Artificial intelligence-based construction project power outage range prediction method

Abstract

The present application relates to power grid dispatching and engineering scheduling data processing technical field, disclose a construction project power cut range prediction method based on artificial intelligence, including: the semantic analysis of construction plan declaration text, extraction equipment identification, operation type and time window and so on scheduling parameter;According to the operation type and equipment two-dimensional coordinate construction space interference area, generate physical operation lockout list;Based on the distribution network connectivity graph model and lockout list deduce power failure island, and evaluate the link residual capacity of edge tie-in switch;When the peak load is greater than the residual capacity, the required operation boundary is trapped in the physical lockout list, trigger time window sliding optimization mechanism, select the recommended construction time window in the candidate time interval, which meets the load transfer requirement and has a legal operation boundary;Generate and issue the power cut range node network list binding time constraint.The present application effectively resolves the conflict between the space safety constraint and the transfer power capacity limit in engineering scheduling, avoids business process deadlock.

Description

Technical Field

[0001] This invention relates to the field of power grid dispatching and project scheduling data processing technology, specifically an artificial intelligence-based method for predicting the scope of power outages in construction projects. Background Technology

[0002] In power distribution network engineering dispatching and construction management, accurately predicting the scope of power outages caused by construction work is fundamental to ensuring the rational allocation of power grid resources and safe on-site operations. Current project dispatching primarily relies on topology analysis within the power distribution management system. This involves disconnecting relevant nodes based on the location of the equipment requiring outages on the circuit connectivity diagram, thereby deducing the power outage area and assessing the feasibility of external power transfer.

[0003] However, actual construction sites often involve the operation of construction machinery, which has strict spatial safety distance requirements. Existing scheduling data processing logic primarily focuses on a single electrical topology level, failing to incorporate the spatial physical interference caused by construction operations into the pre-assessment process. This can lead to the system, when calculating and allocating power transfer operation tasks, potentially assigning switch nodes located within the hazardous area of ​​machinery operation as the operation boundary to the site, thus creating on-site safety hazards.

[0004] Furthermore, when the external network's power transfer capacity is limited, dispatching services need to reduce outage boundaries by cutting off some loads. Because existing forecasting methods lack a joint verification mechanism for spatial safety constraints and electrical capacity constraints, if the boundary cut-off switch derived from the topology happens to be within the construction space interference range, the operation is considered physically unfeasible, and the existing business forecasting process will deadlock because it cannot find a compliant solution. This system, which relies solely on static parameter calculations, cannot resolve conflicts between resource scheduling constraints and spatial safety operation requirements by adjusting business schedules or searching for other feasible time periods. This prevents the system from outputting executable construction time schedules and boundary cut-off solutions, increasing the workload of manual order cancellations and repeated calculations, and reducing the efficiency of overall project scheduling and resource allocation. Summary of the Invention

[0005] The technical problem solved by this invention is that in the existing engineering scheduling and construction scheduling business, the prediction of the power outage range only relies on the deduction of a single electrical topology, without associating the spatial physical interference caused by the operation of construction machinery with the constraints of power transfer capacity. This results in the inability to provide an executable business scheduling and boundary removal solution when resource scheduling constraints and spatial safety operation requirements conflict, thereby causing process lock-in and on-site safety hazards.

[0006] To address the above problems, the present invention provides the following technical solution:

[0007] The first aspect of this invention provides a method for predicting the power outage range of construction projects based on artificial intelligence, comprising the following steps:

[0008] The application text for power distribution network construction plan is parsed to extract the target operating equipment identifier set, construction operation type category, and planned construction time window;

[0009] The safety radius of mechanical operation is determined according to the type of construction operation, and the spatial interference area corresponding to the construction is constructed by combining the two-dimensional geographic coordinates of the target operating equipment.

[0010] Identify device nodes that fall into the spatial interference region but do not belong to the target operation device set, and generate a physical operation lockout list accordingly;

[0011] Based on the distribution network connectivity diagram model and the physical operation interlocking list, a composite isolation section is constructed, an initial power outage island sub-diagram is derived, and the corresponding set of edge tie switches is determined.

[0012] Based on the planned construction time window, the peak load of the initial power outage island sub-map and the remaining support capacity corresponding to the edge tie switch are evaluated;

[0013] When the peak load exceeds the remaining support capacity, determine the required load shedding amount, search for candidate boundary disconnect switches, and filter legal operating boundaries that meet the constraints by combining the physical operation interlock list.

[0014] When there is no legal operation boundary that satisfies the constraints, the construction time window is optimized by sliding search, a feasible recommended construction time window is selected from the candidate time intervals, and the boundary search is performed again.

[0015] The system aggregates data from target operating equipment, spatial interference-related power outage nodes, operating boundary nodes, and construction time windows to generate a list of nodes within the power outage area, which is then sent to the engineering document management terminal.

[0016] Furthermore, the power distribution network construction plan application text is parsed to extract the target operating equipment identifier set, construction operation type category, and planned construction time window. This includes: converting the power distribution network construction plan application text into a text character sequence and inputting it into a pre-trained named entity recognition sequence annotation model; the named entity recognition sequence annotation model adopts a BERT-BiLSTM-CRF network architecture and outputs an annotation sequence result with the same length as the text character sequence; based on the position boundaries and category labels in the annotation sequence result, corresponding text fragments are extracted from the original text character sequence to assemble structured engineering scheduling parameters, including the target operating equipment identifier set, construction operation type category, and planned construction time window.

[0017] Furthermore, the above method also includes: reading the start and end time nodes in the planned construction time window and calculating the construction duration as a resource occupancy scale parameter in the time dimension.

[0018] Furthermore, determining the safe radius of mechanical operations and constructing a spatial interference area by combining two-dimensional geographic coordinates includes: inputting the type of construction operation into a pre-configured mapping dictionary to retrieve the corresponding safe radius parameter of mechanical operations; if no match is found, then using the preset default maximum safe distance parameter; initiating a device location query request to the power distribution network geographic information system to obtain the two-dimensional geographic coordinates of each target operating device in the two-dimensional electronic map layer from the target operating device identifier set.

[0019] Furthermore, the process of constructing the spatial interference region corresponding to the construction is as follows: for each target operating device, an independent two-dimensional spatial interference sub-region is constructed with its two-dimensional geographic coordinates as the center and the mechanical operation safety radius parameter as the radius; the union operation is performed on the two-dimensional spatial interference sub-regions corresponding to all target operating devices to generate the global interference region.

[0020] Furthermore, identifying device nodes falling into the spatial interference region includes: calculating the minimum bounding rectangle of the global interference region and using it as a spatial region retrieval condition to extract the local area device ledger set within this range from the power distribution network geographic information system; and performing spatial collision detection between the two-dimensional geographic coordinates of each device node in the local area device ledger set and the global interference region.

[0021] Furthermore, the process of generating the physical operation lockout list is as follows: from the equipment nodes falling within the global interference area, nodes belonging to the target operation equipment set are removed to obtain the spatial interference associated power outage node set; the spatial interference associated power outage node set is solidified into a structured data form, recording the equipment ledger code, geographical coordinates and safety interference attribute tags of the corresponding equipment, and generating the physical operation lockout list.

[0022] Furthermore, a composite isolation section is constructed and an initial power outage island sub-graph is derived, including: merging the target operating equipment set with the spatial interference-related power outage node set to obtain a composite isolation section node set; in the distribution network connectivity graph model, setting the node state corresponding to each node in the composite isolation section node set to disconnected, and removing associated edges to simulate the main power supply path disconnection state; taking the load-side adjacent nodes downstream of the composite isolation section as the search starting point, running a breadth-first search algorithm along the power flow direction to collect the distribution network nodes that have lost their connection path with the main power supply nodes and their internal connecting branches into an initial power outage island sub-graph.

[0023] Furthermore, the edge tie switch set is determined by: traversing the node connection relationships of the initial power failure island subgraph, screening the tie switches in the whole network graph model that are connected to the interior of the initial power failure island at one end and to the external normal power supply network at the other end, and are in the normally open state, and collecting them into the edge tie switch set.

[0024] Furthermore, the peak load of the initial power outage island sub-map is evaluated, including: extracting the time-series load data of each load node in the initial power outage island sub-map within the planned construction time window from the load forecasting database; summing the load of each node at the same time to obtain the total load time-series curve, and extracting the maximum value within the time window as the peak load forecasting data.

[0025] Furthermore, the remaining support capacity corresponding to the edge tie switches is evaluated, including: obtaining the upper limit threshold of active power of the external live lines connected to each edge tie switch, and the original load forecast time series data within the planned construction time window; subtracting the original load forecast value at the corresponding time from the upper limit threshold of active power, extracting the minimum value as the capacity margin of a single link, and accumulating the capacity margins of the effective links to obtain the remaining capacity margin of the link.

[0026] Furthermore, determining the required load shedding amount and searching for candidate boundary disconnect switches includes: calculating the difference between the peak load forecast data and the remaining capacity margin of the link as the difference in the total required load shedding amount; in the distribution network connectivity graph model, starting from the set of edge tie switches, performing a path backtracking traversal along the network topology towards the interior of the power outage island; when the cumulative strippable load on the branch path reaches or exceeds the difference in the total required load shedding amount, truncating the path search, recording the segmented equipment with operational capability upstream of the current location as candidate boundary disconnect switches, and generating a set of candidate boundary disconnect switches or a combination thereof.

[0027] Furthermore, the legal operation boundary is screened by combining the physical operation interlock list, including: performing a cross-table search between the equipment ledger code of each candidate switch in the candidate boundary disconnect switch set and the physical operation interlock list, and performing a secondary coordinate comparison; if any switch in the candidate boundary disconnect switch or combination is matched in the physical operation interlock list, it is determined that the on-site safe operation conditions are not met, and the next level disconnect switch is searched backwards in the graph model beyond this node until a switch not registered in the interlock list is found and determined as a legal operation boundary.

[0028] Furthermore, the condition for triggering the sliding optimization mechanism is as follows: if no candidate boundary disconnect switch that meets the spatial verification conditions is found when the calculation is performed backward along the path into the interior of the power-loss island to the composite isolation section node, a topological deadlock event is determined to be generated, and the time window reverse sliding optimization mechanism is triggered.

[0029] Furthermore, the process of sliding optimization of the construction time window is as follows: using the construction duration as a fixed window length parameter, an equidistant sliding search is performed on the time axis of the load prediction curve for the target area throughout the day or a preset period with a preset time step to generate a set of candidate time windows.

[0030] Furthermore, a feasible recommended construction time window is selected, including: for each candidate time window in the candidate time window set, calculating its corresponding local peak load prediction data and link remaining capacity margin, and then calculating the difference in the total load shedding amount under each candidate time window; selecting the candidate time window with a legal operating boundary and the lowest difference in the total load shedding amount as the recommended construction time window.

[0031] Furthermore, when selecting a recommended construction time window, if there are multiple candidate time windows with the same difference, priority should be given to the candidate time window with fewer switches corresponding to the legal operation boundary and a time position closer to the original planned construction time window.

[0032] Furthermore, the above method also includes: using the difference parameter of the total required load shedding corresponding to the recommended construction time window to reset the topology calculation constraints and re-execute the boundary search and path inversion process to resolve conflicts.

[0033] Furthermore, a list of nodes within the power outage area is generated, including: performing a graph model vertex union operation on the target operating equipment set, the spatial interference-related power outage node set, and the legal boundary disconnect switch node set determined after spatial interlocking verification, and incorporating the internal nodes enclosed by the boundary to form a comprehensive power outage node set; extracting the attribute parameters of each device in the set and organizing them into a structured relational data table to generate a list of nodes within the power outage area.

[0034] Furthermore, the above method also includes: determining whether the system has triggered the time window sliding optimization mechanism; if not, writing the original planned construction time window into the time constraint field of the power outage range node network list; if it has been triggered, using the recommended construction time window to overwrite and bind the time constraint field, generating the final effective construction time boundary and issuing it.

[0035] A second aspect of the present invention provides an artificial intelligence-based system for predicting the range of power outages in construction projects, comprising:

[0036] The text parsing module is used to perform semantic parsing and parameter extraction on unstructured engineering text, and output the structured business parameters corresponding to the construction plan.

[0037] The spatial computing module communicates with the text parsing module and is used to complete two-dimensional spatial mapping, interference region construction, and spatial collision detection based on structured business parameters.

[0038] The topology evaluation module communicates with the spatial computing module and is used to perform node traversal, power outage range deduction and capacity constraint verification based on the distribution network diagram model.

[0039] The scheduling optimization module communicates with the topology evaluation module and is used to schedule and perform sliding search and optimization adjustment of the construction window from the time dimension when electrical constraints and spatial constraints conflict.

[0040] The decision output module, connected to the above modules, is used to collect processing results to generate a list of nodes in the power outage area, complete time parameter binding, and send the results to the terminal.

[0041] This invention provides an artificial intelligence-based method for predicting the scope of power outages in construction projects. It offers the following advantages:

[0042] 1. This invention maps the spatial safety distance requirements in engineering construction operations and the link capacity limitations in power grid operation into calculable data constraints within the prediction system. By parallel verifying the physical operation lockout list and remaining link capacity, the system can proactively identify safety hazards and resource bottlenecks in the actual execution of engineering scheduling plans. This processing logic avoids business deadlocks and on-site rework caused by traditional manual experience-based approval, improving the compliance of business processes and the rigor of management plans.

[0043] 2. When spatial operational constraints conflict with business resource capacity, the prediction system automatically triggers a time-based sliding search, dynamically mitigating power demand by utilizing the temporal differences in load forecast data within a day or a preset period. This mechanism changes the traditional single-control mode of directly rejecting non-compliant documents during the approval process, and proactively provides the engineering management system with data-verified recommended construction time windows. This not only reduces the number of repeated applications from front-end construction units but also significantly improves the overall allocation efficiency of engineering scheduling resources.

[0044] 3. This invention introduces a natural language processing model to directly perform semantic parsing on the construction plan text input from the front end, extracting the core management elements required for scheduling, and finally integrating the spatial interference results, topology deduction results, and time optimization results into a list of nodes in the power outage area. The entire data processing flow eliminates the manual transcription and comparison steps between cross-business systems, realizing end-to-end data connectivity from front-end manual business submission to back-end terminal command issuance, effectively reducing the cost of manual intervention and the risk of information entry errors in the predictive scheduling process. Attached Figure Description

[0045] Figure 1 This is a structural diagram of the power outage range prediction system of the present invention;

[0046] Figure 2This is a flowchart of the power outage range prediction method of the present invention;

[0047] Figure 3 This is a flowchart of the semantic parsing process for the engineering plan text of this invention;

[0048] Figure 4 This is a schematic diagram illustrating the principle of spatial interference region calculation in this invention.

[0049] Figure 5 This is a schematic diagram illustrating the principle of generating the associated device location and interlock list for the present invention.

[0050] Figure 6 This is a schematic diagram illustrating the network topology deduction and power outage island generation principle of the present invention.

[0051] Figure 7 This is a flowchart of the capacity margin assessment process of the present invention;

[0052] Figure 8 This is a schematic diagram illustrating the dynamic topology reduction determination principle of the present invention;

[0053] Figure 9 This is a schematic diagram of the time window reverse sliding optimization mechanism of the present invention;

[0054] Figure 10 This is a flowchart of the decision compilation and list output process of the present invention;

[0055] Figure 11 This is a schematic diagram of the local topology and computational evolution process of the distribution network according to the present invention;

[0056] Figure 12 This is a curve comparing the intraday time-series load shedding amount and capacity margin before and after the optimization of this invention.

[0057] Among them, 101 is the text parsing module; 102 is the spatial calculation module; 103 is the topology evaluation module; 104 is the scheduling optimization module; and 105 is the decision output module. Detailed Implementation

[0058] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] See attached document Figure 1 The present invention provides an artificial intelligence-based power outage range prediction system for construction projects. The system includes a text parsing module 101, a spatial calculation module 102, a topology evaluation module 103, a scheduling optimization module 104, and a decision output module 105.

[0060] The text parsing module 101 is used to perform semantic parsing and parameter extraction on unstructured engineering text, and output the structured business parameters corresponding to the construction plan.

[0061] The spatial calculation module 102 is communicatively connected to the text parsing module 101 and is used to complete two-dimensional spatial mapping, interference region construction and spatial collision judgment based on structured business parameters.

[0062] The topology evaluation module 103 is communicatively connected to the spatial calculation module 102 and is used to perform node traversal, power outage range deduction and capacity constraint verification based on the power distribution network diagram model.

[0063] The scheduling optimization module 104 is communicatively connected to the topology evaluation module 103, and is used to optimize and adjust the construction window from the time dimension when electrical constraints and spatial constraints conflict.

[0064] The decision output module 105 is used to collect the processing results of the text parsing module 101, spatial calculation module 102, topology evaluation module 103 and scheduling optimization module 104, form a list of nodes in the power outage area, and complete the binding of time parameters and the distribution of results.

[0065] The power outage range prediction system also establishes data connections with the distribution network geographic information system, the distribution management system, and the load prediction database through communication interfaces to obtain spatial location data, network topology data, and load prediction data.

[0066] See attached document Figure 2 The present invention also provides an artificial intelligence-based method for predicting the power outage range of construction projects. This method is based on the aforementioned power outage range prediction system and includes the following steps:

[0067] S10, the text parsing module 101 parses the power distribution network construction plan application text and extracts the target operating equipment identifier set, construction operation type category and planned construction time window;

[0068] S20, the spatial calculation module 102 determines the safe radius of mechanical operation according to the type of construction operation, and constructs the spatial interference area corresponding to the construction based on the two-dimensional geographic coordinates of the target operating equipment;

[0069] S30, the spatial calculation module 102 identifies the device nodes that fall into the spatial interference area and do not belong to the target operation device set, and generates a physical operation lockout list accordingly.

[0070] S40, the topology evaluation module 103 constructs a composite isolation section based on the distribution network connectivity diagram model and the physical operation lockout list, deduces one or more initial power outage island sub-diagrams, and determines the corresponding set of edge tie switches;

[0071] S50, Topology Assessment Module 103 assesses the peak load of each initial power outage island sub-map and the remaining support capacity corresponding to the edge tie switch in conjunction with the planned construction time window;

[0072] S60, when the peak load exceeds the remaining support capacity, the topology evaluation module 103 determines the required load shedding amount, searches for candidate boundary disconnect switches, and combines the physical operation interlock list to screen the legal operation boundaries that meet the constraints.

[0073] S70, when there is no legal operation boundary that satisfies the constraints, the scheduling optimization module 104 performs sliding optimization on the construction time window, selects a feasible recommended construction time window from the candidate time interval, and triggers the topology evaluation module 103 to perform boundary search again.

[0074] S80, the decision output module 105 summarizes the target operation equipment data, spatial interference correlation power outage node data, operation boundary node data and construction time window data, generates a power outage range node network list, and sends it to the engineering document management terminal.

[0075] The following, in conjunction with the accompanying drawings, further explains the specific implementation process of each step of the above method.

[0076] See attached document Figure 3 In power distribution network engineering dispatching scenarios, construction plan application texts are typically manually entered. The text content includes dispatching information such as equipment objects, work types, and planned times. However, its original form is usually unstructured text, making it difficult to directly use for subsequent spatial calculations and scheduling analysis. Therefore, this embodiment uses a text parsing module 101 to convert the construction plan application text into structured engineering parameters.

[0077] In this embodiment, step S10 specifically includes the following steps:

[0078] S101, the text parsing module 101 receives the power distribution network construction plan application text submitted by the front-end document system.

[0079] After receiving the application text, the text parsing module 101 preprocesses the text and generates a file of length [length missing]. Text character sequence Text character sequence It can be represented as ,in, Indicates the total number of characters. Indicates the first One character.

[0080] For basic processing operations such as word segmentation, stop word removal, and encoding standardization involved in Chinese text processing, those skilled in the art can use existing text cleaning techniques to achieve them, and will not be elaborated here.

[0081] S102, the text parsing module 101 processes the text character sequence The input is fed into a pre-trained named entity recognition sequence labeling model.

[0082] In one optional embodiment, the named entity recognition sequence labeling model employs a BERT-BiLSTM-CRF network architecture. Specifically, the text character sequence... First, the word vectors are converted into word vector representations via a BERT embedding layer. Then, the word vector sequence is input into a Bidirectional Long Short-Term Memory (BiLSTM) layer to extract semantic context features before and after each character. Finally, the globally optimal annotation sequence is output via a Conditional Random Field (CRF) layer combined with state transition relations.

[0083] During the model training phase, training samples can be derived from historically archived construction documents of the power supply company. In accordance with the needs of engineering scheduling operations, the entity labeling system can be defined using the BIO annotation method. For example, B-DEV and I-DEV are used to identify the starting and internal positions of equipment name entities, respectively, while B-TIME and I-TIME are used to identify the starting and internal positions of time parameter entities, respectively. During model training, the negative log-likelihood function of the CRF layer can be used as the loss function, and the network parameters are updated through backpropagation until the loss function converges.

[0084] After training, the named entity recognition sequence labeling model can identify each input character. Output the corresponding category label This generates a labeled sequence result with the same length as the input text. By combining character-level boundary tags and category tags, the text parsing module 101 can identify equipment entities, time entities, and other business-related entity fragments in the construction plan text.

[0085] S103, Text parsing module 101 analyzes the labeled sequence results... Position boundaries and category labels in the original text character sequence The corresponding text fragments are extracted and the classification and extraction logic is executed to assemble structured engineering scheduling parameters.

[0086] In this embodiment, the structured engineering scheduling parameters extracted by the text parsing module 101 include at least the target operation equipment identifier set, the construction operation type category, and the planned construction time window.

[0087] The target operating equipment identifier set is used to characterize the equipment objects directly involved in the construction process, denoted as... in, Indicates the number of target operating equipment; the construction operation type category is used to characterize the mechanical operation process characteristics recorded in the construction document, denoted as... The planned construction time window is used to characterize the planned work period for construction applications, denoted as ; ,in, Indicates the start time of the plan. Indicates the end time of the plan.

[0088] To ensure that subsequent spatial calculations, topology deductions, and scheduling optimization processes can be executed normally, if the text parsing module 101 detects missing items in the target operation equipment identifier set, construction operation type category, or planned construction time window, it will generate a data anomaly event and feed it back to the front-end document system to trigger a manual supplementation or correction process.

[0089] S104, Text parsing module 101 calculates the resource occupation time span of the construction project based on the planned construction time window.

[0090] Specifically, the text parsing module 101 reads the planned construction time window. The start time node in and end time node And calculate the duration of construction. The calculation formula is as follows:

[0091] ;

[0092] Construction duration It can be quantified and stored on an hourly basis. This parameter is used to characterize the resource consumption scale of the construction task in the time dimension. In the event of a topology deadlock or when the construction window needs to be adjusted, it can be used as the fixed window length parameter for the time window sliding search performed by the scheduling optimization module 104.

[0093] See attached document Figure 4 In the process of engineering planning, scheduling, and safety management, different types of mechanical operations correspond to different safe operating distances, and their operating range may cause physical interference to surrounding equipment. To address this, the spatial calculation module 102 in this embodiment converts the semantic information of construction operations obtained from text parsing into spatial geometric parameters, and determines the physical interference boundaries corresponding to the construction based on these parameters.

[0094] In this embodiment, step S20 specifically includes the following steps:

[0095] S201, Spatial calculation module 102 receives the construction operation type category output by text parsing module 101. .

[0096] The spatial calculation module 102 is pre-configured with a mapping dictionary between engineering semantics and spatial physical parameters. The mapping dictionary records the correspondence between different construction operation types and the corresponding mechanical operation safety radius parameters. The values ​​of the mechanical operation safety radius parameters can be set with reference to the safety distance thresholds specified for different mechanical operations in current construction safety regulations or power safety work regulations.

[0097] Upon receiving the construction operation type category Then, the spatial calculation module 102 categorizes the construction operation types. Input the mapping dictionary as the query key to retrieve the corresponding mechanical operation safety radius parameter. If the corresponding construction operation type is not matched in the mapping dictionary, the preset default maximum safety distance parameter will be used as the mechanical operation safety radius parameter. This is to ensure that subsequent spatial calculations meet safety constraints.

[0098] S202, Spatial calculation module 102 obtains the target operating device identifier set output by text parsing module 101. ,in For set The number of device nodes included.

[0099] The spatial computing module 102 initiates a device location query request to the power distribution network geographic information system via a data interface. The power distribution network geographic information system returns the target operating devices in set E based on the device ledger codes. Two-dimensional geographic coordinates in a two-dimensional electronic map layer. If coordinate queries fail for some devices due to missing records or abnormal location data, the spatial calculation module 102 records an exception log and removes the corresponding failed device node from subsequent spatial calculation objects.

[0100] The obtained two-dimensional geographic coordinates can be denoted as: ,in, The value range is 1 to n. Indicates the first The lateral projection coordinates of the target operating device. Indicates the first The longitudinal projection coordinates of the target operating equipment. For the data storage structure, coordinate projection transformation, and location retrieval mechanism involved in the power distribution network geographic information system, those skilled in the art can use existing spatial database technologies, and will not be elaborated upon here.

[0101] S203, Spatial calculation module 102 performs geometric topology calculations in a two-dimensional spatial coordinate system.

[0102] Specifically, a two-dimensional plane is used as the projection base map of the power distribution network facilities, and operating equipment is configured for each target. The spatial calculation module 102 uses corresponding two-dimensional geographic coordinates. Using the circle as the center and the mechanical operation safety radius parameter Construct independent two-dimensional spatial interference sub-regions with radius as the base. This process allows for the marking on an electronic map of the maximum hazardous boundary that a single target device might cover during construction.

[0103] The mathematical logic expression for generating independent two-dimensional spatial interference sub-regions is:

[0104] ;

[0105] in, and Represents continuous planar coordinate variables in two-dimensional space.

[0106] S204, Spatial calculation module 102 generates a module containing... After collecting the set of independent spatial interference sub-regions of each element, for all two-dimensional spatial interference sub-regions Perform the union operation on spatial geometric sets.

[0107] This union operation is used to eliminate the overlapping area between the operating radii of adjacent target operating equipment, thereby forming a global interference region that characterizes the overall physical interference range of the current construction project. .

[0108] Global Interference Region The calculation formula is:

[0109] ;

[0110] Through the above spatial mapping and geometric calculation process, the spatial calculation module 102 converts the semantic information of construction operations at the text level into continuous geometric boundaries in two-dimensional space, thereby constructing a static spatial constraint model corresponding to the construction process, providing a spatial calculation basis for subsequent identification of associated equipment, physical operation interlock verification, and power outage range deduction.

[0111] See attached document Figure 5 Because power distribution network equipment is densely distributed in geographical space, other equipment around the target operating equipment, even if not directly subject to a power outage, may still be affected by construction activities as long as they enter the safe radius of mechanical operations, thus constraining on-site operational safety and the delineation of the power outage area. Therefore, in this embodiment, the spatial calculation module 102 further maps the spatial interference area obtained in step S20 into management objects at the equipment ledger level, and generates a physical operation lockout list accordingly.

[0112] In this embodiment, step S30 specifically includes the following steps:

[0113] S301, the spatial computing module 102 initiates a spatial region retrieval request to the power distribution network geographic information system through a data interface to obtain the globally covered interference area. The collection of equipment ledgers for a localized area within the outer enclosure is denoted as... .

[0114] In an optional embodiment, the spatial computing module 102 first calculates the global interference region. The minimum bounding rectangle is used as the spatial region retrieval condition to extract the set of local equipment ledgers located within this range from the power distribution network geographic information system. This method avoids performing a full coordinate traversal of all network device nodes, thereby reducing resource consumption during spatial computation.

[0115] Obtaining a local area equipment ledger collection Then, the spatial computing module 102 reads the set. Each equipment node in China Two-dimensional geographic coordinates And the two-dimensional geographic coordinates are compared with the global interference region generated in step S20. Perform spatial collision detection to determine whether the corresponding device node falls into the global interference region.

[0116] In one optional embodiment, spatial collision detection can be implemented using a ray-crossing algorithm. Specifically, a ray can be emitted from the coordinates of the device node to be detected in any direction, and the parity of the intersection points of the ray with the boundary of the global interference region can be statistically analyzed to determine whether the corresponding device node is located within the global interference region. internal.

[0117] S302, the spatial calculation module 102 performs a conditional filtering operation based on the results of spatial collision detection to extract objects that are affected by physical interference and are associated with power outages.

[0118] Because it falls into the global interference region The equipment nodes within the system typically already include target operating equipment for which power outages have been directly requested. Therefore, it is necessary to distinguish this type of equipment from equipment additionally affected by construction space interference. To this end, the space calculation module 102 calls the target operating equipment identifier set output by the text parsing module 101. It will fall into the global interference region. And does not belong to a set The selected device nodes are defined as the spatial interference-related power outage node set. .

[0119] The mathematical logic expression for this screening process is:

[0120] ;

[0121] in, This represents the set of devices remaining after removing the target operating device. This indicates that the coordinates of the corresponding device node are located in the global interference region. Inside.

[0122] If the filter results contain a set If the result is empty, it indicates that there are no additional interfering devices within the current construction space. In this case, the spatial calculation module 102 can generate an empty data form and pass the result to subsequent processing steps.

[0123] S303, the spatial calculation module 102 will generate a set of spatially interferometrically correlated power outage nodes. The data is solidified into a structured form, generating a physical operation lockout list.

[0124] The physical operation interlock list can record the equipment ledger code, geographical coordinates, and safety interference attribute labels of the spatial interference-related power outage nodes. The physical operation interlock list is used to characterize equipment objects that are within the construction physical interference range and are therefore unsuitable for live operation, power supply switching, or on-site opening and closing operations.

[0125] In subsequent power outage topology simulation and power transfer resource scheduling, the physical operation interlock list is used as a hard constraint. For equipment nodes registered in the physical operation interlock list, the system restricts them at the logical calculation level from being objects that can perform power transfer or live operations, thereby transforming the physical safety distance constraints at the construction site into data constraint rules that can participate in scheduling calculations.

[0126] See attached document Figure 6 After obtaining the physical operation lockout list and determining the construction space constraints, it is also necessary to extrapolate the potential power outage impact range caused by the construction operation from the electrical network level to clarify the power outage area and its subsequent power transfer possibility. To this end, the topology evaluation module 103 in this embodiment comprehensively considers the disconnection impact of the target operation equipment and the spatial interference associated power outage nodes in the distribution network diagram model, extrapolates the formation of the initial power outage island, and further identifies edge tie switches that can be used for external support.

[0127] In this embodiment, step S40 specifically includes the following steps:

[0128] S401, Topology evaluation module 103 obtains the current distribution network connectivity model from the distribution management system. ,in Represents the complete set of nodes in the network. It represents the complete set of all branches in the network.

[0129] In one alternative embodiment, the distribution network connectivity diagram model It is represented by an adjacency matrix or other equivalent graph storage structure, where nodes represent physical devices such as switches and transformers, and branches represent connecting cables between devices.

[0130] The topology evaluation module 103 will output the target operating device set from the text parsing module 101. The set of power outage nodes associated with spatial interference generated by spatial computing module 102 By merging, a set of composite isolation section nodes is obtained. The calculation formula is as follows:

[0131] ;

[0132] After obtaining the set of nodes of the composite isolation section Then, the topology evaluation module 103 evaluates the graph model. Perform graph editing operations to add the collection The node state corresponding to each node is set to disconnected, and the associated edges directly connected to these nodes are removed, so as to simulate the main power supply path disconnection state caused by construction operations and spatial safety avoidance in the graph model.

[0133] S402, Topology evaluation module 103 runs a breadth-first search algorithm to perform network topology connectivity deduction.

[0134] Specifically, the topology evaluation module 103 selects the location of the power node and the topology hierarchy in the distribution network diagram model based on the preset power node locations and topology hierarchy. The downstream load-side adjacent nodes serve as the search starting point, and a hierarchical traversal is performed downstream along the power flow direction.

[0135] During the traversal, all distribution network nodes that have lost their connection to the main power supply nodes and their internal connecting branches are grouped into initial power outage areas, forming an initial power outage island sub-map. ,in, This represents the set of nodes in the initial power-loss island. This represents the set of branches in the initial power outage island. The initial power outage island subgraph represents the original power outage impact range caused by construction operations and spatial safety constraints before external power transfer and dispatch measures are implemented.

[0136] When the composite isolation section forms multiple disconnected power loss areas in the graph model, the topology evaluation module 103 can generate multiple corresponding power loss island sub-graphs and perform edge tie switch identification, capacity verification, boundary search and result summary operations on each power loss island sub-graph.

[0137] For the determination of power flow direction in the distribution network and the data queue processing logic in the breadth-first search algorithm, those skilled in the art can use conventional graph traversal methods to implement them, which will not be elaborated here.

[0138] S403, Topology evaluation module 103 generates initial power-depleted island diagram. Subsequently, a connection status check is performed on the edge of the subgraph to locate external resources that can be used for subsequent scheduling support. When there are multiple power-out island subgraphs, the topology evaluation module 103 performs the above connection status check on each power-out island subgraph and forms a corresponding set of edge tie switches.

[0139] Specifically, the topology evaluation module 103 traverses the initial set of power-depleted island nodes. The connection relationships are used to screen the entire network graph model. A connection branch with one end connected to the interior of the initial power failure island and the other end connected to the external normal power supply network. For normally open tie switches included in this type of connection branch, the topology evaluation module 103 identifies and groups them into an edge tie switch set. .

[0140] If no matching contact switch is found in the investigation results, then the edge contact switch set... An empty value indicates that there are no external support channels available for the current power outage area.

[0141] Edge contact switch set It characterizes the potential entry points for load transfer in the current construction state of the distribution network by changing the operating status of switches, and can serve as a basic operational resource for subsequent capacity margin assessment, load cutting and power outage boundary adjustment.

[0142] See attached document Figure 7 To determine whether the external power supply network can handle the load of the power outage area caused by construction and to avoid overloading of external lines due to power transfer operations, this embodiment uses the topology evaluation module 103 in conjunction with the planned construction time window to perform time-series matching and capacity verification of the load demand of the power outage island and the external support capacity.

[0143] In this embodiment, step S50 specifically includes the following steps:

[0144] S501, Topology evaluation module 103 obtains the planned construction time window from text parsing module 101. The topology evaluation module 103 retrieves the power prediction data of each load node within the initial power outage island sub-map from the load prediction database. When there are multiple initial power outage island sub-maps, the topology evaluation module 103 extracts the corresponding time-series load data for each initial power outage island sub-map and calculates the corresponding peak load prediction data for each.

[0145] In one alternative embodiment, the load forecasting database stores node-level time-series load curves generated based on historical electricity consumption behavior, meteorological information, or other load influencing factors, with a time resolution that can be set to 15 minutes, 30 minutes, or 1 hour.

[0146] Topology evaluation module 103 extracts the initial set of power-depleted island nodes. Each node in the middle During the planned construction window The time-series load data within is denoted as ,in, Indicates the planned construction time window Discrete-time cross-sectional variables within.

[0147] S502, the topology evaluation module 103 performs time-series superposition calculation on the acquired node load data to calculate the peak load prediction data of the entire initial power outage island sub-map.

[0148] Specifically, the topology evaluation module 103 accumulates the load of all load nodes within the initial power outage island at the same time to obtain the time-series curve of the total load of the power outage island, and then calculates the load within the planned construction time window. The maximum value of the total load time-series curve is extracted and recorded as the peak load forecast data. By using the maximum load value within a time window as the evaluation basis, the power transfer capacity can be verified under more unfavorable load conditions.

[0149] Its mathematical formula is:

[0150] ;

[0151] in, Indicates the initial power loss island at time [time]. Total load value, Indicates the interval Perform the maximum value extraction operation within the specified time. and The units of measurement can be uniformly expressed as kilowatts (kW) or megawatts (MW).

[0152] S503, the topology evaluation module 103 evaluates the edge interconnection switch set generated in step S40. Determine the external live lines connected to each interconnecting switch.

[0153] If the edge communication switch set If the value is empty, it indicates that there is no external support channel available for the current power outage area. The topology assessment module 103 can directly set the remaining capacity margin of the link to zero and end the current capacity assessment process.

[0154] If the edge communication switch set If not empty, the topology evaluation module 103 initiates a query request to the power distribution management system to obtain the rated current carrying capacity parameters of the external live lines connected to each edge tie switch, as well as the planned construction time window for these external live lines. The original load forecast time series data is available. The rated current carrying capacity limit parameter can be obtained from the static thermal stability limit value recorded in the equipment ledger.

[0155] To facilitate a unified comparison with load forecast results, the topology evaluation module 103 can convert the upper limit of rated current carrying capacity into an upper limit threshold of active power based on the line's operating voltage level and rated power factor.

[0156] S504, Topology evaluation module 103 calculates the remaining capacity margin of the links corresponding to the set of edge contact switches.

[0157] For edge contact switch set Any switch in The external power lines it is connected to are at all times The available support capacity can be determined by the active power upper limit threshold of the line. Subtract its time The original load Received. Topology evaluation module 103 within the planned construction time window. The minimum available support capacity of the line is extracted and used as the capacity margin for the corresponding single support link.

[0158] If an external power line is within the planned construction window If the minimum available support capacity within the time window is less than zero, it indicates that the line itself is already at risk of overload within the time window. For such lines, the topology evaluation module 103 sets the capacity margin of its individual links to zero and does not include it in the subsequent support capacity.

[0159] Based on this, the topology evaluation module 103 evaluates the edge interconnection switch set. The capacity margins of individual links corresponding to each switch are summed to obtain the remaining link capacity margin used to characterize the overall support capability. .

[0160] In an optional embodiment, if multiple edge interconnection switches share the same external feeder, the same restricted section, or the same upstream power supply for their corresponding external support paths, the topology evaluation module 103 performs joint constraint verification on these edge interconnection switches and only accumulates the independent support capacity that can carry the transferred load in parallel, so as to avoid repeatedly counting the same external support capacity.

[0161] Link remaining capacity margin The mathematical formula for calculation is:

[0162] ;

[0163] in, Indicates switch The upper limit threshold of active power of the connected external power lines. This indicates that the external live line is at time The original load value.

[0164] Through the above capacity assessment process, the system can quantify the available power transfer space in the external network within and outside the construction time window, thereby providing a capacity constraint basis for subsequent load shedding calculation, boundary disconnect switch search, and construction time window optimization.

[0165] See attached document Figure 8 In the process of power transfer dispatching in the distribution network, if the capacity of external support lines is insufficient to handle all the loads in the power outage area, it is necessary to cut off part of the load and redefine the outage boundaries to ensure that the power transfer scheme meets the line capacity constraints. Simultaneously, since some boundary switches may be within the construction space interference range, the boundary reduction process must also meet on-site operational safety requirements. Therefore, this embodiment introduces a physical operation interlocking list into the electrical topology calculation through the topology evaluation module 103 to dynamically reduce the outage boundaries.

[0166] In this embodiment, step S60 specifically includes the following steps:

[0167] S601, Topology evaluation module 103 compares the peak load prediction data of the initial power outage island. With remaining capacity margin of the link .

[0168] when Greater than When this occurs, it indicates that the external network cannot handle all the load transferred to the power outage island, and load shedding is needed to reduce the transfer demand. At this time, the topology evaluation module 103 calculates the difference in the total amount of load shedding required. The calculation formula is as follows:

[0169] ;

[0170] when Less than or equal to When the external support capacity is sufficient to cover the load demand of the power outage island, no load shedding is required. In this case, the topology evaluation module 103 can directly use the edge tie switch as the final operating boundary and the corresponding transfer access location as the basis for determining the power outage range boundary. The subsequent boundary reduction process will not be executed.

[0171] S602, Topology evaluation module 103 uses the set of edge tie switches in the distribution network connectivity graph model. Starting from the point of origin, search for candidate boundary disconnect switches along the electrical distance from farthest to closest.

[0172] Specifically, the topology evaluation module 103 performs a path backtracking traversal along the network topology into the power outage island, and accumulates the load value of each load node on the path level by level during the traversal.

[0173] When the cumulative strippable load on a certain branch path reaches or exceeds the difference between the required total load shedding amount and the total load shedding amount for the first time. At this time, the topology evaluation module 103 cuts off the path search and records the segment equipment with opening and closing operation capability upstream of the current position as a candidate boundary disconnect switch.

[0174] If the cumulative strippable load of a single branch path is insufficient to meet the requirements... This allows combining candidate boundary disconnect switches on multiple branch paths, with the combined cumulative strippable load not less than [amount missing]. As a screening criterion for candidate boundary disconnector switch combinations.

[0175] If a branch path backtracks to a composite isolation section node, its cumulative strippable load is still less than [a certain value]. If the result is negative, it indicates that the branch path cannot provide effective load shedding capability, and the topology evaluation module 103 terminates the power transfer boundary search for the branch path.

[0176] After completing the backtracking calculations for each branch path, the topology evaluation module 103 generates a set of candidate boundary disconnect switches. Furthermore, a set of candidate boundary disconnect switch combinations that satisfy the load shedding constraints is generated for subsequent spatial verification.

[0177] S603, the topology evaluation module 103 introduces the physical operation interlocking list generated in step S30, and evaluates the candidate boundary disconnect switch set. The candidate boundary disconnect switch combinations generated therefrom are then subjected to a secondary coordinate comparison.

[0178] This step is used to verify whether the candidate boundaries obtained at the electrical topology level meet the conditions for safe operation in the field.

[0179] Specifically, the topology evaluation module 103 extracts a set of candidate boundary disconnect switches. The equipment ledger codes for each candidate switch are entered, and a cross-table search is performed against the physical operation interlock list. If a candidate switch is matched with the physical operation interlock list, it indicates that the switch is located within the physical interference range of construction machinery operations and does not meet the safe on-site operation conditions; therefore, it cannot be used as a legal boundary switch.

[0180] When a valid operating boundary is formed by a combination of multiple candidate boundary disconnect switches, the topology evaluation module 103 requires that none of the candidate boundary disconnect switches in the combination be matched by the physical operation blocking list before determining the combination of candidate boundary disconnect switches as a valid operating boundary.

[0181] S604, for the path where the candidate boundary disconnect switch fails the spatial verification, the topology evaluation module 103 skips the non-compliant nodes in the graph model, continues to calculate backwards in the original direction into the power loss island, searches for the next level disconnect switch node, and repeatedly performs physical operation interlocking verification.

[0182] The reverse calculation process continues until a switch node not registered in the physical operation interlock list is found along the line and identified as a legitimate boundary isolating switch.

[0183] If, when tracing back to the composite isolation section node, no candidate switch that meets the spatial verification conditions is found, it is determined that the branch path does not meet the safe transfer conditions, and the power outage range reduction strategy in that direction is abandoned.

[0184] If all candidate boundary disconnect switches or combinations of candidate boundary disconnect switches that meet the load shedding constraints fail the spatial verification, the topology evaluation module 103 outputs the topology deadlock status, which is then used by the scheduling optimization module 104 to trigger the time window reverse sliding optimization mechanism.

[0185] Through the above processing, physical safety space constraints are introduced into the electrical topology calculation process of the power outage boundary, so that the final determined power outage range boundary not only meets the capacity constraints, but also meets the on-site safety operation requirements, thereby avoiding the output of boundary control schemes that cannot be actually implemented.

[0186] See attached document Figure 9 In the process of power distribution network dispatching, the following conflict may occur: On the one hand, to meet the capacity limitations of external energized lines, it is necessary to disconnect some loads and adjust the outage boundaries; on the other hand, the candidate boundary switches derived from electrical topology may all be located within the physical interference area of ​​construction machinery operations, thus lacking on-site operation conditions. In this case, relying solely on topology boundary reduction cannot yield an executable solution, therefore, it is necessary to adjust the construction window from a time perspective to utilize load differences at different times to reduce the pressure on power transfer. To this end, this embodiment uses the scheduling optimization module 104 to perform reverse sliding optimization of the time window.

[0187] In this embodiment, step S70 specifically includes the following steps:

[0188] S701, the scheduling optimization module 104 monitors the boundary reduction calculation status of the topology evaluation module 103 in real time.

[0189] When the scheduling optimization module 104 determines that the candidate boundary disconnector or the combination of candidate boundary disconnectors that meet the capacity stripping conditions cannot constitute a legal operating boundary after completing the spatial verification, and no compliant boundary solution is obtained by calculating backward along the path into the power loss island to the composite isolation section node, the scheduling optimization module 104 generates a topology deadlock event.

[0190] A topology deadlock event indicates that within the current planned construction window, the system is unable to take over all the transferred load through existing external support capacity, nor can it provide a boundary cutting solution that meets on-site safety requirements.

[0191] S702, in response to a topology deadlock event, the scheduling optimization module 104 triggers a time window reverse sliding optimization mechanism.

[0192] Specifically, the scheduling optimization module 104 calls the construction duration output by the text parsing module 101. and the duration of construction Used as a parameter for fixed window length.

[0193] The scheduling optimization module 104 retrieves the full-day or preset cycle load forecast curve of the target area from the power distribution management system, and sets the search start point and end point on the corresponding time axis.

[0194] In an optional embodiment, the search start point can be set to midnight of the current planned date, and the end point can be set to midnight of the current day or a preset time limit to be extended thereafter. The scheduling optimization module 104 uses a preset time step. An equidistant sliding search is performed on the time axis at intervals, where the time step is... It can be consistent with the time resolution of the load forecast curve, such as 15 minutes, 30 minutes or 1 hour.

[0195] The above sliding search generates a set of candidate time windows. ,in, Indicates the number of candidate time windows. Indicates the first There are 10 candidate time windows, and the duration of each candidate time window is 1000. .

[0196] In one alternative embodiment, the candidate time window set Each candidate time window can further meet constraints such as construction management rules, maintenance approval rules, and allowable delay thresholds to ensure that the final recommended construction time window has the conditions for actual implementation.

[0197] S703, Scheduling optimization module 104 targets candidate time window sets. Each candidate time window Then, re-perform the balance calculation of dynamic load and power supply capacity.

[0198] For each candidate time window The scheduling optimization module 104 triggers the topology evaluation module 103 to re-execute the candidate boundary search, candidate boundary combination generation, and physical operation lockout verification based on the load parameters corresponding to the candidate time window.

[0199] The scheduling optimization module 104 calculates the initial power loss island sub-graphs within the candidate time window. Local peak load forecast data And the remaining capacity margin of the corresponding external contact switch set within the corresponding interval. And further calculate the difference in the total load shedding required under this candidate time window. .

[0200] When multiple initial power outage island subgraphs exist, the scheduling optimization module 104 performs the above calculations on each initial power outage island subgraph and forms a corresponding candidate time window feasibility judgment result based on the calculation results of each power outage island. Only when the candidate time window... The scheduling optimization module 104 will only select the candidate time window when there are valid operation boundaries. Record them in the set of feasible candidate time windows.

[0201] Difference in total required load shearing The mathematical formula for calculation is:

[0202] ;

[0203] in, Used to ensure that the difference is not negative, when Less than or equal to hour, A value of zero indicates that no load shedding is required within the candidate time window.

[0204] By performing the above calculations on each candidate time window, the scheduling optimization module 104 can identify time segments with relatively low load levels and higher boundary executability.

[0205] In one optional embodiment, to prevent the sliding search range from exceeding the upper limit of the allowable construction delay, a maximum sliding time span threshold can be set in the time window search logic; when the search range exceeds the threshold, automatic optimization is terminated and a manual intervention prompt is output.

[0206] S704, the scheduling optimization module 104 iterates through the calculation results of all candidate time windows and locks the difference in the total required load shedding. The lowest candidate time window.

[0207] If it exists If a candidate time window is equal to zero and has a corresponding legal operating boundary, it indicates that the internal and external support capacity of the candidate time window can meet the transfer requirements and has an executable boundary scheme. The scheduling optimization module 104 can directly determine it as the recommended construction time window.

[0208] If there are multiple If there are the same candidate time windows, priority will be given to the candidate time windows with fewer switches corresponding to the legal operation boundary and whose time position is closer to the original planned construction time window.

[0209] If no feasible candidate time window exists, the scheduling optimization module 104 will output a manual intervention prompt.

[0210] S705, the scheduling optimization module 104 uses the difference parameter of the required total load shedding corresponding to the recommended construction time window to reset the calculation constraints of the topology evaluation module 103.

[0211] Since the recommended construction time window typically corresponds to a lower load level, the demand for power transfer from the power outage island decreases accordingly, and the available support capacity of external links increases relatively. Based on the recommended construction time window, the topology evaluation module 103 re-executes the boundary search and path calculation process.

[0212] Under the new time constraints, the system can push the operation boundary to a safe switch node far away from the construction interference area, or directly eliminate the load cut-off operation when the external capacity is sufficient, thereby resolving the conflict between spatial operation interlocking and electrical overload by adjusting the construction time window.

[0213] See attached document Figure 10 In order to transform the spatial operation constraints, electrical topology boundaries and construction time parameters obtained from the aforementioned calculation steps into business data that can be used for scheduling execution and on-site operations, this embodiment uses the decision output module 105 to summarize, bind and distribute the multi-source results.

[0214] In this embodiment, step S80 specifically includes the following steps:

[0215] S801, Decision Output Module 105 performs cross-module assembly and assembly operations on data objects.

[0216] Specifically, the decision output module 105 acquires the set of target operating devices output by the text parsing module 101. The set of spatially interferometrically correlated power outage nodes generated by the spatial computing module 102 And the set of legal boundary isolation switch nodes determined by the topology evaluation module 103 after spatial locking verification. .

[0217] Based on this, the decision output module 105 performs a union operation on the vertices of the graph model on the above set, and includes the conventional distribution transformers, branch lines and other related power outage nodes involved in the boundary switch, to form a comprehensive power outage node set. Its mathematical logic expression is:

[0218] ;

[0219] in, Represents the set of boundary isolation switch nodes The set of internal nodes within the topologically connected subset; This refers to the set of distribution network equipment that ultimately needs to be in a state of power outage isolation under the combined constraints of the mechanical operation safety radius and the external network capacity.

[0220] After obtaining the comprehensive set of power outage nodes Subsequently, the decision output module 105 extracts the ledger code, spatial coordinate parameters, and expected switch opening and closing status attributes of each device in the set, and organizes them into a structured relational data table to generate a list of nodes in the power outage area.

[0221] S802, the decision output module 105 binds the time dimension parameters to the generated list of nodes in the power outage area.

[0222] Specifically, the decision output module 105 reads the running status flag of the current processing flow and determines whether the system has triggered a topology deadlock and a time window sliding optimization mechanism during the aforementioned evaluation process.

[0223] If the status flag indicates that the capacity assessment has passed and no deadlock has been triggered, the decision output module 105 extracts the planned construction time window from the original input and writes it into the time constraint field of the power outage range node network list; if the status flag indicates that the conflict resolution process has been executed, the decision output module 105 retrieves the recommended construction time window output in step S70 and overwrites and binds the original time parameters to form the final effective construction time boundary.

[0224] S803, the decision output module 105 sends the list of power outage range nodes with completed time binding to the engineering document management terminal through the network communication interface.

[0225] In one alternative embodiment, the engineering document management terminal may include the main station server of the power distribution network dispatch and control center and the mobile handheld terminal used by the field operation team.

[0226] The project document management terminal generates corresponding switching operation sequences and construction boundary alarm information based on the equipment ledger code, spatial coordinate parameters, expected switch opening and closing status attributes, and time constraint fields in the network list of nodes in the power outage area.

[0227] By issuing a list of nodes in the power outage area, the power distribution network business system can automatically generate safety operation tickets for dispatchers to execute and provide physical operation boundary warning information to on-site construction teams.

[0228] The application layer communication protocol encapsulation, message encryption transmission, and terminal-side data decoding and rendering involved in the result distribution process can be implemented using existing industrial IoT communication technologies by those skilled in the art, and will not be elaborated here.

[0229] To further assist those skilled in the art in understanding the technical solution of the present invention and its practical application effects, the following detailed explanation is provided in conjunction with a specific power distribution network application scenario and the corresponding network topology diagram.

[0230] See attached document Figure 11 In this scenario, a routine line maintenance operation is planned for a certain area of ​​the power distribution network.

[0231] After receiving the unstructured declaration text submitted by the document system, the text parsing module 101 of this system parses out that the target operating device is the sectionalizing switch K1 on the main feeder A line, the construction operation type is "crane material hoisting", and the originally planned construction time window is from 10:00 to 12:00 on the same day. Therefore, the duration parameter of this construction is determined to be 2 hours.

[0232] The spatial calculation module 102 intervenes and, based on the operation type "crane material hoisting," matches a mechanical operation safety radius of 6 meters from the preset dictionary. The system establishes a circular spatial interference area with a radius of 6 meters, centered on the geographical coordinates of sectionalizing switch K1. Through map ledger retrieval, the system finds that the downstream branch load switch K2 of feeder A and the switch K3 of another adjacent branch fall exactly within this 6-meter range. This indicates that during actual construction, due to the crane boom's swing radius, there is a high safety hazard for operators to travel to the locations of K2 and K3 to perform switching operations. Based on this, the system generates a physical operation interlock list and registers the equipment codes of K2 and K3.

[0233] The topology evaluation module 103 simulates disconnecting the target device K1 in the distribution network connectivity model, cutting off the normal power supply path of feeder A. As a result, an initial power outage island sub-diagram is formed downstream of K1. Through network connectivity checks, the system finds a normally open tie switch Tie1 at the end of the power outage island, the other end of which is connected to feeder B, which is normally powered externally. The system retrieves time-series load forecast data and finds that the peak load forecast data in the power outage island reaches 2.5 MW during the originally planned period from 10:00 to 12:00; while feeder B itself is heavily loaded during this period, and its remaining link capacity margin is only 1.5 MW.

[0234] Since the 2.5 MW load demand far exceeds the 1.5 MW capacity margin, at least 1.0 MW of load must be disconnected to meet electrical safety constraints. The topology assessment module 103 starts from tie switch Tie1 and works backward along the topology path into the power outage island to find a boundary switch suitable for load disconnection. The system calculation shows that opening branch load switch K2 would disconnect exactly 1.2 MW of load, meeting the capacity difference requirement. However, when the system checks the physical operation interlock list, it finds that K2 is already registered and is a prohibited device due to physical space constraints, directly rejecting this option. The system continues backtracking to another branch and finds that opening switch K3 would also meet the capacity requirement, but K3 is also affected by the crane's operating radius. There are no other compliant switch nodes along the line, triggering a topology deadlock event at this node. The underlying business situation is that during the originally planned time period, the power grid lacks sufficient capacity to handle the full load, and the site is obstructed by construction machinery, preventing the provision of a safe load disconnection interface.

[0235] Upon detecting a topology deadlock event, the scheduling optimization module 104 immediately initiated a time window reverse sliding optimization mechanism. The system uses a fixed 2-hour window parameter and performs a smooth search on the daily load curve in 30-minute increments. After multiple rounds of recalculation and comparison across time intervals, the system found that within the candidate time window of 14:30 to 16:30, the predicted local peak load of the power outage island had significantly decreased to 1.3 MW because the peak commercial electricity consumption in the area had passed; simultaneously, the remaining capacity margin of external feeder B had increased to 1.8 MW. Within this recommended construction time window, the external capacity was sufficient to fully absorb all the load of the power outage island, and the system did not need to perform any load shedding operations; simply closing the original tie switch Tie1 was sufficient to complete the transfer. This solution avoided the deadlock of having to operate the interference zone switches K2 or K3.

[0236] The decision output module 105 packages and compiles the target operating equipment K1, the affected associated power outage nodes, and the identified tie switch Tie1, and combines this with the updated time constraint of 14:30 to 16:30 to generate a final list of nodes within the power outage area. This list is then distributed to the mobile handheld terminals of the on-site work teams via the network interface. Dispatchers adjusted the work entry time based on this list, successfully completing the power outage maintenance and lossless load transfer without violating any safety operating procedures.

[0237] To verify the effectiveness and technical advantages of the system of the present invention in actual distribution network dispatching operations, this embodiment conducts a comparative experimental analysis based on historical real operation data of a city's distribution network.

[0238] See attached document Figure 12In the experimental verification phase, those skilled in the art selected 2,000 distribution network construction plan documents archived by a power supply company in a certain city in 2025 as the test sample set. The test data simultaneously covered the node ledger of the distribution network connectivity model for the corresponding region, geographic spatial coordinates, and load time-series prediction curves with a 15-minute resolution throughout the year. To form an effective comparison, the experiment introduced a conventional topology assessment strategy as a control group. The conventional topology assessment strategy only calculates capacity based on electrical topology connectivity and fixed time sections, lacking the ability to verify spatial physical interference and dynamically adjust the time dimension.

[0239] The experiment involved concurrent scheduling and simulation calculations for the aforementioned 2000 engineering samples. Statistical results showed that, under the conventional topology evaluation strategy, the boundary disconnect switches derived from the simulations in 284 samples were actually within the safe interference radius of the on-site construction machinery. Because the system failed to identify this spatial conflict, the output operational plans could not be safely executed on-site, resulting in a high rework rate. This system, through mapping analysis using the spatial calculation module 102, successfully intercepted all 284 defective plans with physical operation lockout risks.

[0240] Furthermore, for the intercepted flawed solutions, this system triggers a time-window reverse sliding optimization mechanism. For example... Figure 12 As the curve trend shows, during the regular fixed planned time window (usually concentrated in the morning peak period from 09:00 to 11:00), the remaining capacity margin of the external feeder links is at its lowest point throughout the day. If forced load reduction is implemented during this period, the average load to be cut off in a single construction operation is as high as 850 kilowatts, which greatly affects the reliability of regional power supply. This system uses the construction duration as a sliding window and performs equidistant traversal on the time axis. Figure 12 The optimization trajectory shows that the system automatically and smoothly shifts the operation time window to the off-peak electricity period after 14:30. During this time interval, the remaining capacity margin of the external link recovers significantly. Calculations after resetting the constraints show that the average load shedding amount per operation decreases to below 120 kW, with 215 samples achieving seamless power transfer with zero load shedding due to sufficient capacity.

[0241] Based on comprehensive test data, this system reduced the failure rate of solutions caused by dispatch deadlock by 92.6% while ensuring absolute safety of the on-site operating space, and significantly reduced the total amount of unplanned power outages caused by construction. The experimental data verifies the engineering practical value of introducing two-dimensional spatial physical constraints and dynamic optimization within a time window into electrical topology calculations, proving that the system can provide distribution networks with highly secure and executable automated power outage decision-making solutions.

[0242] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for predicting the range of power outages in construction projects based on artificial intelligence, characterized in that, Includes the following steps: Obtain the construction plan text and extract the target operation equipment identifier set, construction operation type, and planned construction time window from it; Based on the construction operation type and the two-dimensional geographic coordinates corresponding to the target operating equipment identifier set, a spatial interference region is constructed. Identify device nodes that fall into the spatial interference region but do not belong to the target operation device identifier set, and generate a physical operation lockout list; Based on the power distribution network connectivity diagram model, the target operating equipment identifier set and the physical operation lockout list, an initial power outage island sub-diagram and the corresponding edge tie switch set are deduced. Based on the planned construction time window, the peak load of the initial power outage island sub-map and the remaining link capacity margin of the edge tie switch set are evaluated; When the peak load is greater than the remaining capacity margin of the link, the required load shedding amount is calculated, and the legal operation boundary is searched in the distribution network connectivity model in combination with the physical operation block list. When no legal operation boundary satisfying the physical operation lockout list is found, a sliding optimization is performed on the planned construction time window to select a recommended construction time window, and the search for the legal operation boundary is re-executed based on the recommended construction time window; Based on the final determined legal operation boundaries and corresponding construction time windows, a list of nodes in the power outage area is generated and output to the document management terminal.

2. The method for predicting the power outage range of construction projects based on artificial intelligence according to claim 1, characterized in that, The step of obtaining the construction plan text and extracting the target operation equipment identifier set, construction operation type, and planned construction time window from it includes: The construction plan text is converted into a text character sequence, which is then input into a pre-trained named entity recognition model to output the labeled sequence result. Based on the location boundaries and category labels in the labeled sequence results, the corresponding text fragments are extracted and assembled into structured engineering scheduling parameters that include the target operating equipment identifier set, the construction operation type, and the planned construction time window. The start and end times of the planned construction time window are read to calculate the construction duration, which serves as a resource occupancy scale parameter in the time dimension.

3. The method for predicting the power outage range of construction projects based on artificial intelligence according to claim 1, characterized in that, The step of constructing a spatial interference region based on the construction operation type and the two-dimensional geographic coordinates corresponding to the target operating equipment identifier set includes: The mapping dictionary is retrieved according to the construction operation type to obtain the corresponding mechanical operation safety radius parameter; The two-dimensional geographic coordinates of each target operating device in the target operating device identifier set are obtained through a geographic information system. An independent two-dimensional spatial interference sub-region is constructed with each of the aforementioned two-dimensional geographic coordinates as the center and the aforementioned mechanical operation safety radius parameter as the radius. Perform a union operation on all two-dimensional spatial interference sub-regions to generate the spatial interference region.

4. The method for predicting the power outage range of construction projects based on artificial intelligence according to claim 1, characterized in that, The process of identifying device nodes that fall within the spatial interference region but do not belong to the target operation device identifier set, and generating a physical operation lockout list, includes: Extract the local area equipment ledger set that falls within the spatial interference region, and perform spatial collision detection between the geographical coordinates of each equipment node in the local area equipment ledger set and the spatial interference region; Nodes that collide with and belong to the target operating device identifier set are removed to obtain the spatial interference associated power outage node set; The equipment ledger codes of the spatial interference-related power outage node set are solidified into a structured form to generate the physical operation lockout list.

5. The method for predicting the power outage range of construction projects based on artificial intelligence according to claim 4, characterized in that, The initial power outage island sub-graph is derived based on the distribution network connectivity diagram model, the target operating equipment identifier set, and the physical operation interlock list, including: The target operating equipment identifier set is merged with the spatial interference associated power outage node set to obtain the composite isolation section node set. In the power distribution network connectivity model, the node state corresponding to the composite isolation section node set is set to disconnected to cut off the main power supply path; Starting from the downstream load side of the disconnected node, a breadth-first search algorithm is run to collect the nodes and branches that have lost power connection paths into the initial power-loss island sub-graph.

6. The method for predicting the power outage range of construction projects based on artificial intelligence according to claim 1, characterized in that, The evaluation of the peak load of the initial power-out island sub-graph and the remaining link capacity margin of the edge handshake switch set includes: Extract the time-series load data of each load node in the initial power outage island sub-map within the planned construction time window, and sum them up to obtain the maximum value within the time window as the peak load; Obtain the upper limit threshold of active power of the external live lines connected to the edge interconnection switch set, subtract the original load forecast value at the corresponding moment of the planned construction time window, extract the minimum value and accumulate it to obtain the remaining capacity margin of the link.

7. The method for predicting the power outage range of construction projects based on artificial intelligence according to claim 4, characterized in that, The calculation of the required load shedding amount, combined with the physical operation blockade list, and the search for legal operation boundaries in the distribution network connectivity model, includes: In the power distribution network connectivity graph model, starting from the set of edge interconnection switches, a backtracking traversal is performed into the initial power outage island subgraph. When the cumulative strippable load on the branch path reaches the required load removal amount, the segment equipment upstream of the current position is selected as a candidate boundary disconnect switch; If the device register code of the candidate boundary disconnect switch matches the physical operation lockout list, then the search continues beyond that node until a switch not registered in the physical operation lockout list is found and identified as the legal operation boundary.

8. The method for predicting the power outage range of construction projects based on artificial intelligence according to claim 7, characterized in that, When no valid operation boundary satisfying the physical operation lockout list is found, the step of performing sliding optimization on the planned construction time window includes: When the calculation along the path reaches a point where further retreat is not possible and the constraints of the physical operation lockout list are still not escaped, the time window sliding optimization mechanism is triggered. On the time axis of the preset periodic load forecast curve, perform an equidistant sliding search with a preset time step to generate multiple candidate time windows; Calculate the local peak load and link remaining capacity margin corresponding to each candidate time window, and update the corresponding required load shedding amount.

9. The method for predicting the power outage range of construction projects based on artificial intelligence according to claim 8, characterized in that, The step of selecting a recommended construction time window and re-performing the search for the legal operation boundaries based on the recommended construction time window includes: The candidate time window with the lowest required load removal amount and corresponding legal operation boundary is selected as the recommended construction time window; If there are multiple candidate time windows with the same required load shedding amount, the candidate time window with the fewest number of switches on the legal operation boundary and the time position closest to the planned construction time window shall be selected first. The recommended construction time window is used to update the business constraints, and the legal operation boundary is re-searched and established in the distribution network connectivity graph model.

10. The method for predicting the power outage range of construction projects based on artificial intelligence according to claim 9, characterized in that, The step of generating a list of nodes within the power outage area and outputting it to the document management terminal based on the finally determined legal operation boundaries and corresponding construction time windows includes: Perform a graph model vertex union operation on the target operating device identifier set, the spatial interference associated power outage node set, and the finally determined legal operating boundary, and incorporate the internal nodes enclosed by the boundary to form a comprehensive power outage node set; Convert the comprehensive set of power outage nodes into a relational data table to generate the list of nodes in the power outage area; The finalized construction time window is written into the time constraint field of the power outage area node network list and sent to the document management terminal.

Images