Method and device for determining interval power flow of power distribution network scene, and electronic equipment

Abstract

The invention discloses an interval power flow determination method and device for a power distribution network scene and electronic equipment, and relates to the technical field of power distribution networks, and the interval power flow determination method comprises the steps: collecting distributed power generation data of each quarter, and determining a target scene corresponding to each quarter based on the distributed power generation data by adopting a preset clustering algorithm, and based on the distributed power generation data of the target scene, adopting a preset affine algorithm to calculate an initial interval power flow of the target scene, inputting the initial interval power flow into a preset radial basis function neural network model, and outputting a target interval power flow. According to the invention, the technical problem of low accuracy of determining the interval power flow of the power distribution network scene in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network, and in particular to a method for determining interval power flow in a distribution network scenario, a device thereof, and electronic equipment. Background Art

[0002] With the rapid development of power systems dominated by renewable energy, distribution networks have begun to use distributed energy, such as distributed power sources, battery energy storage, and controllable flexible loads. However, due to the uncertainty of distributed power sources and controllable load resources, the homogeneity of distributed resources and loads is relatively serious, which has a great impact on the planning and operation security of distribution networks.

[0003] At present, in order to quickly and accurately analyze the impact of uncertainty generated by distributed power and controllable load resources, most methods such as fuzzy power flow, probabilistic power flow and interval power flow are used to analyze the power flow of uncertainty. Among them, fuzzy power flow characterizes uncertainty by determining the membership density function through fuzzy mathematics, but it is difficult to obtain an accurate membership density function through fuzzy coefficients and fuzzy clustering, resulting in a decrease in accuracy; probabilistic power flow uses random variables to describe uncertainty, such as semi-invariant methods, point estimation and Monte Carlo methods, and the accuracy of probabilistic power flow methods mainly depends on the selected probability density function. Due to the significant uncertainty of distributed power and controllable loads, it is difficult to select an accurate probability density function; interval power flow relies on interval algorithms and is easily affected by dependency problems and warping effects. When the interval algorithm is dealing with multidimensional problems, continuous interval operations will cause the interval calculation results to expand, broaden the boundaries of the interval power flow, and lead to the outbreak of interval errors.

[0004] In the related art, the boundary of the interval power flow is calculated by an affine algorithm. However, the application of the affine algorithm may introduce additional noise elements, which may expand the boundary of the interval power flow and thus affect the scheduling and operation decisions.

[0005] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0006] The embodiments of the present invention provide a method and device for determining interval power flow in a distribution network scenario, and an electronic device thereof, so as to at least solve the technical problem of low accuracy in determining interval power flow in a distribution network scenario in the related art.

[0007] According to one aspect of an embodiment of the present invention, a method for determining interval power flow in a distribution network scenario is provided, comprising: collecting distributed power generation data of each quarter, and based on the distributed power generation data, using a preset clustering algorithm to determine a target scenario corresponding to each quarter; based on the distributed power generation data of the target scenario, using a preset affine algorithm to calculate an initial interval power flow of the target scenario; inputting the initial interval power flow into a preset radial basis function neural network model, and outputting a target interval power flow.

[0008] Furthermore, based on the distributed power generation data, a preset clustering algorithm is used to determine the step of the target scene corresponding to each quarter, including: constructing a wind sequence and a photovoltaic sequence based on the distributed power generation data, wherein the element values ​​in the wind sequence are the wind force at different time points, and the element values ​​in the photovoltaic sequence are the illumination area at different time points; constructing a distance matrix based on the wind sequence and the photovoltaic sequence, wherein the matrix elements in the distance matrix are the minimum cumulative distances between the Euclidean distance and its three nearest neighbors; constructing an objective function based on the distance matrix, and determining the neighborhood distance based on the objective function; determining the minimum number of points, and clustering the wind sequence and the photovoltaic sequence based on the minimum number of points and the neighborhood distance using a preset clustering algorithm to obtain a sample set, wherein the sample set includes: all density-reachable and density-connected data points, and the data points refer to the wind force or illumination area at different time points; determining the wind scene or photovoltaic scene indicated by the data points that do not belong to the sample set as the target scene.

[0009] Furthermore, before calculating the initial interval power flow of the target scenario based on the distributed power generation data of the target scenario by using a preset affine algorithm, it also includes: fitting the noise elements generated by the multiplication operation in the initial affine algorithm to obtain a preset multiplication operation; using the Chebyshev approximation algorithm and the fitting model to eliminate the noise elements generated by the division operation in the initial affine algorithm to obtain a preset division operation; replacing the multiplication operation and the division operation in the initial affine algorithm with the preset multiplication operation and the preset division operation respectively to obtain the preset affine algorithm.

[0010] Furthermore, based on the distributed power generation data of the target scenario, a preset affine algorithm is used to calculate the step of the initial interval power flow of the target scenario, including: determining an electric load node diagram, wherein the electric load node diagram includes: a plurality of load nodes, each of which has a current direction; based on the distributed power generation data of the target scenario, calculating the apparent power and the initial voltage of each of the load nodes; based on the apparent power and the initial voltage, starting from the end node of the electric load node diagram, calculating the affine current to the root node, wherein the end node is the last load node on the electric load node diagram, and the root node is the first load node on the electric load node diagram; based on the affine current and the affine voltage of the previous iteration, calculating the affine voltage of the current iteration, until the absolute difference between the affine voltage of the current iteration and the affine voltage of the previous iteration is less than a preset deviation threshold, thereby obtaining the initial interval power flow.

[0011] Furthermore, before the initial interval flow is input into a preset radial basis function neural network model and the target interval flow is output, it also includes: constructing an initial radial basis function neural network model, wherein the initial radial basis function neural network model includes: a hidden layer, using the radial basis function as the activation function of the neurons in the hidden layer, and each of the neurons in the hidden layer maps the low-dimensional input data to a high-dimensional space through the activation function; using the upper and lower boundary data of the voltage amplitude in the historical interval flow of multiple historical target scenes to train the initial radial basis function neural network model to obtain the trained preset radial basis function neural network model.

[0012] Furthermore, the step of inputting the initial interval power flow into a preset radial basis function neural network model and outputting the target interval power flow includes: using the preset radial basis function neural network model to predict the voltage upper boundary and the voltage lower boundary of the affine voltage in the initial interval power flow to obtain the target voltage upper boundary and the target voltage lower boundary; and determining the target interval power flow based on the target voltage upper boundary and the target voltage lower boundary.

[0013] Furthermore, after the initial interval flow is input into a preset radial basis function neural network model and the target interval flow is output, it also includes: sending the target interval flow of each target scenario to a scheduling terminal, wherein the scheduling terminal determines the power scheduling strategy under the target scenario based on the target interval flow.

[0014] According to another aspect of an embodiment of the present invention, there is also provided an interval flow determination device for a distribution network scenario, comprising: a determination unit, used to collect distributed power generation data of each quarter, and based on the distributed power generation data, use a preset clustering algorithm to determine a target scenario corresponding to each quarter; a calculation unit, used to calculate the initial interval flow of the target scenario based on the distributed power generation data of the target scenario, use a preset affine algorithm; an output unit, used to input the initial interval flow into a preset radial basis function neural network model, and output a target interval flow.

[0015] Further, the determination unit includes: a first construction module, which is used to construct a wind sequence and a photovoltaic sequence based on the distributed power generation data, wherein the element values ​​in the wind sequence are wind forces at different time points, and the element values ​​in the photovoltaic sequence are illumination areas at different time points; a second construction module, which is used to construct a distance matrix based on the wind sequence and the photovoltaic sequence, wherein the matrix elements in the distance matrix are the minimum cumulative distances between the Euclidean distance and its three nearest neighbors; a third construction module, which is used to construct an objective function based on the distance matrix, and determine the neighborhood distance based on the objective function; a first determination module, which is used to determine the minimum number of points, and based on the minimum number of points and the neighborhood distance, cluster the wind sequence and the photovoltaic sequence using a preset clustering algorithm to obtain a sample set, wherein the sample set includes: all data points that are density-reachable and density-connected, and the data points refer to wind forces or illumination areas at different time points; a second determination module, which is used to determine the wind scene or photovoltaic scene indicated by the data points that do not belong to the sample set as the target scene.

[0016] Furthermore, the interval power flow determination device includes: a first processing module, which is used to fit the noise elements generated by the multiplication operation in the initial affine algorithm before calculating the initial interval power flow of the target scenario based on the distributed power generation data of the target scenario using a preset affine algorithm to obtain a preset multiplication operation; a second processing module, which is used to eliminate the noise elements generated by the division operation in the initial affine algorithm using a Chebyshev approximation algorithm and a fitting model to obtain a preset division operation; a first replacement module, which is used to replace the multiplication operation and the division operation in the initial affine algorithm with the preset multiplication operation and the preset division operation respectively to obtain the preset affine algorithm.

[0017] Furthermore, the calculation unit includes: a third determination module, used to determine the power load node diagram, wherein the power load node diagram includes: multiple load nodes, each of which has a current direction; a first calculation module, used to calculate the apparent power and initial voltage of each load node based on the distributed power generation data of the target scenario; a second calculation module, used to calculate the affine current from the end node of the power load node diagram to the root node based on the apparent power and the initial voltage, wherein the end node is the last load node on the power load node diagram, and the root node is the first load node on the power load node diagram; a third calculation module, used to calculate the affine voltage of the current iteration based on the affine current and the affine voltage of the previous iteration, until the absolute difference between the affine voltage of the current iteration and the affine voltage of the previous iteration is less than a preset deviation threshold, thereby obtaining the initial interval flow.

[0018] Furthermore, the interval flow determination device also includes: a fourth construction module, used to construct an initial radial basis function neural network model before inputting the initial interval flow into a preset radial basis function neural network model and outputting the target interval flow, wherein the initial radial basis function neural network model includes: a hidden layer, using the radial basis function as the activation function of the neurons in the hidden layer, and each of the neurons in the hidden layer maps the low-dimensional input data to a high-dimensional space through the activation function; a first training module, used to train the initial radial basis function neural network model using the upper and lower boundary data of the voltage amplitude in the historical interval flows of multiple historical target scenarios to obtain the trained preset radial basis function neural network model.

[0019] Furthermore, the output unit includes: a first prediction module, used to use the preset radial basis function neural network model to predict the voltage upper boundary and the voltage lower boundary of the affine voltage in the initial interval flow, and obtain the target voltage upper boundary and the target voltage lower boundary; a fourth determination module, used to determine the target interval flow based on the target voltage upper boundary and the target voltage lower boundary.

[0020] Furthermore, the interval power flow determination device also includes: a first sending module, which is used to send the target interval power flow of each target scenario to the scheduling terminal after inputting the initial interval power flow into a preset radial basis function neural network model and outputting the target interval power flow, wherein the scheduling terminal determines the power dispatching strategy under the target scenario based on the target interval power flow.

[0021] According to another aspect of an embodiment of the present invention, a computer program product is also provided, including a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the interval power flow determination method for the distribution network scenario described in any one of the above-mentioned items is implemented.

[0022] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement any of the above-mentioned methods for determining interval power flow in the distribution network scenario.

[0023] In the present invention, distributed power generation data of each quarter is collected, and based on the distributed power generation data, a preset clustering algorithm is used to determine the target scenario corresponding to each quarter, and based on the distributed power generation data of the target scenario, a preset affine algorithm is used to calculate the initial interval flow of the target scenario, and the initial interval flow is input into a preset radial basis function neural network model, and the target interval flow is output, thereby solving the technical problem of low accuracy in determining the interval flow of the distribution network scenario in the related art.

[0024] In the present invention, by collecting distributed power generation data of different quarters, and based on the distributed power generation data, a preset clustering algorithm is used to determine the target scenarios (i.e., extreme scenarios) corresponding to different quarters, and then based on the distributed power generation data of the target scenario, a preset affine algorithm can be used to calculate the initial interval flow of the target scenario, and then the initial interval flow of the target scenario is input into a preset radial basis function neural network model to predict the upper and lower boundaries of the interval flow to obtain the target interval flow, which can reduce the boundary error of the determined interval flow, thereby improving the accuracy of the target interval flow boundary. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0026] Figure 1 is a flow chart of an optional interval power flow determination method for a distribution network scenario according to an embodiment of the present invention;

[0027] Figure 2 is a schematic diagram of an optional overall error optimization process according to an embodiment of the present invention;

[0028] Figure 3 is a schematic diagram of an optional radial basis function neural network structure according to an embodiment of the present invention;

[0029] Figure 4 is a flow chart of an optional distribution network extreme scenario analysis method according to an embodiment of the present invention;

[0030] Figure 5 is a schematic diagram of an optional interval power flow determination device for a distribution network scenario according to an embodiment of the present invention;

[0031] Figure 6 It is a hardware structure block diagram of an electronic device (or mobile device) for a method for determining interval power flow in a distribution network scenario according to an embodiment of the present invention. DETAILED DESCRIPTION

[0032] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0033] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0034] It should be noted that the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) collected and involved in the present invention are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data are in compliance with the relevant laws, regulations and standards of the relevant regions, necessary confidentiality measures are taken, and public order and good customs are not violated, and corresponding operation entrances are provided for users to choose to authorize or refuse. For example, an interface is set between the system and the relevant users or organizations. Before obtaining relevant information, it is necessary to send an acquisition request to the aforementioned user or organization through the interface, and obtain relevant information after receiving the consent information fed back by the aforementioned user or organization.

[0035] In the present invention, in order to solve the interval expansion problem caused by non-affine operation in the affine algorithm and improve the calculation efficiency, an iterative method is used to integrate the non-affine operation noise, the affine algorithm is extended to the polar coordinate system, and the Chebyshev and minimum distance approximation are used to achieve the best affine approximation, thereby ensuring accurate tracking of variable interdependencies. Although the noise norm of the non-affine operation can be corrected, the influence of the noise coefficient of independent uncertainty on the central value is ignored. Therefore, the present invention proposes an extreme interval power flow calculation method based on an improved affine algorithm. First, the density-based noisy spatial clustering based on dynamic time warping can be used to identify the extreme scenarios of power output and load demand of distributed power sources, and an improved affine arithmetic method is used to scale the non-affine operation noise of multiplication and division operations into independent noise elements. At the same time, an immune algorithm is used to optimize the noise coefficient related to the extreme scenario, effectively reducing the influence of the non-affine operation noise term on the extreme interval power flow analysis. In addition, a radial basis function neural network algorithm is used to optimize the boundary of the extreme interval power flow, effectively controlling and suppressing the expansion of the upper and lower limits in the extreme interval power flow.

[0036] The present invention is described in detail below in conjunction with various embodiments.

[0037] Embodiment 1

[0038] According to an embodiment of the present invention, an embodiment of a method for determining interval power flow in a distribution network scenario is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.

[0039] Figure 1 is a flow chart of an optional method for determining interval power flow in a distribution network scenario according to an embodiment of the present invention. Figure 1As shown, the method comprises the following steps:

[0040] Step S101 : Collect distributed power generation data for each quarter, and based on the distributed power generation data, use a preset clustering algorithm to determine a target scenario corresponding to each quarter.

[0041] Optionally, distributed power generation data (such as wind power generation data and photovoltaic power generation data) of each quarter can be collected first. Based on the distributed power generation data, a preset clustering algorithm (such as DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm based on DTW (Dynamic Time Warping)) can be used to determine the target scenario corresponding to each quarter (i.e., extreme scenarios in different seasons).

[0042] In the embodiment of the present invention, an extreme scenario refers to a scenario that rarely occurs, for example, a scenario with no continuous wind power generation or no continuous photovoltaic power generation, or a scenario with high power demand of the power grid.

[0043] Step S102, based on the distributed power generation data of the target scenario, a preset affine algorithm is used to calculate the initial interval power flow of the target scenario.

[0044] Optionally, an affine algorithm (i.e., an algorithm that uses the idea of ​​affine transformation to deal with problems related to uncertainty and parameter dependence) can record the dependency between actual parameters (such as node voltage, actual output power of a wind farm) and various uncertain variables (such as wind speed). The affine algorithm includes affine operations (i.e., a linear operation method, including operations such as addition, subtraction, and scalar multiplication) and non-affine operations (i.e., operations that do not satisfy the properties of affine transformation, including operations such as multiplication and division). The result obtained by the affine algorithm is more accurate than the result obtained by the interval operation. However, the non-affine operation in the current affine algorithm will generate noise elements, which will affect the results and complexity of the interval flow. Therefore, this embodiment proposes an improved preset affine algorithm, which can be used according to the distributed power generation data of the target scene to calculate the initial interval flow of the target scene (such as the voltage upper and lower boundaries of the affine voltage).

[0045] Step S103, inputting the initial interval power flow into a preset radial basis function neural network model, and outputting the target interval power flow.

[0046] In the embodiment of the present invention, although the improved affine algorithm can provide a reasonable and accurate upper bound for the extreme interval power flow, the lower bound is still susceptible to changes in the output of the distributed power source and the user load. In order to accurately capture the impact of the output and load size of the distributed power source on the lower bound of the extreme internal power flow and fit the nonlinear relationship, two RBFNNs (i.e., radial basis function neural networks) trained with the upper and lower bounds of the extreme interval power flow amplitude voltage can be used to obtain a preset radial basis function neural network model, and then the upper and lower bounds of the lower voltage of the input initial interval power flow are predicted by the trained preset radial basis function neural network model to obtain an accurate target interval power flow.

[0047] In an embodiment of the present invention, the preset radial basis function neural network model includes an input layer, a hidden layer and an output layer. The preset radial basis function neural network model maps the input data to a high-dimensional space so that the data that is originally linearly inseparable in the low-dimensional space becomes linearly separable in the high-dimensional space, thereby finding the optimal solution or optimal boundary.

[0048] In summary, by collecting distributed power generation data of different quarters and using a preset clustering algorithm based on the distributed power generation data, the target scenarios corresponding to different quarters can be determined, and then the initial interval flow of the target scenario can be calculated based on the distributed power generation data of the target scenario using a preset affine algorithm. After that, the initial interval flow of the target scenario is input into a preset radial basis function neural network model. The preset radial basis function neural network model can output the target interval flow, reducing the boundary error of the interval flow, thereby solving the technical problem of low accuracy in determining the interval flow of the distribution network scenario in the related technology.

[0049] In order to accurately determine the target scene, in the interval power flow determination method of the distribution network scene provided in Example 1 of the present application, based on distributed power generation data, a wind sequence and a photovoltaic sequence are constructed, wherein the element values ​​in the wind sequence are the wind force at different time points, and the element values ​​in the photovoltaic sequence are the illumination area at different time points; based on the wind sequence and the photovoltaic sequence, a distance matrix is ​​constructed, wherein the matrix elements in the distance matrix are the minimum cumulative distances between the Euclidean distance and its three nearest neighbors; based on the distance matrix, an objective function is constructed, and based on the objective function, the neighborhood distance is determined; the minimum number of points is determined, and based on the minimum number of points and the neighborhood distance, a preset clustering algorithm is used to cluster the wind sequence and the photovoltaic sequence to obtain a sample set, wherein the sample set includes: all density-reachable and density-connected data points, and the data points refer to the wind force or illumination area at different time points; the wind scene or photovoltaic scene indicated by the data points that do not belong to the sample set is determined as the target scene.

[0050] Optionally, due to the homogeneity (i.e., similarity or consistency in specific attributes or characteristics) and sequential time changes in the power output of distributed power sources and load demand, for example, some industrial loads have daily or seasonal fluctuation patterns similar to distributed power sources, and the Euclidean distance and Minkowski distance are not sufficient to effectively capture the time dynamics. Therefore, the DTW distance can be used to identify the time dynamics of distributed power and load power output. DTW uses dynamic programming technology to search for all potential alignment paths between two sequences and select the path with the smallest cumulative distance. Based on the distributed power generation data, a wind power sequence (the element values ​​in the wind power sequence are the wind power at different time points) and a photovoltaic sequence (the element values ​​in the photovoltaic sequence are the illumination area at different time points) can be constructed, and then a distance matrix (the matrix elements in the distance matrix are the minimum cumulative distances between the Euclidean distance and its three nearest neighbors) can be constructed based on the distributed power generation data.

[0051] For example, it is assumed that the wind power sequence and the photovoltaic sequence are: sequence Q = (q1, q2, ..., q n ) and the sequence C = (c1, c2, ..., c m ), the sequence lengths are n and m respectively. We can first construct an n×m distance matrix, where each element (i, j) in the distance matrix is ​​the minimum cumulative distance γ(i, j) between the Euclidean distance of (i, j) and its three nearest neighbors {[i-1, j-1], [i, j-1][i-1, j]}, and γ(i, j) can be defined as follows:

[0052] γ(i, j) = d(q i , c j )+min{γ(i-1, j-1), γ(i-1, j), γ(i, j-1)} (1);

[0053] Among them, d(q i , c j ) represents the distance between the i-th point of sequence Q and the j-th point of sequence C, γ(i-1, j-1) represents the minimum cumulative distance of element (i-1, j-1), γ(i-1, j) represents the minimum cumulative distance of element (i-1, j), and γ(i, j-1) represents the minimum cumulative distance of element (i, j-1).

[0054] In the embodiment of the present invention, the objective function of the DTW distance is used to determine the minimum cumulative distance between two time series according to the distance matrix. The objective function can be expressed as: where P is the set of all possible paths in the two sequences, K is the length of the warped path, and w k is the kth element of the distance matrix, Represents the distance between the points corresponding to the kth element of the distance matrix.

[0055] Then, based on the objective function, the neighborhood distance, such as Eps (epsilon, used to define the neighborhood radius of a point), and Minpoints (the minimum number of points, used to define how many points must be within the neighborhood distance of a sample point to consider the point as a core object) can be determined at the same time.

[0056] If a point contains more than the minimum number of points within the neighborhood distance, the point is a core object. Then, according to the determined minimum number of points and the determined neighborhood distance, a preset clustering algorithm (such as the DBSCAN clustering algorithm based on DTW) is used to cluster the wind power sequence and the photovoltaic sequence to obtain a sample set (including: all data points with density reachability and density connection, which refer to the wind power or light area at different time points).

[0057] In the embodiment of the present invention, it is assumed that there are q points, point p is located within the neighborhood distance of point q, and the number of points within the neighborhood distance of point q is greater than the minimum number of points (i.e. and in, represents the neighborhood distance of point q, represents the minimum number of points within the distance of the q-point domain), then the q-point is the core object (i.e., the center point), and the p-point and the points within the core object domain can be density-reachable by the q-point. Within the neighborhood distance of the density-reachable sample sequence, all samples are density-connected to each other.

[0058] In an embodiment of the present invention, clustering algorithms (such as K-means (K-means clustering algorithm, i.e. K-means clustering algorithm), Gaussian mixture model and fuzzy C-means) will misclassify rare events as outliers and exclude them. In this embodiment, a DTW-based DBSCAN clustering algorithm can be used, which can perform clustering without pre-specifying the number of cluster centers, and can obtain a sample set (i.e., cluster clusters) and data points that do not belong to the sample set (i.e., sample points outside the cluster), and the wind scene or photovoltaic scene indicated by the data points that do not belong to the sample set is determined as the target scene.

[0059] In order to accurately determine the preset affine algorithm, in the interval power flow determination method for the distribution network scenario provided in Example 1 of the present application, the noise elements generated by the multiplication operation in the initial affine algorithm are fitted to obtain the preset multiplication operation; the Chebyshev approximation algorithm and the fitting model are used to eliminate the noise elements generated by the division operation in the initial affine algorithm to obtain the preset division operation; the preset multiplication operation and the preset division operation are respectively replaced by the multiplication operation and the division operation in the initial affine algorithm to obtain the preset affine algorithm.

[0060] In this embodiment, the affine algorithm can record the dependency between the actual parameters and various uncertain variables. Due to this additional information, the results obtained by affine operation are often more accurate than those obtained by interval operation. It can be expressed as: Among them, ε i represents noise, ε i The value range of x is [-1, 1]; i ,y i They represent the deviations of the affine variables respectively; x0 and y0 represent the center values ​​respectively; and n represents the total number of noise elements.

[0061] In this embodiment, the affine operation includes an affine operation and a non-affine operation. The affine operation includes operations such as addition, subtraction, and scalar multiplication. The noise ε i (its value range is [-1, 1]), constant λ, deviation of the affine variable (x i ,y i ), center value (x0, y0), and the total number of noise elements n to construct the affine algorithm structure For example, affine operations such as addition and subtraction (i.e. ) and scalar multiplication (i.e. ).

[0062] In the embodiment of the present invention, the multiplication and division of non-affine operations will generate new operation noise terms, which are calculated as follows:

[0063]

[0064]

[0065] Among them, A, B, C, and D represent the derivation coefficients used in non-affine operations, ε n+1 , ε′ n+1 and ε′ n+2 The noise elements are generated by the product operation of non-affine operation and the two noise elements are generated by the division operation.

[0066] In the embodiment of the present invention, since the product operation of the current non-affine operation will generate a noise element ε n+1 , and the division operation will produce a noise element ε′ n+1 and ε′ n+2 , thus affecting the accuracy of the interval power flow results. In this embodiment, the noise elements generated by the multiplication operation in the initial affine algorithm (ie, non-affine operation) can be fitted to obtain the preset multiplication operation:

[0067]

[0068] Among them, m is a parameter, which is used to represent the error term.

[0069] At the same time, the Chebyshev approximation algorithm and the fitting model are used to eliminate the noise elements generated by the division operation in the initial affine algorithm to obtain the preset division operation:

[0070]

[0071]

[0072] Among them, α, β, δ are coefficients, and a and b are The upper and lower boundaries of is the error term in the Chebyshev approximation.

[0073] The preset multiplication operation and the preset division operation are used to replace the multiplication operation and the division operation in the initial affine algorithm respectively, so as to obtain the preset affine algorithm.

[0074] In the embodiment of the present invention, an immune algorithm (used to adjust and optimize noise coefficients related to extreme scenarios in power systems, such as a negative selection algorithm) may be used to optimize noise elements and fitting parameters of non-affine operations.

[0075] Exemplarily, in a certain node system, the fitness function of the immune algorithm can minimize the aggregate error between the actual values ​​and the affine calculated values ​​of 33 nodes, and the parameters of the immune algorithm, such as the number of iterations and the population size, can be set to 20.

[0076] Figure 2 is a schematic diagram of an optional overall error optimization process according to an embodiment of the present invention, such as Figure 2 As shown in the figure, the horizontal axis represents the number of iterations of the immune algorithm (such as 20), and the vertical axis represents the overall errors. Figure 2 The process of significantly reducing the overall error with increasing iteration number is shown, which indicates that the immune algorithm can optimize independent noise elements and improve the accuracy of upper and lower boundaries in extreme interval power flow.

[0077] In order to improve the accuracy of the initial interval power flow, in the interval power flow determination method of the distribution network scenario provided in Example 1 of the present application, an electric load node diagram is determined, wherein the electric load node diagram includes: multiple load nodes, each load node has a current direction; based on the distributed power generation data of the target scenario, the apparent power and initial voltage of each load node are calculated; based on the apparent power and the initial voltage, starting from the end node of the electric load node diagram, the affine current is calculated to the root node, wherein the end node is the last load node on the electric load node diagram, and the root node is the first load node on the electric load node diagram; based on the affine current and the affine voltage of the previous iteration, the affine voltage of the current iteration is calculated until the absolute difference between the affine voltage of the current iteration and the affine voltage of the previous iteration is less than the preset deviation threshold, so as to obtain the initial interval power flow.

[0078] Optionally, the power load node diagram may contain multiple load points, each load node represents a power consumption point, the load nodes are connected by power lines, and each load node has a current direction, and the apparent power of each load node in the power load node diagram can be calculated based on the distributed power generation data of the target scenario. and initial voltage

[0079]

[0080] in, represents the apparent power of the nth load node, represents the initial voltage of the nth load node, S n，n Indicates the influence of the nth noise element of the nth node on the apparent power of the node, ε n represents the nth noise element.

[0081] According to the apparent power and the initial voltage, reverse iteration is performed (i.e., starting from the end node of the power load node diagram (i.e., the last load node on the power load node diagram), and calculating the affine current toward the root node (i.e., the first load node on the power load node diagram)), which can be calculated by the following formula:

[0082]

[0083] in, represents the current of the jth load node during the kth iteration, represents the apparent power of the jth load node, represents the initial voltage of the jth load node during the k-1th iteration, represents the branch current during the kth iteration, represents the load node current during the kth iteration, and A represents the load node branch correlation matrix.

[0084] Then, forward iteration is performed to calculate the voltage drop of each branch (i.e. ), and according to the root node affine voltage, calculate the branch end affine voltage of the iteration, and convert the affine value into a complex number:

[0085]

[0086]

[0087] in, represents the affine voltage at load node j during the kth iteration, represents the affine voltage at load node j-1 during the kth iteration, R j +jX j represents resistance and reactance, represents the branch current during the kth iteration, Re represents the real part of the complex number, Im represents the imaginary part of the complex number, represents the influence of the maximum noise element at load node j on the voltage of the load node during the kth iteration, represents the influence of the minimum noise element at load node j on the voltage of the load node during the kth iteration, It represents the influence of the nth noise element at load node j on the voltage of the load node during the kth iteration.

[0088] Until the absolute difference between the affine voltage of the current iteration and the affine voltage of the previous iteration is less than the preset deviation threshold ε (i.e. ), when the convergence criterion is met, the iteration is stopped and the initial interval power flow (such as the amplitude and phase interval of the output load node voltage) is output, otherwise the iteration continues.

[0089] In the embodiment of the present invention, the affine power flow calculation provides a voltage interval estimate for each node in the power load node diagram. Through the affine power flow calculation, the power flow distribution in the power load node diagram can be analyzed.

[0090] In order to improve the prediction accuracy of the preset radial basis function neural network model, in the interval flow determination method of the distribution network scenario provided in Example 1 of the present application, an initial radial basis function neural network model is constructed, wherein the initial radial basis function neural network model includes: a hidden layer, using the radial basis function as the activation function of the neurons in the hidden layer, and each neuron in the hidden layer maps the low-dimensional input data to the high-dimensional space through the activation function; using the upper and lower boundary data of the voltage amplitude in the historical interval flow of multiple historical target scenarios, the initial radial basis function neural network model is trained to obtain the trained preset radial basis function neural network model.

[0091] Optionally, an initial radial basis function neural network model (i.e., RBFNN (radial basis function neural network) is constructed. The initial radial basis function neural network model includes a hidden layer, wherein the hidden layer includes multiple neurons, and the output of the neurons can be generated by an activation function. In an embodiment of the present invention, a radial basis function (such as a Gaussian function) can be used as the activation function of the neurons in the hidden layer. Each neuron in the hidden layer maps the low-dimensional input data to the high-dimensional space through the activation function, so that the input data that is originally linearly inseparable in the low-dimensional space becomes linearly separable in the high-dimensional space, thereby alleviating and eliminating the conservatism of the extreme interval current. The RBFNN can be expressed by the following formula:

[0092]

[0093] Among them, w i Represents model weight, RBF i represents the Gaussian kernel formed by the Gaussian function, and b represents the model hyperparameter.

[0094] Figure 3 is a schematic diagram of an optional radial basis function neural network structure according to an embodiment of the present invention, such as Figure 3 As shown, the radial basis function neural network includes a data input layer X, a hidden layer, and an output layer Y. The input data is transmitted from the input layer to the hidden layer. The connection between each hidden layer neuron and the neurons in other layers has a weight, such as W (1，1) and bias terms such as b 1 , and each neuron in the hidden layer calculates its radial basis function value (i.e. Where x represents the input vector, σ represents the standard deviation, and exp represents the natural exponential function). The output of the hidden layer (i.e., the radial basis function value) is linearly combined to generate the final network output.

[0095] In an embodiment of the present invention, by collecting the upper and lower boundary data of the voltage amplitude in the historical interval flow of the historical target scene, the initial radial basis function neural network model can be trained using the collected data, and a trained preset radial basis function neural network model (i.e., a model that can realize the prediction of the upper and lower boundaries of the voltage amplitude in the interval flow) can be obtained.

[0096] In order to determine the target interval flow, in the interval flow determination method for the distribution network scenario provided in Example 1 of the present application, a preset radial basis function neural network model is used to predict the voltage upper boundary and the voltage lower boundary of the affine voltage in the initial interval flow to obtain the target voltage upper boundary and the target voltage lower boundary; based on the target voltage upper boundary and the target voltage lower boundary, the target interval flow is determined.

[0097] Optionally, the immune algorithm optimized affine algorithm can provide a reasonably accurate upper boundary for extreme interval power flow, but the lower boundary is still easily affected by the output of distributed power sources and changes in user load. The RBFNN trained with the upper and lower boundary data of the voltage amplitude in the historical interval power flow of the historical target scenario can be integrated into the improved affine algorithm optimized by the immune algorithm. The trained RBFNN (i.e., the preset radial basis function neural network model) can predict the voltage upper boundary and the voltage lower boundary of the affine voltage in the initial interval power flow to obtain the target voltage upper boundary and the target voltage lower boundary. According to the predicted target voltage upper boundary and the target voltage lower boundary, the target interval power flow can be obtained.

[0098] In the embodiment of the present invention, the improved affine algorithm coupled with RBFNN can reduce the error of the lower boundary of the voltage amplitude in the extreme interval power flow. The addition of RBFNN enhances the nonlinear fitting capability and alleviates the inherent over-conservatism in the extreme interval power flow analysis.

[0099] In order to determine the power dispatching strategy under the target scenario, in the interval flow determination method for the distribution network scenario provided in Example 1 of the present application, the target interval flow of each target scenario is sent to the dispatching terminal, wherein the dispatching terminal determines the power dispatching strategy under the target scenario based on the target interval flow.

[0100] Optionally, the predicted target interval flow can provide a scheduling reference for the scheduling terminal. The target interval flow of each target scenario is sent to the scheduling terminal. The scheduling terminal can identify the risks that may cause voltage instability or voltage collapse based on the target interval flow, and can formulate corresponding power scheduling strategies. For example, under extreme weather conditions, the volatility of wind power generation can cause grid voltage instability. The power generation resources can be optimized or the backup power supply and energy storage equipment can be started to protect the stable and safe operation of the power system.

[0101] Figure 4 is a flow chart of an optional distribution network extreme scenario analysis method according to an embodiment of the present invention, such as Figure 4 As shown, the process includes the following: S1: extracting extreme scenarios based on the DBSCAN clustering algorithm of DTW distance; S2: using the improved affine algorithm model (i.e., the improved affine algorithm coupled with RBFNN) to obtain the target interval power flow; S3: scenario analysis, i.e., analyzing the target interval power flow under extreme scenarios (including the accuracy analysis of the improved affine algorithm and the efficiency analysis of the improved affine algorithm calculation).

[0102] In an optional embodiment, the following may be used: Figure 4 The distribution network extreme scenario analysis method flow shown determines the interval power flow of the distribution network scenario.

[0103] In an embodiment of the present invention, the target scenario can be extracted based on the DBSCAN clustering algorithm of the DTW distance, and the improved affine algorithm can be used to calculate the interval power flow under the target scenario. Then, the trained RBFNN model is used to predict the interval power flow under the target scenario to obtain the target interval power flow, which effectively controls and suppresses the expansion of the upper and lower limits in the extreme interval power flow.

[0104] Another optional specific implementation is described in detail below.

[0105] In an embodiment of the present invention, a distribution network extreme scenario analysis method based on DBSCAN clustering and improved affine power flow is proposed. First, DBSCAN based on DTW is used to identify the extreme conditions of power output and load demand of distributed power sources, and then the improved affine arithmetic method is used to fit the noise elements generated by multiplication operations. At the same time, Chebyshev approximation and fitting models are used to eliminate the noise elements in non-affine operation division, and the immune algorithm is applied to optimize the noise elements and fitting parameters of non-affine operations. Then, the RBFNN algorithm is used to optimize the boundaries of extreme interval power flows, which effectively controls and suppresses the expansion of upper and lower limits in extreme interval power flows.

[0106] The DTW-based DBSCAN algorithm proposed in the embodiment of the present invention is used to extract extreme scenarios of distributed power sources and user loads, and then an improved affine algorithm model is used to perform interval power flow analysis under extreme scenarios.

[0107] Extreme scenarios refer to scenarios that rarely occur and have a larger DTW distance from other scenarios, so extreme scenarios can be assigned a -1 mark. The time resolution of extreme cases can be set to 24 hours and the interval to 15 minutes.

[0108] By extracting extreme scenarios of distributed power sources and user loads through the DTW-based DBSCAN algorithm, we can obtain the power fluctuations of photovoltaic power generation, wind power generation and user load in extreme scenarios in spring, summer, autumn and winter respectively. From the power fluctuations, we can see that the output of wind power generation and photovoltaic power generation in extreme cases is much more volatile than in other cases, and the frequency of occurrence is only once. At the same time, the user load power in extreme scenarios is also greater than in other scenarios.

[0109] According to the obtained extreme scenario, the improved affine algorithm can be used to calculate the extreme interval power flow. Taking a certain node system as an example, the amplitude and phase of the node voltage can be specified as the state variables of the extreme interval power flow analysis, and the immune algorithm is used to optimize the noise elements and fitting parameters of the non-affine operation. The fitness function of the immune algorithm aims to minimize the aggregation error between the actual value and the affine calculated value at 33 nodes. The parameters of the immune algorithm can be set (such as the number of iterations is set to 20, the group size is set to 20), and the optimization process is as follows: Figure 2 As shown, the independent noise elements optimized by the immune algorithm significantly improve the accuracy of the upper and lower boundaries in the extreme interval power flow.

[0110] The amplitude and phase of the upper and lower boundaries of the voltage calculated using the improved affine algorithm in extreme scenarios in different seasons (i.e., spring, summer, autumn, and winter) are compared with the actual values. It can be seen that the upper boundary amplitude and phase of the node voltage calculated using the improved affine algorithm are closely consistent with the actual values ​​during the extreme interval flow, and the lower boundary amplitude and phase of the node voltage calculated using the improved affine algorithm show significant volatility, which is mainly caused by distributed power sources and user loads. During peak power load periods, especially when the output of wind power generation and photovoltaic power generation is reduced due to extreme weather conditions, the node voltage amplitude drops significantly. This is because the improved affine algorithm does not fully and accurately optimize the independent noise elements related to wind and photovoltaic power generation and the noise elements related to user volatility. Therefore, in this embodiment, the immune algorithm is continued to be used to optimize the affine algorithm to provide a reasonably accurate upper bound for the extreme interval power flow, but the lower bound is still easily affected by changes in the output of distributed power sources and user loads. In order to accurately capture the impact of distributed power output and load size on the lower bound of the extreme internal power flow and fit the nonlinear relationship, the RBFNN trained using the upper and lower bounds of the extreme interval power flow amplitude voltage can be integrated into the improved affine algorithm optimized by the immune algorithm. The obtained RBFNN can predict the upper and lower bounds of the voltage under the extreme interval power flow.

[0111] The predicted values ​​obtained by RBFNN for the upper and lower boundaries of the voltage under extreme interval power flow are calculated with the actual values, and the error of the predicted values ​​can be obtained. According to the error, the error between the lower limit of the voltage amplitude in the extreme interval power flow prediction in spring and the actual value is greater than that in summer, autumn and winter, while the error between the lower limit of the voltage amplitude in the extreme interval power flow prediction in autumn and winter is smaller than that in summer. Compared with the improved affine operation alone, the improved affine algorithm coupled with RBFNN can reduce the error associated with the lower boundary of the voltage amplitude in extreme interval power flow. RBFNN enhances the nonlinear fitting ability and alleviates the inherent over-conservatism in extreme interval power flow analysis.

[0112] The improved affine algorithm of the embodiment of the present invention is compared with the Monte Carlo method (MCM (Monte Carlo method, a computing technology based on random sampling)). In the Monte Carlo method, the photovoltaic power generation output can be modeled as a normal distribution, the wind power generation output can be modeled as a Weibull distribution, and the user load power can be modeled as a Gaussian distribution. The number of MCM samplings is set to 10000, and the effects of the improved affine algorithm and the Monte Carlo method are compared and analyzed from the perspectives of accuracy and computational efficiency.

[0113] (1) Accuracy comparison analysis: The actual values ​​of the upper and lower boundaries of voltage in extreme scenarios in different seasons (i.e., spring, summer, autumn, and winter), the upper and lower boundary values ​​of voltage calculated by the improved radiation algorithm, and the upper and lower boundary values ​​of voltage calculated by the Monte Carlo method are compared respectively. It can be seen that the lower limit of the voltage amplitude at nodes 20, 21, and 22 obtained by the MCM algorithm deviates significantly from the actual value and the result of the affine algorithm, and the upper boundary of the voltage amplitude obtained by the MCM algorithm is also lower than the actual value and the affine value. Except for nodes 20, 21, and 22 under heavy load, the lower limit of the voltage amplitude of MCM is close to the actual value, and the error between the upper and lower boundaries of the voltage amplitude calculated by MCM and the actual value is greater than that of the improved affine algorithm of the coupled RBFNN. It can be seen that compared with the improved affine algorithm of the coupled RBFNN proposed in the embodiment of the present invention, the sampling of the voltage amplitude boundary of the extreme interval power flow by MCM is too conservative.

[0114] In this embodiment, the accuracy of the extreme interval power flow voltage amplitude boundary is mainly affected by the selection of the distribution function and the random sampling process, and the upper and lower limits of the extreme interval power flow voltage amplitude are intrinsically related to the extreme scenarios of distributed power sources and loads. However, the distribution function used for fitting can include a wide range of scenarios, not just focusing on extreme conditions, and the fitting function cannot accurately represent the actual output and load size of the distributed power source, which may lead to errors, and the MCM random sampling retains the inherent uncertainty of a large number of samples, making it difficult to ensure accuracy.

[0115] (2) Computational efficiency comparison analysis: The improved affine algorithm coupled with RBFNN completed the calculation of the extreme interval power flow of the overall extreme scenario of distributed power generation and load in 245 seconds, and obtained a 96-point curve. The MCM simulation algorithm took 1427 seconds to complete the same task. It can be seen that the computational efficiency of the improved affine algorithm coupled with RBFNN is better than that of the MCM algorithm.

[0116] The following is a detailed description in conjunction with another embodiment.

[0117] Embodiment 2

[0118] The interval power flow determination device for a distribution network scenario provided in this embodiment includes multiple implementation units, and each implementation unit corresponds to each implementation step in the above-mentioned embodiment 1.

[0119] Figure 5 is a schematic diagram of an optional interval power flow determination device for a distribution network scenario according to an embodiment of the present invention, such as Figure 5 As shown, the section power flow determination device may include: a determination unit 50 , a calculation unit 51 , and an output unit 52 .

[0120] The determination unit 50 is used to collect distributed power generation data for each quarter, and based on the distributed power generation data, use a preset clustering algorithm to determine the target scenario corresponding to each quarter;

[0121] A calculation unit 51 is used to calculate the initial interval power flow of the target scenario based on the distributed power generation data of the target scenario by using a preset affine algorithm;

[0122] The output unit 52 is used to input the initial interval power flow into a preset radial basis function neural network model and output the target interval power flow.

[0123] The above-mentioned interval power flow determination device can collect the distributed power generation data of each quarter through the determination unit 50, and based on the distributed power generation data, use a preset clustering algorithm to determine the target scenario corresponding to each quarter, and use a preset affine algorithm based on the distributed power generation data of the target scenario through the calculation unit 51 to calculate the initial interval power flow of the target scenario, and input the initial interval power flow into the preset radial basis function neural network model through the output unit 52 to output the target interval power flow.

[0124] Optionally, the determination unit includes: a first construction module, used to construct a wind sequence and a photovoltaic sequence based on distributed power generation data, wherein the element values ​​in the wind sequence are the wind force at different time points, and the element values ​​in the photovoltaic sequence are the illumination area at different time points; a second construction module, used to construct a distance matrix based on the wind sequence and the photovoltaic sequence, wherein the matrix elements in the distance matrix are the minimum cumulative distances between the Euclidean distance and its three nearest neighbors; a third construction module, used to construct an objective function based on the distance matrix, and determine the neighborhood distance based on the objective function; a first determination module, used to determine the minimum number of points, and based on the minimum number of points and the neighborhood distance, use a preset clustering algorithm to cluster the wind sequence and the photovoltaic sequence to obtain a sample set, wherein the sample set includes: all density-reachable and density-connected data points, and the data points refer to the wind force or illumination area at different time points; a second determination module, used to determine the wind scene or photovoltaic scene indicated by the data points that do not belong to the sample set as the target scene.

[0125] Optionally, the interval power flow determination device includes: a first processing module, which is used to fit the noise elements generated by the multiplication operation in the initial affine algorithm before calculating the initial interval power flow of the target scenario based on the distributed power generation data of the target scenario using a preset affine algorithm to obtain a preset multiplication operation; a second processing module, which is used to eliminate the noise elements generated by the division operation in the initial affine algorithm using a Chebyshev approximation algorithm and a fitting model to obtain a preset division operation; a first replacement module, which is used to replace the multiplication operation and the division operation in the initial affine algorithm with a preset multiplication operation and a preset division operation respectively to obtain a preset affine algorithm.

[0126] Optionally, the calculation unit includes: a third determination module, used to determine the power load node diagram, wherein the power load node diagram includes: multiple load nodes, each load node has a current direction; a first calculation module, used to calculate the apparent power and initial voltage of each load node based on the distributed power generation data of the target scenario; a second calculation module, used to calculate the affine current from the end node of the power load node diagram to the root node based on the apparent power and the initial voltage, wherein the end node is the last load node on the power load node diagram, and the root node is the first load node on the power load node diagram; a third calculation module, used to calculate the affine voltage of the current iteration based on the affine current and the affine voltage of the previous iteration, until the absolute difference between the affine voltage of the current iteration and the affine voltage of the previous iteration is less than a preset deviation threshold, thereby obtaining the initial interval flow.

[0127] Optionally, the interval flow determination device also includes: a fourth construction module, used to construct an initial radial basis function neural network model before inputting the initial interval flow into the preset radial basis function neural network model and outputting the target interval flow, wherein the initial radial basis function neural network model includes: a hidden layer, using the radial basis function as the activation function of the neurons in the hidden layer, and each neuron in the hidden layer maps the low-dimensional input data to the high-dimensional space through the activation function; a first training module, used to train the initial radial basis function neural network model using the upper and lower boundary data of the voltage amplitude in the historical interval flows of multiple historical target scenarios to obtain the trained preset radial basis function neural network model.

[0128] Optionally, the output unit includes: a first prediction module, used to predict the voltage upper boundary and the voltage lower boundary of the affine voltage in the initial interval flow using a preset radial basis function neural network model to obtain a target voltage upper boundary and a target voltage lower boundary; a fourth determination module, used to determine the target interval flow based on the target voltage upper boundary and the target voltage lower boundary.

[0129] Optionally, the interval power flow determination device also includes: a first sending module, which is used to send the target interval power flow of each target scenario to the scheduling terminal after inputting the initial interval power flow into a preset radial basis function neural network model and outputting the target interval power flow, wherein the scheduling terminal determines the power scheduling strategy under the target scenario based on the target interval power flow.

[0130] The above-mentioned interval flow determination device may also include a processor and a memory. The above-mentioned determination unit 50, calculation unit 51, output unit 52, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions.

[0131] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the initial interval power flow is input into the preset radial basis function neural network model by adjusting the kernel parameters to output the target interval power flow.

[0132] The above-mentioned memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one storage chip.

[0133] According to another aspect of an embodiment of the present invention, a computer program product is also provided, including a non-volatile computer-readable storage medium, the non-volatile computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the interval flow determination method for the distribution network scenario described above is implemented.

[0134] When the computer program product is executed on a data processing device, it is suitable for executing an initialization program having the following method steps: collecting distributed power generation data for each quarter, and based on the distributed power generation data, using a preset clustering algorithm to determine the target scenario corresponding to each quarter, based on the distributed power generation data of the target scenario, using a preset affine algorithm to calculate the initial interval power flow of the target scenario, inputting the initial interval power flow into a preset radial basis function neural network model, and outputting the target interval power flow.

[0135] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising one or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by one or more processors, the one or more processors implement the above-mentioned interval flow determination method for the distribution network scenario.

[0136] Figure 6 1 is a hardware structure block diagram of an electronic device (or mobile device) for determining a section power flow in a distribution network scenario according to an embodiment of the present invention. Figure 6 As shown, the electronic device may include one or more processors (e.g., Figure 6 The processor 602a, processor 602b, ..., processor 602n, etc., which may include but are not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 604 for storing data. In addition, it may also include: a display, an input / output interface (I / 0 interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / 0 interface), a network interface, a keyboard, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 6 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 6 More or fewer components as shown, or with Figure 6 Different configurations shown.

[0137] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0138] The embodiments or examples of the present disclosure are not exhaustive, but are only illustrative of some embodiments or examples, and are not intended to be specific limitations on the scope of protection of the present disclosure. In the absence of contradiction, each step in a certain embodiment or example can be implemented as an independent example, and the steps can be combined arbitrarily. For example, the scheme after removing some steps in a certain embodiment or example can also be implemented as an independent example, and the order of the steps in a certain embodiment or example can be arbitrarily exchanged. In addition, the optional methods or optional examples in a certain embodiment or example can be combined arbitrarily; in addition, the various embodiments or examples can be combined arbitrarily, for example, some or all steps of different embodiments or examples can be combined arbitrarily, and a certain embodiment or example can be combined arbitrarily with the optional methods or optional examples of other embodiments or examples.

[0139] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0140] In the several embodiments provided by the present invention, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, 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 units or modules, which can be electrical or other forms.

[0141] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0142] In addition, each functional unit in each embodiment of the present invention may be integrated into one 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 software functional units.

[0143] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-0nly Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk, etc., which can store program code.

[0144] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for determining interval power flow in a distribution network scenario, characterized in that: include: Collect distributed power generation data for each quarter, and based on the distributed power generation data, use a preset clustering algorithm to determine a target scenario corresponding to each quarter; Based on the distributed power generation data of the target scenario, a preset affine algorithm is used to calculate the initial interval power flow of the target scenario; The initial interval power flow is input into a preset radial basis function neural network model, and a target interval power flow is output.

2. The method for determining section power flow according to claim 1, characterized in that: Based on the distributed power generation data, a preset clustering algorithm is used to determine the target scenario corresponding to each quarter, including: Based on the distributed power generation data, a wind power sequence and a photovoltaic sequence are constructed, wherein the element values ​​in the wind power sequence are the wind power at different time points, and the element values ​​in the photovoltaic sequence are the illumination areas at different time points; Based on the wind power sequence and the photovoltaic sequence, a distance matrix is ​​constructed, wherein a matrix element in the distance matrix is ​​a minimum cumulative distance between the Euclidean distance and its three nearest neighbors; Based on the distance matrix, construct an objective function, and based on the objective function, determine the neighborhood distance; Determine the minimum number of points, and based on the minimum number of points and the neighborhood distance, use a preset clustering algorithm to cluster the wind sequence and the photovoltaic sequence to obtain a sample set, wherein the sample set includes: all density-reachable and density-connected data points, and the data points refer to wind power or light area at different time points; The wind scene or photovoltaic scene indicated by the data points that do not belong to the sample set is determined as the target scene.

3. The method for determining section power flow according to claim 1, characterized in that: Before calculating the initial interval power flow of the target scenario based on the distributed power generation data of the target scenario by using a preset affine algorithm, the method further includes: Fitting the noise elements generated by the multiplication operation in the initial affine algorithm to obtain a preset multiplication operation; Using a Chebyshev approximation algorithm and a fitting model, the noise elements generated by the division operation in the initial affine algorithm are eliminated to obtain a preset division operation; The preset multiplication operation and the preset division operation are used to replace the multiplication operation and the division operation in the initial affine algorithm respectively, so as to obtain the preset affine algorithm.

4. The method for determining section power flow according to claim 1, characterized in that: The step of calculating the initial interval power flow of the target scenario based on the distributed power generation data of the target scenario by using a preset affine algorithm comprises: Determine an electric load node diagram, wherein the electric load node diagram includes: a plurality of load nodes, each of the load nodes having a current direction; Based on the distributed power generation data of the target scenario, calculating the apparent power and initial voltage of each of the load nodes; Based on the apparent power and the initial voltage, starting from the end node of the power load node graph, calculating the affine current toward the root node, wherein the end node is the last load node on the power load node graph, and the root node is the first load node on the power load node graph; Based on the affine current and the affine voltage of the previous iteration, the affine voltage of the current iteration is calculated until the absolute difference between the affine voltage of the current iteration and the affine voltage of the previous iteration is less than a preset deviation threshold, thereby obtaining the initial interval power flow.

5. The method for determining section power flow according to claim 1, characterized in that: Before inputting the initial interval power flow into a preset radial basis function neural network model and outputting the target interval power flow, the method further includes: Constructing an initial radial basis function neural network model, wherein the initial radial basis function neural network model comprises: a hidden layer, using the radial basis function as an activation function of neurons in the hidden layer, and each of the neurons in the hidden layer maps low-dimensional input data to a high-dimensional space through the activation function; The voltage amplitude upper and lower boundary data in the historical interval power flow of multiple historical target scenes are used to train the initial radial basis function neural network model to obtain the trained preset radial basis function neural network model.

6. The method for determining section power flow according to claim 1, characterized in that: The step of inputting the initial interval power flow into a preset radial basis function neural network model and outputting the target interval power flow comprises: The preset radial basis function neural network model is used to predict the voltage upper boundary and the voltage lower boundary of the affine voltage in the initial interval power flow to obtain the target voltage upper boundary and the target voltage lower boundary; The target interval power flow is determined based on the target voltage upper boundary and the target voltage lower boundary.

7. The method for determining section power flow according to claim 1, characterized in that: After the initial interval power flow is input into a preset radial basis function neural network model and the target interval power flow is output, the method further includes: The target interval power flow of each target scenario is sent to a dispatching terminal, wherein the dispatching terminal determines a power dispatching strategy under the target scenario based on the target interval power flow.

8. A device for determining interval power flow in a distribution network scenario, characterized in that: include: A determination unit, configured to collect distributed power generation data for each quarter, and determine a target scenario corresponding to each quarter based on the distributed power generation data by using a preset clustering algorithm; A calculation unit, configured to calculate an initial interval power flow of the target scenario based on the distributed power generation data of the target scenario by using a preset affine algorithm; The output unit is used to input the initial interval power flow into a preset radial basis function neural network model and output the target interval power flow.

9. A computer program product, characterized in that It includes a non-volatile computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the interval power flow determination method for the distribution network scenario described in any one of claims 1 to 7.

10. An electronic device, characterized in that: It includes one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the interval power flow determination method for the distribution network scenario described in any one of claims 1 to 7.

Images