A Data Processing Method for Coordinated Situational Awareness and Intelligent Judgment in Main and Distribution Networks

By constructing a global clock-based associated dataset and a power supply affiliation mapping table in the main distribution network, the problems of data heterogeneity and time inconsistency in the main distribution network are solved. This enables accurate splicing and two-way quantitative analysis of the situation characteristics of the main and distribution networks, improving the accuracy and efficiency of situation awareness and intelligent analysis.

CN122839320BActive Publication Date: 2026-11-17STATE GRID ZHEJIANG HAIYAN COUNTY ELECTRIC POWER SUPPLY CO +7
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202611320919.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-08-28
Publication Date
2026-11-17
Estimated Expiration
2046-08-28

AI Technical Summary

Technical Problem

In the main and distribution networks, differences in equipment naming rules, topology description methods, measurement point coding, and time scale definitions make it difficult to directly correlate data after collection, and the inconsistent time base affects the accuracy and efficiency of situational awareness and intelligent judgment.

Method used

Based on the global clock of the main network, an associated dataset is constructed. A device association mapping table is built through the power supply affiliation and time alignment is performed to form a two-way collaborative judgment mechanism, realizing the structured splicing of the status characteristics of the main and distribution networks and a unified time series benchmark.

Benefits of technology

It improves the accuracy and overall efficiency of data fusion in the coordinated situational awareness of the main grid and distribution network, supports two-way quantitative assessment of the impact range of the downward propagation of main grid faults and the upward verification of distribution network anomalies, and provides more comprehensive coordinated situational awareness and intelligent judgment results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122839320B_ABST
    Figure CN122839320B_ABST
Patent Text Reader

Abstract

This invention relates to the field of data processing technology, specifically to a data processing method for collaborative situational awareness and intelligent judgment in a main distribution network. The method includes: constructing a correlated dataset based on multi-source data streams in the main distribution network and using the main network's global clock as a reference; structurally splicing the situational characteristics of the main and distribution networks to obtain a situational feature vector set; constructing a first path for evaluating the impact on the distribution network downwards and a second path for verifying the main network's status upwards through distribution network anomalies, based on the established operating characteristic reference intervals and deviation thresholds in the main distribution network, forming a two-way collaborative judgment mechanism; inputting the situational feature vector set into the first and second paths in parallel, and associating and aggregating the output judgment results according to the same power supply affiliation to obtain a collaborative judgment result dataset; this invention can improve the overall efficiency of data processing links and the fidelity of information transmission in the main distribution network.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a data processing method for coordinated situational awareness and intelligent analysis of main and distribution power grids. Background Technology

[0002] In the coordinated operation and management of the main grid and distribution network, the dispatch center needs to simultaneously acquire multi-source heterogeneous data from both the main grid and distribution network sides for comprehensive situational awareness and intelligent analysis. However, due to the different data modeling standards and communication protocol systems adopted by the main grid and distribution network during their long-term construction, there are fundamental differences between them in equipment naming rules, topology description methods, measurement point coding, and time scale definitions, making it difficult to directly correlate data at the equipment level after collection. Furthermore, the sampling frequency and time base of the main grid data and distribution network data are inconsistent, further exacerbating the difficulty of aligning multi-source information in the spatiotemporal dimensions. These heterogeneities and spatiotemporal inconsistencies at the data level constitute objective technical challenges for data processing in the coordinated situational awareness and intelligent analysis of the main grid and distribution network.

[0003] To address the aforementioned issues, existing solutions in this field attempt to achieve physical aggregation of multi-source data by establishing a unified model extension or intermediate database, or to use linear interpolation and nearest neighbor matching for time alignment. However, existing solutions rely heavily on string matching of device names or manual configuration of mapping tables at the device association level. When the naming rules of the two systems are inconsistent or the ledger updates are lagging, the mapping relationship is difficult to maintain dynamically. At the time alignment level, a uniform interpolation strategy is generally adopted, failing to consider the essential differences in sampling density between the main and distribution network data and the master-slave relationship of the time base, resulting in insufficient alignment accuracy. At the process connection level, situational awareness and intelligent judgment are disconnected, and situational characteristics are not structurally transmitted to the judgment stage, resulting in data redundancy and information loss. At the judgment logic level, they mostly focus on a single voltage level, lacking a two-way quantitative evaluation mechanism for the downward propagation of main network faults and the upward backtracking of distribution network anomalies, making it difficult for judgment results to support dispatchers' overall grasp of the coordinated situation of the main and distribution networks. Summary of the Invention

[0004] This invention provides a data processing method for coordinated situational awareness and intelligent analysis of main and distribution networks to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a data processing method for coordinated situational awareness and intelligent analysis of main and distribution networks, comprising: Step 1: Based on the multi-source data streams in the main distribution network, construct an associated dataset with the main network global clock as the reference, and structurally splice the status features of the main and distribution networks to obtain a status feature vector set; Step 2: Based on the operating characteristic reference range and deviation threshold set in the main distribution network, construct the first path of evaluating the impact of the distribution network downward through the main network and the second path of verifying the status of the main network upward through distribution network anomalies, forming a two-way collaborative judgment mechanism; Step 3: Input the situation feature vector set into the first path and the second path in parallel, and collect the output judgment results according to the same power supply affiliation to obtain a collaborative judgment result dataset.

[0006] Furthermore, the multi-source data streams in the main distribution network include: Acquire main network model data and main network measurement data from the control cloud, distribution network model data and distribution network operation data from the distribution automation terminal, and load data and power outage data from the electricity consumption information collection terminal; Protocol parsing is performed on the main network model data, main network measurement data, distribution network model data, distribution network operation data, load data, and power outage data to obtain the equipment identifier, measurement type, measurement value, and timestamp information fields of the main and distribution networks. The equipment identification, measurement type, measurement value, and timestamp information fields are reorganized according to a preset unified data format to obtain a standardized data stream.

[0007] Furthermore, the construction of the associated dataset includes: In the main distribution network, trace upwards along the parent node of the distribution network topology until the main power supply network equipment is identified, and record the power supply affiliation to form a main distribution equipment association mapping table; Using the timestamp of the main network measurement sequence as the sole benchmark, weighted interpolation is performed on the positions of measurement values ​​corresponding to missing main network reference times in the distribution network measurement sequence. Median convergence is also performed on the cases where multiple distribution network measurement values ​​exist within the window corresponding to the main network reference time, generating a time-aligned distribution network measurement sequence. The time-aligned main and distribution network measurement data are bound according to the main and distribution equipment association mapping table to obtain the associated dataset.

[0008] Furthermore, the structured splicing of the main and distribution network situation characteristics to obtain a situation characteristic vector set includes: Based on the main network measurement data in the associated dataset, calculate the central trend measurement value, dispersion measurement value, and change amplitude measurement value of each measurement value sequence belonging to the same main network device identifier within a preset time window, and combine them to form a main network device status feature sub-vector; Based on the measurement data of distribution network equipment that have a power supply relationship with the main network equipment in the associated dataset, within the same preset time window, for each set of measurement value sequences belonging to the same distribution network equipment identifier, the central trend measure, dispersion measure and change amplitude measure of the measurement value sequence are calculated and combined to form a distribution network equipment status feature sub-vector. The status feature sub-vectors of the main grid equipment and the status feature sub-vectors of the distribution network equipment under the same power supply affiliation are spliced ​​together end to end according to the preset feature dimension order to form the main grid-distribution associated status feature vector. The main and distribution relationship situation feature vectors generated under all power supply affiliations are summarized to form a situation feature vector set.

[0009] Furthermore, the preset time window adopts an adaptive adjustment rule: the millisecond value of the original sampling period of the distribution network equipment is divided by 2 and then rounded up to obtain the duration of the single-sided window.

[0010] Furthermore, the central tendency measure includes at least the arithmetic mean and median of the measured numerical sequence, the dispersion measure includes at least the standard deviation and range of the measured numerical sequence, and the variation measure includes at least the average of the absolute values ​​of the first differences of the measured numerical sequence.

[0011] Furthermore, a two-way collaborative analysis mechanism will be established, including: Read the reference range of normal operation characteristics of the main grid, the reference range of normal operation characteristics of the distribution network, and the deviation threshold that were initially set in the main distribution network; If any metric value in the main network side feature component of the main network-related situation feature vector deviates from the main network normal operation feature reference range, it is determined that the main network device has an abnormal operating status. If any metric value in the distribution network side feature component of the main distribution network correlation status feature vector deviates from the normal operation feature reference range of the distribution network, it is determined that the distribution network equipment has an abnormal operating status. For main grid equipment that is determined to have abnormal operating status, all distribution network equipment that supplies power are traced down according to the power supply affiliation. By comparing the distribution network side characteristic components of the distribution network equipment with the normal operating characteristic reference range and deviation threshold of the distribution network, the impact range and risk level of the main grid fault on the distribution network are quantified. For distribution network equipment that is determined to have abnormal operating status, the power supply affiliation is traced back to the main grid equipment that supplies the power. The impact status of the main grid equipment is verified by comparing the main grid side characteristic components of the main grid equipment with the normal operating characteristic reference range of the main grid.

[0012] Furthermore, the scope and risk level of the quantified main network failure on the distribution network include: Based on the main and distribution equipment association mapping table, query all distribution network equipment powered by abnormal main network equipment; extract the change range metric value of each distribution network equipment, and evaluate the deviation range from the reference range; The number and topological distribution of devices whose deviation exceeds the deviation threshold are statistically analyzed to generate an impact range determination result. For each affected device, a weighted sum is assigned based on its type and preset weights, and the corresponding risk level is matched according to the total risk score obtained from the sum.

[0013] Furthermore, the verification of the impact status of the main network device includes: Based on the power supply affiliation recorded in the main distribution equipment association mapping table, query the main network equipment identifier that supplies power to the distribution network equipment from the association dataset; Extract the main network side feature components corresponding to the main network device from the situation feature vector set; If there is a measurement value in the main network side feature component that deviates from the reference range of the main network's normal operation characteristics, then an indication information of the main network device being affected is generated and the type and magnitude of the deviation measurement value are recorded. If there are no measurement values ​​in the main network side feature components that deviate from the reference range of normal operation of the main network, then an indication information indicating that the main network equipment is not affected is generated, and the indication information is used as the result of the retrospective impact judgment of the distribution network anomaly on the main network.

[0014] Furthermore, the output analysis results are grouped and associated according to the same power supply affiliation, including: Each primary and secondary related situation feature vector is copied into two copies, which are then simultaneously input into the first path and the second path for parallel analysis. Obtain the impact range and risk level results of the first path output, as well as the retrospective impact results of the second path output; Using each power supply affiliation pairing record in the main and auxiliary equipment association mapping table as a collection node, all judgment results belonging to the same pairing are filled into a preset structured template to generate a collaborative judgment result dataset.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention constructs a primary-distribution equipment association mapping by tracing the power supply affiliation of distribution network equipment to primary network equipment, and uses the primary network's global clock as the sole time-series reference to perform differentiated time alignment processing on distribution network measurement data. This fundamentally solves the problem of accurate association between heterogeneous data in the primary and distribution networks at the equipment level and time dimension. This ensures that multi-source data has a unified equipment association foundation and time-series reference before entering the situation feature extraction stage, significantly improving the accuracy and reliability of low-level data fusion in primary-distribution network collaborative situation awareness. Simultaneously, this method structurally splices the situation features of primary network equipment and distribution network equipment under the same power supply affiliation to form a primary-distribution associated situation feature vector. This allows the situation awareness results to directly serve the subsequent intelligent judgment stage in a structured vector form, avoiding secondary extraction and redundant calculation of feature information, and improving the overall efficiency of the data processing link and the fidelity of information transmission.

[0016] 2. This invention constructs a bidirectional collaborative analysis path based on a situational feature vector set. This path simultaneously supports the downward quantification of the impact range and risk level of main grid faults and the upward verification of the main grid's impact status due to distribution network anomalies. By leveraging power supply affiliation, it achieves a bidirectional quantitative expression of the influence relationship between the main and distribution networks. This ensures that the collaborative analysis dataset includes not only information on the impact range of main grid faults on the distribution network but also information on the retrospective impact verification of distribution network anomalies on the main grid. This provides dispatchers with a more comprehensive overall understanding of the collaborative situation of the main and distribution networks, effectively improving the completeness and practical value of intelligent analysis results in collaborative scenarios. The construction of the associated dataset, the generation of situational feature vectors, and the establishment of the bidirectional analysis path in this method are all based on clear data processing logic and quantifiable feature measurement standards. This ensures that the entire situational awareness and intelligent analysis process has good interpretability and traceability, providing solid data support for dispatching decisions. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a data processing method for coordinated situational awareness and intelligent analysis of a main distribution network, provided in an embodiment of the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] This invention provides a method for collaborative situational awareness and intelligent analysis data processing in a main distribution network. The execution entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this invention: a server, a terminal, etc. In other words, the method can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and basic cloud computing services.

[0021] Reference Figure 1 The diagram shown is a flowchart illustrating a data processing method for coordinated situational awareness and intelligent analysis of a main and distribution network according to an embodiment of the present invention. In this embodiment, the data processing method for coordinated situational awareness and intelligent analysis of a main and distribution network includes: Step 1: Based on the multi-source data streams in the main distribution network, construct an associated dataset with the main network global clock as the reference, and structurally splice the status features of the main and distribution networks to obtain a status feature vector set; In this embodiment of the invention, during the coordinated operation of the main and distribution networks, the main network operation data, distribution network operation data, and load and outage data are fused from multiple sources to obtain a standardized data stream of the main and distribution networks, i.e., a multi-source data stream in the main and distribution networks, including: Acquire main network model data and main network measurement data from the control cloud, distribution network model data and distribution network operation data from the distribution automation terminal, and load data and power outage data from the electricity consumption information collection terminal; Protocol parsing is performed on the main network model data, main network measurement data, distribution network model data, distribution network operation data, load data, and power outage data to obtain the equipment identifier, measurement type, measurement value, and timestamp information fields of the main and distribution networks. The equipment identification, measurement type, measurement value, and timestamp information fields are reorganized according to a preset unified data format to obtain a standardized data stream.

[0022] The dispatch cloud is equipped with a main network model storage partition conforming to the IEC61970CIM standard and an IEC60870-5-104 protocol message output interface. The distribution automation terminal is equipped with a DL / T634.5104 protocol communication module, and the electricity information acquisition terminal uses DL / T645-2007 and DL / T698.41 dual communication protocols to transmit data. During the continuous acquisition cycle of the main and distribution networks operating in coordination, the main network model data and the original messages of the main network measurement data are completely read from the storage partition of the dispatch cloud at a fixed acquisition interval of 5 seconds. The distribution network model data is received frame by frame from the network port communication link of each distribution automation terminal. The raw data messages of the distribution network operation are collected and read from each electricity information collection terminal via the front-end communication gateway of the electricity information collection system. All raw messages are completely cached in the local memory storage partition. The cache capacity is preset to 1,000 frames of messages per data source. When the cache capacity threshold is reached, the earliest cached message is overwritten according to the first-in-first-out rule. The entire process completely retains all raw binary messages of the main network model data and main network measurement data from the control cloud, the distribution network model data and distribution network operation data from the distribution automation terminal, and the load data and power outage data from the electricity information collection terminal, without message truncation or discarding.

[0023] For each mainnet model data and mainnet measurement data original 104 protocol message belonging to the scheduling cloud in the cache, a byte-by-byte frame verification operation is performed. The verification benchmark is that the first byte of the message is a fixed 0x68 start character, and the length field byte value matches the actual total payload length of the message. Messages that do not meet the two benchmarks are directly removed from the cache without further parsing. The TCP underlying layer has completed the transmission integrity verification, and no additional application layer check code verification is required. The verified message is split according to the 104 protocol APCI+ASDU layered structure to extract the 12-bit mainnet device site unique code as a temporary device identifier. The type identifier field matches the pre-stored measurement type code lookup table to obtain three measurement types: telemetry, remote signaling, and electrical energy. The time stamp field reads the original time data of year, month, day, hour, minute, second, and millisecond from the 7-byte CP56Time2a structure as a temporary timestamp. The measurement value data field reads 4 bytes of floating-point binary value as a temporary measurement value, completing the extraction of all valid fields of a single mainnet message.

[0024] For each original DL / T645 message containing load data and power outage data from the electricity information collection terminal in the cache, a byte-by-byte frame verification operation is performed. The verification benchmark is the message preamble 4 bytes (0xFE wake-up code), the 5th byte (0x68 frame start character), and the checksum byte within the frame equal to the unsigned sum of all data bytes. Messages that do not meet the benchmark are directly removed from the cache without further parsing. Messages that pass the verification are split into the 12-bit communication address field of the electricity meter, the data identifier field, the time-of-use load value field, and the power outage event timestamp field. The electricity meter communication address is used as a temporary setting. The system uses a backup identifier and a data identifier to match a pre-stored electricity consumption type coding table to obtain two measurement types: positive active load and power outage event. The power outage event timestamp field and the load acquisition timestamp field read the original time data of year, month, day, hour, and minute as temporary timestamps. The time-of-use load value field reads a 4-digit decimal metering value as a temporary measurement value. This completes the extraction of all valid fields from a single electricity consumption message. After three types of message hierarchical parsing operations, four types of information fields are uniformly obtained: temporary equipment identifier, temporary measurement type, temporary measurement value, and temporary timestamp, which are not standardized in the main distribution network.

[0025] A globally unified data reassembly template is pre-built. The fixed field order in the template is: Equipment Unique Identifier, Data Type Label, Measurement Value, and Standardized Timestamp. The Equipment Unique Identifier within the template uses a globally unique 17-character string encoding. This encoding is formed by sequentially concatenating two segments: a geographic grid code and an equipment type code. The geographic grid code consists of 13 characters, directly concatenated from a 6-character district / county code, a 4-character feeder code, and a 3-character distribution area code. All characters use only decimal digits 0 to 9. The equipment type code consists of 4 decimal digits. In the encoding lookup table, 0001 corresponds to a power substation, 0002 to a busbar, 0003 to a main transformer, 0004 to a feeder, 0005 to a distribution transformer, 0006 to a switch, and 0007 to a meter. After concatenation of the two encoding segments, a unique 17-character Equipment Unique Identifier is formed. The data type label within the template uses a fixed 3-digit decimal number. The coding is as follows: 001 corresponds to main grid remote measurement, 002 corresponds to main grid remote signaling, 103 corresponds to distribution network current, 104 corresponds to distribution network switch position, 105 corresponds to application load, and 106 corresponds to power outage event. The first digit of the code distinguishes the grid level, with 0 representing the main grid and 1 representing the distribution network. The last two digits are assigned to new measurement types in ascending order of natural numbers. Distribution network reactive power is tagged with 107, and distribution network switch fault signal is tagged with 108. The measurement values ​​in the template are uniformly converted to 64-bit floating-point decimal format with two decimal places. The standardized timestamps in the template are uniformly converted to the preset format, which is the total number of milliseconds from 00:00 UTC on January 1, 1970 to the time of collection, represented by 15 decimal digits. This value itself does not carry time zone information. The temporary timestamps extracted from various original messages are converted to UTC time according to their respective time zone references before calculating the total number of milliseconds. The local Beijing time is converted to UTC time according to UTC+8 time zone.

[0026] The parsed temporary device identifier is matched against a pre-stored main distribution network equipment code mapping table. The table stores the correspondence between the temporary communication addresses of all main network substations, buses, main transformers, distribution network feeders, distribution transformers, switches, and metering meters and the global 17-bit unified device identifier. After a successful match, the corresponding 17-bit character is extracted as the unified device identifier required for standardized reassembly. The 3-digit code rule in the parsed temporary measurement type reference template is converted into the corresponding numeric character as the data type label. The parsed temporary measurement value is truncated to a fixed number of decimal places according to the fixed floating-point conversion rule, discarding any decimal places exceeding two decimal places without rounding to obtain the standardized measurement value. The parsed temporary timestamp is calculated field by field according to the UTC time conversion standard to generate a 15-digit decimal standardized timestamp. The converted unified device identifier, data type label, measurement value, and standardized timestamp are concatenated and encapsulated according to the fixed order specified in the unified data reassembly template. Each encapsulated result is an independent structured data unit.

[0027] The system continuously performs multi-source raw message acquisition, protocol-based frame-by-frame verification and parsing, temporary field extraction, unified template field mapping and conversion, and structured encapsulation operations. The independent structured data units generated in each round of encapsulation are continuously written into the time-series data stream storage partition according to the acquisition time sequence. All structured data units in the time-series data stream storage partition are completely connected to form a continuous and uninterrupted standardized data stream. Each data unit in the standardized data stream completely contains four fixed fields: unified device identifier, data type label, measurement value, and standardized timestamp. There are no structured data units with missing fields, disordered field order, inconsistent units, or inconsistent time scale benchmarks mixed in. The standardized data stream is the final product of the multi-source data fusion operation and is completely output to the subsequent device association mapping processing stage.

[0028] It should be noted that the aforementioned 5-second fixed acquisition interval is the data reading interval at which the system receives raw data packets of the distribution network operation from the network port communication link of each distribution automation terminal frame by frame. The raw sampling period of the distribution network equipment (such as the distribution automation terminal) refers to the time interval at which the equipment itself measures and samples the power grid operating status; these are different concepts. In this embodiment, the distribution automation terminal is equipped with a DL / T634.5104 protocol communication module, and its typical raw sampling period is 3 seconds. This raw sampling period is determined by the internal sampling clock of the terminal equipment and is independent of the 5-second fixed data reading interval on the system side. The specific value of the raw sampling period can be obtained by reading the timestamp difference between two adjacent data units of the same distribution network equipment in the standardized data stream.

[0029] When the system receives raw data packets of the distribution network operation from the network port communication link of each distribution automation terminal frame by frame at a fixed data reading interval of 5 seconds, for distribution network equipment with an original sampling period of 3 seconds, the equipment has completed one or more rounds of independent sampling between two adjacent system reading times. The system reading operation obtains the latest sampled data frame in the equipment's cache at that reading time. Sampled data frames that have been generated internally by the equipment but not retrieved at the system reading time are overwritten and managed by the terminal equipment according to the local first-in-first-out rule and are not used as input data for the timing alignment process on the system side. In this embodiment, the distribution network measurement sequence targeted by the timing alignment processing is the distribution network measurement data sequence actually obtained by the system side at a 5-second reading interval. The weighted interpolation and median convergence operations are both carried out based on this actual obtained sequence, and no retrospective recovery is performed on the historical sampled data that has been overwritten locally by the terminal equipment.

[0030] In the main distribution network, trace upwards along the parent node of the distribution network topology until the main power supply network equipment is identified, and record the power supply affiliation to form a main distribution equipment association mapping table; Using the timestamp of the main grid measurement sequence as the sole benchmark, weighted interpolation is performed on the positions of measurement values ​​corresponding to missing main grid reference times in the distribution network measurement sequence. Median convergence processing is also performed on cases where multiple distribution network measurement values ​​exist within the window corresponding to the main grid reference time, generating a time-aligned distribution network measurement sequence. When the number of distribution network measurement data entries within the window is 0, the representative value of the distribution network measurement corresponding to that main grid reference time is marked as null and does not participate in the subsequent generation of situation feature vectors. The time-aligned main and distribution network measurement data are bound according to the main and distribution equipment association mapping table to obtain the associated dataset.

[0031] Weighted interpolation is used for numerical estimation scenarios where the main grid reference time has no corresponding sampling time in the distribution network measurement sequence. Median convergence is used for denoising and aggregation scenarios where multiple distribution network sampling values ​​exist within the window corresponding to the main grid reference time. These two methods are suitable for different data loss and redundancy situations. The processed distribution network measurement values ​​are uniformly used as input data for subsequent situational feature calculations. When there is no preceding adjacent sampling time for the main grid reference time, the measurement value of the nearest subsequent adjacent sampling time is used as the estimate; when there is no subsequent adjacent sampling time for the main grid reference time, the measurement value of the nearest preceding adjacent sampling time is used as the estimate.

[0032] The structured splicing of the main and distribution network situation features yields a situation feature vector set, including: Based on the main network measurement data in the associated dataset, calculate the central trend measurement value, dispersion measurement value, and change amplitude measurement value of each measurement value sequence belonging to the same main network device identifier within a preset time window, and combine them to form a main network device status feature sub-vector; Based on the measurement data of distribution network equipment that have a power supply relationship with the main network equipment in the associated dataset, within the same preset time window, for each set of measurement value sequences belonging to the same distribution network equipment identifier, the central trend measure, dispersion measure and change amplitude measure of the measurement value sequence are calculated and combined to form a distribution network equipment status feature sub-vector. The status feature sub-vectors of the main grid equipment and the status feature sub-vectors of the distribution network equipment under the same power supply affiliation are spliced ​​together end to end according to the preset feature dimension order to form the main grid-distribution associated status feature vector. The main and distribution relationship situation feature vectors generated under all power supply affiliations are summarized to form a situation feature vector set.

[0033] The preset time window adopts an adaptive adjustment rule: the original sampling period of the distribution network equipment in milliseconds is divided by 2 and then rounded up to obtain the duration of the single-sided window.

[0034] Retrieve the complete static topology connection data table within the power distribution network topology storage partition. Each record in the data table stores a 17-bit globally unique device identifier for a single distribution network device, a 17-bit globally unique device identifier for the device's direct superior parent node, and an electrical connection validity status determination bit. A value of 1 for the electrical connection validity status determination bit indicates physical electrical connection and allows for tracing upwards; a value of 0 indicates a disconnected line and does not participate in the tracing operation. Extract the 17-bit globally unique device identifier built into all data units marked as distribution network devices within the standardized data stream as the unique search key for the distribution network device to be traced. Use this search key to search the distribution network topology connection data table to obtain the globally unique identifier of the corresponding parent node device and the electrical connection validity status determination bit. Only for parent node devices with a determination bit value of 1, continue searching upwards for matching. Repeat the loop operation of searching the globally unique identifier of the parent node and determining the valid connection until the retrieved globally unique identifier of the parent node matches the unified device identifier corresponding to any type of equipment code in the main network equipment lookup table (substation, bus, main transformer). Terminate the upward tracing process of the current distribution network device. When the parent node's electrical connection mark is 0, terminate the tracing and start the main network equipment lookup process. The distribution equipment association mapping table is written with topology anomaly markers. Distribution transformers with dual main transformers and dual power supply will generate two sets of independent power supply affiliation records. The two sets of affiliation relationships will generate independent main-distribution association situation feature vectors. During the bidirectional analysis phase, the two sets of vectors are analyzed separately in parallel and the analysis results are output independently. Throughout the process, the corresponding records of the globally unique identifiers of distribution network equipment and the corresponding globally unique identifiers of power supply main network equipment obtained in each round of retrieval are retained. A structured storage carrier for the main-distribution equipment association mapping table is pre-allocated. The carrier has three fixed storage fields: the unified identifier of the distribution network equipment, the unified identifier of the main network equipment, and the power supply affiliation. A valid power supply affiliation marker with a fixed value of 1 indicates that the electrical power supply affiliation is real and valid. After each group of traces is terminated, the globally unique identifiers of the distribution network equipment and the globally unique identifiers of the main power supply network equipment are completely filled into the corresponding rows and columns of the carrier and a power supply affiliation valid marker with a value of 1 is written. After all distribution network equipment completes the step-by-step upward tracing and record writing operations, all valid stored records inside the carrier are integrated to generate a complete main and distribution equipment association mapping table. The main and distribution equipment association mapping table, as the unique structured product of the equipment power supply affiliation, is completely stored in the memory index partition for use in the timing alignment process.

[0035] All data units marked as main network devices are extracted from the standardized data stream and grouped according to the same 17-bit globally unique device identifier. All structured data units corresponding to the same globally unique identifier of the main network device are independently divided into a group of main network device measurement sequences. Within a single group of main network device measurement sequences, the data units are rearranged in ascending order according to the 15-bit standardized timestamp value built into each data unit. The standardized timestamp carried by each data unit in each group of main network device measurement sequences is extracted and copied one by one into an independent one-dimensional array storage unit. The timestamp values ​​within the array are kept in ascending order without being disordered. After all the standardized timestamps corresponding to a single group of main network device measurement sequences are completely entered into the one-dimensional array storage unit, an independent main network time reference sequence is generated. All grouped main network device measurement sequences generate their own independent main network time reference sequences. All main network time reference sequences are uniformly stored in the timing benchmark storage partition as a globally unique timing comparison benchmark. The UTC time millisecond-level timestamp conversion rule of the main network global clock is completely consistent with the generation stage of the standardized data stream, and there are no differences in time zone or millisecond conversion benchmarks.

[0036] Using the unified identifier of the distribution network equipment in each record of the main distribution equipment association mapping table as the search key, all structured data units marked as distribution network equipment and matching the globally unique identifier in the standardized data stream are retrieved. All structured data units corresponding to the globally unique identifier of the same distribution network equipment are independently divided into a group of distribution network equipment measurement sequences. The measurement sequences of a single distribution network equipment are rearranged in ascending order according to the built-in standardized timestamp values. The unified identifier of the main network equipment with a power supply relationship with the distribution network equipment in the main distribution equipment association mapping table is read, and the globally unique identifier of the main network equipment is retrieved. The main network time reference sequence is a complete one-dimensional array. The standardized timestamp value of each data unit in the distribution network equipment measurement sequence is sequentially compared with each stored timestamp value in the one-dimensional array of the main network time reference sequence. This completes the recording of all main network reference times in the main network time reference sequence that do not form an equivalent match with any timestamp value in the distribution network equipment measurement sequence. Simultaneously, it records all measurement value entries in the distribution network equipment measurement sequence whose standardized timestamps fall within the corresponding interval for each reference time in the main network time reference sequence. This completes a one-to-one matching comparison operation between a set of distribution network equipment measurement sequences and all times in the corresponding main network time reference sequence. For any main network reference time in the main network time reference sequence that does not find an equivalent distribution network timestamp, all standardized timestamps in the current distribution network equipment measurement sequence are traversed in ascending order. The largest timestamp with a value less than the main network reference time is selected as the previous adjacent sampling time, and the smallest timestamp with a value greater than the main network reference time is selected as the next adjacent sampling time. The built-in measurement value of the data unit corresponding to the previous adjacent sampling time is read and recorded as the previous value, and the built-in measurement value of the data unit corresponding to the next adjacent sampling time is recorded as the built-in measurement value. The difference between the standard timestamp of the main network reference time and the previous adjacent sampling time is recorded as the second time distance. The difference between the standard timestamp of the next adjacent sampling time and the main network reference time is recorded as the third time distance. The first time distance and the second time distance are added together to obtain the total time interval. The first weighted component is obtained by multiplying the previous value and the second time distance. The second weighted component is obtained by multiplying the second value and the first time distance. The first weighted component and the second weighted component are added together and divided by the total time interval to obtain a unique value as the measurement estimate of the distribution network equipment at the main network reference time.

[0037] The generated measurement estimates are bound and stored with the main network reference time and the unified identifier of the distribution network equipment. This completes the weighted interpolation generation operation for a single missing measurement value. The entire weighted interpolation calculation process uses only time distance as the weight and does not introduce other external correction coefficients. The calculation benchmark is derived solely from the timestamps and measurement values ​​of the adjacent valid sampling times of the distribution network equipment itself. For any main network reference time within the main network time reference sequence, the original sampling period value of the distribution network equipment is read. This original sampling period value is determined by the timestamp difference between two adjacent data units of the distribution network equipment in the standardized data stream. The millisecond value of the original sampling period is divided by 2 and rounded up to obtain the single-sided window duration. All data units containing built-in measurement values ​​within the distribution network equipment measurement sequence whose standardized timestamp values ​​are greater than or equal to the main network reference time minus the single-sided window duration, and less than or equal to the main network reference time plus the single-sided window duration, are extracted. All the selected measurement values ​​are sorted in ascending order of magnitude. The total number of measurement value entries after sorting is counted. If the total number of entries is odd, the measurement value located at the position where the total number of entries plus one divided by two is selected as the median. When the total number of items is even, the median is obtained by adding the two measured values ​​(total number of items divided by two plus one) and (total number of items divided by two plus one). This median is used as the representative measurement value for the distribution network equipment at the corresponding main network reference time. The representative measurement value is then bound and stored with the corresponding main network reference time and a unified identifier for the distribution network equipment. This completes the median convergence processing operation for multiple sampled values ​​within a single window. After weighted interpolation to generate measurement estimates and median convergence to generate representative measurement values, the complete filled values ​​for each main network time reference sequence at the corresponding time are obtained. The network measurement entries are integrated with the corresponding network measurement entries at the main network reference time within the original main network equipment measurement sequence into a single time-aligned data unit. Each time-aligned data unit contains five fixed fields: unified identifier for main network equipment, unified identifier for distribution network equipment, standardized timestamp of unified main network reference time, main network measurement value, and distribution network measurement value. There are no time-aligned data units with missing distribution network measurement values ​​at the same main network reference time. After all main network and distribution network measurement sequences have completed time-series alignment processing, a complete set of time-aligned main network equipment measurement data and distribution network equipment measurement data is output.

[0038] Read each pair of equipment in the main and distribution equipment association mapping table where the power supply affiliation valid flag is 1. Use the unified identifier of the main network equipment and the unified identifier of the distribution network equipment in the record as the joint search key. After the time-aligned processing is completed, retrieve the time-aligned main network equipment measurement data and distribution network equipment measurement data set. Extract all main network measurement entries and distribution network measurement entries that match the joint search key. Write the main network measurement values ​​and distribution network measurement values ​​corresponding to the same main network reference time into the same binding record carrier. The carrier's built-in fields include the unified identifier of the main network equipment, the unified identifier of the distribution network equipment, the time tag (i.e., the standardized timestamp of the main network reference time), the main network measurement value, the distribution network measurement value, and the equipment association tag (i.e., the power supply affiliation pairing code). The equipment association tag is generated by combining the global unique identifier of the main network with a fixed separator and the global unique identifier of the distribution network to form a unique code. After all power supply affiliation pairings have completed the equipment-level association binding operation, all binding records are integrated to generate an association dataset with spatiotemporal association tags. The association dataset is the final product of the time-alignment and equipment association binding process and is completely output to the situation feature extraction process.

[0039] The duration of the complete intercept window in the situational characteristic calculation stage is directly derived from the duration of the single-sided window. The duration of the complete intercept window is equal to the duration of the single-sided window multiplied by 2. In this embodiment, the typical value of the original sampling period of the distribution automation terminal is 3 seconds. Dividing this original sampling period value by 2 and rounding up yields a single-sided window duration of 1500 milliseconds. The corresponding duration of the complete intercept window is fixed at 3000 milliseconds. The window sliding step size is assigned to 1500 milliseconds, which is equal to the duration of the single-sided window. There is a 50% overlap in the time intervals of two adjacent complete intercept windows. All binding records carrying the same unified identifier of the main network device are read from the associated dataset. The timestamp falling within the current 3000 milliseconds is intercepted based on the standardized timestamp of the built-in timestamp of each binding record. The system captures all mainnet measurement values ​​corresponding to the bound records within a millisecond window interval, forming an independent measurement value sequence belonging to the same mainnet device identifier. For this measurement value sequence, it performs separate calculations for central trend, dispersion, and amplitude of change. The central trend measure includes two types of indicators: arithmetic mean and median. The arithmetic mean is calculated by successively summing all measurement values ​​in the sequence, counting the total number of measurement value entries in the sequence, and dividing the sum by the total number of entries to obtain the arithmetic mean. The median is calculated by sorting all measurement values ​​in the sequence in ascending order. If the total number of entries is odd, the median is selected from the middle value after sorting; if the total number of entries is even, the median is selected from the middle value after sorting. The median is obtained by summing the two middle values ​​in the sorted sequence and dividing by two. The dispersion measure includes standard deviation and range. The standard deviation is calculated by subtracting each value in the sequence from the previously calculated arithmetic mean, then raising each difference to the squared difference. All squared differences are summed, and the variance is obtained by dividing the sum by the total number of measurement values. The standard deviation is then calculated by taking the square root of the variance. The range is calculated by extracting the maximum and minimum values ​​from the sequence, and subtracting the minimum from the maximum. The magnitude of variation is measured by the average of the absolute values ​​of the first-order differences. The process involves sequentially reading adjacent sets of measurement values ​​from the sequence of measurement values ​​belonging to the same main network device identifier, arranged in ascending time order. The measurement value at a later time step is subtracted from the measurement value at the previous time step to obtain a single set of first-order difference raw values. For each set of first-order difference raw values, the absolute value is taken to obtain the first-order difference absolute value. All first-order difference absolute values ​​within the window are successively summed to obtain the total absolute value. The total number of entries within the window that can generate first-order difference absolute values ​​is counted (i.e., the total number of measurement value entries minus one). The sum of absolute values ​​is divided by the total number of entries to obtain the average of the first-order difference absolute values. The arithmetic mean, median, standard deviation, range, and average of the first-order difference absolute values ​​calculated from the measurement value sequence of the same main network device are then arranged in a fixed order.Five sets of measurement values ​​are consecutively encapsulated to form a one-dimensional ordered array. This one-dimensional ordered array is defined as a sub-vector of the main network device status characteristics. Each set of main network device measurement value sequences uniquely corresponds to a set of main network device status characteristic sub-vectors. The order of the measurement values ​​within each sub-vector remains consistent and fixed throughout the process without any changes.

[0040] Read all binding records in the associated dataset that have a power supply affiliation with the aforementioned main network equipment and carry a unified identifier for a single distribution network equipment. Based on the standardized timestamp of the built-in time tag of each binding record, extract all distribution network measurement values ​​within a 3000-millisecond complete extraction window interval that is completely consistent with the main network feature calculation. This forms an independent measurement value sequence belonging to the same distribution network equipment identifier. Completely replicate the complete calculation process of the three types of measurement values ​​of the main network equipment, sequentially solve the arithmetic mean and median corresponding to the distribution network measurement value sequence to form the central trend measurement value, solve the standard deviation and range to form the dispersion measurement value, and solve the average of the absolute values ​​of the first-order differences to form the variation range measurement value. Encapsulate the five sets of distribution network measurement values ​​in a fixed order that is completely consistent with the main network sub-vector to form a one-dimensional ordered array. This one-dimensional ordered array is defined as the distribution network equipment status feature sub-vector. Under the same 3000-millisecond complete extraction window, the same distribution network equipment uniquely generates a set of distribution network equipment status feature sub-vectors. The measurement dimension order inside the distribution network sub-vector strictly corresponds to the main network sub-vector without any dimension misalignment.

[0041] The preset fixed feature dimension splicing order rule is that all dimensions of the main grid equipment status feature sub-vector under the same power supply affiliation are used as high-order feature components, and all dimensions of the distribution network equipment status feature sub-vector are used as low-order feature components. There is no dimension interleaving or dimension swapping splicing method. Read one set of main grid equipment status feature sub-vectors and one set of distribution network equipment status feature sub-vectors under the same power supply affiliation pair. First, copy all ordered measurement values ​​inside the main grid sub-vector completely. Then, copy all ordered measurement values ​​inside the distribution network sub-vector continuously after the end of all dimensions of the main grid. The two segments of ordered measurement values ​​are spliced ​​and integrated without gaps at the beginning and end into a single extended one-dimensional ordered array. This extended one-dimensional ordered array is defined as the main-distribution associated status feature vector. A single set of main-distribution power supply affiliation pair uniquely generates a set of main-distribution associated status feature vectors under a single 3000-millisecond complete truncation window. The first five dimensions of the vector are fixed to correspond to the three types of measurement values ​​of the main grid, and the last five dimensions are fixed to correspond to the three types of measurement values ​​of the distribution network. The dimension positions are bound without confusion.

[0042] Traverse all power supply affiliation pairs in the main and distribution equipment association mapping table where the valid power supply affiliation flag is equal to 1. For each pair, extract a complete 3000-millisecond capture window and independently generate the corresponding main and distribution affiliation situation feature vector. The duration of each window sliding forward is uniformly 1500 milliseconds. Generate the main and distribution affiliation situation feature vector for the current window only when the number of distribution network measurement data entries contained in the current 3000-millisecond complete capture window is greater than or equal to 3. If the number of distribution network measurement data entries is less than 3, no situation feature vector is generated for the window and the process skips to the next window. All independent main and distribution affiliation situation feature vectors generated by all power supply affiliation pairs and all valid 3000-millisecond complete capture windows are completely summarized and stored in a unified vector storage partition. All vectors maintain a one-dimensional ordered array storage format and the internal dimension arrangement order remains unchanged. The complete set of all ordered vectors after the summary is defined as the situation feature vector set. The situation feature vector set is the final product of the multi-dimensional situation feature structure splicing link and is completely output to the two-way collaborative analysis and processing link.

[0043] Step 2: Based on the operating characteristic reference range and deviation threshold set in the main distribution network, construct the first path of evaluating the impact of the distribution network downward through the main network and the second path of verifying the status of the main network upward through distribution network anomalies, forming a two-way collaborative judgment mechanism; In this embodiment of the invention, a two-way collaborative judgment mechanism is formed, including: Read the reference range of normal operation characteristics of the main grid, the reference range of normal operation characteristics of the distribution network, and the deviation threshold that were initially set in the main distribution network; If any metric value in the main network side feature component of the main network-related situation feature vector deviates from the main network normal operation feature reference range, it is determined that the main network device has an abnormal operating status. If any metric value in the distribution network side feature component of the main distribution network correlation status feature vector deviates from the normal operation feature reference range of the distribution network, it is determined that the distribution network equipment has an abnormal operating status. For main grid equipment that is determined to have abnormal operating status, all distribution network equipment that supplies power are traced down according to the power supply affiliation. By comparing the distribution network side characteristic components of the distribution network equipment with the normal operating characteristic reference range and deviation threshold of the distribution network, the impact range and risk level of the main grid fault on the distribution network are quantified. For distribution network equipment that is determined to have abnormal operating status, the power supply affiliation is traced back to the main grid equipment that supplies the power. The impact status of the main grid equipment is verified by comparing the main grid side characteristic components of the main grid equipment with the normal operating characteristic reference range of the main grid.

[0044] The quantitative analysis of the impact scope and risk level of main network failures on distribution networks includes: Based on the main and distribution equipment association mapping table, query all distribution network equipment powered by abnormal main network equipment; extract the change range metric value of each distribution network equipment, and evaluate the deviation range from the reference range; The number and topological distribution of devices whose deviation exceeds the deviation threshold are statistically analyzed to generate an impact range determination result. For each affected device, a weighted sum is assigned based on its type and preset weights, and the corresponding risk level is matched according to the total risk score obtained from the sum.

[0045] The verification of the impact status of the main network device includes: Based on the power supply affiliation recorded in the main distribution equipment association mapping table, query the main network equipment identifier that supplies power to the distribution network equipment from the association dataset; Extract the main network side feature components corresponding to the main network device from the situation feature vector set; If there is a measurement value in the main network side feature component that deviates from the reference range of the main network's normal operation characteristics, then an indication information of the main network device being affected is generated and the type and magnitude of the deviation measurement value are recorded. If there are no measurement values ​​in the main network side feature components that deviate from the reference range of normal operation of the main network, then an indication information indicating that the main network equipment is not affected is generated, and the indication information is used as the result of the retrospective impact judgment of the distribution network anomaly on the main network.

[0046] During the scheduling initialization phase, three types of fixed judgment parameters are written into the parameter fixed storage partition: the main network normal operation characteristic reference range, the distribution network normal operation characteristic reference range, and the globally unified deviation threshold. The main network normal operation characteristic reference range sets independent upper and lower limits for five types of metrics: arithmetic mean, median, standard deviation, range, and the average of the first-order difference absolute value. The main network arithmetic mean reference range is limited to 0.9 times to 1.1 times the rated load. The main network median reference range is completely consistent with the arithmetic mean range. The upper limit of the main network standard deviation reference range is set to 3% of the rated load value, and the lower limit is fixed at zero. The upper limit of the main network range reference range is set to 8% of the rated load value, and the lower limit is fixed at zero. The upper limit of the main network first-order difference absolute value reference range is set to 0.5% of the rated load value, and the lower limit is fixed at zero. The distribution network normal operation characteristic reference range simultaneously sets dedicated upper and lower limits for the five types of metrics. The reference range for the arithmetic mean and median of the distribution network is limited to 0.92 to 1.08 times the rated output of the distribution transformers. The upper limit of the standard deviation of the distribution network is 4% of the rated output of the distribution transformers, and the lower limit is zero. The upper limit of the range of the distribution network is 10% of the rated output of the distribution transformers, and the lower limit is zero. The upper limit of the average value of the absolute value of the first-order difference of the distribution network is 0.6% of the rated output of the distribution transformers, and the lower limit is zero. The globally unified deviation threshold is fixed at 15%. This deviation threshold is determined based on the statistical calibration of historical fault data of the target distribution network. By statistically analyzing the samples of the deviation amplitude of the distribution network equipment during the downward transmission of faults from the main grid in the past three years, the lower quartile of the deviation amplitude of the samples exceeding the standard is taken as the threshold benchmark. After being corrected by the experience of dispatching and operation experts, it is determined to be 15%. This deviation threshold is used to determine the boundary of the deviation amplitude after the change amplitude measurement value exceeds the reference range. After the three types of parameters are written, they are permanently stored in the parameter fixed storage partition for direct reading and calling throughout the entire process of bidirectional collaborative analysis. Among them, the rated output of the distribution transformer refers to the rated output of the distribution transformer.

[0047] The first five ordered values ​​of each independent master-slave related situational feature vector in the situational feature vector set are read to form the main network side feature component. Five sets of measurement values ​​are extracted sequentially from the main network side feature component: arithmetic mean, median, standard deviation, range, and the average of the absolute values ​​of the first-order differences. Each extracted measurement value is compared with the lower limit and upper limit of the corresponding dimension's main network normal operation feature reference interval in the parameter-fixed storage partition. If a single measurement value is greater than or equal to the lower limit of the corresponding dimension and less than or equal to the upper limit of the corresponding dimension, it is determined that the measurement value has not deviated from the interval. If a single measurement value is less than the lower limit of the corresponding dimension or greater than the upper limit of the corresponding dimension, it is determined that the measurement value has deviated from the interval. When any set of measurement values ​​in the main network side feature component corresponding to a master-slave related situational feature vector is determined to be deviating from the interval, an abnormal flag value of 1 is written into the corresponding status storage unit of the unified identifier of the main network device bound to the vector, thereby determining that the main network device has an abnormal operating status. When all measurement values ​​of the main network side feature component have not deviated from the corresponding interval, a flag value of 0 is written into the status storage unit.

[0048] The last five ordered values ​​of each independent main-distribution associated situational feature vector in the situational feature vector set are read to form the distribution network side feature component. The arithmetic mean, median, standard deviation, range, and average of the first-order difference absolute value are extracted sequentially from the distribution network side feature component. Each extracted metric value is compared with the lower limit and upper limit of the corresponding dimension's distribution network normal operation feature reference interval in the parameter fixed storage partition. If a single metric value is greater than or equal to the lower limit of the corresponding dimension and less than or equal to the upper limit of the corresponding dimension, it is determined that the metric value has not deviated from the interval. If a single metric value is less than the lower limit of the corresponding dimension or greater than the upper limit of the corresponding dimension, it is determined that the metric value has deviated from the interval. When any metric value in the distribution network side feature component corresponding to a single main-distribution associated situational feature vector is determined to be deviating from the interval, an abnormal flag value of 1 is written into the corresponding status storage unit of the distribution network equipment uniformly identified by the vector, thereby determining that the distribution network equipment has an abnormal operating status. When all metric values ​​of the distribution network side feature component have not deviated from the corresponding interval, a flag value of 0 is written into the status storage unit.

[0049] The system retrieves the unified identifiers of all main network devices with a flag value of 1 in the device status storage unit. Using these unified identifiers as the search key, it traverses all records in the main-distribution device association mapping table with a valid power supply affiliation flag of 1. It extracts the unified identifiers of distribution network devices stored in all records to form a set of distribution network identifiers to be verified. All distribution network devices in the set are all subordinate distribution network devices covered by the current abnormal main network device. The system retrieves the situation feature vector set and uses the unified identifiers of each group of distribution network devices in the set as the search index to match and extract the last five ordered values ​​of the corresponding main-distribution association situation feature vector as the exclusive distribution network side feature component of the distribution network device. All matched distribution network side feature components are completely retained for subsequent deviation amplitude evaluation. In the case of two sets of power supply affiliation records for the same distribution network device in the dual main transformer dual power supply scenario, the two records generate distribution network side feature components independently and participate in the subsequent deviation amplitude evaluation independently. The two sets of evaluation results are stored independently without merging or averaging.

[0050] The fifth ordered value of each distribution network side feature component, i.e., the change amplitude metric, is extracted separately. The upper limit reference value corresponding to the change amplitude dimension in the distribution network normal operation feature reference interval within the parameter-fixed storage partition is retrieved. The positional relationship between the change amplitude metric of the distribution network equipment and the value in the distribution network change amplitude normal operation feature reference interval [0, transformer rated output × 0.6%] is determined. The complete step-by-step numerical calculation process is as follows: First, read the upper limit value of the change amplitude dimension in the distribution network normal operation feature reference interval; second, determine whether the change amplitude metric of the distribution network equipment is greater than the upper limit of the interval. If the change amplitude metric is greater than the upper limit of the interval (transformer rated output × 0.6%), then subtract the upper limit of the interval from the current metric value to obtain the result. The absolute difference of positive exceedance is divided by the upper limit of the interval and then multiplied by 100% to obtain the percentage deviation range. If the change range measurement value is between the lower limit of the interval (0) and the upper limit of the interval (0.6% of the rated output of the distribution transformer), then no deviation range calculation is performed, and it is directly determined that the distribution network equipment is not affected by the main grid fault. For distribution network equipment with exceedance behavior and a calculated percentage deviation range, the calculated deviation range is compared with the globally unified 15% deviation threshold. If the deviation range value is greater than 15%, the distribution network equipment is determined to be affected by the main grid fault, and if the deviation range value is less than or equal to 15%, the distribution network equipment is determined to be not affected by the main grid fault.

[0051] Iterate through all distribution network devices in the set of distribution network identifiers to be verified, and perform the calculation of the change range deviation and threshold comparison operation for each one. Count the total number of unified identifiers of all distribution network devices with deviation values ​​greater than 15%. This total number is directly used as the number of affected distribution network devices. Retrieve the geographic grid code bound to the unified identifier of each distribution network device in the distribution network topology storage partition, and collect all the geographic grid codes corresponding to the distribution network devices identified as affected. Group and classify the geographic grid codes according to the first six digits of the administrative region, the middle four digits of the feeder line, and the last three digits of the distribution area. The grouping and classification results are directly used as the spatial distribution information of the affected distribution network devices in the topology. The collected number of affected distribution network devices and spatial distribution information are uniformly written into the structured record carrier to generate the result of the determination of the impact range of the main network fault on the distribution network.

[0052] For each distribution network device identified as affected, a weighted sum is calculated based on its device type and preset weights to obtain a comprehensive risk score. The weight values ​​are determined according to the scope of the device failure and the importance of the load: tie switches, as key nodes in the distribution network topology, will directly cause power outages on multiple downstream lines due to their failure, resulting in the widest impact, and are assigned a weight of 5 per unit; industrial distribution transformers supply higher load levels, resulting in greater power outage losses, and are assigned a weight of 3 per unit; residential distribution transformers supply relatively lower load levels, and are assigned a weight of 1 per unit. The risk level thresholds are set based on the total number of devices and the scale of the target distribution network topology: the total number of distribution transformers and tie switches in this distribution network is approximately 80. Based on the typical proportions of tie switches, industrial distribution transformers, and residential distribution transformers, the theoretical maximum risk score is approximately 160. 50% of the theoretical maximum value is taken as the first-level high-risk threshold, i.e., 80, and 12.5% ​​of the theoretical maximum value is taken as the second-level medium-risk threshold, i.e., 20. The remaining range is assigned a third-level low-risk threshold. The weight values ​​of all affected distribution network devices are summed one by one to obtain the comprehensive risk score. When the comprehensive risk score is greater than or equal to 80, it is matched with the first-level high-risk level; when the comprehensive risk score is greater than or equal to 20 and less than 80, it is matched with the second-level medium-risk level; when the comprehensive risk score is greater than or equal to 1 and less than 20, it is matched with the third-level low-risk level; when the comprehensive risk score is equal to 0, it is determined to be risk-free and does not match any risk level. The matched risk level code is written into the structured record carrier that stores the impact range determination results. The impact range determination content and the level code are integrated to generate the risk level determination result of the main grid fault on the distribution network. The complete determination data of the impact range and risk level of the main grid fault downward quantification diffusion are uniformly collected to form the complete output data of the downward analysis path.

[0053] The system retrieves the unified identifiers of all distribution network devices with a marked value of 1 in the device status storage unit. Using this unified identifier as the search key, it iterates through all records in the main distribution device association mapping table with a valid power supply affiliation marked as 1. The system extracts the unified identifiers of the main network devices stored in these records to form a main network identifier to be verified. Using this identifier as the search index, it retrieves the situation feature vector set and extracts the first five ordered values ​​from the corresponding main distribution associated situation feature vector as the main network-side feature component specific to this main network device. It then sequentially extracts five sets of measurement values ​​from the main network-side feature component: arithmetic mean, median, standard deviation, range, and the average of the first-order difference absolute value. Each extracted measurement value is compared with the lower and upper limits of the corresponding dimension's normal operation feature reference interval in the parameter-fixed storage partition. If a single measurement value is less than the lower limit or greater than the upper limit of the corresponding dimension, it is determined to be off-interval. Each set of measurement values ​​determined to be off-interval is recorded. The absolute difference between the corresponding dimension name, the current measurement value, and the corresponding interval boundary reference value is recorded. For measurement values ​​greater than the upper limit of the interval, the absolute difference between the measurement value and the upper limit reference value is recorded. For measurement values ​​less than the lower limit of the interval, the absolute difference between the measurement value and the lower limit reference value is recorded. All deviation dimension names and corresponding absolute differences are uniformly written into the structured storage carrier to form the indication information of the main network device being affected. When all five sets of measurement values ​​in the feature components of the main network side meet the condition of being greater than or equal to the lower limit value of the corresponding dimension and less than or equal to the upper limit value of the corresponding dimension, the fixed character "not affected" is written into the structured storage carrier as the indication information of the main network device not being affected. The complete structured carrier storing the main network device identifier, indication information, deviation dimension and magnitude is defined as the retrospective impact judgment result of the distribution network anomaly on the main network. A unique set of retrospective impact judgment results is generated for each abnormal distribution network device. After all abnormal distribution network devices have completed the upward retrospective verification, all retrospective impact judgment results are collected to form the complete output data of the upward analysis path.

[0054] The impact range and risk level determination results of the main network faults on the distribution network output from the downward analysis path, and the retrospective impact determination results of the distribution network anomalies on the main network output from the upward analysis path, are all collected and integrated. The complete determination data of the two types of paths are uniformly stored in the temporary partition of the bidirectional judgment results. For the case where the same distribution network equipment corresponds to two sets of power supply affiliation records that generate two sets of judgment results independently in the dual main transformer dual power supply scenario, the two sets of results are associated and stored according to the unified identifier of the same distribution network equipment during the result collection stage. If both sets of results are abnormal or both are normal, they are directly written into the final result dataset. If one set of results is abnormal and the other set of results is normal, the result is not written into the final result dataset. A dual-source inconsistency marker is written into the result record of the distribution network equipment, and the abnormal result is taken as the final judgment conclusion of the equipment. At the same time, two sets of original judgment data are retained for dispatchers to review and verify. Based on reading fixed reference intervals and deviation thresholds, main network equipment abnormal judgment logic, distribution network equipment abnormal judgment logic, main network fault downward quantification analysis logic, and distribution network abnormal upward backtracking verification logic, a bidirectional collaborative judgment analysis path with two types of analysis capabilities is fully constructed. The logical system that fully collects all judgment output data is uniformly defined as the bidirectional collaborative judgment mechanism. The bidirectional collaborative judgment mechanism is completely output as the product of this step to the final judgment dataset aggregation stage.

[0055] Step 3: Input the situation feature vector set into the first path and the second path in parallel, and collect the output judgment results according to the same power supply affiliation to obtain a collaborative judgment result dataset.

[0056] In this embodiment of the invention, the output judgment results are grouped together according to the same power supply affiliation, including: Each primary and secondary related situation feature vector is copied into two copies, which are then simultaneously input into the first path and the second path for parallel analysis. Obtain the impact range and risk level results of the first path output, as well as the retrospective impact results of the second path output; Using each power supply affiliation pairing record in the main and auxiliary equipment association mapping table as a collection node, all judgment results belonging to the same pairing are filled into a preset structured template to generate a collaborative judgment result dataset.

[0057] The system retrieves the complete one-dimensional ordered array data of each independent master-distribution related situation feature vector stored in the situation feature vector set. It then performs a complete copy operation on all dimensional measurement values ​​of a single master-distribution related situation feature vector to generate two identical vector copies. These two vector copies retain the same dimensional arrangement order, equipment power supply affiliation pairing and binding identifiers, and timing window marking information as the original master-distribution related situation feature vectors. Neither vector copy has undergone any dimensional deletion, value tampering, or identifier misalignment. The first vector copy is completely imported into the first path for evaluating the impact of the distribution network downwards from the main grid, performing a full-process analysis and judgment operation. Simultaneously, the second vector copy is completely imported into the second path for verifying the main grid status upwards from the distribution network anomaly, performing a full-process analysis and judgment operation. The analysis and judgment operations of the two vector copies are started synchronously, executed independently, and do not occupy computing resources, achieving a parallel analysis and judgment operation mode.

[0058] The system continuously reads the standardized structured data output after the first path's complete operation. This standardized structured data is defined as the impact range and risk level results. The impact range and risk level results internally store the power supply affiliation pairing code bound to the corresponding vector copy, the total number of affected distribution network devices, the geographic grid three-level grouping and distribution information of the affected distribution network devices, the comprehensive risk score calculated by weighted summation for a single affected device, and the matched first-level high-risk, second-level medium-risk, or third-level low-risk level code. Simultaneously, the system reads the standardized structured data output after the second path's complete operation. This standardized structured data is defined as the retrospective impact results. The retrospective impact results internally store the power supply affiliation pairing code bound to the corresponding vector copy, the unified identifier of the main power grid device, the indication information of whether the main grid device is affected or not, the measurement dimension name of the deviation from the reference interval, and the percentage magnitude value of the deviation interval. The impact range and risk level results output by the first path and the retrospective impact results output by the second path each independently carry the power supply affiliation pairing code that corresponds one-to-one with the original main distribution network associated situation feature vector. There is no judgment result data without a pairing code.

[0059] Retrieve the complete master-distribution equipment association mapping table stored in the memory index partition. Read each power supply affiliation pairing record in the master-distribution equipment association mapping table with a valid power supply affiliation flag value of 1. Concatenate the unified identifier of the distribution network equipment and the unified identifier of the master network equipment stored in a single power supply affiliation pairing record to generate a unique power supply affiliation pairing code as the unique retrieval index of the aggregation node. Retrieve all judgment result data in the first path output of all impact range and risk level results and the second path output of all retrospective impact results where the power supply affiliation pairing code is completely consistent with the current aggregation node retrieval index. Fill in the fields of all successfully matched impact range and risk level results and all retrospective impact results in the order of fixed fields in the preset structured template. The fixed fields in the preset structured template are, in order, the aggregation node power supply affiliation pairing code, the unified identifier of the master network equipment, the unified identifier of the distribution network equipment, the time sequence window timestamp, the complete dataset of impact range and risk level results, the complete dataset of retrospective impact results, and dual source. The power supply inconsistency flag bit, with a default value of 0 for dual-source power supply inconsistency flag bit, indicates that the abnormal judgment status of the two sets of judgment results under the same pair is consistent. If there are two independent power supply traceability records under the same power supply belonging pair and the abnormal judgment status of the two sets of records is inconsistent, the value is assigned to 1 and the complete content of the two sets of original judgment results is retained simultaneously. After all power supply belonging pair records with a valid power supply belonging flag value of 1 in the main and distribution equipment association mapping table have completed the judgment result filling and entry operation, the preset structured templates that have been filled in are integrated and summarized to generate a unified set. This unified set is defined as the collaborative judgment result dataset. The collaborative judgment result dataset, as the final product of the parallel judgment and result collection of situation feature vectors, is completely output to the dispatch situation display and risk decision support stage. The collaborative judgment result dataset fully carries two types of bidirectional judgment information: the quantification of the downward impact of the main grid and the upward backtracking of distribution network anomalies. It eliminates the defects of situation information loss caused by single unidirectional judgment and fully realizes the technical effect of global quantitative assessment of the collaborative situation of the main and distribution networks.

Claims

1. A method for processing data of power distribution network coordination situation awareness and intelligent research and judgment, characterized in that, The method includes: Step 1: Based on the multi-source data streams in the main distribution network, construct an associated dataset with the main network global clock as the reference, and structurally splice the status features of the main and distribution networks to obtain a status feature vector set; Step 2: Based on the operating characteristic reference range and deviation threshold set in the main distribution network, construct the first path of evaluating the impact of the distribution network downward through the main network and the second path of verifying the status of the main network upward through distribution network anomalies, forming a two-way collaborative judgment mechanism; Step 3: Input the situation feature vector set into the first path and the second path in parallel, and collect the output judgment results according to the same power supply affiliation to obtain the collaborative judgment result dataset; The construction of the associated dataset includes: In the main distribution network, trace upwards along the parent node of the distribution network topology until the main power supply network equipment is identified, and record the power supply affiliation to form a main distribution equipment association mapping table; Using the timestamp of the main network measurement sequence as the sole benchmark, weighted interpolation is performed on the positions of measurement values ​​corresponding to missing main network reference times in the distribution network measurement sequence. Median convergence is also performed on the cases where multiple distribution network measurement values ​​exist within the window corresponding to the main network reference time, generating a time-aligned distribution network measurement sequence. Bind the time-aligned main and distribution network measurement data according to the main and distribution equipment association mapping table to obtain the associated dataset; The structured splicing of the main and distribution network situation features yields a situation feature vector set, including: Based on the main network measurement data in the associated dataset, calculate the central trend measurement value, dispersion measurement value, and change amplitude measurement value of each measurement value sequence belonging to the same main network device identifier within a preset time window, and combine them to form a main network device status feature sub-vector; Based on the measurement data of distribution network equipment that have a power supply relationship with the main network equipment in the associated dataset, within the same preset time window, for each set of measurement value sequences belonging to the same distribution network equipment identifier, the central trend measure, dispersion measure and change amplitude measure of the measurement value sequence are calculated and combined to form a distribution network equipment status feature sub-vector. The status feature sub-vectors of the main grid equipment and the status feature sub-vectors of the distribution network equipment under the same power supply affiliation are spliced ​​together end to end according to the preset feature dimension order to form the main grid-distribution associated status feature vector. The main and distribution relationship situation feature vectors generated under all power supply affiliations are summarized to form a situation feature vector set.

2. The data processing method for coordinated situational awareness and intelligent analysis of a main and distribution network as described in claim 1, characterized in that, The multi-source data streams in the main distribution network include: Acquire main network model data and main network measurement data from the control cloud, distribution network model data and distribution network operation data from the distribution automation terminal, and load data and power outage data from the electricity consumption information collection terminal; Protocol parsing is performed on the main network model data, main network measurement data, distribution network model data, distribution network operation data, load data, and power outage data to obtain the equipment identifier, measurement type, measurement value, and timestamp information fields of the main and distribution networks. The equipment identification, measurement type, measurement value, and timestamp information fields are reorganized according to a preset unified data format to obtain a standardized data stream.

3. The data processing method for coordinated situational awareness and intelligent analysis of a main distribution network as described in claim 1, characterized in that, The preset time window adopts an adaptive adjustment rule: the original sampling period of the distribution network equipment in milliseconds is divided by 2 and then rounded up to obtain the duration of the single-sided window.

4. The data processing method for coordinated situational awareness and intelligent analysis of a main distribution network as described in claim 1, characterized in that, The central tendency measure includes at least the arithmetic mean and median of the measured numerical sequence, the dispersion measure includes at least the standard deviation and range of the measured numerical sequence, and the variation measure includes at least the average of the absolute values ​​of the first-order differences of the measured numerical sequence.

5. The data processing method for coordinated situational awareness and intelligent analysis of a main distribution network as described in claim 1, characterized in that, A two-way collaborative analysis mechanism has been established, including: Read the reference range of normal operation characteristics of the main grid, the reference range of normal operation characteristics of the distribution network, and the deviation threshold that were initially set in the main distribution network; If any metric value in the main network side feature component of the main network-related situation feature vector deviates from the main network normal operation feature reference range, it is determined that the main network device has an abnormal operating status. If any metric value in the distribution network side feature component of the main distribution network correlation status feature vector deviates from the normal operation feature reference range of the distribution network, it is determined that the distribution network equipment has an abnormal operating status. For main grid equipment that is determined to have abnormal operating status, all distribution network equipment that supplies power are traced down according to the power supply affiliation. By comparing the distribution network side characteristic components of the distribution network equipment with the normal operating characteristic reference range and deviation threshold of the distribution network, the impact range and risk level of the main grid fault on the distribution network are quantified. For distribution network equipment that is determined to have abnormal operating status, the power supply affiliation is traced back to the main grid equipment that supplies the power. The impact status of the main grid equipment is verified by comparing the main grid side characteristic components of the main grid equipment with the normal operating characteristic reference range of the main grid.

6. The data processing method for coordinated situational awareness and intelligent analysis of a main distribution network as described in claim 5, characterized in that, The quantitative analysis of the impact scope and risk level of main network failures on distribution networks includes: Based on the main and distribution equipment association mapping table, query all distribution network equipment powered by abnormal main network equipment; extract the change range metric value of each distribution network equipment, and evaluate the deviation range from the reference range; The number and topological distribution of devices whose deviation exceeds the deviation threshold are statistically analyzed to generate an impact range determination result. For each affected device, a weighted sum is assigned based on its type and preset weights. The corresponding risk level is then matched based on the total risk score obtained from the summation.

7. The data processing method for coordinated situational awareness and intelligent judgment of a main distribution network as described in claim 6, characterized in that, The verification of the impact status of the main network device includes: Based on the power supply affiliation recorded in the main distribution equipment association mapping table, query the main network equipment identifier that supplies power to the distribution network equipment from the association dataset; Extract the main network side feature components corresponding to the main network device from the situation feature vector set; If there is a measurement value in the main network side feature component that deviates from the reference range of the main network's normal operation characteristics, then an indication information of the main network device being affected is generated and the type and magnitude of the deviation measurement value are recorded. If there are no measurement values ​​in the main network side feature components that deviate from the reference range of normal operation of the main network, then an indication information indicating that the main network equipment is not affected is generated, and the indication information is used as the result of the retrospective impact judgment of the distribution network anomaly on the main network.

8. The data processing method for coordinated situational awareness and intelligent analysis of a main distribution network as described in claim 1, characterized in that, The output analysis results are grouped and associated according to the same power supply affiliation, including: Each primary and secondary related situation feature vector is copied into two copies, and simultaneously input into the first path and the second path for parallel analysis. Obtain the impact range and risk level results of the first path output, as well as the retrospective impact results of the second path output; Using each power supply affiliation pairing record in the main and auxiliary equipment association mapping table as a collection node, all judgment results belonging to the same pairing are filled into a preset structured template to generate a collaborative judgment result dataset.

Citation Information

Patent Citations

  • N-1 calculating method for combined calculation of main networks and distribution networks and automatic spare-power switching actions

    CN103679546A

  • Main and distribution topology through power supply path tracking method based on real-time power flow

    CN117767552A