A real-time data service system and method based on network configuration service

Abstract

This invention discloses a real-time data service system and method based on distribution network services, relating to the field of distribution network service technology. The system consists of five layers: a data acquisition layer that collects real-time measurements, models, and graphics of distribution networks in various regions; an upper-level aggregation layer that stitches together the boundary topology of high- and low-voltage regional power grids to build a master model covering the entire domain and all voltage levels; a model management layer that stores device parameter topology and manages versions; a cloud platform layer that provides unified data, model, and computational support; and an application service layer that carries various distribution network service applications. The model management layer can dynamically trim the master model across the entire domain in three dimensions based on three types of parameters—geography, voltage, and scenario—based on the service request to generate a dedicated logical view. This invention effectively reduces memory and computational consumption, and master model updates only incrementally synchronize related views, improving resource utilization and system response speed under multi-service concurrency.

Description

Technical Field

[0001] This invention relates to the field of power distribution network service technology, and in particular to a real-time data service system and method based on power distribution network services. Background Technology

[0002] With the deepening construction of new power systems, massive and flexible resources such as distributed power sources, electric vehicles, and energy storage devices are being widely integrated, and the distribution network is gradually transforming from a passive unidirectional radial network to an active bidirectional interactive network. The characteristics of power grid operation are becoming increasingly complex, the number of dispatching objects is increasing significantly, and the need for collaborative management across voltage levels and regions is becoming increasingly urgent, placing higher demands on the model management capabilities of real-time data services for distribution network operations.

[0003] Currently, the field of power grid dispatching has widely deployed main and distribution network model splicing technology, which splices the models of the main grid and distribution network at the boundary nodes to construct a full-voltage-level power grid model, realizing unified modeling and sharing of the main and distribution networks in the digital space. However, the aforementioned full-voltage-level power grid model includes a complete set of equipment from UHV to low-voltage distribution networks. The model is large in scale and complex in structure, covering hundreds of thousands of equipment nodes and topological connections. In actual operation, different business applications (such as dispatch monitoring, fault analysis, and power transfer path search) only need to use the geographical areas, voltage levels, and topological subsets relevant to their own business scope in the complete model, rather than the full data. Because the existing system lacks a mechanism for dynamic tailoring and customized assembly of the full-voltage-level power grid model on demand, each business application can only load the complete full-voltage-level power grid model or a pre-fixed fixed model subset. This results in each application occupying a large amount of memory and computing resources to load redundant data, leading to low resource utilization. Furthermore, it is difficult to meet the differentiated and real-time requirements of different model views in multi-business concurrent scenarios. Summary of the Invention

[0004] To address the lack of mechanisms in existing technologies for dynamically tailoring and customizing the master model of a power grid across all voltage levels, which leads to redundant data loading, low resource utilization, and difficulty in meeting the differentiated real-time requirements of multi-service concurrent scenarios, this invention provides a real-time data service system and method based on distribution network services. This system significantly reduces memory usage and computational overhead, while ensuring that changes to the master model are only incrementally synchronized to related views, thus significantly improving resource utilization and system response efficiency in multi-service concurrent scenarios. The specific technical solution is as follows: This invention provides a real-time data service system based on power distribution network services, comprising: The data acquisition layer is deployed in each lower-level regional power grid to collect real-time measurement data, model files, and graphics of the local main and distribution networks. The upper-level aggregation layer is used to receive data uploaded by each lower-level regional power grid, and uses the model of the first regional power grid as the skeleton and the model of at least one second regional power grid as the branch to perform topology splicing at the boundary nodes to construct the main model of the power grid at all voltage levels. The model management layer is used to store the parameter attributes and topology connections of each device in the full voltage level power grid master model, and to perform master model version management; The cloud platform layer is used to provide unified data, model, and computing environment support for upper-layer applications. The application service layer, deployed on top of the cloud platform layer, is used to provide business application functions to power grid business personnel. The model management layer is configured to: receive a model request instruction from the application service layer, perform three-dimensional dynamic trimming on the full voltage level power grid master model according to the model request instruction, generate a logical view instance for application services, and provide the logical view instance to the application service layer for use.

[0005] Preferably, the model request instruction includes geographical range parameters, voltage level range parameters, and business analysis scenario parameters; The model management layer includes a three-dimensional pruning unit; the three-dimensional pruning unit is used to perform the following operations: A spatial index grid is constructed using the geographical range parameters, and the power grid equipment falling within the spatial index grid and its topological relationships are extracted. Based on the voltage level range parameters, all power grid equipment within the voltage level range is extracted, and N levels of coupling boundary equipment are traced upwards and downwards along the topological connection relationship. The coupling boundary equipment is extracted as model boundary nodes, where N is a preset positive integer. The business type is determined based on the business analysis scenario parameters, and a topological connectivity subgraph associated with the current business type is extracted.

[0006] Preferably, the model management layer further includes: The logical view management unit is used to assign a unique view identifier to each generated logical view instance and record the mapping relationship between the logical view instance and the source object in the full voltage level power grid master model; The change impact domain calculation unit is used to calculate the range of equipment affected by the model change when the main model of the full voltage level power grid changes, and to find an instance containing the range of equipment in the stored logical view instances, and send an incremental update instruction to the found logical view instance. The view incremental update unit is used to receive the incremental update instruction and synchronize the changed content to the corresponding logical view instance according to the mapping relationship; The view access frequency statistics unit is used to monitor the access frequency and recent access time of each logical view instance when it is called by the application service layer. The hierarchical cache management unit is used to divide each logical view instance into a hot data area, a warm data area, and a cold data area according to the access frequency and the recent access time; the hot data area stores logical view instances whose access frequency is higher than a first threshold or whose recent access time is within a preset duration; the cold data area stores logical view instances whose access frequency is lower than a second threshold and whose recent access time exceeds a preset duration. The preloaded prediction unit is used to build an application access behavior prediction model based on historical access sequences and predict the logical view instance that may be called in the next moment according to the access context of the current application service layer.

[0007] Preferably, the application service layer includes: The panoramic operation monitoring unit is used to acquire the logical view instance provided by the model management layer, and overlay real-time measurement data on the logical view instance to generate a visual monitoring interface. The intelligent analysis and decision-making unit is used to call the logical view instance provided by the model management layer corresponding to the current analysis scope when a fault signal or maintenance plan trigger instruction is received. Based on the equipment model, topology connection relationship and real-time measurement data in the logical view instance, it performs main and distribution network fault collaborative analysis, power supply path dynamic analysis and intelligent planning and optimization of load transfer path. The operation status prediction unit is used to call the logical view instance provided by the model management layer, use the topological connection relationship in the logical view instance as a constraint, extract historical operation section data to form a sample set, use a clustering algorithm to divide the sample set into scenarios, and establish a dynamic prediction model based on association rules to predict the future operation status trend. The risk management unit is used to call the logical view instance provided by the model management layer based on the prediction results output by the operation status prediction unit and combined with the current operation data, to overlay and display a comparison view of the current operation status and the predicted operation status on the logical view instance, to calculate the comprehensive risk score of each device according to the preset risk assessment rules, to identify the device whose comprehensive risk score exceeds the preset risk threshold as a weak link and to generate a risk report.

[0008] Preferably, in the operating state prediction unit, the clustering algorithm is a topological constraint weighted extended clustering algorithm, which performs the following operations: Using historical operational data of the distribution network as the sample set, each sample point includes the electrical quantity measurement vector at the corresponding time. and topology diagram Define any two sample points and The combined distance between for: in, The standard Euclidean distance for the measured data. Graph editing distance for topology, This is a dynamic adjustment coefficient that adaptively adjusts based on the degree of drastic change in the power grid topology during the current period. The optimal number of clusters was determined by using the weighted sum of the silhouette coefficient and the Davidson-Bourdin index as the clustering evaluation index. Based on the comprehensive distance and the optimal number of clusters, the historical operation data of the distribution network are initially clustered to form multiple typical operation scenario clusters. Based on association rule mining algorithms, association rules are established between measurement data, load data, network topology and operating status within each scenario cluster; When new real-time data flows in, the clustering affiliation is dynamically adjusted based on the comprehensive distance between the real-time data and each scene cluster and the matching degree of the association rules; the operating status trend at the next moment is predicted based on the historical evolution pattern of the cluster to which it belongs and the post-confidence of the association rules.

[0009] Preferably, the intelligent analysis and decision-making unit includes: The fault collaborative monitoring and processing module is used to establish a unified fault information model for the main and distribution networks. It maps the heterogeneous fault signals of the main network fault detection device and the distribution network fault indicator into a unified fault event object. Based on the fault device identifier in the fault event object, it locates the fault device in the logical view instance and analyzes the scope of impact, thereby realizing the panoramic fusion of fault information and the collaborative handling process. The power supply path dynamic analysis module is used to search all reachable load points from the specified power point and generate power supply paths based on the topological connection relationship in the logical view instance. It calculates the electrical distance and load margin of each power supply path, overlays and renders the calculation results on the visualization monitoring interface, and highlights the power supply paths whose load rate exceeds the preset threshold. The intelligent load transfer path planning module is used to automatically search for feasible load transfer and restoration paths within a range that meets preset constraints when a power grid failure or planned maintenance causes a power outage.

[0010] Preferably, the intelligent transfer path planning module includes: The path search submodule is used to abstract the power grid topology into a weighted directed graph based on graph theory, and to search for a set of feasible transfer paths within the range of preset constraints using a dynamic impedance weighted shortest path algorithm. In the weighted directed graph, the overall cost weight (Cost) is the sum of the costs of each edge. Defined as: In the formula, For the edge The per-unit impedance value, For the edge The current load rate, For the edge The historical number of operations of the corresponding switching equipment. For the edge The reliability coefficient of the corresponding equipment; , , , These are weighting coefficients that are dynamically adjusted based on the current operating scenario. The safety verification submodule is used to simulate the switch change operation involved in each feasible power transfer path in the power transfer path set and calculate the power flow distribution of the entire power grid after power transfer. The calculated voltage of each node, current of each branch and equipment load rate under N-1 condition are compared with the corresponding preset limits, and the feasible power transfer path that meets the limit requirements in all three items is marked as a safe path. The optimization decision submodule is used to perform multi-objective optimization evaluation of the safety path with network loss increment, number of switching operations, feeder load rate variance and power restoration time as optimization objectives, generate a comprehensive score and select the optimal power transfer scheme. The simulation and exercise submodule is used to construct a digital twin environment based on the device model and topology connection relationship in the logical view instance, load the optimal power transfer scheme in the digital twin environment and simulate the execution of the switching operation sequence, recalculate the power flow distribution and update the device status display after each switching change operation, and feed back the simulation execution results to the optimization decision submodule.

[0011] Preferably, the optimization decision submodule is specifically used for: Establish a multi-objective function ,in The goal is to minimize the increase in network loss before and after the transfer. The objective is to minimize the number of switching operations involved in the power transfer process. The objective is to minimize the variance of the load rate of each feeder after the transfer. The objective is to minimize the estimated power restoration time. The NSGA-II algorithm is used to solve the above multi-objective optimization problem to obtain the Pareto optimal solution set; The entropy-weighted TOPSIS method is used to make multi-attribute decisions on the Pareto optimal solution set and select the optimal transfer scheme. Preferably, the panoramic operation monitoring unit includes: The hierarchical monitoring module is used to construct a hierarchical monitoring view that includes power grid geographical wiring diagrams, plant wiring diagrams, and equipment bay details. In response to zoom or pan operations on any view, it synchronously refreshes the display range and display accuracy of other views. It stores view combinations and initial zoom ratios corresponding to different business positions through preset monitoring templates, and automatically loads the monitoring template corresponding to the user's position when the user logs in. The indicator display module is used to read measurement data from the real-time database at a preset period, calculate the running indicator value according to the pre-configured indicator calculation formula, compare the running indicator value with the preset alarm threshold range, render and display the comparison result on the monitoring view with the corresponding color or graphic icon, configure the corresponding drill-down path for each running indicator, and in response to the click operation of any running indicator, drill down along the drill-down path to display the detailed composition data of the running indicator. The alarm fusion module is used to perform real-time correlation, merging, suppression, and root cause analysis of multi-source heterogeneous alarms based on the alarm information model, forming a fault event chain and providing handling guidance.

[0012] This invention also provides a real-time data processing method based on power distribution network services, applicable to a layered architecture system comprising a data acquisition layer, an upper-level aggregation layer, a model management layer, a cloud platform layer, and an application service layer, comprising the following steps: The data acquisition layer collects real-time measurement data, model files, and graphics of the main and distribution networks of each lower-level regional power grid. The upper-level aggregation layer receives data uploaded by each lower-level regional power grid, and uses the model of the first regional power grid as the skeleton and the model of at least one second regional power grid as the branch, and performs topology splicing at the boundary nodes to construct the main model of the power grid at all voltage levels. The model management layer stores the parameter attributes and topology connections of each device in the full voltage level power grid master model and performs master model version management. The model management layer receives model request instructions from the application service layer, and performs three-dimensional dynamic trimming on the full voltage level power grid master model according to the model request instructions to generate a logical view instance for application services. The logical view instance is provided to the application service layer so that the application service layer can execute power grid business application functions based on the logical view instance; wherein the cloud platform layer provides computing environment support for the application service layer.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention discloses a real-time data processing system and method based on distribution network services. The system receives model request instructions from the application service layer through the model management layer. Based on the three-dimensional parameters of geographical range, voltage level and business scenario, the system dynamically prunes the main power grid model of all voltage levels and generates lightweight logical view instances for each business application as needed. This allows each application to load only a subset of equipment and topology relationships related to itself, significantly reducing memory usage and computational overhead. At the same time, each application's logical view maintains a mapping relationship with the main model. When the main model changes, only incremental synchronization to the associated views is required, without a full update. This significantly improves the system response efficiency and resource utilization in multi-service concurrent scenarios. Attached Figure Description

[0014] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0015] Figure 1 This is a schematic diagram of the overall interactive framework of a real-time data service system based on power distribution network services, provided for an embodiment of the present invention.

[0016] Figure 2 This invention provides a schematic diagram of the composition and interaction principle of a real-time data service system based on power distribution network services, as provided in an embodiment of the invention.

[0017] Figure 3 A flowchart illustrating the model management layer operation mechanism provided in this embodiment of the invention.

[0018] Figure 4 This is a schematic diagram illustrating the composition of the application service layer business function modules provided in this embodiment of the invention.

[0019] Figure 5 This is a flowchart of a real-time data processing method based on power distribution network services, provided as an embodiment of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] It should be understood that, when used in this specification, the terms “comprising” and “including” indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0022] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0023] It should also be further understood that the term "and / or" as used in this specification refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes such combinations.

[0024] Please refer to the following examples. Figures 1 to 5 .

[0025] Please see Figure 1 and Figure 2 This invention provides a real-time data service system based on power distribution network services, including a data acquisition layer, an upper-level aggregation layer, a model management layer, a cloud platform layer, and an application service layer. The data acquisition layer is deployed in various lower-level regional power grids, such as municipal power grids. It connects to various automation systems in the local main and distribution power grids via protocol adapters, including SCADA (Supervisory Control And Data Acquisition), distribution automation systems, and electricity consumption information acquisition systems. The data acquisition layer acquires real-time measurement data, equipment model files, and power grid graphical files from each automation system at preset acquisition intervals, such as 1 to 5 seconds. The real-time measurement data includes electrical measurements such as voltage, current, power, and frequency; the equipment model files contain equipment parameters, rated values, and information about the respective substations; and the power grid graphical files include geographical wiring diagrams, substation wiring diagrams, and single-line diagrams. After acquisition, the data acquisition layer performs local preprocessing on the acquired data.

[0026] In this embodiment, preprocessing includes: removing invalid values ​​from real-time measurement data and marking values ​​exceeding the reasonable measurement range as invalid; standardizing the data format by converting data from different sources into a unified data structure; repackaging model files and graphic files according to a preset standard data format, and adding a collection timestamp and source identifier to each data entry. After preprocessing, the data acquisition layer uploads the packaged data to the upper-level aggregation layer via a dedicated power grid or fiber optic communication link. The upper-level aggregation layer is communicatively connected to the data acquisition layer. After receiving data uploaded by each lower-level regional power grid, it uses the model of the first regional power grid as the skeleton and the model of at least one second regional power grid as branches, and performs topology splicing at the boundary nodes to construct the main model of the power grid at all voltage levels.

[0027] In this embodiment, the first regional power grid is the regional power grid with the highest voltage level or the largest power supply range, such as a 500kV provincial backbone power grid. The second regional power grid is other regional power grids that are topologically connected to the first regional power grid, such as 220kV or 110kV distribution networks in various cities. Boundary nodes are the endpoints of the interconnecting transformers or interconnecting lines between the first and second regional power grids. Specifically, during the interconnection process, the upper-level aggregation layer identifies the voltage levels and topological connections of the equipment in each regional power grid model, locates the positions of each boundary node in the first regional power grid model, and connects the equipment corresponding to the boundary nodes in the second regional power grid model as branches, forming a tree-like full-voltage-level power grid master model with the first regional power grid as the root node and the second regional power grid as the branch nodes. The full-voltage-level power grid master model includes all voltage-level power grid equipment and complete topological connections from ultra-high voltage 500kV to low voltage 380V.

[0028] The model management layer, connected to the upper-level aggregation layer, receives the full-voltage-level power grid master model constructed by the upper-level aggregation layer and stores it in a relational database. The parameter attributes of each device are stored in a device table structure, and the topology connections are stored in a connection relationship table between devices. The model management layer also performs master model version management: each time the model is changed, a new version number is generated, and the change content, change time, and personnel involved are recorded. The cloud platform layer, connected to the model management layer, is used to provide unified data, model, and computing environment support for upper-layer applications. The application service layer, deployed on top of the cloud platform layer, is used to provide business application functions to power grid business personnel. The model management layer is configured to: receive a model request instruction from the application service layer, the model request instruction carrying geographical range parameters, voltage level range parameters, and business analysis scenario parameters required by the application; and, based on the model request instruction, dynamically trim the full voltage level power grid master model in three dimensions—spatial, voltage, and business—to generate a logical view instance for application services, and provide the logical view instance to the application service layer for use.

[0029] Specifically, the model management layer includes a three-dimensional pruning unit; this unit receives model request instructions from the application service layer. The model request instructions carry the geographical range parameters, voltage level range parameters, and business analysis scenario parameters required by the application. Based on these three-dimensional parameters, the three-dimensional pruning unit performs three-dimensional dynamic pruning on the full-voltage-level power grid master model, generating a logical view instance for the application service. The three-dimensional dynamic pruning includes spatial dimension pruning, voltage dimension pruning, and business dimension pruning, and the pruning order of the three dimensions can be adjusted. The following explanation uses the order of spatial dimension pruning first, then voltage dimension pruning, and finally business dimension pruning as an example.

[0030] A spatial index grid is constructed using the geographic range parameters. Power grid equipment falling within the spatial index grid and their topological relationships are extracted. The spatial index grid is established using Geohash encoding. In practical implementation, the three-dimensional pruning unit is based on the geographical range parameter in the model request instruction in the spatial dimension. The geographical range parameter can be a range of latitude and longitude coordinates, such as a certain degree east longitude or a certain degree north latitude, or a regional administrative code, such as the administrative division code of a prefecture-level city. The three-dimensional pruning unit constructs a spatial index grid based on this geographical range parameter, specifically using Geohash encoding. The three-dimensional pruning unit converts the geographical range parameter into the corresponding Geohash string level and selects a Geohash encoding length that matches the required geographical range precision; the longer the encoding length, the finer the grid division. The three-dimensional pruning unit uses the spatial grid within this encoding range as the spatial pruning boundary, extracting all power grid equipment falling within this spatial index grid and their topological relationships in the full-voltage level power grid master model. For equipment that crosses grid boundaries, such as transmission lines spanning two grids, the three-dimensional pruning unit retains the entire equipment and also extracts the associated equipment in the grid at its opposite end.

[0031] Based on the voltage level range parameters, all power grid equipment within the voltage level range is extracted, and N levels of coupling boundary equipment are traced upwards and downwards along the topological connection relationship. The coupling boundary equipment is extracted as model boundary nodes, where N is a preset positive integer. In specific implementation, the three-dimensional pruning unit is based on the voltage level range parameter in the model request instruction in the voltage dimension. The voltage level range parameter can specify a single voltage level, such as 220kV, or specify a voltage level range, such as 110kV to 500kV. The three-dimensional pruning unit extracts all grid devices that match the voltage level range parameter from the device set obtained after spatial pruning. At the same time, in order to maintain the topological integrity of the pruned logical view and enable boundary devices to still perform electrical connection analysis with the external power grid, the three-dimensional pruning unit also traces N levels of coupled boundary devices upwards and downwards along the topological connection relationship. In this embodiment, N is a preset positive integer, which is set by the system operation and maintenance personnel according to actual needs. For example, N=1 indicates tracing the devices directly connected to the boundary devices at the upstream and downstream levels, and N=2 indicates tracing two levels. The specific tracing method is as follows: For devices located at the pruning boundary after voltage dimension pruning, i.e., devices that have a direct connection relationship with unextracted high-voltage or low-voltage level devices, the three-dimensional pruning unit traces upwards along the topological connection relationship to the higher voltage level side N-level devices and downwards to the lower voltage level side N-level devices, extracting all traced coupling boundary devices. Based on the device connection relationship table in the full voltage level power grid master model, the three-dimensional pruning unit starts from the boundary devices and queries their power supply side parent nodes and power receiving side child nodes level by level, adding all devices along the query path to the extraction set.

[0032] The business type is determined based on the business analysis scenario parameters, and a topological connectivity subgraph associated with the current business type is extracted. The business type includes scheduling monitoring, transfer path search, state estimation calculation, fault analysis, or risk assessment.

[0033] In practical implementation, at the business dimension, the three-dimensional pruning unit is based on the business analysis scenario parameters in the model request instruction. These parameters are generated by the application service layer according to the specific business application type, and may include types such as scheduling monitoring, transfer path search, state estimation calculation, fault analysis, and risk assessment. The three-dimensional pruning unit determines the business type based on these parameters and extracts the topological connectivity subgraph associated with the current business type.

[0034] If the business type is dispatch monitoring, the three-dimensional pruning unit extracts monitoring equipment such as substations, distribution lines, distribution transformers, circuit breakers and disconnect switches with voltage levels of 35kV and above from the pruned equipment set, and extracts all topological connected subgraphs within this range according to the power supply relationship, so that any device can be traced back to its power supply point.

[0035] If the business type is a power transfer path search, the three-dimensional pruning unit extracts power transfer-related equipment such as power supply points, feeders, tie switches, and load switches from the pruned equipment set, and extracts the interconnections between feeders using feeders as the basic unit, forming a searchable ring network topology subgraph. Power supply points include substation buses.

[0036] An embodiment of the present invention provides a real-time data service system based on distribution network services. By receiving model request instructions from the application service layer through the model management layer, the system dynamically prunes the main power grid model for all voltage levels based on three-dimensional parameters such as geographical range, voltage level, and business scenario. It generates lightweight logical view instances for each business application as needed, so that each application only loads a subset of equipment and topology relationships related to itself, significantly reducing memory usage and computational overhead. At the same time, each application's logical view maintains a mapping relationship with the main model. When the main model changes, only incremental synchronization to the associated views is required, without a full update, which significantly improves the system response efficiency and resource utilization in multi-service concurrent scenarios. Meanwhile, by using Geohash encoding to construct a spatial index grid for geographical range extraction in the spatial dimension, tracing N-level coupled boundary devices along topological connections in the voltage dimension, and extracting topological connectivity subgraphs associated with business types in the business dimension, the topological integrity and electrical connectivity of the pruned logical view are ensured. The pruned logical view can still independently perform complete power grid analysis functions such as power flow calculation and fault analysis. Different business applications can obtain their own customized logical view instances based on the same full-voltage-level power grid master model. The logical view instances are isolated from each other and do not affect each other, meeting the differentiated needs of model views in multi-business concurrent scenarios and avoiding the resource contention and mutual interference problems when multiple applications share the same model copy in traditional solutions.

[0037] Please see Figure 3 In a preferred embodiment of the present invention, the model management layer further includes a logical view management unit, a change impact domain calculation unit, a view incremental update unit, a view access frequency statistics unit, a hierarchical cache management unit, and a preloading prediction unit. The logical view management unit is used to manage each logical view instance generated by the three-dimensional pruning unit throughout its entire lifecycle. Specifically, each time the three-dimensional pruning unit completes a dynamic pruning and generates a logical view instance, the logical view management unit assigns a unique view identifier to that instance. This identifier can use a globally unique identifier format to ensure that it is not repeated throughout the entire system. Simultaneously, the logical view management unit records the mapping relationship between the logical view instance and the source objects in the main model of the full voltage level power grid. This mapping relationship is stored in the form of a mapping table. Each row of the mapping table records a reference pointer between a device object in the logical view instance and the corresponding source object in the main model, ensuring that any device in the logical view can be traced back to its original device in the main model. This mapping table is stored separately from the model data of the logical view instance, providing a lookup basis for subsequent incremental updates.

[0038] The Change Impact Domain Calculation Unit monitors change events in the main model of the power grid across all voltage levels. When maintenance personnel modify equipment parameters, add or delete equipment, or adjust topology connections in the main model, the main model's version management mechanism triggers a model change event. After capturing this event, the Change Impact Domain Calculation Unit obtains the set of equipment objects involved in the change and calculates the scope of equipment affected by the change based on the topology connections in the main model. The calculation method is as follows: Using the changed equipment as the center, a search is conducted outwards along the topology connections for a preset order, including all equipment falling within the search range. For the changed equipment itself, the impact domain only includes the equipment itself. After determining the scope of impact, the Change Impact Domain Calculation Unit traverses all logical view instances stored in the logical view management unit, checking each logical view instance's mapping table to see if it contains any equipment within the scope of impact. If it does, the logical view instance is determined to be affected by the current main model change, and an incremental update command is sent to the found logical view instance. The incremental update command includes the change type, the identifier of the source object involved in the change, the changed value, and the change timestamp.

[0039] The incremental update unit receives incremental update commands from the change impact domain calculation unit and synchronizes the changed content to the corresponding logical view instance according to the mapping relationship recorded by the logical view management unit. Specifically, the incremental update unit parses the incremental update command, extracts the change type and source object identifier, and finds the target object corresponding to the source object identifier in the logical view instance according to the mapping relationship table. If the change type is parameter modification, the incremental update unit writes the changed value into the parameter field of the corresponding target object in the logical view instance; if the change type is device addition, the incremental update unit adds the corresponding device object and its topology connection relationship in the logical view instance; if the change type is device deletion, the incremental update unit removes the corresponding device object and its association relationship from the logical view instance. After the incremental update is completed, the incremental update unit reports the update completion status to the logical view management unit, and the logical view management unit updates the version identifier of the logical view instance accordingly. Therefore, each logical view instance does not need to be fully regenerated and trimmed when the main model changes; only incremental synchronization of the changed content is required, significantly reducing model maintenance overhead.

[0040] The view access frequency statistics unit is used to monitor in real time the access frequency and most recent access time of each logical view instance when called by the application service layer. Specifically, whenever the application service layer initiates a call request to a logical view instance, the view access frequency statistics unit records the timestamp of that call. The view access frequency statistics unit uses a preset statistical period, such as every 5 minutes as a statistical window, to count the cumulative number of calls to each logical view instance within the current window as the access frequency, and records the time of the most recent call to each instance as the most recent access time. The statistical results are updated to the hierarchical cache management unit in real time.

[0041] The tiered cache management unit is used to divide logical view instances into hot data zones, warm data zones, and cold data zones for hierarchical storage based on the access frequency and recent access time provided by the view access frequency statistics unit. The division rules are as follows: logical view instances with an access frequency higher than a first threshold or a recent access time within a preset duration are assigned to the hot data zone (e.g., instances accessed more than 10 times per minute or accessed within the last 5 minutes); logical view instances with an access frequency lower than a second threshold and a recent access time exceeding a preset duration are assigned to the cold data zone (e.g., instances accessed less than once per hour and not accessed within the last hour); the remaining instances are assigned to the warm data zone. Logical view instances in the hot data zone reside in memory to ensure fast access for frequently accessed instances; logical view instances in the cold data zone are stored on disk to free up memory resources; logical view instances in the warm data zone are scheduled between memory and disk on demand. The tiered cache management unit monitors the access statistics changes of each instance in real time, and automatically migrates the instance between memory and disk when the access frequency or recent access time of an instance crosses the partition threshold. For example, when an instance in the cold data area is accessed, the instance's last access time is updated. If the updated time meets the conditions for the hot data area, the tiered cache management unit loads the instance from the disk into memory and assigns it to the hot data area. When an instance in the hot data area has not been accessed for a long time, the tiered cache management unit writes the instance from memory to the disk and assigns it to the cold data area.

[0042] The preloading prediction unit is used to build an application access behavior prediction model based on historical access sequences. It predicts the logical view instance that might be called in the next moment based on the current application service layer's access context and loads the prediction results from disk into memory in advance. Specifically, the preloading prediction unit continuously collects the call sequences of each logical view instance from the application service layer, forming a historical access sequence sorted by time. The preloading prediction unit uses a Markov chain-based prediction method: taking the M most recently accessed logical view instances as the current state, it calculates the conditional probability of each instance being called again in the current state in the historical sequence, and selects the instance with the highest conditional probability as the prediction result. M is a preset positive integer, for example, M=3. The preloading prediction unit sends the prediction results to the hierarchical cache management unit, which loads the prediction results from disk into memory in advance before they are actually loaded into memory. Therefore, when the application service layer actually initiates a call to the predicted instance, the instance is already resident in memory, eliminating disk read latency and further improving system response speed.

[0043] This embodiment records the mapping relationship between logical view instances and main model source objects through a logical view management unit, calculates the device range affected by main model changes and locates affected instances through a change impact domain calculation unit, and incrementally updates the view based on the mapping relationship to incrementally synchronize the changed content to the corresponding logical view instances. This achieves incremental synchronization of associated views rather than full redistribution when the main model changes, significantly reducing model maintenance overhead and data transmission volume. Simultaneously, a hierarchical cache management unit divides logical view instances into hot, warm, and cold zones for hierarchical storage based on access frequency and recent access time. A pre-loading prediction unit predicts instances that may be called at the next moment based on historical access sequences and pre-loads them into memory, ensuring that high-frequency access instances reside in memory and low-frequency instances are stored on disk, achieving optimal allocation of cache resources under limited memory conditions. The synergistic effect of these mechanisms shortens the average model loading time on the application side, improving resource utilization and system response speed in multi-service concurrent scenarios.

[0044] Please see Figure 4 In a preferred embodiment of the present invention, the application service layer includes: The panoramic operation monitoring unit is used to acquire the logical view instance provided by the model management layer, and overlay real-time measurement data on the logical view instance to generate a visual monitoring interface. The panoramic operation monitoring unit includes a hierarchical monitoring module, an indicator display module, and an alarm fusion module. The hierarchical monitoring module is used to construct hierarchical monitoring views, including power grid geographic wiring diagrams, substation wiring diagrams, and equipment bay detail diagrams. The power grid geographic wiring diagram renders equipment icons on an electronic map according to the actual geographical location of the equipment. The substation wiring diagram shows the connection relationships of busbars, circuit breakers, and disconnect switches within a substation. The equipment bay detail diagram shows the detailed wiring and parameters of equipment within a single bay. These three views are linked through equipment object identifiers. When a user performs a zoom or pan operation on any view, the hierarchical monitoring module synchronously refreshes the display range and resolution of other views, ensuring that each view always focuses on the same geographic area or the same equipment range. The hierarchical monitoring module also stores view combinations and initial zoom ratios corresponding to different business positions through preset monitoring templates. For example, the monitoring template for dispatching duty personnel defaults to displaying the provincial geographic wiring diagram, while the monitoring template for substation inspection personnel defaults to displaying the substation wiring diagrams of their assigned substations. When a user logs in, the hierarchical monitoring module automatically loads the corresponding monitoring template based on the user's job identifier. The indicator display module reads measurement data from the real-time database at preset intervals and calculates the operating indicator values ​​according to pre-configured indicator calculation formulas. These formulas can be pre-configured by maintenance personnel using a formula editor; for example, line load rate equals line current divided by line rated current, and transformer load rate equals transformer apparent power divided by transformer rated capacity. The indicator display module compares the calculated operating indicator values ​​with preset alarm threshold ranges. These alarm threshold ranges can be configured as multi-level ranges, such as normal range, warning range, and alarm range. Based on the comparison results, the indicator display module renders and displays the results on the monitoring view using corresponding colors or graphic icons; for example, green for normal, yellow for warning, and red for alarm. The indicator display module also configures corresponding drill-down paths for each operating indicator. These drill-down paths define the data link from the summary indicator to the detailed indicator; for example, drilling down from the substation load rate to the load rates of each feeder, and then down to the load rates of each distribution transformer. When a user clicks on any operating indicator, the indicator display module drills down along the drill-down path to display the detailed constituent data of that indicator. The alarm fusion module is used to perform real-time correlation, merging, suppression, and root cause analysis of multi-source heterogeneous alarms based on the alarm information model, forming a fault event chain and providing handling guidance. Specifically, the alarm fusion module receives alarm signals from various sources such as SCADA systems, protection information management systems, and fault recording systems. The alarm information model defines a unified data structure and attribute fields for alarm objects, including alarm device identifier, alarm type, alarm time, and alarm level. The alarm fusion module processes multi-source heterogeneous alarms according to the alarm information model as follows: First, it merges multiple alarms of the same type generated by the same device in adjacent time periods into one alarm record; second, for derivative alarms caused by the same reason, it suppresses them without repeated reporting; then, by analyzing the topological connection relationship between alarm devices and the chronological order of alarm times, it identifies the root cause alarm from the merged alarm records and strings the root cause alarm and its related alarms together to form a complete fault event chain; finally, based on the type of fault event chain, it matches the corresponding handling guidance information from the contingency plan database and pushes it to the dispatcher. The intelligent analysis and decision-making unit is used to call the logical view instance provided by the model management layer corresponding to the current analysis scope when a fault signal or maintenance plan trigger instruction is received. Based on the equipment model, topology connection relationship and real-time measurement data in the logical view instance, it performs main and distribution network fault collaborative analysis, power supply path dynamic analysis and intelligent planning and optimization of load transfer path.

[0045] The intelligent analysis and decision-making unit includes a fault collaborative monitoring and processing module, a power supply path dynamic analysis module, and an intelligent power transfer path planning module. The fault collaborative monitoring and processing module is used to establish a unified fault information model for the main and distribution networks. The fault information model defines a unified data structure and attribute fields for fault event objects, including fault device identifier, fault type, fault occurrence time, fault phase, and fault electrical quantity characteristics. The module receives heterogeneous fault signals from main network fault detection devices and distribution network fault indicators, mapping these signals from different sources to unified fault event objects according to the fault information model. After mapping, the module locates the fault device in the logical view instance based on the fault device identifier in the fault event object and analyzes its impact range centered on the fault device. The impact range analysis method is as follows: in the topological connection relationship of the logical view instance, starting from the fault device, traverse all reachable devices downstream along the power supply direction until an open-loop point or tie switch is encountered, marking all devices passed during the traversal as the outage impact range. The module overlays and renders the fault device location and outage impact range onto the visualization monitoring interface of the panoramic operation monitoring unit, and marks the fault device in the logical view instance, achieving panoramic fusion of fault information and collaborative handling processes. The power supply path dynamic analysis module uses the topological connections in the logical view instance as its data foundation. Starting from a specified power source point, it traverses all reachable load points in the logical view instance using a breadth-first search algorithm to generate the power supply path corresponding to each load point. The power supply path is represented by the sequence of devices traversed from the power source point to the load point. The module calculates the electrical distance and load margin of each power supply path. The electrical distance is the sum of the impedances of all devices on the power supply path, and the load margin is the difference between the available capacity of the most constrained device on the power supply path and the current load. The module renders the calculated power supply paths, electrical distances, and load margins in an overlay manner on the visualization monitoring interface of the panoramic operation monitoring unit. Power supply paths with load rates exceeding a preset threshold are highlighted with a high-key color or flashing indicator to alert the dispatcher to the risk of overload. The intelligent load transfer path planning module is used to automatically search for feasible load transfer and restoration paths within a range that meets preset constraints when a power grid failure or planned maintenance causes a power outage.

[0046] The intelligent transfer path planning module includes a path search submodule, a safety verification submodule, an optimization decision submodule, and a simulation and exercise submodule.

[0047] The path search submodule is used to search for a set of feasible transfer paths within a certain range, satisfying preset constraints, when a power grid failure or planned maintenance causes a power outage. This is achieved by abstracting the power grid topology into a weighted directed graph based on graph theory. Specifically, the path search submodule uses circuit breakers, disconnectors, and tie switches in the logical view instance as edges of the graph and bus nodes as vertices. The weighted directed graph contains a comprehensive cost weight (Cost) for each edge. Defined as: In the formula, For the edge The per-unit impedance value, For the edge The current load rate, For the edge The historical number of operations of the corresponding switching equipment. For the edge The reliability coefficient of the corresponding equipment; , , , These are weighting coefficients that are dynamically adjusted based on the current operating scenario. For example, in a heavy-load scenario, the path search submodule is increased. The value should prioritize paths with low load rates; in scenarios where device health is poor, increase the value. The value is selected based on prioritizing paths with high reliability. Preset constraints include electrical constraints (node ​​voltage amplitude limits), topological constraints (network connectivity and open-loop operation), capacity constraints (line current carrying capacity and transformer capacity), operational constraints (limits on the number of switch operations and protection setting coordination), and reliability constraints (power restoration time limits). The path search submodule uses a heuristic A / B algorithm. The algorithm performs path search, using the comprehensive cost weight as the path cost, while employing a heuristic pruning strategy to reduce the search space and supporting parallel computing acceleration.

[0048] The safety verification submodule simulates the switching operations involved in each feasible power transfer path found by the path search submodule, and calculates the power flow distribution of the entire power grid after the power transfer based on the device parameters and topology connections in the logical view instance. The safety verification submodule uses the Newton-Raphson method for power flow calculation, obtaining the voltage magnitude and phase angle of each node, and the active and reactive power of each branch after the power transfer. The safety verification submodule compares the calculated node voltages with preset upper and lower voltage limits, compares the branch currents with preset current carrying capacity limits, and performs an N-1 safety check to determine if the load rate of remaining devices exceeds the limit if any device is taken out of operation. The safety verification submodule marks feasible power transfer paths that meet all three comparison results as safe paths. The optimization decision submodule is used to perform multi-objective optimization evaluation of the safety path with network loss increment, number of switching operations, feeder load rate variance and power restoration time as optimization objectives, generate a comprehensive score and select the optimal power transfer scheme. The multi-objective optimization evaluation includes establishing a multi-objective function. ,in The goal is to minimize the increase in network loss before and after the transfer. The objective is to minimize the number of switching operations involved in the power transfer process. The objective is to minimize the variance of the load rate of each feeder after the transfer. The objective is to minimize the estimated power restoration time. The NSGA-II (Non-dominated Sorting Genetic Algorithm II) algorithm is used to solve the above multi-objective optimization problem, and the Pareto optimal solution set is obtained, that is, there is no set of solutions where all objectives are better than other solutions; Specifically, the execution process of the NSGA-II algorithm is as follows: The first step is to initialize the population.

[0049] The optimization decision submodule encodes each safe path marked by the safety verification submodule into an individual in the population, with each individual corresponding to a complete transfer scheme. Binary encoding is used, where each bit represents the operating state of a switch device: 0 indicates no operation, and 1 indicates operation. Let the total number of switches involved in the safe path be L, then each individual is a binary string of length L. The population size is set to the total number of safe paths. If the number of safe paths exceeds a preset limit, for example, more than 100, then 100 paths are randomly selected as the initial population.

[0050] The second step is to calculate the values ​​of each objective function.

[0051] The optimization decision-making submodule calculates the values ​​of the corresponding transfer scheme for each individual in the population on the four optimization objectives.

[0052] Assume the total active power loss of the entire network before the transfer is After the transfer, the total active power loss of the entire network is Increase in network loss Defined as: Let the total number of switch position changes involved in the power transfer scheme be... Number of switch operations Defined as: Assume the load rates of each feeder after the power transfer are respectively Its mean is Feeder load factor variance Defined as: Assume the operation time of each switch in the power transfer scheme is... The protection device has an activation delay of Power restoration time Defined as: The third step is quick non-dominated sorting.

[0053] The optimization decision submodule performs a non-dominated sort on all individuals in the current population. The rule for the non-dominated sort is: for any two individuals in the population... and If an individual The performance indicators on all four optimization objectives are no worse than those of the individual. And superior to individuals in at least one objective. Then it is called an individual Dominant Individual The optimization decision submodule iterates through all individuals in the population, assigns individuals not dominated by any other individual to the first non-dominated layer, removes individuals from the first non-dominated layer from the population, and repeats the above process for the remaining individuals to divide the population into the second non-dominated layer, the third non-dominated layer, and so on.

[0054] The fourth step is to calculate the congestion distance.

[0055] For individuals within the same non-dominated layer, the optimization decision submodule calculates the crowding distance for each individual. The crowding distance measures the density of an individual in the target space; a larger crowding distance indicates sparser solutions and better diversity around the individual. Let the... The value of each optimization objective is... All individuals within the non-dominated layer are categorized as follows: Sort in ascending order, and set the crowding distance between the first and last individuals to infinity. For intermediate individuals... In its first The contribution of congestion on each objective is: in, and Individuals The last individual and the first individual in the first... Indicator values ​​for each target, and These are the first (i) in the non-dominated layer. The maximum and minimum values ​​of each target. Individual The congestion distance is the sum of its congestion contributions across all targets: The fifth step is selection, crossover, and mutation.

[0056] The optimization decision-making submodule employs a binary tournament selection strategy to select parent individuals from the current population: each time, two individuals are randomly selected from the population, prioritizing the individual with the smaller non-dominated layer number; if the non-dominated layer numbers are the same, the individual with the greater crowding distance is selected. This selection process is repeated to select parent individuals sufficient for the population size. Then, the optimization decision-making submodule uses a single-point crossover operator to generate offspring individuals: two parent individuals are randomly selected, and gene exchange occurs at a randomly selected crossover point to generate two offspring individuals, with the crossover probability preset to a certain value. After crossover, the optimization decision submodule uses a bit-flipping mutation operator to mutate the offspring individuals: for each gene in the offspring individuals, a preset mutation probability is applied. The value of this bit is flipped, i.e., 0 becomes 1 and 1 becomes 0. After mutation, offspring population is generated. Step 6: Elite strategy merging.

[0057] The optimization decision submodule merges the parent and offspring populations into a merged population twice the size of the original population. It then performs fast non-dominated sorting and crowding distance calculation on the merged population again. Next, the optimization decision submodule selects individuals to fill the next generation population according to the non-dominated layer number in ascending order. When filling a certain non-dominated layer, if the number of individuals in that layer exceeds the remaining filling slots, it prioritizes selecting individuals with larger crowding distances until the next generation population size reaches the preset population size. The seventh step is the iteration termination judgment. The optimization decision submodule repeats steps two through six until the preset maximum number of iterations is reached. For example... The iteration terminates when the Pareto optimal solution set no longer changes across multiple generations of the population. After iteration terminates, the first non-dominated layer individuals in the current population constitute the Pareto optimal solution set. In other words, there is no set of solutions that are superior to all other solutions in all objectives. Each solution in the Pareto optimal solution set has its own advantages and disadvantages in each of the four optimization objectives, but there is no solution that is superior to another solution in all objectives.

[0058] After the iteration terminates, the optimization decision submodule uses the entropy weight TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) method to make multi-attribute decisions on the Pareto optimal solution set and select the unique optimal transfer scheme from the Pareto optimal solution set.

[0059] The entropy-weighted TOPSIS method is used to make multi-attribute decisions on the Pareto optimal solution set. The decision matrix is ​​constructed based on the values ​​of each Pareto solution on each optimization objective. The entropy-weighted method is used to objectively calculate the weight of each optimization objective based on the dispersion of the values ​​of each objective. Then, the TOPSIS method is used to calculate the distance between each Pareto solution and the positive and negative ideal solutions. Finally, the relative closeness between each solution and the positive and negative ideal solutions is calculated. The Pareto solution with the largest relative closeness is selected as the optimal transfer scheme. The execution process of the entropy weight TOPSIS method is as follows: The first step is to construct the decision matrix.

[0060] The optimization decision submodule uses each solution in the Pareto optimal solution set as a decision scheme and four optimization objectives as decision attributes to construct... Decision matrix ,in Let be the number of solutions in the Pareto optimal solution set. The decision matrix is... Line number Column elements Indicates the first The first scheme is in the Indicator values ​​for each optimization objective: The second step is data normalization.

[0061] Because the four optimization objectives have different dimensions, the optimization decision-making submodule normalizes the decision matrix. All four objectives are cost-based, meaning smaller values ​​are better. The optimization decision-making submodule uses the range normalization method to calculate the normalized index value. For the... For each objective, take the minimum value in that column of the decision matrix. and maximum value Normalized index value for: After normalization, all objectives are uniformly transformed into positive indicators where larger values ​​are better, and .

[0062] The third step is to calculate the information entropy and entropy weight.

[0063] The optimization decision submodule calculates the first... Information entropy of a target First calculate the... Under the first goal The characteristic proportion of each scheme : when At that time, it was agreed Then calculate the information entropy. : Information entropy The smaller the value, the greater the dispersion of the target's indicator value across different options, and the more decision-making information it contains. The optimization decision submodule calculates the entropy weight of each target based on the information entropy. : Entropy weight satisfy That is, the objective weights of each optimization objective.

[0064] The fourth step is to determine the positive ideal solution and the negative ideal solution.

[0065] The optimization decision submodule determines the positive and negative ideal solutions for each objective based on the normalized decision matrix. Positive ideal solution The solution that maximizes the value of each objective is the negative ideal solution. The solution that minimizes the value across all objectives: The fifth step is to calculate the relative proximity.

[0066] The optimization decision submodule calculates the Euclidean distance between each solution and the positive and negative ideal solutions. Distance between each solution and the ideal solution The distance between the sum and the negative ideal solution They are respectively: Then calculate the relative similarity of each scheme. : Relative closeness The value ranges from 0 to 1. The larger the value, the closer the solution is to the positive ideal solution and the further away it is from the negative ideal solution, which means the better the overall performance.

[0067] Step 6: Select the optimal solution.

[0068] The optimization decision-making submodule selects the solution with the highest relative similarity as the optimal solution: The results are output in the form of a decision report, which includes detailed operational steps for the optimal solution, specific values ​​for each optimization objective, relative proximity, and weight information for each objective.

[0069] By combining the NSGA-II algorithm and the entropy-weighted TOPSIS method, the optimization decision submodule can automatically select the optimal solution from multiple feasible transition paths.

[0070] The simulation and exercise submodule is used to construct a digital twin environment based on the device model and topology connection relationship in the logical view instance, load the optimal power transfer scheme in the digital twin environment and simulate the execution of the switching operation sequence, recalculate the power flow distribution and update the device status display after each switching change operation, and feed back the simulation execution results to the optimization decision submodule.

[0071] The operation status prediction unit is used to call the logical view instance provided by the model management layer, use the topological connection relationship in the logical view instance as a constraint, extract historical operation section data to form a sample set, use a clustering algorithm to divide the sample set into scenarios, and establish a dynamic prediction model based on association rules to predict the future operation status trend. In this embodiment, the clustering algorithm is a topologically constrained weighted extended clustering algorithm; the running state prediction unit performs the following operations: Using historical operational data of the distribution network as the sample set, each sample point includes the electrical quantity measurement vector at the corresponding time. and topology diagram Define any two sample points and The combined distance between for: in, The standard Euclidean distance for the measured data. Graph editing distance for topology, This is a dynamic adjustment coefficient that adaptively adjusts based on the degree of change in the power grid topology during the current period; for example, when the power grid topology changes frequently, Decreasing the value increases the weight of topological similarity; when the power grid topology is relatively stable, Increasing the value increases the weight of electrical quantity similarity.

[0072] The optimal number of clusters was determined by using the weighted sum of the silhouette coefficient and the Davidson-Bourdin index as the clustering evaluation index. Specifically, for different cluster number K values, the running state prediction unit calculates the silhouette coefficient Silhouette(K) and the Davidson-Bourdin index DBI(K) respectively, to... As a comprehensive evaluation index, the following was selected: The maximized value of K is taken as the optimal number of clusters, where The preset weighting coefficient has a value range of 0 to 1; Based on the comprehensive distance and the optimal number of clusters, the K-medoids (K-medoids Algorithm) algorithm is used to perform initial clustering of historical operation data of the distribution network, forming multiple typical operation scenario clusters; Based on association rule mining algorithms, association rules are established between measurement data, load data, network topology and operating status within each scenario cluster; In practice, association rules are represented in the form of X→Y, where X is the antecedent of the rule, including real-time electrical measurements such as voltage, current, power, and frequency, power and current operating parameters of various loads, and topological information such as grid equipment connection relationships and switch status. Y is the consequent of the rule, including operating status identifiers such as overload, low voltage, voltage exceeding limits, and equipment abnormality, and their confidence levels. The operating status prediction unit sets minimum support and minimum confidence thresholds and mines association rules that meet the threshold conditions from each scenario cluster.

[0073] When new real-time data flows in, the operation status prediction unit dynamically adjusts its cluster affiliation based on the comprehensive distance between the real-time data and each scene cluster, as well as the matching degree of the association rules. Specifically, the operation status prediction unit calculates the comprehensive distance between the real-time data and the cluster centers of each scene cluster, and simultaneously calculates the matching degree of each measurement value in the real-time data with the antecedent of the association rules for each scene cluster. The scene cluster with the closest comprehensive distance and the highest matching degree is taken as the cluster to which the real-time data belongs. Based on the historical evolution pattern of the assigned cluster and the confidence degree of the consequent of the association rules, the operation status prediction unit predicts the operation status trend at the next moment. For example, if the current real-time data belongs to the heavy load scene cluster, and the association rules of this cluster show that the confidence degree of voltage over-limit at the next moment is 0.85 when the load rate exceeds 80%, then the operation status prediction unit outputs a voltage over-limit risk warning.

[0074] The risk management unit, based on the prediction results output by the operation status prediction unit and combined with current operation data, calls the logical view instance provided by the model management layer to overlay and display a comparison view between the current and predicted operation statuses on the logical view instance. The risk management unit calculates a comprehensive risk score for each device according to preset risk assessment rules. These rules include: overload risk score calculated based on the ratio of the device's current load rate to its rated capacity; voltage exceedance risk score calculated based on the deviation between the current voltage amplitude and the rated voltage; and power supply reliability risk score calculated based on the device's historical failure rate and predicted failure probability. The risk management unit weights and sums these three risk scores to obtain a comprehensive risk score. Devices with comprehensive risk scores exceeding preset risk thresholds are identified as weak points, and a risk report containing a list of weak points, risk level, recommended inspection cycle, and pre-control operation suggestions is generated and submitted to the dispatch management platform for review and handling by business personnel.

[0075] Please see Figure 5 The present invention also provides a real-time data processing method based on power distribution network services, applicable to a layered architecture system comprising a data acquisition layer, an upper-level aggregation layer, a model management layer, a cloud platform layer, and an application service layer, comprising the following steps: Step S1: Collect real-time measurement data, model files, and graphics of the main and distribution networks of each lower-level regional power grid through the data acquisition layer; Step S2: Receive data uploaded by each lower-level regional power grid through the upper-level aggregation layer, and construct the main model of the power grid with the model of the first regional power grid as the skeleton and the model of at least one second regional power grid as the branch at the boundary nodes. Step S3: Store the parameter attributes and topology connections of each device in the full voltage level power grid master model through the model management layer, and perform master model version management; Step S4: Receive a model request instruction from the application service layer through the model management layer, and perform three-dimensional dynamic trimming on the full voltage level power grid master model according to the model request instruction to generate a logical view instance for application services. Step S5: Provide the logical view instance to the application service layer so that the application service layer can execute power grid business application functions based on the logical view instance; wherein the cloud platform layer provides computing environment support for the application service layer.

[0076] This embodiment has the same technical effect as a real-time data processing system based on distribution network services, and will not be described again here.

[0077] Those skilled in the art will recognize that the units of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of the invention.

[0078] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical functional division. In actual implementation, there may be other division methods, such as multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored.

[0079] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0080] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0081] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the specification of the present invention.

Claims

1. A real-time data service system based on power distribution network services, characterized in that, include: The data acquisition layer is deployed in each lower-level regional power grid to collect real-time measurement data, model files, and graphics of the local main and distribution networks. The upper-level aggregation layer is used to receive data uploaded by each lower-level regional power grid, and uses the model of the first regional power grid as the skeleton and the model of at least one second regional power grid as the branch to perform topology splicing at the boundary nodes to construct the main model of the power grid at all voltage levels. The model management layer is used to store the parameter attributes and topology connections of each device in the full voltage level power grid master model, and to perform master model version management; The cloud platform layer is used to provide unified data, model, and computing environment support for upper-layer applications. The application service layer, deployed on top of the cloud platform layer, is used to provide business application functions to power grid business personnel. The model management layer is configured to: receive a model request instruction from the application service layer, perform three-dimensional dynamic trimming on the full voltage level power grid master model according to the model request instruction, generate a logical view instance for application services, and provide the logical view instance to the application service layer for use.

2. The real-time data service system based on distribution network services according to claim 1, characterized in that, The model request instruction includes geographical range parameters, voltage level range parameters, and business analysis scenario parameters; The model management layer includes a three-dimensional pruning unit; the three-dimensional pruning unit is used to perform the following operations: A spatial index grid is constructed using the geographical range parameters, and the power grid equipment falling within the spatial index grid and its topological relationships are extracted. Based on the voltage level range parameters, all power grid equipment within the voltage level range is extracted, and N levels of coupling boundary equipment are traced upwards and downwards along the topological connection relationship. The coupling boundary equipment is extracted as model boundary nodes, where N is a preset positive integer. The business type is determined based on the business analysis scenario parameters, and a topological connectivity subgraph associated with the current business type is extracted.

3. A real-time data service system based on distribution network services according to claim 2, characterized in that, The model management layer also includes: The logical view management unit is used to assign a unique view identifier to each generated logical view instance and record the mapping relationship between the logical view instance and the source object in the full voltage level power grid master model; The change impact domain calculation unit is used to calculate the range of equipment affected by the model change when the main model of the full voltage level power grid changes, and to find an instance containing the range of equipment in the stored logical view instances, and send an incremental update instruction to the found logical view instance. The view incremental update unit is used to receive the incremental update instruction and synchronize the changed content to the corresponding logical view instance according to the mapping relationship; The view access frequency statistics unit is used to monitor the access frequency and recent access time of each logical view instance when it is called by the application service layer. The hierarchical cache management unit is used to divide each logical view instance into a hot data area, a warm data area, and a cold data area according to the access frequency and the recent access time; the hot data area stores logical view instances whose access frequency is higher than a first threshold or whose recent access time is within a preset duration; the cold data area stores logical view instances whose access frequency is lower than a second threshold and whose recent access time exceeds a preset duration. The preloaded prediction unit is used to build an application access behavior prediction model based on historical access sequences and predict the logical view instance that may be called in the next moment according to the access context of the current application service layer.

4. A real-time data service system based on distribution network services according to claim 1, characterized in that, The application service layer includes: The panoramic operation monitoring unit is used to acquire the logical view instance provided by the model management layer, and overlay real-time measurement data on the logical view instance to generate a visual monitoring interface. The intelligent analysis and decision-making unit is used to call the logical view instance provided by the model management layer corresponding to the current analysis scope when a fault signal or maintenance plan trigger instruction is received. Based on the equipment model, topology connection relationship and real-time measurement data in the logical view instance, it performs main and distribution network fault collaborative analysis, power supply path dynamic analysis and intelligent planning and optimization of load transfer path. The operation status prediction unit is used to call the logical view instance provided by the model management layer, use the topological connection relationship in the logical view instance as a constraint, extract historical operation section data to form a sample set, use a clustering algorithm to divide the sample set into scenarios, and establish a dynamic prediction model based on association rules to predict the future operation status trend. The risk management unit is used to call the logical view instance provided by the model management layer based on the prediction results output by the operation status prediction unit and combined with the current operation data, to overlay and display a comparison view of the current operation status and the predicted operation status on the logical view instance, to calculate the comprehensive risk score of each device according to the preset risk assessment rules, to identify the device whose comprehensive risk score exceeds the preset risk threshold as a weak link and to generate a risk report.

5. A real-time data service system based on distribution network services according to claim 4, characterized in that, In the operational status prediction unit, the clustering algorithm is a topologically constrained weighted extended clustering algorithm, which performs the following operations: Using historical operational data of the distribution network as the sample set, each sample point includes the electrical quantity measurement vector at the corresponding time. and topology diagram Define any two sample points and The combined distance between for: ； in, The standard Euclidean distance for the measured data. Graph editing distance for topology, This is a dynamic adjustment coefficient that adaptively adjusts based on the degree of drastic change in the power grid topology during the current period. The optimal number of clusters was determined by using the weighted sum of the silhouette coefficient and the Davidson-Bourdin index as the clustering evaluation index. Based on the comprehensive distance and the optimal number of clusters, the historical operation data of the distribution network are initially clustered to form multiple typical operation scenario clusters. Based on association rule mining algorithms, association rules are established between measurement data, load data, network topology and operating status within each scenario cluster; When new real-time data flows in, the clustering affiliation is dynamically adjusted based on the comprehensive distance between the real-time data and each scene cluster and the matching degree of the association rules; the operating status trend at the next moment is predicted based on the historical evolution pattern of the cluster to which it belongs and the post-confidence of the association rules.

6. A real-time data service system based on distribution network services according to claim 4, characterized in that, The intelligent analysis and decision-making unit includes: The fault collaborative monitoring and processing module is used to establish a unified fault information model for the main and distribution networks. It maps the heterogeneous fault signals of the main network fault detection device and the distribution network fault indicator into a unified fault event object. Based on the fault device identifier in the fault event object, it locates the fault device in the logical view instance and analyzes the scope of impact, thereby realizing the panoramic fusion of fault information and the collaborative handling process. The power supply path dynamic analysis module is used to search all reachable load points from the specified power point and generate power supply paths based on the topological connection relationship in the logical view instance. It calculates the electrical distance and load margin of each power supply path, overlays and renders the calculation results on the visualization monitoring interface, and highlights the power supply paths whose load rate exceeds the preset threshold. The intelligent load transfer path planning module is used to automatically search for feasible load transfer and restoration paths within a range that meets preset constraints when a power grid failure or planned maintenance causes a power outage.

7. A real-time data service system based on distribution network services according to claim 6, characterized in that, The intelligent transfer route planning module includes: The path search submodule is used to abstract the power grid topology into a weighted directed graph based on graph theory, and to search for a set of feasible transfer paths within the range of preset constraints using a dynamic impedance weighted shortest path algorithm. In the weighted directed graph, the overall cost weight (Cost) is the sum of the costs of each edge. Defined as: ； In the formula, For the edge The per-unit impedance value, For the edge The current load rate, For the edge The historical number of operations of the corresponding switching equipment. For the edge The reliability coefficient of the corresponding equipment; , , , These are weighting coefficients that are dynamically adjusted based on the current operating scenario. The safety verification submodule is used to simulate the switch change operation involved in each feasible power transfer path in the power transfer path set and calculate the power flow distribution of the entire power grid after power transfer. The calculated voltage of each node, current of each branch and equipment load rate under N-1 condition are compared with the corresponding preset limits, and the feasible power transfer path that meets the limit requirements in all three items is marked as a safe path. The optimization decision submodule is used to perform multi-objective optimization evaluation of the safety path with network loss increment, number of switching operations, feeder load rate variance and power restoration time as optimization objectives, generate a comprehensive score and select the optimal power transfer scheme. The simulation and exercise submodule is used to construct a digital twin environment based on the device model and topology connection relationship in the logical view instance, load the optimal power transfer scheme in the digital twin environment and simulate the execution of the switching operation sequence, recalculate the power flow distribution and update the device status display after each switching change operation, and feed back the simulation execution results to the optimization decision submodule.

8. A real-time data service system based on distribution network services according to claim 7, characterized in that, The optimization decision submodule is specifically used for: Establish a multi-objective function ,in The goal is to minimize the increase in network loss before and after the transfer. The objective is to minimize the number of switching operations involved in the power transfer process. The objective is to minimize the variance of the load rate of each feeder after the transfer. The objective is to minimize the estimated power restoration time. The NSGA-II algorithm is used to solve the above multi-objective optimization problem to obtain the Pareto optimal solution set; The entropy-weighted TOPSIS method is used to make multi-attribute decisions on the Pareto optimal solution set and select the optimal transfer scheme.

9. A real-time data service system based on distribution network services according to claim 4, characterized in that, The panoramic operation monitoring unit includes: The hierarchical monitoring module is used to construct a hierarchical monitoring view that includes power grid geographical wiring diagrams, plant wiring diagrams, and equipment bay details. In response to zoom or pan operations on any view, it synchronously refreshes the display range and display accuracy of other views. It stores view combinations and initial zoom ratios corresponding to different business positions through preset monitoring templates, and automatically loads the monitoring template corresponding to the user's position when the user logs in. The indicator display module is used to read measurement data from the real-time database at a preset period, calculate the running indicator value according to the pre-configured indicator calculation formula, compare the running indicator value with the preset alarm threshold range, render and display the comparison result on the monitoring view with the corresponding color or graphic icon, configure the corresponding drill-down path for each running indicator, and in response to the click operation of any running indicator, drill down along the drill-down path to display the detailed composition data of the running indicator. The alarm fusion module is used to perform real-time correlation, merging, suppression, and root cause analysis of multi-source heterogeneous alarms based on the alarm information model, forming a fault event chain and providing handling guidance.

10. A real-time data processing method based on power distribution network services, applied to a layered architecture system comprising a data acquisition layer, an upper-level aggregation layer, a model management layer, a cloud platform layer, and an application service layer, characterized in that... Includes the following steps: The data acquisition layer collects real-time measurement data, model files, and graphics of the main and distribution networks of each lower-level regional power grid. The upper-level aggregation layer receives data uploaded by each lower-level regional power grid, and uses the model of the first regional power grid as the skeleton and the model of at least one second regional power grid as the branch, and performs topology splicing at the boundary nodes to construct the main model of the power grid at all voltage levels. The model management layer stores the parameter attributes and topology connections of each device in the full voltage level power grid master model and performs master model version management. The model management layer receives model request instructions from the application service layer, and performs three-dimensional dynamic trimming on the full voltage level power grid master model according to the model request instructions to generate a logical view instance for application services. The logical view instance is provided to the application service layer so that the application service layer can execute power grid business application functions based on the logical view instance; wherein the cloud platform layer provides computing environment support for the application service layer.

Images