Power distribution network division data model based on power grid load flow spatial distribution and improved algorithm thereof

Abstract

The invention discloses a power distribution network division data model based on power grid load flow spatial distribution and a data algorithm thereof, which introduces a newly constructed load flow spatial distribution data model on the basis of vectorization clustering processing of power distribution network load data. According to the data model, a geometrical and visualized data structure is constructed through a space field of a load vector, data optimization is carried out in the mass center initialization process of a K-means algorithm and / or the mass center iteration process based on the Euler distance, and adjustability and optional comparison of different space distribution scales are adapted. And optimizing grid division of the load-complementary power distribution network under different space distribution scales. According to the method, the grid division efficiency and effect of the power distribution network can be greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid big data, and in particular to a distribution network partitioning data model based on the spatial distribution of power grid load flows and the improvement and construction of an algorithm therein. Background Art

[0002] With the rapid growth of power system scale and power load, power load forecasting and load analysis have become important issues in power system management and operation. Load analysis is to identify and detect potential load characteristics from the load data of the power system, thereby providing a basis for effective load management and power grid planning. Load area clustering and load complementation of distribution networks are very important links. As an important tool for allocating power resources, load complementation and grid division of distribution networks play an important role in balancing loads and ensuring the safety and stability of the power grid.

[0003] Traditional distribution network grid division methods often cannot effectively handle the needs and limitations of complex distribution networks, resulting in low distribution efficiency. Machine learning clustering algorithms, such as the K-means algorithm, have become powerful tools to solve this problem due to their simplicity and ease of implementation. However, current load area clustering methods are mainly based on load clustering technology and spatial clustering technology. These methods usually require the number of clusters to be manually specified in advance, which is not always feasible or optimal in practical applications. In addition, these methods usually do not consider the spatial information of load distribution, and therefore may not capture some non-random characteristics of load distribution. In particular, the existing K-means algorithm has too many single algorithm processes and lacks independent secondary evaluation indicators, especially indicators of adjustability and optional comparison that are suitable for different spatial distribution scales. This has caused many difficult-to-overcome defects in the effect of algorithm execution, making the effect of optimization and distribution network grid division sometimes good and sometimes bad, and its randomness and bias are difficult to eliminate within the algorithm.

[0004] To this end, we have developed a new data processing idea by introducing parallel indicators that take into account the real geographical distribution properties of the power grid, namely a data model and data analysis method based on a two-dimensional load vector field. We hope to propose and construct a new data model, including the data analysis and processing algorithms embedded within it. Summary of the invention

[0005] The technical problem to be solved by the present invention is to overcome various deficiencies of the prior art and provide a distribution network partitioning data model based on the spatial distribution of power grid load flow, including data analysis and processing algorithms embedded therein.

[0006] In order to solve the above technical problems, the technical solutions adopted by the present invention are as follows.

[0007] A distribution network partitioning data model based on the spatial distribution of power grid load flow is proposed. This data model introduces a newly constructed load flow spatial distribution data model based on the vectorized clustering processing of distribution network load data. This data model constructs a geometric and intuitive data structure with the spatial field of load vectors, and performs data optimization in the centroid initialization of the K-means algorithm and / or the centroid iteration process based on Euler distance. It adapts to the adjustability and optional comparison of different spatial distribution scales, and optimizes the distribution network grid division with load complementarity under different spatial distribution scales.

[0008] As a preferred technical solution of the present invention, the following data processing process is constructed: first, the load data of the distribution network is imported and converted into a clusterable data vector; then, the spatial field data model of the load vector is introduced to construct an improved algorithm, and the initial centroid of the k-means algorithm is obtained and / or the centroid is updated; finally, the above data process is repeated to iterate until the division of all grids reaches a predetermined standard and / or overflows a preset upper limit of the number of iterations, and the distribution network is gridded according to the centroid distribution.

[0009] As a preferred technical solution of the present invention, the acquisition of the initial centroid and / or the update of the centroid adopts a fusion algorithm based on load vector balanced complementarity, introducing a newly constructed load flow spatial distribution data model, and based on this data model, the regional distribution balance of the load vector is characterized by the spatial distribution geometric characteristics of the load vector and its numerical connotation, thereby optimizing the selection of the initial centroid.

[0010] As a preferred technical solution of the present invention, specifically, the introduction of load vector regional distribution representation adopts the following data model: for the clusterable vectors obtained by transforming the power grid load data after preprocessing, a vector field is constructed according to its true geographical distribution in the power grid system, especially fitted as a vector field in a continuous two-dimensional space. Considering the adjustability and optional comparison of adapting to different spatial distribution scales, the continuous two-dimensional space can be divided into different comparable subspace regions according to different geometric shapes, and then a load vector balanced complementary algorithm in two-dimensional space is constructed on the boundaries of these delineated subspace regions, that is, firstly, the flow of the load vector on the boundary of the subspace region is calculated, which is a finite differential, and then the finite flow differential of the load vector on the boundary of the entire region is summed to obtain the load flow and the load flow of the entire subspace region, and finally, balanced and complementary load characterization parameters are constructed according to the obtained load flow and the geometric properties of the subspace region itself, and the selection of the initial centroid is optimized based on the obtained load characterization parameters and their spatial distribution variation to reduce the randomness and bias of the initial selection, avoid the poor clustering effect of the original K-means algorithm due to improper initial selection, and finally more effectively realize the load complementary grid division of the distribution network.

[0011] As a preferred technical solution of the present invention, for the load characterization parameters, the global load flow on the entire subspace area is characterized by an average distribution of the entire area based on the geometric properties of the selected subspace area itself, thereby constructing the load characterization parameters.

[0012] As a preferred technical solution of the present invention, the global load flow allows it to adopt different specific characterization methods according to the selected mathematical algorithm and the subsequent centroid update process of the k-means algorithm. It is preferred to use the cumulative amount of the segmented load flow on the entire subspace area on the entire boundary to numerically characterize the global load flow.

[0013] As a preferred technical solution of the present invention, the geometric properties of the subspace region itself allow it to adopt different specific representation methods according to the selected mathematical algorithm and the subsequent centroid update process of the k-means algorithm. We select the size attribute of the spatial region as the core representation item of its own attributes based on a tentative evaluation of the data value, and generally abandon its shape attribute and weaken other geometric topological attributes.

[0014] As a preferred technical solution of the present invention, under the most application-valuable simplified processing, the geometric properties of the subspace region itself are preferably numerically represented by the two-dimensional area of ​​the subspace region.

[0015] As a preferred technical solution of the present invention, the full boundary accumulation of the segmented load flow in the entire subspace area is characterized by an average distribution of the entire area on the geometric area attribute of the selected subspace area itself. Specifically, the ratio of the two is taken to construct the load characterization parameter.

[0016] As a preferred technical solution of the present invention, by comparing the spatial sub-region divisions of different scales and different distribution types and the load characterization parameters thereunder, the selection of the initial centroid is optimized based on the balance of its spatial distribution;

[0017] The spatial sub-region divisions of different scales and different distribution types can be constructed through a callable external data set, which stores pre-established power grid geographic region division paradigms; this callable external data set is an open data set. In addition to the pre-made region division paradigms derived from conventional geometric region division models, new region division paradigms obtained from the outside or new region division paradigms newly created in any data processing can be added to the data set at any time to achieve data expansion and update.

[0018] The spatial distribution balance of the load characterization parameters can be measured using common data distribution balance, including the difference of the load characterization parameters under addition and subtraction operations, the variation ratio of their absolute values, variance analysis, or other common algorithms.

[0019] As a preferred technical solution of the present invention, the centroid is updated in the k-means algorithm, and the centroid with the minimum weight is selected for updating based on the weight calculation of the Euler distance; and iteration is performed.

[0020] As a preferred technical solution of the present invention, for the centroid update process, the above-mentioned centroid initialization optimization algorithm is iteratively applied to the centroid update process for optimization; wherein, self-spatial region selection and data attribute representation and related data processing are all optional, but it must be ensured that the specific data options applied to the initial centroid optimization and the centroid update optimization in the single data optimization process remain consistent.

[0021] The present invention also includes a data structure and / or database generated according to the above-mentioned load complementary distribution network grid division method, including a data structure and / or database stored locally, on a server or in the cloud.

[0022] The present invention also includes an executable program / software or a class program / software application according to the above-mentioned load complementary distribution network grid division method or including the above-mentioned data structure and database, including executable software, programs, applications and their systems on PC, mobile or other conventional platforms or conventional types.

[0023] The beneficial effect of adopting the above technical solution is that the present invention improves the efficiency and accuracy of distribution network grid division by improving the K-means clustering algorithm and adopting a more reasonable initialization method and update method. In summary, considering the inherent randomness and problems of the classical algorithm, we have formulated a new data processing idea, by introducing parallel indicators that consider the real geographical distribution attributes of the power grid, that is, a spatial analysis method based on a two-dimensional load vector field, considering the load distribution characteristics from the perspective of two-dimensional space, dividing the two-dimensional space into multiple subspace regions, calculating the load flow parameters of each subspace region for evaluating the load clustering characteristics, etc., thereby constructing the new improved k-means analysis method of the present invention; the selection of the initial centroid based on the load characterization parameters and their spatial distribution variation optimization can reduce the randomness and bias of the initial selection, avoid the poor clustering effect of the original K-means algorithm due to improper initial selection, and finally more effectively realize the load complementary grid division of the distribution network.

[0024] It can be seen that compared with the existing algorithms, the present invention can greatly reduce the randomness and deviation of the initial selection, while optimizing the efficiency of the centroid update. The overall algorithm convergence is significantly compacted and the efficiency (quantity) and effect (quality) of the distribution network grid division are greatly improved. DETAILED DESCRIPTION

[0025] In the description of the following embodiments, specific details such as specific system structures, technologies, etc. are provided for illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0026] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or their collections. It should also be understood that the term "and / or" used in the present specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0027] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0028] In addition, in the description of the present application specification and the attached claims, the terms "first", "second", "third", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance. The reference to "one embodiment" or "some embodiments" described in the present application specification means that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in the different parts of this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0029] Example 1

[0030] The improvement of the algorithm of the present invention is centered on the assistance team of power grid researchers and data experts, and a new parallel data indicator is introduced to try to compensate for and eliminate the inherent defects of the original algorithm, so as to reduce the randomness of the initialization centroid selection and certain imbalances in the centroid update, enhance the stability of the algorithm, improve the segmentation effect and efficiency, and realize the grid division of the distribution network in a complex environment.

[0031] In terms of technical ideas, the present invention introduces a newly constructed load flow spatial distribution data model based on the vectorized clustering processing of distribution network load data. This data model constructs a geometric and intuitive data structure with the spatial field of the load vector, performs data optimization in the centroid initialization of the K-means algorithm and / or the centroid iteration process based on the Euler distance, adapts to the adjustability and optional comparison of different spatial distribution scales, and optimizes the distribution network grid division of load complementarity under different spatial distribution scales. Based on the above technical ideas, the following dominant data processing process is constructed: first, the distribution network load data is imported and converted into a clusterable data vector; then the spatial field data model of the load vector is introduced to construct an improved algorithm to obtain the initial centroid of the k-means algorithm and / or update the centroid; the centroid update is based on the weight calculation of the Euler distance, and the centroid with the minimum weight is selected for update and iteration; finally, the above data process is repeated to iterate until the division of all grids reaches the predetermined standard and / or overflows the preset upper limit of the number of iterations, and the distribution network is gridded according to the centroid distribution.

[0032] Obviously, the core lies in the acquisition and update of the initial centroid, which is also the core of our research and development. The acquisition and / or update of the initial centroid adopts a fusion algorithm based on load vector balanced complementarity, introduces a newly constructed load flow spatial distribution data model, and uses the spatial distribution geometric characteristics and numerical connotations of the load vector on this data model to characterize the regional distribution balance of the load vector, thereby optimizing the selection of the initial centroid. Specifically, the introduction of load vector regional distribution representation adopts the following data model: for the clusterable vectors obtained by transforming the power grid load data after preprocessing, a vector field is constructed according to its actual geographical distribution in the power grid system, especially fitted as a vector field in a continuous two-dimensional space. Considering the adjustability and optional comparison to adapt to different spatial distribution scales, the continuous two-dimensional space can be divided into different comparable subspace areas according to different geometric shapes, and then a load vector balanced complementary algorithm in two-dimensional space is constructed on the boundaries of these delineated subspace areas, that is, the flow of the load vector on the boundary of the subspace area is first calculated, which is a finite differential, and then the finite flow differential of the load vector on the boundary of the entire area is summed to obtain the load flow sum of the entire subspace area, and finally, the balanced and complementary load characterization parameters are constructed according to the obtained load flow and the geometric properties of the subspace area itself. At this time, we can optimize the selection of the initial centroid based on the balance of its spatial distribution by comparing the spatial sub-region divisions of different scales and different distribution types and the load characterization parameters under them; the spatial distribution balance of the load characterization parameters can be measured by common data distribution balance, including the difference of the load characterization parameters under addition and subtraction operations, the variation ratio of its absolute value, variance analysis, or other common algorithms. The selection of the initial centroid is optimized based on the obtained load characterization parameters and their spatial distribution variation to reduce the randomness and bias of the initial selection, avoid the poor clustering effect of the original K-means algorithm due to improper initial selection, and finally more effectively realize the load complementary grid division of the distribution network. Further, for the centroid update process, it is obvious that the above-mentioned centroid initialization optimization algorithm can and should be iteratively applied to the centroid update process for optimization; among them, including the self-spatial region selection and its data attribute representation and related data processing are all optional, but it must be ensured that the specific data options applied to the initial centroid optimization and centroid update optimization in the single data optimization process are consistent.

[0033] Example 2

[0034] Spatial sub-region divisions of different scales and distribution types can be constructed through a callable external dataset, which stores pre-established power grid geographic region division paradigms; this callable external dataset is an open dataset. In addition to the pre-made region division paradigms derived from conventional geometric region division models, new region division paradigms obtained from the outside or newly created new region division paradigms in any data processing can be added to the dataset at any time to achieve data expansion and update.

[0035] Example 3

[0036] The load characterization parameter characterizes the global load flow in the entire subspace region by an average distribution of the entire region on the geometric properties of the selected subspace region itself. In theory and practice, the global load flow allows it to be represented in different specific ways according to the selected mathematical algorithm and the subsequent centroid update process of the k-means algorithm. Here, based on the simplicity of data and the operation effect of the simulation, we use the cumulative amount of the segmented load flow on the entire boundary of the entire subspace region to numerically characterize the global load flow.

[0037] Similarly, the geometric properties of the subspace region itself, in theory and practice, also allow it to be represented in different specific ways according to the selected mathematical algorithm and the subsequent centroid update process of the k-means algorithm. We choose the size attribute of the spatial region as the core representation item of its own attributes based on the tentative evaluation of the data value, and generally abandon its shape attribute and weaken other geometric topological attributes, because for the industrial process (not scientific research process) of power grid division, too much introduction of geometric topological attributes will make the data model too complicated, but the benefits are very weak.

[0038] Therefore, under the most application-valued simplified processing, the geometric properties of the subspace region itself are preferably numerically characterized by the two-dimensional area of ​​the subspace region. In this way, the full boundary accumulation of the segmented load flow on the entire subspace region can be characterized by the average distribution of the entire region on the geometric area properties of the selected subspace region itself. Naturally, it is conceivable to take the ratio of the two to construct the load characterization parameters, which is the most conventional and most commonly used data selection.

[0039] Example 4

[0040] An optimization for the algorithm iteration endpoint. In this area, we set a dual indicator, that is, considering the iteration to the predetermined standard or overflowing the preset upper limit of the number of iterations. Calculate the difference between the result of each clustering and the previous result. When the difference is lower than a certain threshold or the iteration reaches a certain number of times, stop the iteration; this two-layer design process considers the accuracy of the segmentation while avoiding unlimited iterations, saving the running time of the algorithm. Finally, the termination condition is reached, the algorithm stops iterating, and the final centroid and grid division results are obtained.

[0041] Example 5

[0042] We built several data execution processes and their reference codes in the intermediate simulation. In the example, we built a temporary simulation process for the medium term. It is worth noting that it contains several tentative subsequent processing steps. After the distribution network grid division is completed, the grid can be further optimized; the feasibility and maturity optimization of several tentative subsequent processing steps are not within the priority scope of this simulation.

[0043] Further optimization and completion of the code can refer to the following data processing process example (it can be assumed that there are n groups of load data in two-dimensional space, each group of data corresponds to a point). First, the power grid load data is preprocessed and converted into a clusterable vector. According to its actual geographical distribution in the power grid system, it is constructed as a vector field and fitted as a vector field in a continuous two-dimensional space. Considering the adjustability and optional comparison of adapting to different spatial distribution scales, the continuous two-dimensional space can be divided into different comparable subspace regions according to different geometric shapes. Calculate the finite differential of the load vector flow on the boundary of the subspace region. The finite flow differential of the load vector on the boundary of the entire region is summed to obtain the load flow sum of the entire subspace region, and finally, the balanced and complementary load characterization parameters are constructed based on the obtained load flow and the two-dimensional area of ​​the subspace region. The load characterization parameters obtained in the previous step and their spatial distribution variation are input into the K-means++ algorithm for the selection of the initial point. In this way, the newly created data process takes into account the changes in the flow in the subspace region, thereby greatly reducing the randomness and bias of the initial selection, and thus a more reasonable distribution network grid division can be generated.

[0044] Furthermore, the code is constructed to deeply match the core improved algorithm, and an algorithm instance that is almost close to the usable range is constructed. You can refer to the following data processing process example for further adjustment and optimization:

[0045] Step 1: Data preprocessing and vector field construction. Let P = {p1, p2, ..., p n} is a collection of power grid load data. After preprocessing, the load data is converted into a clusterable vector data set, denoted as S = {s1, s2, …, s n}. Each load vector s i =(s ix ,s iy ), where s ix and iy Represent the x component and y component of the load vector in two-dimensional space respectively. Interpolation methods such as bilinear interpolation can be used to construct a (continuous) load vector field F(x,y)=(u(x,y),v(x,y)). For a given two-dimensional space point (x,y), the nearest neighbor interpolation method can be used to find the load vector s closest to the point (x,y) from the load vector data set S. neighbor , and use its x and y components as the component values ​​of the load vector field at that point.

[0046] Step 2: Subspace region division. Based on the vector field F(x,y), the two-dimensional space is divided into multiple subspace regions sub-D = {D1, D2, ..., D m}, each subspace region D i They are all geometric shapes (such as rectangles, polygons, etc.); here we can use geometric partitioning algorithms (analytic geometry algorithms) to divide the two-dimensional space into different subspace regions.

[0047] Step 3: Calculate the two-dimensional flow on the boundary of the selected region. For each selected subspace region D i , calculate the finite two-dimensional flow of the load vector on several segmented segments on its boundary; let D i The boundary of is composed of line segments L1, L2, ..., L k Each line segment L j Through the two endpoints P j1 and P j2 Furthermore, in order to calculate L i The finite two-dimensional flow on L i Select a series of discrete points R j = {r j1 ,r j2 ,…,r jn}, where r ji Indicates that it is located on line segment L j For the discrete point r ji , calculate the corresponding load vector field F(r ji )=(u ji, v ji ). Naturally, the next step is to calculate the discrete points r ji The unit tangent vector T at ji , by calculating the line segment L j The slope k j And calculate its tangent angle θ j , we can get T ji=(cos(θ j ), sin(θ j )). Then, the load vector field F(r ji ) and the unit tangent vector T ji Do the dot product and get the finite two-dimensional flow Q ji =u ji *cos(θ j )+v ji *sin(θ j ).

[0048] Step 4: Sum the finite flow of each segment. For each selected subspace region D i , we take all discrete points r on its boundary ji The finite flow Q ji Sum the load flow to get the total Qsum i =Σ(Q ji ).

[0049] Step 5: Calculate the load characterization parameters. For each subspace region D i , with the total load flow Qsum i and area A i Calculate the load characterization parameter L i =Qsum i / A i Load characterization parameter L i It represents the load distribution density of the subspace area, and based on the data process constructed above, it can be seen that it can be used to evaluate the degree of load aggregation and dispersion.

[0050] Among them, for the continuous fitting of the load vector field, the nearest neighbor interpolation method can be implemented using the NearestNeighbors of the sklearn library. You can first fit the load vector dataset S using the NearestNeighbors class, then use the kneighbors method to obtain the index of the load vector closest to the coordinate (x, y), and finally return the nearest neighbor load vector found as output. When implementing the code, just make sure to replace the load vector dataset S with the actual load dataset and replace x and y with the actual coordinates.

[0051] In addition, according to the tips of our data team, we can also use the SciPy library to perform nearest neighbor interpolation, using the functions in the scipy.interpolate module to implement it; we can use the NearestNDInterpolator class to create a nearest neighbor interpolator object, and then use the interpolator object to interpolate the given new point and return the corresponding interpolated value. .

[0052] Example 6

[0053] The present invention also includes a data structure and / or database generated according to the above-mentioned load complementary distribution network grid division method, including a data structure and / or database stored locally, on a server or in the cloud.

[0054] The present invention also includes an executable program / software or a class program / software application according to the above-mentioned load complementary distribution network grid division method or including the above-mentioned data structure and database, including executable software, programs, applications and their systems on PC, mobile or other conventional platforms or conventional types.

[0055] Example 7

[0056] The scientific research team of the present invention includes researchers from multiple units and technical fields. On the one hand, the implementation method of the present invention is applicable to distribution networks of different scales and complexities. On the other hand, the improved method based on the K-means algorithm of the present invention can also be extended to other fields, such as data mining, image processing, etc., and has broad application potential.

[0057] In the above embodiments, the description of each embodiment has its own emphasis. For the part that is not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In each embodiment, the hardware implementation of the technology can directly adopt existing intelligent devices, including but not limited to industrial computers, PCs, smart phones, handheld stand-alone machines, floor-standing stand-alone machines, etc. Its input device preferably adopts a screen keyboard, its data storage and calculation module adopts existing memory, calculator, controller, its internal communication module adopts existing communication ports and protocols, and its remote communication adopts existing GPRS networks, World Wide Internet, etc. It can be clearly understood by technicians in the relevant field that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other and are not used to limit the protection scope of this application. The specific working process of the units and modules in the above system can refer to the corresponding process in the above method embodiment, which will not be repeated here.

[0058] In the embodiments provided by the present invention, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms. The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0059] Each functional unit in each embodiment of the present invention may be integrated into a processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of a software functional unit. If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it may be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and may also instruct the relevant hardware to complete it through a computer program. The computer program may be stored in a computer-readable storage medium, and the computer program may be executed by the processor to implement the steps of the above-mentioned various method embodiments. Among them, the computer program includes computer program code, and the computer program code may be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0060] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A distribution network partitioning data model based on the spatial distribution of power grid load flow, characterized by: This data model introduces a newly constructed load flow spatial distribution data model based on the vectorized clustering processing of distribution network load data. This data model constructs a geometric and intuitive data structure with the spatial field of load vectors, and optimizes data in the centroid initialization of the K-means algorithm and / or the centroid iteration process based on Euler distance. It adapts to the adjustability and optional comparison of different spatial distribution scales, and optimizes the distribution network grid division with load complementarity under different spatial distribution scales.

2. The distribution network partitioning data model based on the spatial distribution of power grid load flow according to claim 1 is characterized in that: The following data processing process is constructed: first, the distribution network load data is imported and converted into a clusterable data vector; then, the spatial field data model of the load vector is introduced to construct an improved algorithm to obtain the initial centroid of the k-means algorithm and / or update the centroid; finally, the above data process is repeated to iterate until the division of all grids reaches a predetermined standard and / or overflows a preset upper limit of the number of iterations, and the distribution network is gridded according to the centroid distribution.

3. The distribution network partitioning data model based on the spatial distribution of power grid load flow according to claim 2 is characterized in that: The acquisition of the initial centroid and / or the update of the centroid adopts a fusion algorithm based on load vector balanced complementarity, introducing a newly constructed load flow spatial distribution data model. On this data model, the spatial distribution geometric characteristics of the load vector and its numerical connotation are used to characterize the regional distribution balance of the load vector, thereby optimizing the selection of the initial centroid.

4. The distribution network partitioning data model based on the spatial distribution of power grid load flow according to claim 3 is characterized in that: Specifically, the introduction of load vector regional distribution representation adopts the following data model: for the clusterable vectors obtained by transforming the power grid load data after preprocessing, a vector field is constructed according to its actual geographical distribution in the power grid system, especially fitted as a vector field in a continuous two-dimensional space. Considering the adjustability and optional comparison to adapt to different spatial distribution scales, the continuous two-dimensional space can be divided into different comparable subspace regions according to different geometric shapes, and then a load vector balanced complementary algorithm in two-dimensional space is constructed on the boundaries of these delineated subspace regions, that is, first calculate the flow of the load vector on the boundary of the subspace region, which is a finite differential, and then sum the finite flow differentials of the load vector on the boundary of the entire region to obtain the load flow sum of the entire subspace region, and finally construct balanced and complementary load characterization parameters according to the obtained load flow and the geometric properties of the subspace region itself, and optimize the selection of the initial centroid based on the obtained load characterization parameters and their spatial distribution variation.

5. The distribution network partitioning data model based on the spatial distribution of power grid load flow according to claim 4 is characterized in that: For the load characterization parameter, the global load flow on the entire subspace area is characterized by an average distribution of the entire area based on the geometric properties of the selected subspace area itself, thereby constructing the load characterization parameter.

6. The distribution network partitioning data model based on the spatial distribution of power grid load flow according to claim 5 is characterized in that: The global load flow is allowed to adopt different specific representation methods according to the selected mathematical algorithm and the subsequent centroid update process of the k-means algorithm, and the global load flow is numerically represented by the cumulative amount of the segmented load flow on the entire boundary in the entire subspace area; The geometric properties of the subspace region itself allow it to adopt different specific representation methods according to the selected mathematical algorithm and the subsequent centroid update process of the k-means algorithm. According to the tentative evaluation of the data value, the size attribute of the spatial region is selected as the core representation item of its own attributes, while generally abandoning its shape attribute and weakening other geometric topological attributes; Under application-oriented simplified processing, the geometric properties of the subspace region itself are preferably numerically represented by the two-dimensional area of ​​the subspace region.

7. The distribution network partitioning data model based on the spatial distribution of power grid load flow according to claim 6 is characterized by: The full boundary accumulation of the segmented load flow in the entire subspace area is characterized by an average distribution of the entire area on the geometric area attribute of the selected subspace area itself. Specifically, the ratio of the two is taken to construct the load characterization parameter.

8. The distribution network partitioning data model based on the spatial distribution of power grid load flow according to claim 7 is characterized in that: By comparing the spatial sub-region divisions of different scales and distribution types and the load characterization parameters under them, the selection of the initial centroid is optimized based on the balance of their spatial distribution. The spatial sub-region divisions of different scales and different distribution types are constructed through a callable external data set, in which a pre-established grid geographic region division paradigm is stored; this callable external data set is an open data set, and in addition to the pre-established region division paradigm derived from the conventional geometric region division model, a new region division paradigm obtained from the outside or a new region division paradigm newly created in any data processing can be added to the data set at any time to realize data expansion and update; The spatial distribution balance of the load characterization parameters can be measured using common data distribution balance, including the difference of the load characterization parameters under addition and subtraction operations, the variation ratio of their absolute values, variance analysis, or other common algorithms.

9. The distribution network partitioning data model based on the spatial distribution of power grid load flow according to claim 2 is characterized in that: In the k-means algorithm, the centroid is updated, and the centroid with the minimum weight is selected for update based on the weight calculation of Euler distance; and it is iterated; Furthermore, for the centroid update process, the above-mentioned centroid initialization optimization algorithm is iteratively applied to the centroid update process for optimization; wherein, the self-space region selection and its data attribute representation and related data processing are all optional, and when selecting, it is necessary to ensure that the specific data options applied to the initial centroid optimization and the centroid update optimization in the single data optimization process remain consistent.

10. A data structure and / or database generated by the method for grid division of a load complementary distribution network according to any one of claims 1 to 9, comprising a data structure and / or database stored locally, on a server or in the cloud.