Space-time association method and system for multi-source data of power distribution network

Abstract

The invention discloses a time-space association method and system for multi-source data of a power distribution network, belongs to the technical field of multi-source data processing of the power distribution network, and aims to avoid omission of real anomalies and misjudgment of normal fluctuations during data screening through a data dynamic management and control mechanism, so that acquired data are more real and reliable, and the accuracy of data screening is improved. Multi-source data can be accurately aligned to the same space-time dimension by associating the data with the space-time label, so that the relevance among the multi-source data is improved, a core support is provided for means such as multi-modal feature fusion during management and optimization of the power distribution network, meanwhile, the data is traceable through the composite index, the reliability of the data is ensured, and the method is suitable for popularization and application. The problem that the actual adaptability of a final governance scheme is poor due to the fact that space-time relevance of multi-source data is insufficient during governance of an existing power distribution network is solved.

Description

A method and system for spatiotemporal correlation of multi-source data in power distribution networks Technical Field

[0001] This invention relates to the field of multi-source data processing technology for power distribution networks, specifically to a method and system for spatiotemporal correlation of multi-source data in power distribution networks. Background Technology

[0002] With the continuous expansion of the power distribution network, the widespread access of distributed energy and energy storage devices, and the impact of extreme weather disasters, the operating environment of the power distribution network is becoming increasingly complex, leading to a growing difficulty in power distribution network management. Currently, power distribution network management and protection setting mainly rely on two types of technologies. One is to increase redundancy through hardware investment such as adding new substations and upgrading lines. However, this solution is costly, has a long construction period, cannot dynamically adapt to demand fluctuations and the uncertainty of distributed energy output, and does not involve data-level optimization. The second is to explore protection setting systems based on multi-source data. However, due to the strong heterogeneity and dispersion of multi-source data (operation, demand side, environment, historical data), insufficient spatiotemporal correlation and fusion, and limited accuracy in abnormal data removal and missing value repair, the reliability of feature extraction is low. As a result, the optimization schemes generated by traditional optimization governance models are difficult to implement and have poor practical adaptability. Therefore, a more reliable multi-source data processing solution is needed.

[0003] Chinese patent, publication number CN111680862A, publication date: September 8, 2020, discloses a risk early warning method for distribution network multi-source data fusion based on spatiotemporal grid association. This method integrates multi-source information from the distribution network, divides a selected area of ​​the distribution network into grids, encodes each grid unit, integrates multi-source information from the distribution network under freezing disasters, constructs a model of the relationship between meteorological loads and component failure probabilities, performs spatiotemporal correlation processing based on the divided spatiotemporal grids and distribution network information data, and filters out risk grids based on meteorological load data within the distribution network grids to establish a visualization system for freezing disaster risk early warning of the distribution network. Although it considers the impact of time and space on multi-source data analysis to some extent, it does not align the multi-source data with the spatiotemporal dimensions, thus still exhibiting spatiotemporal deviations when analyzing and issuing early warnings. Summary of the Invention

[0004] This invention addresses the problem of poor practical adaptability of existing distribution network governance solutions due to insufficient spatiotemporal correlation of multi-source data. It provides a method and system for spatiotemporal correlation of multi-source data in distribution networks. Through a dynamic data management mechanism, it ensures that real anomalies are not overlooked and normal fluctuations are not misjudged during data screening, making the collected data more authentic and reliable. By associating spatiotemporal tags with the data, multi-source data can be accurately aligned to the same spatiotemporal dimension, thereby improving the correlation between multi-source data. This provides core support for multimodal feature fusion and other methods in distribution network governance and optimization. Furthermore, the composite index ensures data traceability and reliability.

[0005] In a first aspect, one technical solution provided in this embodiment of the invention is: a spatiotemporal correlation method for multi-source data in a power distribution network, comprising the following steps: S1, collecting relevant core data during the operation of the power distribution network and preprocessing it; S2, filtering the relevant core data based on a dynamic data management and control mechanism to obtain target core data; S3, determining the corresponding spatiotemporal anchor points based on the spatiotemporal characteristics and business attributes of the target core data, and generating corresponding spatiotemporal tags based on the spatiotemporal anchor points; S4, organizing and sorting the spatiotemporal tags and storing them in the cloud, and constructing a corresponding spatiotemporal composite index mechanism to query the core data.

[0006] This solution utilizes data preprocessing combined with dynamic anomaly filtering to efficiently remove common anomalies such as format errors, preventing low-quality data from entering subsequent spatiotemporal correlation processes. Furthermore, by incorporating dynamic scenario thresholds, it adapts to the heterogeneous characteristics of multi-source data, ensuring that neither genuine anomalies are overlooked nor normal fluctuations are misjudged during data filtering, thereby improving data accuracy. Spatiotemporal anchors and labels address the issue of spatiotemporal misalignment in multi-source data, accurately aligning multi-source data to the same time dimension. This transforms multi-source data into a correlated dataset, providing core support for multimodal feature fusion and the construction of business scenario feature groups. Through spatiotemporal correlation of multi-source data, the solution improves the modeling accuracy of the demand-side energy supply system and the response speed of the protection setting model when optimizing the distribution network, enabling the effective implementation of performance optimization solutions.

[0007] As a preferred option, in S1, relevant core data on the operation of the distribution network are collected, including the following steps: deploying edge computing devices at distributed power generation grid connection points, energy storage access points, line sectionalizing switches, and areas with concentrated user loads in the distribution network; and collecting distribution network operation data, demand-side resource data, and environmental data as relevant core data.

[0008] In this solution, edge devices are deployed at key nodes such as distributed power grid connection points and energy storage access points to achieve local data acquisition and preprocessing, avoiding the long-distance transmission delay of centralized acquisition and meeting the millisecond-level data requirements of the distribution network. The deployment locations directly target the key scenarios in which distribution network data is generated, forming a complete data chain from grid status to resource processing and environmental impact, providing reliable basic data for subsequent spatiotemporal correlation, and also meeting application scenarios such as multimodal feature extraction of data after spatiotemporal correlation.

[0009] Preferably, in S2, the data dynamic management and control mechanism specifically includes the following steps: collecting historical operating data of the distribution network; the historical operating data includes normal operating data, abnormal operating data, and historical environmental data; comparing the normal operating data and abnormal operating data in the historical operating data to obtain the static abnormal threshold of the corresponding data; analyzing the historical environmental data in the historical operating data to construct the core operating scenarios of the distribution network, including extreme weather scenarios, equipment maintenance scenarios, load change scenarios, and general scenarios; determining the current operating scenario of the distribution network based on the relevant core data collected during the operation of the distribution network; and dynamically adjusting the static abnormal threshold based on the current operating scenario to obtain the dynamic abnormal threshold.

[0010] In this solution, thresholds are generated by comparing historical normal and abnormal data, which aligns with the actual operating patterns of the power distribution network and avoids deviations caused by subjective settings. It also focuses on core scenarios such as extreme weather and equipment maintenance, thus adapting to the different characteristics of data fluctuations and avoiding misjudgments and omissions caused by a single standard. By adjusting the thresholds in real time according to the current operating scenario, it tolerates reasonable fluctuations under extreme weather conditions while strictly controlling anomalies in general scenarios, thereby improving data accuracy. The high-quality output data can be directly connected to the determination of spatiotemporal anchor points and tag generation, reducing secondary processing costs and providing reliable data support for protection setting and vulnerability optimization.

[0011] As a preferred option, in S2, the relevant core data is filtered to obtain the target core data, including the following steps: determining the current operating scenario of the distribution network based on the relevant core data during the operation of the distribution network; adjusting the upper / lower limit of the static threshold of the corresponding data based on the current operating scenario of the distribution network to obtain the dynamic abnormal threshold of the corresponding data; classifying the relevant core data that exceeds the dynamic abnormal threshold as abnormal data, and performing secondary verification on the abnormal data. If the verification fails, the corresponding data is removed to obtain the target core data.

[0012] In this solution, by adjusting the thresholds in a scenario-based manner, the data evaluation is made more consistent with the actual operation of the power distribution network, avoiding the one-size-fits-all problem under a single static standard, and making the thresholds more in line with the fluctuation characteristics of the data. Through the isolation forest secondary verification, it is possible to effectively distinguish between real anomalies and false anomalies, thereby significantly reducing the probability of misjudgment and missed judgment, ensuring the reliability of the target core data, and providing high-quality data support for subsequent spatiotemporal correlation.

[0013] Preferably, in S3, determining the corresponding spatiotemporal anchor points based on the spatiotemporal characteristics and business attributes of the target core data includes the following steps: the spatiotemporal characteristics include the acquisition frequency and spatial range; when acquiring core data based on the acquisition frequency, the corresponding edge computing device records the acquisition time in real time to obtain the time anchor points of various core data; extracting the equipment ID and equipment coordinates corresponding to the distribution network operation data from the equipment asset ledger system and geographic information system built into the distribution network as the operation data spatial anchor points; extracting the corresponding demand resource ID from the demand-side resource management system built into the distribution network as the resource data spatial anchor points; dividing the distribution network into several sub-regions based on the power supply range, and using the sub-region IDs corresponding to different weather types in the environmental data as environmental data spatial anchor points; associating the time anchor points of various core data with the corresponding spatial anchor points to obtain the spatiotemporal anchor points of the target core data.

[0014] In this solution, firstly, edge devices record the collection time in real time according to the collection frequency to ensure a unified time base for data of different frequencies, laying the foundation for subsequent spatiotemporal alignment. Secondly, anchor points are extracted from a dedicated system according to data type, and runtime data is bound to device ID and coordinates, resource data is associated with resource ID, and environmental data is associated with partition ID, avoiding data without attribution or incorrect attribution. Through spatiotemporal anchor point association, the originally scattered heterogeneous data has a unified association benchmark, thereby breaking down the heterogeneous barriers of multi-source data and providing reliable data support for the subsequent generation of spatiotemporal labels.

[0015] As a preferred embodiment, in S3, generating corresponding spatiotemporal labels based on spatiotemporal anchor points includes the following steps: setting time labels using a date plus sampling time format; setting spatial labels using coordinates plus device ID / required resource ID / sub-region ID format; importing data from the time anchor points of various core data into the time labels, and importing various IDs and their coordinates from the spatial anchor points of various core data into the spatial labels; assigning a unique identifier to the time label and spatial label of each core data point, and associating the time labels and spatial labels of various core data points based on the unique identifier to obtain spatiotemporal labels.

[0016] In this solution, by standardizing the format of time and spatial labels, the confusion of label formats for different types of data is avoided, clearing the way for subsequent spatiotemporal association. The imported data directly reuses the spatiotemporal information of the spatiotemporal anchor points, avoiding secondary entry errors and ensuring that the labels correspond one-to-one with the core data. By using unique identifiers to strongly bind spatiotemporal labels to data, the time and spatial labels of each data item are firmly associated, ensuring that the data can be accurately traced to a specific time and physical location, and eliminating spatiotemporal misalignment.

[0017] As a preferred option, in S4, the spatiotemporal labels are organized, sorted, and then stored in the cloud, including the following steps: all spatiotemporal labels are spatiotemporally aligned, and sorted according to the order of the unique identifiers of the spatiotemporal labels to obtain a spatiotemporally aligned dataset, and the spatiotemporally aligned dataset is stored in the cloud.

[0018] In this solution, spatiotemporal alignment can eliminate the problem of heterogeneous misalignment of multi-source data. At the same time, sorting by unique identifiers makes the data logic clear and facilitates quick location of related data during subsequent queries. Cloud storage not only ensures the security and stability of massive spatiotemporal aligned datasets, but also supports cross-scenario sharing and access. It also lays a solid foundation for the construction of subsequent spatiotemporal composite indexes and accurate queries, thereby significantly improving data access efficiency and strongly supporting rapid decision-making in distribution network protection settings and vulnerability optimization.

[0019] As a preferred option, spatiotemporal dimension alignment is performed on all spatiotemporal labels, including the following steps: unifying the spatiotemporal labels based on a preset frequency, aligning all spatiotemporal labels and their corresponding core data to the standard timestamp of the preset frequency; associating various types of core data with the same spatial range based on the device ID / required resource ID / sub-region ID of the spatial labels in the spatiotemporal labels to obtain a spatial aggregated dataset; and mapping the spatial aggregated dataset to the standard timestamp to complete the spatiotemporal dimension alignment.

[0020] In this solution, data from different acquisition frequencies are aligned to a standard timestamp by unifying the frequency, thereby eliminating time deviations. At the same time, data from different IDs are associated with the same spatial range to form a spatially aggregated dataset, avoiding data ownership confusion. The final spatiotemporal alignment allows scattered and heterogeneous data to form a spatiotemporally unique and closely related dataset, which greatly improves data fusion efficiency and lays the foundation for subsequent sorting and storage, composite index construction, and fast querying. It completely solves the problem of spatiotemporal asynchrony of multi-source data and strongly supports accurate decision-making in distribution network protection settings and vulnerability optimization.

[0021] As a preferred embodiment, in S4, the spatiotemporal composite indexing mechanism specifically includes the following steps: using the timestamp in the spatiotemporal label as the primary index and various spatial IDs as secondary indexes to search for all core data at the corresponding spatial ID on the corresponding timestamp.

[0022] In this solution, the primary index is used to quickly locate the target time range, and the secondary index is used to accurately locate the corresponding spatial location. The dual filtering significantly narrows the query range, thereby avoiding full data traversal and significantly improving query response speed. At the same time, it can quickly aggregate all core data in the same time and space, meeting the efficient data access needs of scenarios such as distribution network fault tracing and operation analysis, providing rapid data support for protection setting and vulnerability optimization, and ensuring timely decision-making.

[0023] Secondly, one technical solution provided in this embodiment of the invention is: a multi-source data spatiotemporal correlation system for power distribution networks, comprising a data acquisition and processing module, a dynamic management and control module, a spatiotemporal tag generation module, and an organization module; the data acquisition module is used to collect relevant core data during the operation of the power distribution network and perform preprocessing; the dynamic management and control module, based on a data dynamic management and control mechanism, filters relevant core data to obtain target core data; the spatiotemporal tag generation module determines the corresponding spatiotemporal anchor point based on the spatiotemporal characteristics and business attributes of the target core data, generates corresponding spatiotemporal tags based on the spatiotemporal anchor points, and assigns values; the organization module organizes and sorts the assigned spatiotemporal tags and stores them in the cloud, and constructs a corresponding spatiotemporal composite index mechanism to query the core data.

[0024] In this solution, a corresponding system is built to integrate the spatiotemporal correlation method, thereby enabling human-computer interaction and improving the user experience.

[0025] The beneficial effects of this invention are as follows: Through a dynamic data management and control mechanism, this invention can ensure that real anomalies are not overlooked and normal fluctuations are not misjudged when filtering data, thus making the collected data more authentic and reliable. By associating spatiotemporal labels with data, multi-source data can be accurately aligned to the same spatiotemporal dimension, thereby improving the correlation between multi-source data and providing core support for multimodal feature fusion and other methods in power distribution network governance and optimization. At the same time, the composite index makes the data traceable, ensuring the reliability of the data.

[0026] The above description of the invention is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0027] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings. The drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings.

[0028] Figure 1 is a flowchart of a spatiotemporal correlation method for multi-source data in a power distribution network according to the present invention; Figure 2 is a block diagram of a spatiotemporal correlation system for multi-source data in a power distribution network according to the present invention. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only one preferred embodiment of this invention and are only used to explain this invention. They do not limit the scope of protection of this invention. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0030] Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations (or steps) can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but it may also have additional steps not included in the figures; the process may correspond to a method, function, procedure, subroutine, subroutine, etc.

[0031] Example 1: As shown in Figure 1, in order to solve the problem that the actual adaptability of the final governance scheme is poor due to the insufficient spatiotemporal correlation of multi-source data in the existing distribution network governance, this example provides a spatiotemporal correlation method for multi-source data of distribution network, including the following steps: S1: Collect relevant core data of distribution network operation and perform preprocessing.

[0032] In this embodiment, the collection of relevant core data during the operation of the distribution network includes the following steps: deploying edge computing devices at distributed power generation grid connection points, energy storage access points, line sectionalizing switches, and areas with concentrated user loads in the distribution network; and collecting distribution network operation data, demand-side resource data, and environmental data as relevant core data.

[0033] This embodiment deploys edge devices at key nodes such as distributed power grid connection points and energy storage access points to achieve local data acquisition and preprocessing, avoiding the long-distance transmission delay of centralized acquisition and meeting the millisecond-level data requirements of the distribution network. The deployment location directly targets the key scenarios in which distribution network data is generated, forming a complete data chain from grid status to resource processing and environmental impact, providing reliable basic data for subsequent spatiotemporal correlation, and also meeting application scenarios such as multimodal feature extraction of data after spatiotemporal correlation.

[0034] S2: Based on the dynamic data management and control mechanism, relevant core data is filtered to obtain the target core data.

[0035] In this embodiment, the data dynamic management and control mechanism specifically includes the following steps: collecting historical operating data of the distribution network; the historical operating data includes normal operating data, abnormal operating data, and historical environmental data; comparing the normal operating data and abnormal operating data in the historical operating data to obtain the static abnormal threshold of the corresponding data; analyzing the historical environmental data in the historical operating data to construct the core operating scenarios of the distribution network, including extreme weather scenarios, equipment maintenance scenarios, load change scenarios, and general scenarios; determining the current operating scenario of the distribution network based on the relevant core data collected during the operation of the distribution network; and dynamically adjusting the static abnormal threshold based on the current operating scenario to obtain the dynamic abnormal threshold.

[0036] Specifically, for example, the main monitoring indicator for distribution network operation data is line current, and its static anomaly threshold can be set to a current magnitude within 70%-130A of the rated current as normal data, and exceeding this range as abnormal data; the main monitoring indicator for demand-side resource data is photovoltaic output fluctuation, and its static anomaly threshold can be set to a fluctuation range within 20% / min as normal data, and exceeding this range as abnormal data; the main monitoring indicator for environmental data is solar irradiance, and its static anomaly threshold can be set to a solar irradiance range of 0-2000W / m². 2 Data within the specified range is considered normal; data outside this range is considered abnormal. Other general indicators include data format, and the static anomaly threshold is typically whether key fields such as timestamps and device IDs are missing. The monitoring indicators mentioned above are for reference only; in actual applications, multiple indicators can be selected for judgment based on specific needs.

[0037] In this embodiment, the relevant core data is filtered to obtain the target core data, including the following steps: determining the current operating scenario of the distribution network based on the relevant core data of the distribution network during operation; adjusting the upper / lower limit of the static threshold of the corresponding data based on the current operating scenario of the distribution network to obtain the dynamic anomaly threshold of the corresponding data; classifying the relevant core data that exceeds the dynamic anomaly threshold as abnormal data, performing secondary verification on the abnormal data based on the isolated forest algorithm, and removing the corresponding data if the verification fails to obtain the target core data.

[0038] Specifically, for example, in extreme weather scenarios, the trigger condition is receiving extreme weather warnings from meteorological departments. In this case, the abnormal fluctuation threshold for operational data is relaxed by 50%, and the fluctuation threshold for demand-measured resource data is relaxed by 50%, continuing until one hour after the warning information is dispelled, at which point it automatically reverts to the static threshold. In equipment maintenance scenarios, the trigger condition is receiving maintenance instructions such as line maintenance or testing from edge nodes. In this case, the judgment of data anomalies in the equipment under maintenance is disabled, and only the integrity of the data format is checked. The static threshold is then restored after the maintenance instructions are completed. In load surge scenarios, the trigger condition is a change in user load exceeding 30% within 5 minutes. In this case, the abnormal fluctuation threshold for load data is temporarily relaxed to 40%, continuing for 10 minutes, and then gradually restored to the static threshold by reducing the threshold by 5% every 2 minutes. The trigger conditions, threshold adjustment, and restoration mechanisms for the above scenarios include, but are not limited to, the methods described above, and can be set according to actual needs.

[0039] Abnormal data obtained after filtering based on dynamic thresholds can be subjected to secondary verification using methods including but not limited to the Isolation Forest algorithm.

[0040] This embodiment adjusts the thresholds based on specific scenarios to make the data assessment more consistent with the actual operation of the power distribution network, avoiding the one-size-fits-all problem under a single static standard and making the thresholds more in line with the fluctuation characteristics of the data. Through the isolation forest secondary verification, it can effectively distinguish between real anomalies and false anomalies, thereby significantly reducing the probability of misjudgment and missed judgment, ensuring the reliability of the target core data, and providing high-quality data support for subsequent spatiotemporal correlation.

[0041] S3: Determine the corresponding spatiotemporal anchor points based on the spatiotemporal characteristics and business attributes of the target core data, and generate corresponding spatiotemporal tags based on the spatiotemporal anchor points.

[0042] In this embodiment, determining the corresponding spatiotemporal anchor points based on the spatiotemporal characteristics and business attributes of the target core data includes the following steps: the spatiotemporal characteristics include the acquisition frequency and spatial range; when acquiring core data based on the acquisition frequency, the corresponding edge computing device records the acquisition time in real time to obtain the time anchor points of various core data; extracting the equipment ID and equipment coordinates corresponding to the power distribution network operation data from the equipment asset ledger system and geographic information system built into the power distribution network as the operation data spatial anchor points; extracting the corresponding demand resource ID from the demand-side resource management system built into the power distribution network as the resource data spatial anchor points; dividing the power distribution network into several sub-regions based on the power supply range, and using the sub-region IDs corresponding to different weather types in the environmental data as the environmental data spatial anchor points; associating the time anchor points of various core data with the corresponding spatial anchor points to obtain the spatiotemporal anchor points of the target core data.

[0043] Specifically, because distribution network operation data focuses on the real-time status of power grid equipment such as lines, switches, and transformers, its acquisition frequency is generally 50Hz, and its spatial range is generally concentrated on line nodes and switching equipment. Its core business data mainly includes voltage, current, and power flow. When determining its spatiotemporal anchor point, the clock is recorded synchronously while acquiring the corresponding data, using a timestamp in the format "YYYY-MM-DD HH:MM:SS.fff" as the time anchor point for that piece of operation data. For example, if a piece of distribution network operation data is: the current of line L102 collected at 08:30:00.123 on 2025-09-01 is 380A, then the time anchor point for this data is 2025-09-01. 08:30:00.123; The spatial anchor point is the unique ID of the device and its error packets. The unique ID of the device, such as the line number, switch number, and transformer number, is unique throughout the entire power grid and is associated with operation and maintenance records and maintenance plans. For example, the spatial anchor point of a certain distribution network operation data point mentioned above is L102, and its coordinates are expressed in latitude and longitude, which can be queried from the distribution network's built-in GIS geographic information system. The acquisition method for the spatiotemporal anchor points of other types of data is similar.

[0044] First, this embodiment records the acquisition time in real time by the edge device according to the acquisition frequency, ensuring a unified time base for data of different frequencies and laying the foundation for subsequent spatiotemporal alignment. Second, it extracts anchor points from a dedicated system according to data type, binds running data to device ID and coordinates, associates resource data with resource ID, and corresponds environmental data to partition ID, avoiding data without attribution or incorrect attribution. Through spatiotemporal anchor point association, the originally scattered heterogeneous data has a unified association benchmark, thereby breaking down the heterogeneous barriers of multi-source data and providing reliable data support for the subsequent generation of spatiotemporal labels.

[0045] In this embodiment, generating corresponding spatiotemporal tags based on spatiotemporal anchor points includes the following steps: setting time tags using a date plus sampling time format; setting spatial tags using coordinates plus device ID / required resource ID / sub-region ID format; importing data from the time anchor points of various core data into the time tags, and importing various IDs and their coordinates from the spatial anchor points of various core data into the spatial tags; assigning a unique identifier to the time tag and spatial tag of each core data piece, and associating the time tags and spatial tags of various core data pieces based on the unique identifier to obtain spatiotemporal tags.

[0046] This embodiment avoids the confusion of label formats for different types of data by unifying the formats of time tags and spatial tags, thus clearing the way for subsequent spatiotemporal association. The imported data directly reuses the spatiotemporal information of the spatiotemporal anchor points, avoiding secondary entry errors and ensuring that the tags correspond one-to-one with the core data. By using unique identifiers to strongly bind spatiotemporal tags to data, the time tags and spatial tags of each piece of data are firmly associated, ensuring that the data can be accurately traced to a specific time and physical location, and eliminating spatiotemporal misalignment.

[0047] S4: After organizing and sorting the spatiotemporal tags, store them in the cloud and build a corresponding spatiotemporal composite index mechanism to query the core data.

[0048] In this embodiment, the spatiotemporal labels are organized, sorted, and then stored in the cloud, including the following steps: all spatiotemporal labels are aligned in spatiotemporal dimensions, and sorted according to the order of the unique identifiers of the spatiotemporal labels to obtain a spatiotemporal aligned dataset, and the spatiotemporal aligned dataset is stored in the cloud.

[0049] This embodiment eliminates the problem of heterogeneous misalignment of multi-source data by aligning it in the spatiotemporal dimension. At the same time, sorting by unique identifiers makes the data logic clear and facilitates quick location of related data during subsequent queries. Cloud storage not only ensures the security and stability of massive spatiotemporally aligned datasets, but also supports cross-scenario sharing and access. It also lays a solid foundation for the construction of subsequent spatiotemporal composite indexes and accurate queries, thereby greatly improving data access efficiency and strongly supporting rapid decision-making in distribution network protection settings and vulnerability optimization.

[0050] In this embodiment, spatiotemporal dimension alignment of all spatiotemporal labels includes the following steps: unifying the spatiotemporal labels based on a preset frequency, aligning all spatiotemporal labels and their corresponding core data to a standard timestamp of the preset frequency; associating various types of core data with the same spatial range based on the device ID / required resource ID / sub-region ID of the spatial labels in the spatiotemporal labels to obtain a spatial aggregated dataset; and mapping the spatial aggregated dataset to a standard timestamp to complete the spatiotemporal dimension alignment.

[0051] Specifically, the main operation of frequency unification is to down-convert high-frequency data and up-convert low-frequency data. Taking the sampling frequency of distribution network operation data as 50Hz, demand-side resource data as 10Hz, and environmental data as 1Hz as an example, if the preset frequency is set to 10Hz, then the spatiotemporal labels corresponding to the distribution network operation data need to be down-converted. The specific process is as follows: group the data according to the timestamp, take every 5 50Hz data as a group, and take the average value within the group as the value of the 10Hz data. For example, the average of the 5 current values ​​from 08:30:00.000 to 08:30:00.099 is taken as the 10Hz data at 08:30:00.100.

[0052] The spatiotemporal labels corresponding to the environmental data are upsampled using linear interpolation. Based on two adjacent 1Hz data points, nine interpolation points are inserted in between to obtain 10Hz data. For example, the illumination at 08:30:00.000 is 1000W / m². 2 At 08:30:01.000, the illumination was 1050W / m². 2 Therefore, the interpolation value for 08:30:00.100 is 1005 W / m², and for 08:30:00.200 it is 1010 W / m². 2 And so on; after the operation is completed, align the spatiotemporal labels corresponding to all data to the standard 10Hz timestamp to ensure that there are operational, demand-side, and environmental data that can be associated under the same timestamp.

[0053] The specific process of spatial alignment is as follows: Data with the same device ID and demand resource ID in the geographic location labels are filtered out. For example, the current data of line L102 and the output data of photovoltaic PV001 on line L102 are associated to form the line L102 device dataset. Then, data with the same sub-region ID in the geographic location labels are filtered out. For example, the solar irradiance data of region R08, the output data of all photovoltaics in region R08, and the user load data of region R08 are associated to form the region R08 aggregate dataset. The region R08 aggregate dataset and the line L102 device dataset are used as a spatial aggregate dataset and mapped to a standard timestamp to finally obtain the spatiotemporal aligned dataset. The table below shows the spatiotemporal aligned dataset output once every 100ms: Table 1. Example of spatiotemporal aligned dataset Table Time Label Spatial Label Distribution Network Operation Data Demand Side Resource Data Environmental Data 2025-09-01 08:30:00.000 L102 / R08 380A 850kW 1000W / m 2 2025-09-01 08:30:00.100L102 / R08382A855kW1005W / m 2 The table shows the distribution network operation data (current data, demand-side data, photovoltaic output of photovoltaic power generation equipment with resource number PV001) and environmental data (sunlight intensity) in sub-region R08 of line L102 at 08:30:00.000 and 08:30:00.100 on September 1, 2025.

[0054] This embodiment aligns data from different acquisition frequencies to a standard timestamp by unifying the frequency, thereby eliminating time deviations. At the same time, it associates data within the same spatial range according to various IDs, forming a spatially aggregated dataset and avoiding data ownership confusion. The final spatiotemporal alignment allows scattered and heterogeneous data to form a spatiotemporally unique and closely related dataset, which greatly improves data fusion efficiency and lays the foundation for subsequent sorting and storage, composite index construction, and fast querying. It completely solves the problem of spatiotemporal asynchrony of multi-source data and strongly supports accurate decision-making in distribution network protection settings and vulnerability optimization.

[0055] In this embodiment, the spatiotemporal composite indexing mechanism specifically includes the following steps: using the timestamp in the spatiotemporal label as the primary index and various spatial IDs as secondary indexes to search for all core data at the corresponding spatial ID on the corresponding timestamp.

[0056] This embodiment uses a primary index to quickly locate the target time range and a secondary index to accurately pinpoint the corresponding spatial location. The dual filtering significantly narrows the query range, thereby avoiding full data traversal and significantly improving query response speed. At the same time, it can quickly aggregate all core data in the same time and space, meeting the efficient data access needs of scenarios such as distribution network fault tracing and operation analysis. It provides rapid data support for protection setting and vulnerability optimization, ensuring timely decision-making.

[0057] Example 2: As shown in Figure 2, this example also provides a spatiotemporal correlation system for multi-source data in a power distribution network, including a data acquisition and processing module, a dynamic management and control module, a spatiotemporal tag generation module, and an organization module. The data acquisition module is used to collect relevant core data during the operation of the power distribution network and perform preprocessing. The dynamic management and control module, based on a data dynamic management and control mechanism, filters relevant core data to obtain target core data. The spatiotemporal tag generation module determines the corresponding spatiotemporal anchor points based on the spatiotemporal characteristics and business attributes of the target core data, generates corresponding spatiotemporal tags based on the spatiotemporal anchor points, and assigns values. The organization module organizes and sorts the assigned spatiotemporal tags and stores them in the cloud, and constructs a corresponding spatiotemporal composite index mechanism to query the core data. By constructing a corresponding system to implement the spatiotemporal correlation method in this solution, human-computer interaction is realized, improving the user experience.

[0058] As can be seen from the above embodiments, it has at least the following substantial effects: (1) By combining data preprocessing with dynamic anomaly screening, the present invention can efficiently remove common abnormal data such as format errors, avoid low-quality data from entering the subsequent spatiotemporal correlation process, and further combine dynamic scene thresholds to adapt to the heterogeneous characteristics of multi-source data. When screening data, it will not miss real anomalies, nor misjudge normal fluctuations, thereby improving the accuracy of data; (2) The present invention solves the problem of spatiotemporal misalignment of multi-source data through spatiotemporal anchors and spatiotemporal labels, and can accurately align multi-source data to the same time dimension, thereby transforming multi-source data into a correlated dataset, providing core support for multimodal feature fusion and business scenario feature group construction; (3) The present invention improves the modeling accuracy of demand-side energy supply system and the response speed of protection setting model when optimizing the management of distribution network through spatiotemporal correlation of multi-source data, so that the performance optimization scheme can be effectively implemented.

[0059] The specific embodiments described above are preferred embodiments of the spatiotemporal correlation method and system for multi-source data in power distribution networks of the present invention, and are not intended to limit the specific scope of the present invention. The scope of the present invention includes but is not limited to the specific embodiments described above. All equivalent changes made in accordance with the shape and structure of the present invention are within the protection scope of the present invention.

Claims

1. A spatiotemporal correlation method for multi-source data in a power distribution network, characterized in that: Includes the following steps: S1. Collect and preprocess relevant core data during the operation of the power distribution network; S2. Based on the data dynamic management and control mechanism, relevant core data are filtered to obtain target core data; S3. Based on the spatiotemporal characteristics and business attributes of the target core data, the corresponding spatiotemporal anchor points are determined, and corresponding spatiotemporal tags are generated based on the spatiotemporal anchor points; S4. After organizing and sorting the spatiotemporal tags, store them in the cloud and build a corresponding spatiotemporal composite index mechanism to query the core data.

2. The spatiotemporal correlation method for multi-source data in a power distribution network according to claim 1, characterized in that: In S1, relevant core data for the operation of the distribution network are collected, including the following steps: deploying edge computing devices at distributed power generation grid connection points, energy storage access points, line sectionalizing switches, and areas with concentrated user loads in the distribution network; collecting distribution network operation data, demand-side resource data, and environmental data as relevant core data.

3. The spatiotemporal correlation method for multi-source data in a power distribution network according to claim 1, characterized in that: In S2, the data dynamic management and control mechanism specifically includes the following steps: collecting historical operating data of the distribution network; the historical operating data includes normal operating data, abnormal operating data, and historical environmental data; comparing the normal operating data and abnormal operating data in the historical operating data to obtain the static abnormal threshold of the corresponding data; analyzing the historical environmental data in the historical operating data to construct the core operating scenarios of the distribution network, including extreme weather scenarios, equipment maintenance scenarios, load change scenarios, and general scenarios; determining the current operating scenario of the distribution network based on the relevant core data collected during the operation of the distribution network; and dynamically adjusting the static abnormal threshold based on the current operating scenario to obtain the dynamic abnormal threshold.

4. The spatiotemporal correlation method for multi-source data in a power distribution network according to claim 3, characterized in that: In S2, the relevant core data is filtered to obtain the target core data, including the following steps: determining the current operating scenario of the distribution network based on the relevant core data during the operation of the distribution network; adjusting the upper / lower limit of the static threshold of the corresponding data based on the current operating scenario of the distribution network to obtain the dynamic abnormal threshold of the corresponding data; classifying the relevant core data that exceeds the dynamic abnormal threshold as abnormal data, and performing secondary verification on the abnormal data. If the verification fails, the corresponding data is removed to obtain the target core data.

5. The spatiotemporal correlation method for multi-source data in a distribution network according to claim 2, characterized in that: In S3, determining the corresponding spatiotemporal anchor points based on the spatiotemporal characteristics and business attributes of the target core data includes the following steps: the spatiotemporal characteristics include the acquisition frequency and spatial range; when acquiring core data based on the acquisition frequency, the corresponding edge computing device records the acquisition time in real time to obtain the time anchor points of various core data; extracting the equipment ID and equipment coordinates corresponding to the distribution network operation data from the equipment asset ledger system and geographic information system built into the distribution network as the operation data spatial anchor points; extracting the corresponding demand resource ID from the demand-side resource management system built into the distribution network as the resource data spatial anchor points; dividing the distribution network into several sub-regions based on the power supply range, and using the sub-region IDs corresponding to different weather types in the environmental data as environmental data spatial anchor points; associating the time anchor points of various core data with the corresponding spatial anchor points to obtain the spatiotemporal anchor points of the target core data.

6. The spatiotemporal correlation method for multi-source data in a power distribution network according to claim 5, characterized in that: In S3, generating corresponding spatiotemporal labels based on spatiotemporal anchor points includes the following steps: setting time labels using the format of date plus sampling time; setting spatial labels using the format of coordinates plus device ID / required resource ID / sub-region ID; importing data from the time anchor points of various core data into the time labels, and importing various IDs and their coordinates from the spatial anchor points of various core data into the spatial labels; assigning a unique identifier to the time label and spatial label of each core data, and associating the time labels and spatial labels of various core data based on the unique identifier to obtain spatiotemporal labels.

7. The spatiotemporal correlation method for multi-source data in a distribution network according to claim 6, characterized in that: In S4, the spatiotemporal labels are organized, sorted, and then stored in the cloud. This includes the following steps: aligning all spatiotemporal labels in terms of spatiotemporal dimensions, sorting them based on the order of their unique identifiers to obtain a spatiotemporal aligned dataset, and storing the spatiotemporal aligned dataset in the cloud.

8. The spatiotemporal correlation method for multi-source data in a distribution network according to claim 7, characterized in that: Spatiotemporal alignment of all spatiotemporal labels includes the following steps: unifying the spatiotemporal labels based on a preset frequency, aligning all spatiotemporal labels and their corresponding core data to a standard timestamp of the preset frequency; associating various types of core data with the same spatial range based on the device ID / required resource ID / sub-region ID of the spatial labels in the spatiotemporal labels to obtain a spatial aggregated dataset; and mapping the spatial aggregated dataset to a standard timestamp to complete the spatiotemporal alignment.

9. The spatiotemporal correlation method for multi-source data in a power distribution network according to claim 1, characterized in that: In S4, the spatiotemporal composite indexing mechanism specifically includes the following steps: using the timestamp in the spatiotemporal label as the primary index and various spatial IDs as secondary indexes to search for all core data at the corresponding spatial ID on the corresponding timestamp.

10. A spatiotemporal correlation system for multi-source data in a power distribution network, applicable to the spatiotemporal correlation method for multi-source data in a power distribution network as described in any one of claims 1-9, characterized in that: It includes a data acquisition and processing module, a dynamic control module, a spatiotemporal tag generation module, and an organization module; the data acquisition module is used to collect relevant core data during the operation of the power distribution network and perform preprocessing. The dynamic management and control module filters relevant core data based on a data dynamic management and control mechanism to obtain target core data; the spatiotemporal tag generation module determines the corresponding spatiotemporal anchor points based on the spatiotemporal characteristics and business attributes of the target core data, generates corresponding spatiotemporal tags based on the spatiotemporal anchor points and assigns values; the sorting module sorts and sorts the assigned spatiotemporal tags and stores them in the cloud, and builds a corresponding spatiotemporal composite index mechanism to query the core data.

Images