A power distribution network multi-resource integration planning method and system based on a graph neural network

Abstract

The present application relates to the technical field of power distribution network planning, and more particularly to a power distribution network multi-resource integrated planning method and system based on a graph neural network. The method comprises the following steps: collecting the operation and geographic data of the distribution transformers, feeder switches, 10kV feeders and each load access point in the target planning area to form a basic planning data set; constructing a power distribution network topology graph and using a graph neural network to identify overloaded feeder sections, low-voltage end sections and overloaded transformer nodes; then, according to the spatial position and power supply correlation of each problem area, determining the distribution transformer capacity expansion layout, tie switch addition and line extension reconstruction path; and finally, optimizing and outputting the target area planning scheme through power flow checking. The present application identifies the weak links of the power distribution network through a graph neural network and cooperatively optimizes the capacity expansion layout, tie transfer and line extension path, thereby achieving the precision, collaboration and economy of power distribution network planning and significantly improving the power supply capacity and overall operation reliability.

Description

Technical Field

[0001] This invention relates to the field of distribution network planning technology, and in particular to a multi-resource fusion planning method and system for distribution networks based on graph neural networks. Background Technology

[0002] During the operation of a distribution network, heavily loaded feeder sections, low-voltage end sections, and transformer overload nodes often coexist, and are spatially adjacent to each other and coupled in terms of power supply relationships. Existing planning methods typically address individual problems by increasing transformer capacity, adding switches, or upgrading lines, lacking the ability to conduct a comprehensive analysis based on the spatial distribution and power supply relationships of various problem points. This easily leads to fragmented planning measures, unreasonable resource allocation, and the possibility that localized solutions may still trigger new overload or low-voltage problems in adjacent areas, resulting in poor overall planning effectiveness. Summary of the Invention

[0003] Therefore, it is necessary to provide a method and system for multi-resource fusion planning of distribution networks based on graph neural networks to solve at least one of the above-mentioned technical problems.

[0004] To achieve the above objectives, a multi-resource fusion planning method for distribution networks based on graph neural networks is proposed. The upstream substation is electrically connected to a 10kV feeder, the 10kV feeder is electrically connected to a feeder switch, the 10kV feeder is electrically connected to the high-voltage side of the distribution transformer, and the low-voltage side of the distribution transformer is electrically connected to the load connection point. The method includes the following steps: Step S1: Obtain the equipment ledger data of the existing distribution transformers, feeder switches and 10kV feeders in the target planning area, and obtain the historical electricity consumption data, geographical location data and load forecast data of each load access point, and organize them to obtain the basic planning dataset; Step S2: Construct a distribution network topology based on the electrical connection relationship between the distribution transformer, 10kV feeder and load access point, and use a preset graph neural network based on the basic planning dataset to identify the feeder heavy load section, terminal low voltage section and transformer overload node in the distribution network topology. Step S3: Determine the planning results, including the location of distribution transformer capacity expansion, tie switch addition, and line extension and modification path, based on the spatial distribution location and power supply relationship of the feeder heavy load section, terminal low voltage section, and transformer overload node. Step S4: Perform a flow verification on the planning results to determine the planning scheme for the target area.

[0005] This invention also provides a graph neural network-based multi-resource fusion planning system for distribution networks, used to execute the graph neural network-based multi-resource fusion planning method for distribution networks described above. The graph neural network-based multi-resource fusion planning system for distribution networks includes: The data acquisition module is used to obtain the equipment ledger data of the existing distribution transformers, feeder switches and 10kV feeders in the target planning area, as well as the historical electricity consumption data, geographical location data and load forecast data of each load access point, and organize them to obtain the basic planning dataset. The topology identification module is used to construct a distribution network topology map based on the electrical connection relationship between distribution transformers, 10kV feeders and load access points, and to identify the feeder heavy-load sections, terminal low-voltage sections and transformer overload nodes in the distribution network topology map using a preset graph neural network based on the basic planning dataset. The planning and decision-making module is used to determine the planning results, including the location of distribution transformer capacity expansion, the location of tie switch addition, and the route of line extension and renovation, based on the spatial distribution of feeder heavy load sections, terminal low voltage sections, and transformer overload nodes and power supply relationships. The verification output module is used to perform power flow verification on the planning results in order to determine the planning scheme for the target area.

[0006] The beneficial effects of this invention lie in its ability to construct a basic data system for distribution network planning that integrates equipment ledger data, historical load data, geographical location data, and load forecast data, thereby possessing both electrical connection and spatial distribution attributes. Furthermore, it utilizes graph neural networks to identify and associate heavily loaded feeder sections, low-voltage terminal sections, and transformer overload nodes. Compared to traditional planning methods that rely on manual experience or single-index threshold judgments, this approach more accurately uncovers the topological transmission relationships and load coupling relationships between distribution network nodes, improving problem location accuracy. Further, this invention performs overlapping scanning of the power transfer range of heavily loaded feeder sections and transformer overload nodes to form a load conflict superposition area. Based on load release priorities, it dynamically determines the locations for increasing distribution transformer capacity and adding tie switches, enabling feeder... The coordinated implementation of load shedding and distribution station load shedding avoids the problems of localized improvement and overall imbalance caused by single-point renovations in traditional planning. Simultaneously, by starting with the low-voltage section at the end and ending with the remaining capacity node, impedance reduction path tracing is performed. This comprehensively considers tower conditions, road corridor conditions, and available channel conditions to generate line extension and renovation paths, effectively reducing the length of new lines, construction difficulty, and overall investment costs. Furthermore, this invention updates the load distribution of adjacent areas in real time after each planning measure is completed and cyclically executes subsequent planning decisions, enabling the planning process to have dynamic iterative optimization capabilities. This significantly improves the overall power supply capacity, voltage quality, and network operation economy of the target area, thereby achieving coordinated planning for distribution network expansion, transfer, reinforcement, and end-point management. It has high engineering practical value and promotional value. Attached Figure Description

[0007] Figure 1 This is a flowchart illustrating the steps of a multi-resource fusion planning method for power distribution networks based on graph neural networks. Figure 2 This is a diagram showing the electrical connection relationships of the distribution network resources in a target planning area, as illustrated in one embodiment. Figure 3 This is a schematic diagram of a segment-by-segment nested process in one embodiment; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0008] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0009] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0010] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0011] To achieve the above objectives, please refer to Figures 1 to 3 A multi-resource fusion planning method for distribution networks based on graph neural networks is proposed. The upstream substation is electrically connected to a 10kV feeder, the 10kV feeder is electrically connected to a feeder switch, the 10kV feeder is electrically connected to the high-voltage side of the distribution transformer, and the low-voltage side of the distribution transformer is electrically connected to the load connection point. The method includes the following steps: Step S1: Obtain the equipment ledger data of the existing distribution transformers, feeder switches and 10kV feeders in the target planning area, and obtain the historical electricity consumption data, geographical location data and load forecast data of each load access point, and organize them to obtain the basic planning dataset; Step S2: Construct a distribution network topology based on the electrical connection relationship between the distribution transformer, 10kV feeder and load access point, and use a preset graph neural network based on the basic planning dataset to identify the feeder heavy load section, terminal low voltage section and transformer overload node in the distribution network topology. Step S3: Determine the planning results, including the location of distribution transformer capacity expansion, tie switch addition, and line extension and modification path, based on the spatial distribution location and power supply relationship of the feeder heavy load section, terminal low voltage section, and transformer overload node. Step S4: Perform a flow verification on the planning results to determine the planning scheme for the target area.

[0012] In one embodiment, the power distribution network of a new eastern district of a city can be used as the target planning area for illustration and reference. Figure 2 The target planning area consists of an upstream substation, 10kV feeders, feeder switches, distribution transformers, and load connection points. The upstream substation is the 110kV Dongcheng substation belonging to the power grid. The Dongcheng substation supplies power to the new area through six 10kV feeders. Each 10kV feeder is laid along the municipal roads and connected to various branch lines via feeder switches. At the end of each branch line, multiple distribution transformers are connected. The low-voltage side of the distribution transformers connects to residential users, commercial users, and new energy vehicle charging loads.

[0013] In this embodiment, the target planning area is approximately 12 square kilometers, including 64 distribution transformers, 38 feeder switches, and 1,820 low-voltage load access points. The southwest area is a newly developed residential area, the eastern area is a commercial area, and the southern area has a centralized charging station for new energy vehicles. The overall load growth rate in recent years is approximately 11.6%.

[0014] The project acquires equipment ledger data for existing distribution transformers, feeder switches, and 10kV feeders within the target planning area, along with historical electricity consumption data, geographical location data, and load forecast data for each load connection point. This data is then compiled into a basic planning dataset. Specifically, the equipment ledger data includes: distribution transformer capacity, model, commissioning time, and historical load rate; 10kV feeder conductor model, line length, and rated current carrying capacity; feeder switch installation location, section attributes, and operating status; and outgoing line capacity information from the upstream substation.

[0015] For example, the No. 27 distribution transformer has a capacity of 800kVA and a maximum load rate of 96% in the past year; the No. 5 feeder has a total length of 8.7km and a rated current carrying capacity of 320A.

[0016] Historical electricity consumption data includes electricity consumption curves at 15-minute intervals for each load access point over the past three years, used to reflect the characteristics of residential evening peak hours, commercial midday peak hours, and the high fluctuation characteristics of charging stations at night.

[0017] The geographic location data was obtained using the AutoNavi map coordinate system, including road corridor locations, equipment coordinate locations, land development boundaries, and cable laying channel information.

[0018] The load forecast data is based on urban construction planning and application information. It is predicted that the maximum load in the region will increase from 31.2MW to 42.5MW in the next three years, with the southwest region having the highest growth rate of 23%. Based on the above data, a basic planning dataset is formed.

[0019] The distribution network topology is constructed based on the electrical connection relationships between distribution transformers, 10kV feeders and load access points. Based on the basic planning dataset, a pre-set graph neural network is used to identify the feeder overload sections, terminal low-voltage sections and transformer overload nodes in the distribution network topology.

[0020] Specifically, the upstream substation, feeder node, feeder switch node, distribution transformer node, and load access point node are used as nodes in the graph structure, and the cable connection relationship, switch connection relationship, and upstream and downstream power supply relationship are used as edges in the graph structure, thus constructing a distribution network topology graph containing 1927 nodes and 2146 edges.

[0021] The load rate, voltage deviation, historical growth rate, peak-to-valley difference, adjacent power flow contribution, and spatial density of each node are used as node features and input into a preset graph neural network model. The topological correlation features are extracted through multi-layer neighborhood aggregation operations, and the risk identification results are output.

[0022] The identification results indicate that there is a heavily loaded section in the middle section of feeder line 5, with a predicted peak summer load rate of η=108%. There is a low-voltage section at the end of feeder No. 3 in the Cuihuyuan residential area, with the predicted minimum user-end voltage being U=198V; there is an overload node at distribution transformer No. 27, with the predicted maximum operating load rate reaching L=121%.

[0023] Based on the spatial distribution and power supply relationship of the feeder heavy load section, the terminal low voltage section and the transformer overload node, the planning results include the location of distribution transformer capacity expansion, the location of tie switch addition, and the route of line extension and renovation. Specifically, a comprehensive decision was made based on the distance relationship between each abnormal point and the existing network, the transfer relationship, and the road conditions: (1) For the overload problem of the No. 27 distribution transformer, its capacity was increased from 800kVA to 1250kVA, and a 630kVA box-type transformer was added in the center of the newly built residential area in the southwest to share the increased load of residents; (2) For the overload problem of feeder No. 5, a tie switch was added at the intersection of feeder No. 5 and feeder No. 6, so that part of the load of feeder No. 5 could be switched to feeder No. 6 for power supply; (3) For the low voltage problem at the end, 1.6km of 10kV cable line was laid along the planned road, and part of the load in the southwest area was changed from the original end power supply method to the nearest feeder No. 6. Based on this, a comprehensive planning result including capacity expansion, tie switch layout and line extension path was formed.

[0024] The planning results were verified by power flow analysis to determine the planning scheme for the target area. Specifically, the planning results were imported into the distribution network power flow calculation model and verified according to the predicted maximum load scenario for the next three years. The results showed that: the maximum feeder load rate decreased from 108% to 79%; the minimum terminal voltage increased from 198V to 221V; the load rate of distribution transformer No. 27 decreased from 121% to 73%; and each feeder met the power transfer requirements under single-circuit fault conditions. Therefore, the above planning results were determined as the formal planning scheme for the target area.

[0025] Preferably, step S2, which uses a pre-defined graph neural network based on the basic planning dataset to identify the feeder overload sections, terminal low-voltage sections, and transformer overload nodes in the distribution network topology, includes: A distribution network topology map is constructed based on the equipment connection relationships in the basic planning dataset, and the load rate, capacity margin and terminal voltage parameters of each node are extracted as graph feature data. The graph feature data is input into a preset graph neural network, and the operating features of adjacent nodes are transmitted step by step according to the power supply path direction to generate the associated evaluation values ​​of each feeder section and each distribution transformer. The feeder heavy load sections are determined based on the associated evaluation values ​​of each feeder section, the terminal low voltage sections are determined based on the associated evaluation values ​​of the terminal nodes, and the transformer overload nodes are determined based on the associated evaluation values ​​of each distribution transformer.

[0026] In one embodiment, the power distribution network of the eastern new district of a certain city is still used as the target planning area for explanation. After completing the basic planning dataset, a preset graph neural network is used to identify the feeder overload sections, terminal low-voltage sections, and transformer overload nodes in the power distribution network topology map based on the basic planning dataset.

[0027] Specifically, a distribution network topology is first constructed based on the equipment connection relationships in the basic planning dataset. The six 10kV outgoing lines of the 110kV Dongcheng substation are used as source nodes, and the feeder segment nodes, feeder switch nodes, distribution transformer nodes, and low-voltage load access point nodes are used as network nodes. Actual cable connections, switch connections, and upstream / downstream power supply relationships are used as graph edges to form a distribution network topology with power supply direction attributes. In this embodiment, a total of 1927 nodes and 2146 connection edges are generated.

[0028] After the topology graph is constructed, the operational characteristics of each node are further extracted as graph feature data. The load rate of each node is obtained as the ratio of real-time load power to rated capacity; for example, the current load rate of distribution transformer No. 27 is 96%. Capacity margin is calculated as the ratio of remaining available capacity to rated capacity; for example, feeder No. 5 has a rated current carrying capacity of 320A and a current of 286A, so the capacity margin is 10.6%. Terminal voltage parameters are obtained from smart meters at the feeder terminals; for example, the voltage at the Cuihuyuan terminal of feeder No. 3 is 205V. In addition, historical growth rate, peak-valley difference coefficient, and seasonal fluctuation factor can be introduced as auxiliary features to form a graph feature data matrix.

[0029] Subsequently, the graph feature data is input into a pre-defined graph neural network model. This model employs a multi-layer neighborhood aggregation structure, transmitting features step-by-step along the power supply path. Specifically, it propagates operational status information from upstream substation nodes to feeder nodes, then to distribution transformer nodes, and finally to end-load nodes, while simultaneously aggregating downstream load pressure and end-voltage attenuation information in the reverse direction. Through multiple rounds of iterative calculations, each feeder section, distribution transformer, and end-node obtains a correlation evaluation value reflecting the upstream and downstream coupling relationship.

[0030] For example, the model output results show that: the correlation evaluation value of a certain section in the middle of feeder No. 5 reached 0.93, indicating that the section was significantly affected by the combined effects of upstream power supply pressure and downstream concentrated load; the correlation evaluation value of the Cuihuyuan node at the end of feeder No. 3 reached 0.89, indicating that the voltage risk at the end was relatively high; and the correlation evaluation value of distribution transformer No. 27 reached 0.95, indicating that it was continuously operating under high load and had an overload trend.

[0031] Furthermore, the feeder heavy-load sections are determined based on the correlation evaluation values ​​of each feeder section. When the correlation evaluation value of a feeder section is higher than a preset threshold of 0.85, and the predicted load rate exceeds 95% of the rated capacity, it is determined to be a feeder heavy-load section. In this embodiment, approximately 1.4 km of the middle section of feeder No. 5 is identified as a heavy-load section, with a predicted peak summer load rate of 108%.

[0032] Low-voltage sections at the end of a line are determined based on the correlation evaluation value of the end node. When the correlation evaluation value of the end node is higher than a preset threshold of 0.80 and the predicted user-end voltage is lower than 210V, its location is determined to be a low-voltage section at the end of the line. In this embodiment, the Cuihuyuan residential area at the end of feeder line 3 is identified as a low-voltage section at the end of the line, with a predicted minimum user-end voltage of 198V.

[0033] Transformer overload nodes are determined based on the correlation evaluation values ​​of each distribution transformer. When the correlation evaluation value of a distribution transformer exceeds a preset threshold of 0.90, and the predicted load rate exceeds 100% of the rated capacity, it is identified as a transformer overload node. In this embodiment, distribution transformer No. 27 is identified as an overload node, with a predicted maximum operating load rate reaching 121%.

[0034] As an example of the present invention, reference is made to Figure 2 As shown, step S3 in this example includes: Step S31: Overlap scan the adjacent feeder overloaded sections and transformer overload nodes according to the power supply transfer range to form load conflict superposition areas, and calculate the load release priority of each load conflict superposition area; Step S32: Search the remaining capacity nodes around the corresponding load conflict superposition area in sequence according to the load release priority, determine the location of the distribution transformer capacity expansion or the location of the tie switch addition, so that the released load amount and the load to be transferred are dynamically matched. Step S33: Starting from the low-voltage section at the end and ending at the determined remaining capacity node, perform impedance reduction path finding to determine the line extension and renovation path; Step S34: After completing the location of each distribution transformer capacity expansion, tie switch addition, or line extension modification, update the load distribution of adjacent areas in real time, and repeat steps S32 and S33 until the overall planning result of the target planning area is obtained.

[0035] In one embodiment, the power distribution network of the eastern new district of a certain city is still used as the target planning area for explanation. After identifying the heavy-load section of feeder No. 5, the low-voltage section at the end of feeder No. 3, and the overload node of distribution transformer No. 27, the overall planning result of the target planning area is further generated based on their spatial distribution and power supply relationship.

[0036] Specifically, the overloaded sections of adjacent feeders and the overloaded nodes of transformers are first scanned over the same area according to the power supply transfer range to form a load conflict superposition area. The so-called power supply transfer range refers to the area where the load can be redistributed through transfer, capacity expansion, or new line construction under the constraints of the existing switch structure, line channel conditions, and allowable power supply radius. In this embodiment, after spatial superposition analysis of the overloaded section in the middle of feeder No. 5 and the power supply area where distribution transformer No. 27 is located, it was found that the two have a common influence range of about 1.8 square kilometers in the southwest residential development area. The residential load in this area increases significantly during the evening peak, forming the main load conflict superposition area.

[0037] Subsequently, the load release priority was calculated for each area with overlapping load conflicts. The load release priority was determined by comprehensively considering factors such as the current overload level, load growth rate, user importance level, and the difficulty of implementing modifications. For example, the southwest residential area was assigned a first-level load release priority because the predicted load rate of distribution transformer No. 27 reached 121%, and new residents would be connected within a year. Although the eastern commercial area experienced localized peak loads, these could be alleviated through staggered operation, so its load release priority was lower than that of the southwest area.

[0038] After determining the load release priority, the remaining capacity nodes around the corresponding load conflict superposition area are searched sequentially according to the priority order. Remaining capacity nodes include distribution transformer nodes with remaining power supply margin, nearby feeder connection points, and nodes where additional tie switches can be added. In this embodiment, the load rate of the southern section of feeder 6 in the southwest area was found to be only 62%, indicating a relatively high remaining capacity. Furthermore, there are reserved locations for ring main unit installations along the planned road; therefore, this area was prioritized as the load receiving node.

[0039] Based on the remaining capacity node, the locations for increasing the capacity of distribution transformers or adding tie switches will be further determined to dynamically match the released load with the load to be transferred. Specifically, the capacity of distribution transformer No. 27 will be increased from 800kVA to 1250kVA, which can release approximately 320kW of the original overload pressure; at the same time, a tie switch will be added at the intersection of feeder No. 5 and feeder No. 6 to transfer approximately 1.1MW of residential load in the southwest area to feeder No. 6. The released load and the load to be transferred will be calculated in real time, and when the two are close to balance, further capacity expansion in this area will be stopped.

[0040] Regarding low-voltage management at the end of the line, the impedance reduction path was selected starting from the Cuihuyuan low-voltage section of feeder No. 3 and ending at the remaining capacity node of feeder No. 6. During the path selection process, factors such as road corridor, cable channel length, number of bends along the route, and total line impedance were considered, prioritizing paths with gradually decreasing total impedance. In this embodiment, a new 10kV cable line of approximately 1.6km in length was ultimately determined to be laid along the planned road in the new area, switching some of the loads originally powered by the end of feeder No. 3 to the near end of feeder No. 6, thereby reducing line voltage drop and improving the voltage level at the end of the line.

[0041] After completing the location of each distribution transformer capacity expansion, tie switch installation, or line extension modification, the load distribution of adjacent areas is updated in real time. For example, after the capacity expansion of distribution transformer No. 27 and the commissioning of the tie switch, the load rate of the middle section of feeder No. 5 decreased from 108% to 86%, and the load rate of feeder No. 6 increased from 62% to 74%; after the new line was connected, the minimum voltage at the end of feeder No. 3 increased from 198V to 218V. Subsequently, the remaining capacity and demand for improvement in the remaining areas are recalculated, and the remaining capacity search and line routing process continues until there are no new heavy load, low voltage, or overload risk points in the entire area.

[0042] The final overall planning results for the target planning area include: increasing the capacity of distribution transformer No. 27 to 1250kVA, adding one 630kVA box-type transformer in the southwest residential area, adding one tie switch between feeder No. 5 and feeder No. 6, and laying 1.6km of new 10kV cable lines.

[0043] Preferably, the overlapping scan of adjacent feeder overload sections and transformer overload nodes according to the power supply transfer range is specifically as follows: Using the first and last switch nodes of each feeder heavy-load section as boundaries, the transferable power supply radius is progressively extended to both sides along the power supply path to form the corresponding feeder unloading scan zone; Centered on each transformer overload node, the corresponding transformer shunt scanning zone is formed by expanding outwards in stages according to the low-voltage outgoing line direction and the connection relationship between adjacent transformer areas. The feeder load release scanning zone and the transformer load detachment scanning zone are overlapped segment by segment according to their spatial location, and the difference in the acceptable capacity of the two is compared simultaneously to screen out the overlapping sections that simultaneously meet the path connectivity condition and the capacity acceptance condition. Boundary splicing is performed on consecutive adjacent overlapping sections to form a load conflict superposition area.

[0044] In one embodiment, the power distribution network of the eastern new district of a certain city is still used as the target planning area for explanation. After the heavy-load section of feeder No. 5 and the overload node of distribution transformer No. 27 have been identified, in order to accurately determine the key areas for load transfer, the heavy-load sections of adjacent feeders and the overload nodes of transformers are further overlapped according to the range of power transfer to form a load conflict superposition area.

[0045] Specifically, the transferable power supply radius is first extended outwards along the power supply path from the first and last switch nodes of each feeder's heavy-load section as boundaries, forming a corresponding feeder load release scanning zone. In this embodiment, the heavy-load section of feeder No. 5 is located between switch K12 and switch K18, with a line length of approximately 1.4 km. Using switch nodes K12 and K18 as boundary points, the scan is carried out segment by segment along the upstream and downstream feeder directions, respectively. Taking into account the conductor current carrying capacity, allowable transfer distance, and the operational feasibility of switches along the line, the transferable power supply radius is set to 0.9 km upstream and 1.2 km downstream, thus forming a feeder load release scanning zone covering the southwest residential area and adjacent road corridors.

[0046] Subsequently, with each transformer overload node as the center, the load balancing scan zone is expanded outward in stages according to the direction of the low-voltage outgoing lines and the connection relationships between adjacent transformer areas, forming a corresponding transformer load balancing scan zone. In this embodiment, distribution transformer No. 27 is located in the center of the southwest residential development area, with a total of 4 outgoing lines on the low-voltage side, respectively connecting to residential buildings, public service facilities, and underground parking garage charging loads. Centered on distribution transformer No. 27, the load balancing scan zone is expanded outward by 300m along the direction of each of the 4 low-voltage outgoing lines, and combined with the low-voltage connection points already reserved between adjacent transformer areas No. 26 and No. 28, the primary load-bearing range is further expanded to the surrounding area, forming a transformer load balancing scan zone covering the surrounding residential plots.

[0047] Furthermore, the feeder load release scanning zone and the transformer load detachment scanning zone are overlapped segment by segment according to their spatial location, and the difference in their acceptable capacity is compared simultaneously to screen out overlapping sections that simultaneously meet the path connectivity condition and the capacity acceptance condition. The path connectivity condition is used to determine whether the existing road corridor, cable corridor, or switch connection relationship supports load transfer; the capacity acceptance condition is used to determine whether the adjacent feeder or transformer substation has the margin to accept the new load. In this embodiment, after overlap analysis, three overlapping sections are formed along the planned road on the south side of the southwest area, with a total length of approximately 860m. Among them, the section near the access point of feeder branch No. 6 has a remaining capacity of approximately 1.35MW, which can accept the load released by distribution transformer No. 27 and part of the load transferred by feeder No. 5, and is therefore determined to be an effective overlapping section.

[0048] Subsequently, the boundaries of consecutive overlapping sections were spliced ​​together to form a load conflict superposition area. The three effective overlapping sections were then spliced ​​and integrated according to road continuity and power supply channel integrity, ultimately forming a load conflict superposition area of ​​approximately 1.8 square kilometers. This area is simultaneously affected by the heavy load pressure of feeder No. 5 and the overload pressure of distribution transformer No. 27, and has the engineering conditions for power transfer and capacity expansion via feeder No. 6; therefore, it was identified as a priority area for remediation.

[0049] Preferably, the methods for obtaining path connectivity conditions and capacity carrying capacity conditions include: Extract the current line routing, switch connection point location, and reserved channel data between the feeder unloading scan zone and the transformer shunt scan zone to determine whether there is a through-path power supply connection between the two scan zones; if there is a through-path power supply connection, then the path connectivity condition is met. Calculate the remaining power supply capacity, release load capacity, and conductor current carrying margin at both ends of the passable power supply connection path to form a capacity carrying parameter set; The difference between the remaining power supply capacity and the releaseable load capacity in the capacity acceptance parameter set is calculated, and the conductor current carrying margin is compared with the preset safety margin threshold. When the remaining power supply capacity is not less than the releaseable load capacity, and the conductor current carrying margin is greater than or equal to the preset safety margin threshold, the capacity acceptance condition is determined to be met.

[0050] In one embodiment, the power distribution network of the eastern new district of a certain city is still used as the target planning area for explanation. After forming the load release scanning zone of feeder No. 5 and the load shunting scanning zone of distribution transformer No. 27, in order to accurately screen the effective overlapping sections where load transfer can be implemented, the path connectivity conditions and capacity acceptance conditions are further determined.

[0051] Specifically, the existing line routing, switch connection point locations, and reserved passage data between the feeder unloading scan zone and the transformer shunt scan zone are first extracted to determine whether there is a passable power supply connection path between the two scan zones. The existing line routing data comes from the power distribution GIS system and includes 10kV cable corridors, overhead line paths, and road laying directions; the switch connection point location data includes the locations of existing ring main units, sectionalizing switches, and potentially retrofitted disconnectors; and the reserved passage data includes underground integrated pipe corridors in the new area, cable duct resources, and available construction passages in road green belts.

[0052] In this embodiment, data extraction revealed that the southern boundary of the load release scanning zone of feeder 5 is approximately 420m from the nearest point of feeder branch 6. A double-sided cable duct has been constructed along the planned road between them, and a foundation for an expandable ring main unit is located there. The northern side of the load shunting scanning zone of transformer 27 is directly adjacent to this road corridor. Therefore, it is determined that there is a continuous and penetrable power supply connection path between the load release scanning zone of feeder 5, the load shunting scanning zone of transformer 27, and the receiving node of feeder 6, satisfying the path connectivity condition.

[0053] Subsequently, the remaining power supply capacity, releaseable load capacity, and conductor current carrying margin of the nodes at both ends of the connectable power supply connection path are calculated to form a capacity acceptance parameter set. Among them, the remaining power supply capacity refers to the available margin of the receiving node after meeting the existing load; the releaseable load capacity refers to the load that needs to be transferred or distributed in the area to be treated; and the conductor current carrying margin refers to the remaining current carrying space of the conductor in the connection path after deducting the current operating current from the rated current carrying capacity of the conductor.

[0054] In this embodiment, the current maximum load of the southern section of feeder No. 6 is 7.4MW, and the allowable operating capacity is 8.8MW, so the remaining power supply capacity is 1.4MW; the total load to be transferred from the heavy-load section of distribution transformer No. 27 and feeder No. 5 is about 1.18MW; the proposed new connecting line adopts YJV22-8.7 / 15kV-3×300 cable with a rated current carrying capacity of about 430A, and the current equivalent operating current is about 265A, corresponding to a conductor current carrying capacity margin of about 165A, thus forming a capacity acceptance parameter set.

[0055] Furthermore, the difference between the remaining power supply capacity and the releaseable load capacity in the capacity acceptance parameter set is calculated, and the conductor current carrying capacity is compared with the preset safety margin threshold. In this embodiment, the remaining power supply capacity of 1.4MW minus the releaseable load capacity of 1.18MW yields a capacity difference of 0.22MW, which is greater than zero; at the same time, the conductor current carrying capacity of 165A is higher than the preset safety margin threshold of 80A. Therefore, it is determined that the receiving side can not only accept the load to be transferred, but also that the line still maintains sufficient operational safety margin.

[0056] When the remaining power supply capacity is not less than the releaseable load capacity, and the conductor current carrying capacity is greater than or equal to the preset safety margin threshold, the capacity acceptance condition is determined to be met. Therefore, in this embodiment, the connection path simultaneously meets the path connectivity condition and the capacity acceptance condition, and is identified as an effective load transfer channel, and is included in the priority planning scope for subsequent tie switch addition locations and line extension and modification paths.

[0057] Preferably, the feeder load release scanning band and the transformer load detachment scanning band are interleaved segment by segment according to their spatial positions, including: The geographical coordinate order of the feeder load release scanning zone and the transformer load detachment scanning zone is determined based on geographical location data; The boundary points of each strip segment of the feeder load release scanning zone and the transformer load shunting scanning zone are extracted in order of geographic coordinates, and the corresponding feeder segment sequence and load shunting segment sequence are obtained. Taking the first segment of the feeder segment sequence as a reference, the sequence moves segment by segment along the power supply extension direction, and simultaneously calls the adjacent off-load segment sequence for corresponding comparison: when the projection range of the feeder segment and the off-load segment intersects or is adjacent, the overlapping segment information of the corresponding position is recorded, and the comparison continues to the next segment. Adjacent segments that have continuously obtained overlapping segment information are overlapped one by one.

[0058] In one embodiment, the power distribution network of the eastern new district of a certain city is still used as the target planning area for explanation. This can be referred to... Figure 3 After the No. 5 feeder load release scanning zone and the No. 27 distribution transformer load shunting scanning zone have been formed, in order to accurately identify the spatially overlapping areas between the two that have collaborative governance value, the feeder load release scanning zone and the transformer load shunting scanning zone are further overlapped segment by segment according to their spatial positions to determine the subsequent load conflict superposition area.

[0059] Specifically, the geographical coordinate order of the feeder load release scanning zone and the transformer shunt scanning zone is first determined based on geographical location data. The geographical location data adopts the coordinate system of the New Area Power Distribution GIS Platform and is uniformly converted into planar projected coordinates. The feeder load release scanning zone of No. 5 is sorted from west to east along the line direction, and the transformer shunt scanning zone of No. 27 is sorted from the center to the periphery according to the extension direction of the low-voltage outgoing line, thus forming a comparable spatial coordinate order relationship.

[0060] In this embodiment, the load shedding scan zone of feeder No. 5 extends along Jinqiao Avenue in a northeast-southwest direction, with a total length of approximately 2.1 km; the load shunting scan zone of distribution transformer No. 27 is distributed in a fan shape around the center of the southwest residential area, mainly covering the north and south residential clusters. After coordinate sorting, the two scan zones can be uniformly mapped to the same geographic reference plane for segment-by-segment analysis.

[0061] Subsequently, the boundary points of each strip segment of the feeder load release scanning zone and the transformer load balancing scanning zone were extracted sequentially using geographic coordinates, resulting in the corresponding feeder segment sequence and load balancing segment sequence. Specifically, the feeder load release scanning zone was discretized into 11 feeder segments, F1 to F11, based on a length of 200m. Simultaneously, the transformer load balancing scanning zone was divided into 8 load balancing segments, T1 to T8, based on the influence range of low-voltage outgoing lines and road / street boundaries. The coordinates of the start and end boundary points, width range, and power supply coverage direction were recorded for each segment.

[0062] For example, feeder segment F4 is located near the intersection of Jinqiao Avenue and Cuihu Road, covering an area of ​​about 60m on both sides of the road; sectionalizing segment T3 corresponds to the first phase of the Southwest Residential Area, with its boundary about 180m from the No. 27 distribution transformer.

[0063] Furthermore, taking the first segment of the feeder segment sequence as a reference, the sequence moves segment by segment along the power supply extension direction, and simultaneously calls adjacent load segment sequences for corresponding comparison. When the projection ranges of the feeder segment and the load segment intersect or are adjacent, the overlapping section information at the corresponding position is recorded, and the comparison continues to the next segment.

[0064] In this embodiment, the scan proceeds downstream segment by segment, starting from feeder segment F1. When the scan reaches F4, its projection range is adjacent to the boundary of the load distribution segment T2, with a minimum distance of only 18m, so the first overlapping segment is recorded. When the scan reaches F5 and F6, it substantially intersects and covers load distribution segments T3 and T4 respectively, with intersection widths reaching 45m and 63m, and is recorded as a continuous overlapping segment. When the scan reaches F7, the distance to the subsequent load distribution segment exceeds 120m, no longer meeting the adjacent condition, so the recording of this continuous segment is stopped.

[0065] Subsequently, adjacent segments with continuously obtaining overlapping section information are sequentially overlaid. Specifically, the adjacent overlapping sections formed by F4-T2, F5-T3, and F6-T4 are spliced ​​and integrated according to road continuity and power supply accessibility to form a continuous overlay strip with a total length of approximately 610m and a width between 45m and 70m. This continuous overlay strip is simultaneously within the load release capacity coverage of feeder No. 5 and the load shunting demand of distribution transformer No. 27, thus meeting the conditions for serving as a load transfer channel and a new interconnection node deployment area.

[0066] Preferably, step S33 includes: Extract the location of existing poles and towers, road corridors and available channels around the low-voltage section at the end, establish the starting path search domain, and set the identified remaining capacity nodes as target nodes; The initial path impedance value is calculated based on the line length, number of turns, and channel occupancy of each candidate extension direction within the starting path search domain, forming a candidate path set; High-impedance path branches are eliminated segment by segment according to the path impedance value from large to small, and the incremental impedance value of the subsequent segment is superimposed on the remaining path in real time to perform impedance-decreasing path finding. When a candidate path is connected to the target node, the comprehensive impedance values ​​of each connected path are compared, and the connected path with the smallest comprehensive impedance value is selected as the line extension and renovation path.

[0067] In one embodiment, the power distribution network of the eastern new district of a certain city is still used as the target planning area for explanation. After identifying a low-voltage section at the end of feeder No. 3 in the Cuihuyuan residential area and determining that the southern branch node of feeder No. 6 has remaining capacity, in order to solve the problem of low voltage at the end users, the impedance reduction path finding method is further adopted to determine the line extension and renovation path.

[0068] Specifically, the locations of existing power poles, road corridors, and available access routes around the low-voltage section at the end of the line were first extracted to establish a starting path search domain, and the identified remaining capacity nodes were set as target nodes. The low-voltage section at the end of the line is located east of the Cuihu Garden residential area and is currently powered by the end of feeder line 3. During the summer evening peak, the lowest voltage at the user end drops to 198V. After retrieving GIS data, the locations of 12 existing power poles, 4 municipal road corridors, and available access routes such as underground cable ducts and reserved passages in green belts were obtained.

[0069] In this embodiment, the power distribution access boundary of Cuihu Garden residential area is taken as the starting point, and a path search domain is established within a radius of 1.8km. At the same time, the southern node of feeder No. 6, which has a remaining power supply capacity of 1.4MW, is set as the target node to take over the new load after the line extension.

[0070] Subsequently, the initial path impedance values ​​are calculated based on the line length, number of turns, and channel occupancy of each candidate extension direction within the starting path search domain, forming a candidate path set. Specifically, multiple candidate paths are generated from the starting point in the three main road directions: north, west, and south. Each path is assigned a comprehensive impedance parameter; the longer the line length, the more turns, and the tighter the channel occupancy, the higher the impedance value. For example: Path A runs south along Cuihu Road and connects to Jinqiao Avenue, with a total length of 1.62km and 3 turns; some sections require shared trenches with communication pipelines. Path B runs directly along the western side of the residential area to a planned branch road, with a total length of 1.34km and 2 turns; the road channel is relatively wide. Path C detours along the northern green belt and connects to the main road, with a total length of 1.91km and 4 turns. After calculation, the initial impedance value for path A is 8.6, for path B it is 6.2, and for path C it is 10.1, thus forming the candidate path set.

[0071] Furthermore, high-impedance path branches are eliminated segment by segment according to their path impedance values, from largest to smallest. For the remaining paths, incremental impedance values ​​from subsequent segments are added in real time for impedance-decreasing path finding. High-impedance branches corresponding to path C are eliminated first, while paths A and B continue to extend forward. When the path continues to extend through intersections, narrow passages, or new corner locations, incremental impedance values ​​are added in real time; when entering wide, straight sections, a low incremental impedance state is maintained.

[0072] In this embodiment, when path A passes through the existing stormwater and sewage pipe network at the south entrance of Jinqiao Avenue, the impedance value increases by 1.9; path B only increases the impedance value by 0.7 in the planned branch road section on the west side. Therefore, path B continues to maintain the advantage of lower overall impedance.

[0073] Subsequently, when the candidate paths connect to the target node, the overall impedance values ​​of each connecting path are compared, and the connecting path with the lowest overall impedance value is selected as the line extension and renovation path. It was ultimately confirmed that both path A and path B can connect to the target node in the southern section of feeder 6. Path A has an overall impedance value of 11.3, while path B has an overall impedance value of 7.5. Therefore, path B is selected as the optimal line extension and renovation path.

[0074] The final determined route for the line extension and renovation is as follows: Starting from the power distribution access point on the west side of the Cuihu Garden residential area, approximately 1.34 km of 10kV cable will be laid along the planned road on the west side, connecting to the ring main unit node on the south section of feeder No. 6. After implementation, some loads in the Cuihu Garden residential area will be supplied from the end of feeder No. 3 to the near end of feeder No. 6. The minimum voltage at the end is expected to increase from 198V to 221V, while simultaneously reducing the voltage drop at the end of feeder No. 3, thus achieving the goal of addressing low-voltage sections at the end of the line.

[0075] Preferably, data on the location of existing power poles, road corridors, and available passages around the low-voltage section at the end are extracted to establish the starting path search domain, including: Read the coordinate data of existing poles and towers within a preset range around the end low-voltage section, and form a set of candidate pole and tower points according to their distance from the end low-voltage section; Extract road centerline data, road boundary data, and roadside laying condition data within a preset range to generate a road corridor strip area; Data on open ground passages, existing power lines, and shared pole passages within a preset range are obtained, and building encroachment areas, prohibited construction areas, and obstruction areas are eliminated to form a set of usable passages; The candidate tower point set, the road corridor strip area and the available passage set are spatially superimposed to generate several passable path units; Using the boundary of the terminal low-pressure section as the starting boundary, each passable path unit is connected and expanded to form the starting path search domain.

[0076] In one embodiment, the power distribution network of the eastern new district of a certain city is still used as the target planning area for explanation. A low-voltage section has been identified at the end of feeder 3 in the Cuihu Garden residential area, requiring a new line to connect to feeder 6, which has remaining capacity. To improve the efficiency of selecting the line extension path, the locations of existing poles, road corridors, and available channels around the low-voltage section are first extracted to establish a starting path search domain, providing a feasible path basis for subsequent impedance reduction path finding.

[0077] Specifically, the coordinate data of existing poles within a preset range around the end low-voltage section are first read, and a set of candidate pole points is formed according to their distance from the end low-voltage section. Taking the boundary of the end low-voltage section of Cuihuyuan residential area as the center, a range with a radius of 2km is selected as the preset search range. The coordinate information of existing 10kV and 0.4kV poles is retrieved from the power distribution GIS database, and a total of 27 poles are read, including 14 10kV poles and 13 low-voltage mixed poles.

[0078] Subsequently, each pole was sorted according to its shortest distance to the boundary of the low-voltage section, forming a candidate pole set. For example, pole G1 is 82m from the low-voltage section boundary and is located at the intersection of the road on the west side of the community; pole G2 is 135m from the boundary and is located at the intersection of Cuihu Road and Jinqiao Branch Road; pole G3 is 210m from the boundary and is close to the planned access road. Poles that are closer to the boundary and meet the conditions for renovation were given priority in the candidate sequence.

[0079] Secondly, data on road centerlines, road boundaries, and roadside laying conditions within a preset range are extracted to generate a road corridor strip area. The road data originates from the New Area's municipal planning platform and includes nine roads such as Cuihu Road, Jinqiao Avenue, the planned branch road on the west side, and internal secondary arterial roads. The road centerlines and both sides of the road's red line boundaries are read, and the roadside laying conditions are evaluated based on factors such as sidewalk width, green belt width, underground pipeline density, and excavation permit level.

[0080] In this embodiment, Cuihu Road has sidewalk space of more than 1.5m on both sides, which can be used for cable duct construction; the green belt of the planned branch road on the west side is continuous, suitable for the construction of new cable trenches; while the underground communication pipelines in the eastern section of Jinqiao Avenue are dense, and the laying is restricted. Based on this, the two sides of the road that can be constructed will be expanded to form a strip area, generating a set of road corridor strip areas.

[0081] Furthermore, data on open ground passages, existing power lines, and shared pole passages within a preset range are acquired, and areas encroached upon by buildings, prohibited construction zones, and areas obstructed by obstacles are eliminated to form a set of usable passages. The open ground passages include vacant plots of land outside residential areas, green space setback areas, and temporary construction passages; existing power lines include existing cable ducts and overhead line corridors; shared pole passages include continuous sections of existing poles that meet the clearance requirements.

[0082] For example, a 160-meter-long green setback zone on the northwest side of the community can be used as an open passageway; the existing double-circuit cable ducts on the western section of Cuihu Road can be directly utilized; and the four consecutive poles at the south entrance of the residential area are suitable for adding 10kV lines. At the same time, areas encroached by residential buildings, restricted construction areas around schools, and areas obstructed by large billboards were automatically eliminated, ultimately forming a collection of usable passageways with a total area of ​​approximately 36,000 square meters.

[0083] Subsequently, the candidate pole and tower point set, the road corridor strip area, and the set of usable passages are spatially overlaid to generate several passable path units. Various resources are uniformly mapped using a spatial grid, and continuously traversable construction units are identified, forming path units P1 to P18. Among them, unit P3 is a composite unit of "pole and tower + road corridor," and unit P7 is a composite unit of "cable duct + green corridor," possessing higher construction priority.

[0084] Finally, using the boundary of the terminal low-voltage section as the starting boundary, each passable path unit is expanded to form the starting path search domain. Starting from the western boundary of the low-voltage section in the Cuihu Garden residential area, the connectivity between adjacent path units is detected sequentially, and continuously reachable path units are expanded and connected layer by layer.

[0085] In this embodiment, path units P2, P3, P7, P11 and P15 can be connected sequentially from the starting boundary to form a continuous search channel leading to the southern node of feeder line 6; some eastward extension paths are terminated due to dense buildings or obstacles.

[0086] Preferably, the connectivity expansion of each passable path unit, starting from the boundary of the terminal low-pressure section, includes: The boundary of the low-pressure section at the end is divided into multiple boundary trigger points, and preset search angles are projected outwards for each point. The passable path units within each search angle are selected as the first round of response units. The passage priority is calculated based on the distance, angle, and channel margin between each first-round response unit and its corresponding boundary trigger point, and the expansion is initiated sequentially according to the passage priority. For passable path units that have been extended, search for adjacent path units along their extension direction, and simultaneously suppress the access of path units in the foldback direction and the direction of repeated coverage. The accessible path units covered by each retained extension direction are merged at the boundary to form the starting path search domain.

[0087] In one embodiment, the scenario of addressing the low-voltage section at the end of the power distribution network in the Cuihuyuan residential area of ​​a certain city's eastern new district is still used for illustration. The spatial overlay of the surrounding candidate tower point set, the road corridor strip area, and the set of usable passageways has been completed, forming multiple passable path units. To prioritize the extension of the line along passageways with better construction conditions, shorter distances, and more reasonable directions, the boundary of the low-voltage section at the end is further used as the starting boundary to connect and expand each passable path unit, forming a starting path search domain.

[0088] Specifically, the boundary of the terminal low-voltage section is first divided into multiple boundary trigger points, and preset search angles are projected outwards for each point. Passable path units within each search angle are selected as the first-round response units. Based on the outer contour of the terminal low-voltage section of the Cuihu Garden residential area, the boundary is divided into 12 boundary trigger points at equal intervals, denoted as B1 to B12, with a distance of approximately 25m between each adjacent trigger point.

[0089] Subsequently, a preset search angle is projected outward from each boundary trigger point. In this embodiment, each trigger point is set with a 120° sector search range, with the main search direction pointing to the southwest and west areas where nodes with remaining capacity are located, while also taking into account the road passage to the south. Within each sector angle range, passable path units adjacent to the trigger point are selected as the first round of response units.

[0090] For example, trigger point B3, located on the western boundary, searches for path units P2 and P3; trigger point B5, located in the southwest corner, searches for path units P4 and P7; and trigger point B8, located on the southern boundary, searches for path unit P6. Although several trigger points on the eastern side project search angles, they do not form response units due to the dense buildings and lack of effective passageways.

[0091] Secondly, the passage priority is calculated based on the distance, angle, and channel margin between each initial response unit and its corresponding boundary trigger point, and the expansion is initiated sequentially according to the passage priority. A comprehensive priority evaluation rule is established: the shorter the distance, the smaller the angle, and the wider the channel, the higher the priority.

[0092] In this embodiment, path unit P3 is only 18m from trigger point B3, with an entry direction deviation angle of 12° and an effective construction width of 2.6m, giving it the highest overall priority. Path unit P7 is 31m from trigger point B5, with a deviation angle of 20° and a width of 2.2m, giving it the next highest priority. Although path unit P6 is relatively close, it needs to detour through the parking lot and the passage width is only 1.1m, therefore its priority is lower. Therefore, the expansion is initiated sequentially in the order of P3→P7→P2→P4→P6.

[0093] Furthermore, for the traversable path unit that has been activated for expansion, the system searches for adjacent path units along its extension direction, while simultaneously suppressing the access of path units in the turnback direction and overlapping coverage direction. Specifically, when path unit P3 is activated, it searches westward along its road direction to find adjacent path unit P9, and continues to connect to P11; if it detects that an adjacent unit needs to turn eastward back to its original starting direction, it is determined to be a turnback direction and suppressed.

[0094] Meanwhile, if a path unit has already been covered by another high-priority extension chain, it is marked as a duplicate coverage unit and will not be included again. For example, if P11 has already been preferentially occupied by the P3 extension chain, it will not be included again when it is subsequently extended from P7 to P11, in order to avoid path redundancy.

[0095] During the expansion process, the overall direction is continuously advanced towards the remaining capacity nodes in the southern section of feeder 6, making the search path clearly target-oriented.

[0096] Finally, the accessible path units covered by each retained extension direction are fused at their boundaries to form the starting path search domain. The P3, P9, P11, and P15 units covered by the P3 extension chain are spatially fused with the P7, P10, and P14 units covered by the P7 extension chain, and overlapping boundaries are eliminated to form a continuous reachable region.

[0097] Of particular importance, step S32 includes: The load conflict superposition areas are sorted from high to low according to the load release priority, and the load conflict superposition area with the highest priority in the current ranking is selected as the processing area for this round. The search is expanded outward in a ring shape from the boundary of the current disposal area, and adjacent feeder nodes, distribution transformer nodes and connectable nodes are extracted in sequence to form a set of remaining capacity nodes. A node access priority sequence is generated by combining the carrying capacity margin of each remaining capacity node, the access path length, and the convenience of power supply switching. Location adaptation is performed one by one according to the node access priority sequence. When the corresponding node has the capacity compensation condition, it is determined as the location for the distribution transformer capacity expansion. When the corresponding node has the load transfer condition, it is determined as the location for the tie switch addition. After each location is determined, the load to be transferred is reduced synchronously and the released load is accumulated. Subsequent remaining capacity nodes are continuously called up until the released load and the load to be transferred are dynamically matched.

[0098] In one embodiment, taking the power distribution network of the eastern new district of a certain city as an example, it was identified that the heavily loaded section of feeder No. 5 and the overload node of distribution transformer No. 27 formed a load conflict superposition area in the southwest residential development area, and therefore, load release was prioritized. The load conflict superposition areas were sorted according to their load release priority from high to low, and the area with the highest priority was selected as the area to be handled in this round. In this embodiment, the southwest residential area simultaneously had a feeder load rate of 108%, a transformer load rate of 121%, and a load growth rate of 23%, and was therefore identified as the area to be handled in this round.

[0099] Subsequently, the search was expanded outward in a ring shape from the boundary of the area, sequentially extracting adjacent feeder nodes, distribution transformer nodes, and connectable nodes to form a set of nodes with remaining capacity. The surrounding search yielded feeder branch node N1 (No. 6), road intersection connecting node K1, and distribution transformer node T22 (No. 22). Among them, N1 has a remaining capacity of 1.4MW, K1 is suitable for switch installation, and T22 is suitable for capacity expansion.

[0100] Next, based on the load capacity, access path length, and power transfer convenience of each node, a node access priority sequence is generated. After evaluation, N1 has the shortest access distance and the largest capacity, so it is ranked first; K1 has the conditions for load transfer, so it is ranked second; T22 is ranked third.

[0101] Subsequently, location adaptation is performed one by one according to the node access priority sequence. When a node has the capacity compensation condition, it is determined as the location for increasing the capacity of the distribution transformer; when a node has the load transfer condition, it is determined as the location for adding a tie switch. In this embodiment, a 630kVA box-type transformer is added near N1; a tie switch is added at K1; and T22 is increased from 630kVA to 1000kVA.

[0102] After each location is determined, the load to be transferred is simultaneously reduced and the released load is accumulated. Subsequent remaining capacity nodes are continuously activated until the released load and the load to be transferred achieve a dynamic match. In this embodiment, approximately 2.21MW of load is released, meeting the 2.18MW load demand for transfer in this area. After implementation, the load rate of distribution transformer No. 27 drops to 73%, and the load rate of feeder No. 5 drops to 82%.

[0103] Of particular importance is the trend verification of the planning results, which includes: Based on the planning results, construct a planning network model for the target area and load the predicted load data for each load access point; Power flow calculations are performed along the power supply direction of each feeder to obtain the load results of each line section and the voltage results of each node. Based on the load and voltage results, adjustments were made to the locations of distribution transformer capacity expansion points, tie switch addition points, and line extension and modification paths. The adjusted planning results are used to perform power flow calculations again to determine the planning scheme for the target area.

[0104] In one embodiment, the power distribution network of the eastern new district of a certain city is still taken as an example. After completing the expansion and installation of distribution transformers, the addition of tie switches, and the extension and renovation of lines, the power flow of the planning results is checked.

[0105] First, a planning network model for the target area was constructed based on the planning results, and the projected load data for the next three years was loaded. The maximum load in the area increased from 31.2MW to 42.5MW. The model included data on the addition of a 630kVA box-type transformer, the expanded No. 22 and No. 27 distribution transformers, the addition of a tie switch, and the laying of 1.34km of new lines.

[0106] Subsequently, power flow calculations were performed along the power supply direction of each feeder to obtain the load results for each line section and the voltage results for each node. The initial calculation showed that the load rate of the last section of feeder No. 5 was still 83%, and the voltage of some terminal nodes in the southwest area was 218V.

[0107] Based on the load and voltage results, the planning results were adjusted to transfer some of the newly added residential loads to feeder line 6, move the line extension access point to a closer branch node, and optimize the segment location of the tie switch.

[0108] Subsequently, power flow calculations were performed again on the adjusted planning results. The results showed that the maximum feeder load rate dropped to 79%, the minimum terminal voltage increased to 221V, and the load rates of all distribution transformers were below 80%. Based on this, the planning results were determined as the official planning scheme for the target area.

[0109] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0110] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A power distribution network multi-resource integration planning method based on a graph neural network, characterized in that, The upstream substation is electrically connected to the 10kV feeder, the 10kV feeder is electrically connected to the feeder switch, the 10kV feeder is electrically connected to the high-voltage side of the distribution transformer, and the low-voltage side of the distribution transformer is electrically connected to the load connection point. The method includes the following steps: Step S1: Obtain the equipment ledger data of the existing distribution transformers, feeder switches and 10kV feeders in the target planning area, and obtain the historical electricity consumption data, geographical location data and load forecast data of each load access point, and organize them to obtain the basic planning dataset; Step S2: Construct a distribution network topology based on the electrical connection relationship between the distribution transformer, 10kV feeder and load access point, and use a preset graph neural network based on the basic planning dataset to identify the feeder heavy load section, terminal low voltage section and transformer overload node in the distribution network topology. Step S3: Determine the planning results, including the location of distribution transformer capacity expansion, tie switch addition, and line extension and modification path, based on the spatial distribution location and power supply relationship of the feeder heavy load section, terminal low voltage section, and transformer overload node. Step S4: Perform a flow verification on the planning results to determine the planning scheme for the target area.

2. The multi-resource fusion planning method for distribution networks based on graph neural networks according to claim 1, characterized in that, In step S2, based on the basic planning dataset, a pre-set graph neural network is used to identify the feeder overload sections, terminal low-voltage sections, and transformer overload nodes in the distribution network topology, including: A distribution network topology map is constructed based on the equipment connection relationships in the basic planning dataset, and the load rate, capacity margin and terminal voltage parameters of each node are extracted as graph feature data. The graph feature data is input into a preset graph neural network, and the operating features of adjacent nodes are transmitted step by step according to the power supply path direction to generate the associated evaluation values ​​of each feeder section and each distribution transformer. The feeder heavy load sections are determined based on the associated evaluation values ​​of each feeder section, the terminal low voltage sections are determined based on the associated evaluation values ​​of the terminal nodes, and the transformer overload nodes are determined based on the associated evaluation values ​​of each distribution transformer.

3. The multi-resource fusion planning method for distribution networks based on graph neural networks according to claim 1, characterized in that, Step S3 includes the following steps: Step S31: Overlap scan the adjacent feeder overloaded sections and transformer overload nodes according to the power supply transfer range to form load conflict superposition areas, and calculate the load release priority of each load conflict superposition area; Step S32: Search the remaining capacity nodes around the corresponding load conflict superposition area in sequence according to the load release priority, determine the location of the distribution transformer capacity expansion or the location of the tie switch addition, so that the released load amount and the load to be transferred are dynamically matched. Step S33: Starting from the low-voltage section at the end and ending at the determined remaining capacity node, perform impedance reduction path finding to determine the line extension and renovation path; Step S34: After completing the location of each distribution transformer capacity expansion, tie switch addition, or line extension modification, update the load distribution of adjacent areas in real time, and repeat steps S32 and S33 until the overall planning result of the target planning area is obtained.

4. The multi-resource fusion planning method for distribution networks based on graph neural networks according to claim 3, characterized in that, The overlapping scan of adjacent feeder overloaded sections and transformer overload nodes is performed according to the power supply transfer range, specifically as follows: Using the first and last switch nodes of each feeder heavy-load section as boundaries, the transferable power supply radius is progressively extended to both sides along the power supply path to form the corresponding feeder unloading scan zone; Centered on each transformer overload node, the corresponding transformer shunt scanning zone is formed by expanding outwards in stages according to the low-voltage outgoing line direction and the connection relationship between adjacent transformer areas. The feeder load release scanning zone and the transformer load detachment scanning zone are overlapped segment by segment according to their spatial location, and the difference in the acceptable capacity of the two is compared simultaneously to screen out the overlapping sections that simultaneously meet the path connectivity condition and the capacity acceptance condition. Boundary splicing is performed on consecutive adjacent overlapping sections to form a load conflict superposition area.

5. The multi-resource fusion planning method for distribution networks based on graph neural networks according to claim 4, characterized in that, The methods for obtaining path connectivity conditions and capacity carrying capacity conditions include: Extract the current line routing, switch connection point location, and reserved channel data between the feeder unloading scan zone and the transformer shunt scan zone to determine whether there is a through-path power supply connection between the two scan zones; if there is a through-path power supply connection, then the path connectivity condition is met. Calculate the remaining power supply capacity, release load capacity, and conductor current carrying margin at both ends of the passable power supply connection path to form a capacity carrying parameter set; The difference between the remaining power supply capacity and the releaseable load capacity in the capacity acceptance parameter set is calculated, and the conductor current carrying margin is compared with the preset safety margin threshold. When the remaining power supply capacity is not less than the releaseable load capacity, and the conductor current carrying margin is greater than or equal to the preset safety margin threshold, the capacity acceptance condition is determined to be met.

6. The multi-resource fusion planning method for distribution networks based on graph neural networks according to claim 4, characterized in that, The feeder load release scan zone and the transformer load detachment scan zone are sequentially overlapped according to their spatial positions, including: The geographical coordinate order of the feeder load release scanning zone and the transformer load detachment scanning zone is determined based on geographical location data; The boundary points of each strip segment of the feeder load release scanning zone and the transformer load shunting scanning zone are extracted in order of geographic coordinates, and the corresponding feeder segment sequence and load shunting segment sequence are obtained. Taking the first segment of the feeder segment sequence as a reference, the sequence moves segment by segment along the power supply extension direction, and simultaneously calls the adjacent off-load segment sequence for corresponding comparison: when the projection range of the feeder segment and the off-load segment intersects or is adjacent, the overlapping segment information of the corresponding position is recorded, and the comparison continues to the next segment. Adjacent segments that have continuously obtained overlapping segment information are overlapped one by one.

7. The multi-resource fusion planning method for distribution networks based on graph neural networks according to claim 3, characterized in that, Step S33 includes: Extract the location of existing poles and towers, road corridors and available channels around the low-voltage section at the end, establish the starting path search domain, and set the identified remaining capacity nodes as target nodes; The initial path impedance value is calculated based on the line length, number of turns, and channel occupancy of each candidate extension direction within the starting path search domain, forming a candidate path set; High-impedance path branches are eliminated segment by segment according to the path impedance value from large to small, and the incremental impedance value of the subsequent segment is superimposed on the remaining path in real time to perform impedance-decreasing path finding. When a candidate path is connected to the target node, the comprehensive impedance values ​​of each connected path are compared, and the connected path with the smallest comprehensive impedance value is selected as the line extension and renovation path.

8. The multi-resource fusion planning method for distribution networks based on graph neural networks according to claim 7, characterized in that, Extract the locations of existing power poles, road corridors, and available passages around the low-voltage section at the end, and establish the starting path search domain, including: Read the coordinate data of existing poles and towers within a preset range around the end low-voltage section, and form a set of candidate pole and tower points according to their distance from the end low-voltage section; Extract road centerline data, road boundary data, and roadside laying condition data within a preset range to generate a road corridor strip area; Data on open ground passages, existing power lines, and shared pole passages within a preset range are obtained, and building encroachment areas, prohibited construction areas, and obstruction areas are eliminated to form a set of usable passages; The candidate tower point set, the road corridor strip area and the available passage set are spatially superimposed to generate several passable path units; Using the boundary of the terminal low-pressure section as the starting boundary, each passable path unit is connected and expanded to form the starting path search domain.

9. The multi-resource fusion planning method for distribution networks based on graph neural networks according to claim 8, characterized in that, Starting from the boundary of the terminal low-pressure section, the connectivity expansion of each passable path unit includes: The boundary of the low-pressure section at the end is divided into multiple boundary trigger points, and preset search angles are projected outwards for each point. The passable path units within each search angle are selected as the first round of response units. The passage priority is calculated based on the distance, angle, and channel margin between each first-round response unit and its corresponding boundary trigger point, and the expansion is initiated sequentially according to the passage priority. For passable path units that have been extended, search for adjacent path units along their extension direction, and simultaneously suppress the access of path units in the foldback direction and the direction of repeated coverage. The accessible path units covered by each retained extension direction are merged at the boundary to form the starting path search domain.

10. A multi-resource fusion planning system for power distribution networks based on graph neural networks, characterized in that, For executing the graph neural network-based multi-resource fusion planning method for distribution networks as described in claim 1, the graph neural network-based multi-resource fusion planning system for distribution networks includes: The data acquisition module is used to obtain the equipment ledger data of the existing distribution transformers, feeder switches and 10kV feeders in the target planning area, as well as the historical electricity consumption data, geographical location data and load forecast data of each load access point, and organize them to obtain the basic planning dataset. The topology identification module is used to construct a distribution network topology map based on the electrical connection relationship between distribution transformers, 10kV feeders and load access points, and to identify the feeder heavy-load sections, terminal low-voltage sections and transformer overload nodes in the distribution network topology map using a preset graph neural network based on the basic planning dataset. The planning and decision-making module is used to determine the planning results, including the location of distribution transformer capacity expansion, the location of tie switch addition, and the route of line extension and renovation, based on the spatial distribution of feeder heavy load sections, terminal low voltage sections, and transformer overload nodes and power supply relationships. The verification output module is used to perform power flow verification on the planning results in order to determine the planning scheme for the target area.

Images