limits defined in the safety constraint model. All operation sequences in which the state quantities obtained by the power grid flow calculation model during the simulation process always satisfy all constraints in the safety constraint model, as well as the final power grid state after each sequence is executed, are recorded as feasible handling measures. The strategy set is used to calculate the total number of load points and total load capacity that can be restored from the power outage state within the scope described in the fault impact domain description document after the operation sequence is completed. This serves as the power restoration range indicator. The number of specific operation instructions contained in each operation sequence is counted as the operation step quantity indicator. Based on the power flow calculation model, data such as line load rate and voltage deviation recorded during the simulation execution of each operation sequence are used. Combined with equipment reliability parameters, the probability and severity of secondary risks such as equipment overload and cascading failures that may be caused during the execution of the operation sequence are calculated. After weighted summation, the overall risk assessment index is obtained. The power restoration range indicator, operation step quantity indicator, and overall risk assessment index are normalized respectively. Different weight coefficients are assigned to each index according to the scheduling priority strategy. The weighted comprehensive score of each operation sequence is calculated. All operation sequences are sorted from high to low according to the score, and the operation sequence with the highest score is selected as the optimal sequence.

[0085] In practical implementation, the switching status, active power output, reactive power output, and load values ​​of each device in the affected equipment status dataset are injected into the power grid flow calculation model. The initial power flow distribution within the current fault influence domain is calculated, reflecting the power flow and voltage status of the power grid before any operation is performed after the fault occurs. The handling measures extracted from similar historical cases are categorized into switching operation, output adjustment, and load control. Switching operation measures correspond to the opening and closing commands of circuit breakers or disconnectors; output adjustment measures correspond to the commands to change the active and reactive power output setpoints of generators or reactive power compensation devices; and load control measures correspond to the commands to disconnect or restore interruptible loads. Each type of measure is described as a rule for changing the parameters or status of the corresponding device in the power grid model. For example, the switching operation measure "closing the bus tie switch CB-101" is described as a rule to change the "switching status" attribute of device CB-101 in the power grid model from "open" to "closed". In the simulation environment, the rules for changing power flow are applied sequentially to the power flow calculation model according to the operation order in the candidate operation set. After each rule is applied, the power flow calculation is re-performed to obtain a new power grid state. The power flow calculation employs numerical methods such as the Newton-Raphson method or the fast decoupling method to solve the power flow equations. After each re-performance of the power flow calculation, the calculation results are compared and verified with the line current carrying limit, transformer capacity limit, and voltage upper and lower limits defined in the safety constraint model. The line current carrying limit is determined based on the conductor type and operating environment; the transformer capacity limit is determined based on the rated capacity on the equipment nameplate; and the voltage upper and lower limits are determined based on the power grid operation regulations. Essentially, the set of feasible disposal strategies is created by recording all operation sequences in which the state variables obtained from the power flow calculation model during the simulation process consistently satisfy all constraints in the safety constraint model, as well as the final power grid state after each sequence is executed. Each element in the feasible disposal strategy set contains an ordered list of operation instructions and a corresponding snapshot of the final power grid state.

[0086] For each operation sequence in the set of feasible handling strategies, after the operation sequence is completed, the total number of load points and the total load capacity that can be restored from the power outage state within the range described in the fault impact domain description document are counted as indicators of the power restoration range. The power outage state of a load point is determined based on whether its power supply bus voltage is lower than the minimum operating voltage or whether the upstream switch is in the open state. The number of specific operation instructions contained in each operation sequence is counted as an indicator of the number of operation steps. Each specific operation instruction corresponds to an independent remote control or remote adjustment command that can be executed by the dispatch control system. Based on the data such as line load rate and voltage deviation recorded during the simulation execution of each operation sequence using the power grid power flow calculation model, combined with equipment reliability parameters, the probability and severity of secondary risks such as equipment overload and cascading failures that may be triggered during the execution of the operation sequence are calculated, and a weighted sum is obtained to obtain the overall risk assessment index. In some embodiments, the calculation of the overall risk assessment index can be expressed as:

[0087]

[0088] in: Represents the overall risk assessment value. This represents the total number of all transmission lines and transformer branches involved in the simulation. Representing the The maximum load rate of the branch during the simulation process, Representing the The risk weighting coefficient for each branch is set based on the branch type, historical failure rate, and importance in the power grid. This represents the total number of all nodes involved in the simulation. Representing the The maximum voltage deviation per unit value of each node during the simulation process. Representing the The voltage risk weighting coefficient for each node. This can be understood as follows: the power restoration range indicator, the number of operation steps indicator, and the overall risk assessment indicator are normalized separately. Normalization uses a minimum-maximum method to map each indicator value to the [0,1] interval, and different weighting coefficients are assigned to each indicator according to the scheduling priority strategy. For example, in the emergency power restoration phase, the weighting coefficient for the power restoration range indicator... It may be set to 0.6, the weighting coefficient for the number of operation steps indicator. The weighting coefficient of the overall risk assessment indicator is set to 0.2. The value is set to 0.2. A weighted composite score is calculated for each operation sequence. All operation sequences are sorted from highest to lowest score, and the operation sequence with the highest score is selected as the optimal sequence. Optionally, refer to Table 1 for the evaluation index data of different operation sequences in the feasible disposal strategy set.

[0089] Table 1: Evaluation Indicators for Feasible Disposal Strategies

[0090]

[0091] In some embodiments, based on the data in Table 1, it is assumed that the weighting coefficients are set to... , , The operation sequence Seq-01 has the highest overall score, therefore it is selected as the optimal sequence.

[0092] See Figure 4 In the simulation and deduction phase of the power grid fault handling strategy, the heatmap visually presents the pass rate distribution of various constraint indicators for different operation sequences (Seq-01 to Seq-05) during the power flow calculation and security constraint verification stages. Specifically, the horizontal dimension of the graph corresponds to five core security constraint items: line current carrying capacity limit, transformer capacity limit, voltage upper and lower limits, switch status validity, and power flow convergence; the vertical dimension represents candidate operation sequences. During the verification process, the constraint pass rate corresponding to each operation sequence is quantified by a color gradient (green → yellow → red): green represents a pass rate ≥ 98% (e.g., the transformer capacity limit and switch status validity pass rates of Seq-01 reach 100%), yellow corresponds to 94%-96% (e.g., the voltage upper and lower limits pass rate of Seq-02 is 94%), and orange / red indicates low pass rate items (e.g., the voltage upper and lower limits pass rate of Seq-03 is only 92%). The data distribution clearly shows that the validity of the switch state maintains a 100% pass rate across all sequences, reflecting the topological logic compliance of the operation sequences. However, the voltage upper and lower limits are weak points in constraint verification (with a pass rate of less than 95% for multiple sequences), requiring a stronger priority for voltage regulation measures in subsequent strategy optimization. The value of this heatmap lies in its spatial presentation of multi-dimensional pass rates, providing a quantitative basis for safety constraints in selecting the optimal operation sequence. It is a key visualization tool for assessing the feasibility of fault handling strategies.

[0093] In one embodiment of the present invention, each simulated operation in the optimal sequence is mapped to a standard remote control and remote adjustment command template of the dispatch control system, and specific equipment numbers, target state values ​​or adjustment values ​​are filled in to generate machine-readable raw control commands. For each raw control command, the real-time operating conditions of the power grid that must be met when it is actually executed in the power grid are analyzed, and they are converted into a series of logical judgment statements to form a list of preconditions for the execution of the command. For each raw control command, a feedback verification step is designed after the command is issued. The feedback verification step includes checking whether the equipment status has changed to the target status and whether the relevant electrical measurement values ​​have entered the expected range within a specified time. The raw control command, its corresponding list of preconditions, and the feedback verification step are combined into a complete step unit according to the execution order in the optimal sequence. All step units are assembled in sequence, and a fault overview, scope of impact, and overall handling objectives are added at the beginning, and a description of the expected power grid status after the handling is completed is added at the end. Finally, a fault handling plan document is packaged to generate a fault handling plan document.

[0094] In practical implementation, each simulated operation in the optimal sequence is mapped to a standard remote control and remote adjustment command template of the dispatch control system. The standard remote control command template is used to control the opening and closing operations of switchgear, and the standard remote adjustment command template is used to adjust the setpoint of generator or on-load tap-changing transformer. The specific equipment number, target status value or adjustment value is filled in to generate machine-readable raw control commands. For example, the simulated operation "close the bus tie switch CB-101" is mapped and filled into a remote control selection execution command that conforms to the IEC60870-5-101 protocol. Its message body contains the equipment address code of "CB-101" and the control command code of "closing". For each original control command, the real-time operating conditions of the power grid that must be met during its actual execution are analyzed and converted into a series of logical judgment statements, forming a list of preconditions for the execution of the command. For example, for the original control command "close the bus tie switch CB-101", the list of preconditions may include "the remote / local control handle of switch CB-101 is in the 'remote' position", "the spring energy storage mechanism of switch CB-101 has completed energy storage", and "the isolating switches QS-101 and QS-102 on both sides of switch CB-101 are both in the closed position". In some embodiments, these preconditions are verified by querying the device status telemetry data in the real-time database or by obtaining feedback by sending query commands to the device. For each original control command, a feedback verification step is designed after the command is issued. This step includes checking whether the equipment status has changed to the target status and whether relevant electrical measurements have entered the expected range within a specified time. For example, after issuing a closing command to switch CB-101, the feedback verification step is set to check within 10 seconds whether the remote signaling position signal transmitted from the switch is "closed" and whether the bus voltage connected to the switch has returned to the rated range. In practice, the length of the specified time is determined based on the equipment type and the average response time of the communication network. The signal required for feedback verification comes from the real-time data acquisition of the dispatch automation system.

[0095] It can be understood that the original control commands, their corresponding preconditions, and feedback verification steps are combined into a complete step unit according to the execution order in the optimal sequence. A step unit is the smallest independently verifiable and executable operational logic block in the fault handling plan. All step units are assembled sequentially, with a fault overview, impact scope, and overall handling objective appended at the beginning, and a description of the expected power grid state after handling appended at the end. Finally, a fault handling plan file is generated. The fault overview briefly describes the fault time, equipment, and type; the impact scope references key information from the fault impact domain description file; and the overall handling objective states the core result achieved in this handling, such as "restoring power supply to all loads of the ABC substation except the faulty bus." The description of the expected power grid state includes quantitative indicators such as the voltage level of key nodes and the load rate of the main supply line. In some embodiments, the fault handling plan file is encapsulated in a structured format such as XML or JSON to facilitate parsing and execution by the dispatch control system or intelligent operation ticket system. Optionally, after the fault handling plan file is generated, it is assigned a unique plan identifier and associated with the initial fault event record that triggered this handling, forming new case knowledge stored in the historical fault case library.

[0096] See Figure 5 In the fault handling strategy simulation stage of the smart grid dispatch and control system, the correlation between historical case matching degree and handling effect was quantified. Specifically, the figure uses case matching degree (%) as the horizontal axis and handling effect score as the vertical axis. The blue solid line represents the average handling effect under different matching degrees, the light gray area shows the effect fluctuation range, and the red dashed line is the trend line. When the case matching degree increases from 60% to 100%, the average handling effect score increases from 65±8 to 96±2, showing a significant positive correlation trend. At the same time, the effect fluctuation range (error term) decreases as the matching degree increases (from ±8 to ±2), indicating that the higher the matching degree, the stronger the stability of the handling effect. This relationship provides a quantitative basis for the case retrieval stage of fault handling strategy: when searching the historical case database, cases with higher matching degrees should be selected first to improve the certainty and quality of the handling effect.

[0097] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0098] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent fault handling in a smart grid dispatching and control system, characterized in that, The method includes: Receive asynchronous multi-source monitoring data streams from different monitoring levels of the power grid, and import the asynchronous multi-source monitoring data streams into a distributed message queue for caching and traffic shaping. For the original monitoring data entries cached in the distributed message queue, based on the unique code of the power grid equipment and the data generation time, time sequence alignment and data splicing operations are performed to construct a real-time data record with complete spatiotemporal attributes. From the real-time data records, characteristic data representing abnormal power grid states are identified and separated. Based on a pre-set fault feature template library, pattern matching and feature extraction are performed on the characteristic data to generate a preliminary fault feature vector set, including: Iterate through the various operational status measurements in the real-time data records, compare them with the normal operating threshold range of the corresponding equipment or line, and mark all data points that exceed the limit and their exceeding the limit. Load predefined fault feature templates from the historical database. The fault feature templates define the abnormal change patterns of multiple related state variables in the time dimension and the correlation propagation patterns in the spatial dimension under various typical faults. The set of marked out-of-limit data points is analyzed by sliding window on the time axis to capture abrupt changes, step jumps, and oscillation trends of state variables, and similarity calculation is performed with the time patterns defined in the fault feature template. The set of data points that exceed the limit and their positional relationship in the power grid topology that meet the similarity conditions are matched with the spatial association pattern defined in the fault feature template. If both the spatiotemporal patterns are successfully matched, the corresponding feature code, fault type identifier and confidence weight are extracted from the template and encapsulated into a fault feature unit. All successfully matched fault feature units are summarized, and each unit corresponds to a multi-dimensional preliminary fault feature vector. All vectors constitute the preliminary fault feature vector set. By combining the power grid geographic information layer with the topological connectivity, the preliminary fault feature vector set is mapped to the physical power grid structure. Based on the mapping results, the potential impact domain boundary of the fault is calculated, and a fault impact domain description file is generated, including: Obtain a digital geographic information layer of the power grid that includes the geographic coordinates and electrical connections of transmission lines, substations, distribution transformers, and load points; The power grid equipment identifier associated with each fault feature vector in the preliminary fault feature vector set is located in the power grid digital geographic information layer to determine the precise geographical coordinates of the fault occurrence. Centered on the precise geographical coordinates, and based on the fault type identifier and confidence weight contained in the preliminary fault feature vector set, the corresponding electrical distance attenuation coefficient and load importance weight coefficient are selected from the preset impact propagation model. On the digital geographic information layer of the power grid, a breadth-first search is performed along the topological connection edge with the fault point as the center. The influence intensity is calculated in combination with the electrical distance attenuation coefficient. The search stops when the influence intensity attenuates to below the threshold or when an electrical disconnection point is encountered. All nodes and edges traversed constitute the potential influence range. The outer polygon boundary coordinates of the potential impact range, the list of affected devices within the range, and the estimated impact level of each device are structured and organized, and output as the fault impact domain description file. Based on the spatial range defined by the fault impact domain description file, the status data of all operating equipment within the spatial range are filtered from the real-time data records. Combined with a historical fault case library, multiple rounds of iterative handling strategy deduction are performed to ultimately generate a fault handling plan containing a specific sequence of operation instructions, including: Parse the fault impact domain description file and extract the bounding polygon boundary coordinates of the potential impact range recorded therein; Using the boundary coordinates of the circumscribed polygon as a spatial filtering condition, spatial queries are performed on the real-time data records to quickly retrieve the current operating status data of all power grid equipment whose geographical location falls within the polygon range, forming a dataset of the status of affected equipment. From the historical fault case database, using the current fault type, the main affected equipment type, and the current power grid operation mode as combined query conditions, historical similar cases are retrieved, and the handling measures taken in the historical similar cases and the changes in the power grid status after their implementation are extracted. Using the affected equipment status dataset as the initial power grid state and the handling measures in the historical similar cases as the candidate operation set, each candidate operation is simulated and executed in the power grid power flow calculation model and the security constraint model. During the simulation, the power flow distribution, node voltage, and line load rate of the power grid are evaluated in real time to determine whether they meet the safety operation constraints, and the simulation operation sequence and its resulting final state are recorded. From all simulated operation sequences that meet safety constraints, the optimal sequence is selected by evaluating and ranking them based on a comprehensive objective function that maximizes the range of power restoration, minimizes the number of operation steps, and minimizes overall risk. Each simulated operation in the optimal sequence is converted into a specific remote control instruction recognizable by the scheduling and control system, along with the preconditions and verification steps for executing the remote control instruction. These are then arranged in sequence to form the final fault handling plan.

2. The intelligent fault handling method for a smart grid dispatching and control system according to claim 1, characterized in that, Receive asynchronous multi-source monitoring data streams from different monitoring levels of the power grid, and import the asynchronous multi-source monitoring data streams into a distributed message queue for caching and traffic shaping, including: Listen to and receive real-time data messages from the power grid dispatching master station, substation automation system, distribution automation terminal and distributed energy monitoring unit; The received real-time data packets are parsed according to the protocol, the communication packet header is stripped, and the payload containing device code, timestamp, measurement value and data quality identifier is extracted; The payload is written into message topic channels of different priorities according to its source monitoring level and data urgency label. The distributed message queue dynamically adjusts memory allocation according to the consumption rate of each topic channel. Each payload written to the distributed message queue is appended with a globally unique streaming identifier, and the time it enters the queue is recorded to form a raw data buffer stream with a time sequence mark.

3. The intelligent fault handling method for a smart grid dispatching and control system according to claim 2, characterized in that, For the raw monitoring data entries cached in the distributed message queue, based on the unique code of the power grid equipment and the data generation time, time alignment and data concatenation operations are performed to construct a real-time data record with complete spatiotemporal attributes, including: The raw data cache stream with time-series tags is consumed sequentially from the designated topic channel of the distributed message queue; Using the unique code of the power grid equipment as the primary key, multiple data entries from different monitoring levels targeting the same power grid equipment are aggregated within a preset time alignment window; The aggregated data entries are sorted in ascending order according to their data generation time, and similar measurement data with time deviations within the tolerance range are processed using a weighted fusion algorithm based on data quality identification to form a unique state snapshot of the power grid equipment within the time window. The unique state snapshot of each power grid device is associated and bound with its corresponding geographical coordinates and topology node information to generate a real-time data record containing time, space, device attributes, and operating status.

4. The intelligent fault handling method for a smart grid dispatching and control system according to claim 1, characterized in that, Using the affected equipment status dataset as the initial power grid state and the handling measures in the historical similar cases as the candidate operation set, each candidate operation is simulated and executed in the power grid power flow calculation model and the security constraint model, including: The switching status, active power output, reactive power output, and load value of each device in the affected equipment status dataset are injected into the power grid power flow calculation model to calculate the initial power flow distribution within the current fault influence domain. The measures extracted from the historical similar cases are classified into switch operation type, output adjustment type, and load control type. Each type of measure is described as a rule for changing the corresponding equipment parameters or states in the power grid model. In the simulation environment, the change rules are applied to the power flow calculation model in the order of the operation in the candidate operation set. After each rule is applied, the power flow calculation is performed again to obtain the new power grid state. After each power flow calculation, the calculation results are compared and verified with the line current carrying limit, transformer capacity limit, and voltage upper and lower limits defined in the security constraint model. Record all operation sequences in which the state variables obtained by the power flow calculation model always satisfy all constraints in the security constraint model during the simulation execution process, as well as the final power grid state after each sequence is executed, as a set of feasible disposal strategies.

5. The intelligent fault handling method for a smart grid dispatching and control system according to claim 4, characterized in that, From all simulated operation sequences that meet safety constraints, the optimal sequences are selected based on a comprehensive objective function that maximizes the power restoration range, minimizes the number of operation steps, and minimizes overall risk. These sequences include: For each operation sequence in the set of feasible handling strategies, the total number of load points and the total load capacity that can be restored from the power outage state within the range described in the fault impact domain description file after the operation sequence is completed are counted as indicators of the power restoration range. The number of specific operation instructions contained in each operation sequence is counted as an indicator of the number of operation steps. Based on the line load rate and voltage deviation data recorded during the simulation of each operation sequence by the power grid power flow calculation model, and combined with equipment reliability parameters, the probability and severity of secondary risks of equipment overload and cascading failures that may be caused during the execution of the operation sequence are calculated, and the overall risk assessment index is obtained after weighted summation. The power restoration range index, the number of operation steps index, and the overall risk assessment index are normalized respectively, and different weight coefficients are assigned to each index according to the scheduling priority strategy. Calculate the weighted comprehensive score for each operation sequence, sort all operation sequences from highest to lowest score, and select the operation sequence with the highest score as the optimal sequence.

6. The intelligent fault handling method for a smart grid dispatching and control system according to claim 5, characterized in that, Each simulated operation in the optimal sequence is converted into a specific remote control instruction recognizable by the scheduling and control system, along with the preconditions and verification steps for executing the remote control instruction. These are then arranged sequentially to form the final fault handling plan, including: Each simulated operation in the optimal sequence is mapped to a standard remote control and remote adjustment instruction template of the scheduling and control system, and the specific equipment number, target status value or adjustment value is filled in to generate a machine-readable raw control instruction. For each of the original control commands, analyze the real-time operating conditions of the power grid that must be met when the command is actually executed, and convert them into a series of logical judgment statements to form a list of preconditions for the execution of the command. For each of the original control commands, a feedback verification step is designed after the command is issued. The feedback verification step includes checking whether the equipment status has changed to the target status and whether the relevant electrical measurement values ​​have entered the expected range within a specified time. The original control command, its corresponding precondition list, and the feedback verification steps are combined into a complete step unit according to the execution order in the optimal sequence. Assemble all the steps in sequence, and add a fault overview, scope of impact, and overall handling objectives at the beginning, and add a description of the expected power grid state after handling is completed at the end. Finally, package them to generate the fault handling plan file.

7. A fault intelligent handling system for a smart grid dispatch and control system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent fault handling method of the smart grid dispatch and control system according to any one of claims 1 to 6.

Images