A method for dynamic grid division of a city photovoltaic cluster based on space-time coupling tensor

Abstract

The application discloses a kind of city area photovoltaic cluster dynamic grid division methods based on space-time coupling tensor, comprising: synchronously collecting the active power time series and geographic coordinates of all photovoltaic nodes in city area;Build the heterogeneous space adjacency matrix considering electrical topology constraint, calculate the static space correlation degree between any two photovoltaic nodes;Build the time-shift cross-correlation matrix that captures the moving characteristics of weather system, quantify the dynamic time correlation degree between nodes;Perform space-time feature tensor fusion, obtain global space-time affinity matrix;Photovoltaic nodes are divided using improved spectral clustering algorithm, and the optimal grid number is adaptively determined by contour coefficient, finally generate photovoltaic cluster grid division scheme at this moment;According to grid drift threshold, trigger grid reconstruction.The application can effectively consider physical topology constraint and meteorological time delay characteristics, realize the accurate aggregation of photovoltaic cluster, significantly improve the accommodation capacity and regulation flexibility of distribution network to distributed energy.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of power grids and relates to distributed photovoltaic management control technology, in particular to a municipal photovoltaic cluster dynamic grid division method based on space-time coupling tensors. BACKGROUND

[0002] With the deepening of the "double carbon" strategy, the penetration rate of distributed photovoltaics in urban distribution networks continues to rise, showing the characteristics of many points, wide surfaces, scattered layout and small individual capacity. The massive access of distributed photovoltaics has changed the distribution network from a traditional passive network to an active network, and its inherent intermittency and volatility have brought serious challenges to the power quality and safe and stable operation of the power grid. In order to achieve efficient management and regulation of massive scattered photovoltaics, the strategy of cluster division and hierarchical regulation has become the industry consensus, that is, by dividing photovoltaic sites with similar geographical locations or electrical characteristics into several grids, the grid is aggregated and managed as a unit.

[0003] However, the existing photovoltaic cluster grid division technology has significant limitations in practical application. First, the existing spatial division dimension is relatively single, mostly only based on geographical Euclidean distance or administrative division for simple physical clustering, ignoring the actual electrical topology constraints of the distribution network. Due to the complex line layout of urban distribution networks, two photovoltaic sites adjacent in geographical location may belong to different feeders or even substations, and simple geographical clustering will result in low electrical coupling within the grid, making it difficult to achieve effective voltage and power flow coordinated control.

[0004] Secondly, and more importantly, the existing technology generally uses instantaneous correlation analysis when dealing with the time correlation of photovoltaic output fluctuations. This method assumes that photovoltaic fluctuations within the same region occur synchronously. However, in actual scenarios, the severe fluctuations in photovoltaic output are mainly caused by cloud cover, and clouds have a moving characteristic driven by atmospheric wind fields. This means that power drops at upstream sites will be transmitted to downstream sites after a certain time delay. The existing static correlation calculation method cannot capture the space-time delay characteristics caused by cloud movement, resulting in nodes with strong causal relationships but time differences being incorrectly divided into different grids, severely weakening the ability of the cluster to suppress fluctuations through mutual aid.

[0005] Therefore, there is an urgent need to develop a dynamic grid division method that can simultaneously consider electrical-geographical dual constraints and accurately capture the space-time delay characteristics caused by cloud movement. SUMMARY

[0006] Invention purposes: In order to solve the problem of regulation mismatch caused by ignoring the cloud movement time delay and electrical constraints in the existing photovoltaic cluster division, a kind of municipal photovoltaic cluster dynamic grid division method based on space-time coupling tensor is provided, by fusing geographical-electricity heterogeneous adjacency matrix and time shift cross correlation matrix capturing fluctuation lag, the space-time drift characteristics of photovoltaic output fluctuation are accurately quantified, the adaptive dynamic reconstruction of grid is realized, and the cluster regulation flexibility and fluctuation suppression ability of distribution network to distributed photovoltaic are significantly improved.

[0007] Technical scheme: In order to achieve the above purpose, the present application provides a kind of municipal photovoltaic cluster dynamic grid division method based on space-time coupling tensor, comprising the following steps:

[0008] S1: establish municipal photovoltaic monitoring database, in the preset historical sampling period, the active power time series of all photovoltaic nodes in municipal area and corresponding geographic coordinates are synchronously collected;

[0009] S2: according to the data collected in step S1, construct the heterogeneous space adjacency matrix considering the electrical topology constraint, the static space correlation degree between any two photovoltaic nodes is calculated by fusing geographical Euclidean distance and electrical connection impedance;

[0010] S3: according to the data collected in step S1, construct the time shift cross correlation matrix capturing the moving characteristics of weather system, the maximum cross correlation coefficient and its corresponding time lag between photovoltaic output sequences are calculated by sliding time window, the dynamic time correlation degree between nodes is quantified;

[0011] S4: execute space-time feature tensor fusion, map the static space correlation degree and dynamic time correlation degree to global space-time affinity matrix through nonlinear coupling function;

[0012] S5: based on global space-time affinity matrix, photovoltaic nodes are divided by using improved spectral clustering algorithm, and the optimal grid number is adaptively determined by using contour coefficient, and finally the photovoltaic cluster grid division scheme at this moment is generated;

[0013] S6: set grid drift threshold, when the grid center offset caused by the change of space-time characteristics of newly collected data exceeds the threshold, trigger grid reconstruction, otherwise keep the current division scheme.

[0014] Further, the construction method of heterogeneous space adjacency matrix in step S2 comprises:

[0015] A1: based on geographic coordinates , photovoltaic node physical adjacency graph is constructed by using Delaunay triangulation algorithm;

[0016] A2: for the node pair which is physically adjacent, define its space distance measure For:

[0017]

[0018] wherein, is the geographic Euclidean distance between nodes and ; is the electrical distance between two nodes in the distribution network topology, i.e., the equivalent impedance modulus value; and are the maximum geographic distance and the maximum electrical distance within the city area, respectively; is the geographic-electrical weight coefficient, whose value range is ;

[0019] A3: The distance metric is converted into spatial similarity using the Gaussian kernel function, and the matrix element is obtained:

[0020]

[0021] wherein, is the spatial scale decay parameter; the heterogeneous spatial adjacency matrix is obtained according to the matrix element .

[0022] Further, the construction method of the time-shift cross-correlation matrix in the step S3 includes:

[0023] B1: The power normalized sequences and of any two photovoltaic nodes and are defined, and the cross-correlation function is defined, wherein, is the time lag, and the value range is ;

[0024] B2: The maximum cross-correlation coefficient and the corresponding optimal lag time in the allowed lag range are calculated:

[0025]

[0026] B3: The dual time correlation degree containing fluctuation similarity and response synchrony is constructed:

[0027]

[0028] wherein, is the time synchronization tolerance constant, and the formula ensures that only the nodes with similar fluctuation patterns and similar occurrence times have high time correlation degree; the dual time correlation degree​ obtaining the time-shift cross-correlation matrix .

[0029] Further, the step S4 adopts a weighted coupling strategy based on Hadamard product to obtain a global spatio-temporal affinity matrix , where the element is calculated by the formula:

[0030]

[0031] wherein is a coupling coefficient for adjusting the preference of spatio-temporal correlation; is a connectivity indicator function, which is 1 only when there exists a physical connection path between the nodes and within the power supply range of the power transformation stations at each voltage level of the power distribution network, otherwise 0, so as to ensure that the divided grid does not physically split the power grid structure.

[0032] Further, the process of dividing the photovoltaic nodes in the step S5 by using the improved spectral clustering algorithm includes:

[0033] C1: calculating the degree matrix of the global spatio-temporal affinity matrix , where is a diagonal matrix, and the diagonal elements ;

[0034] C2: constructing a normalized Laplacian matrix :

[0035]

[0036] wherein is an identity matrix;

[0037] C3: calculating the eigenvectors corresponding to the first eigenvalues of the normalized Laplacian matrix , composing an eigenvector matrix , and normalizing the row vectors of ;

[0038] C4: regarding each row in the eigenvector matrix as a sample point, clustering it by using the K-means++ algorithm, and the cluster label obtained is the grid number to which each photovoltaic node belongs.

[0039] Further, the method of adaptively determining the optimal number of grids by using the contour coefficient in the step S5 includes:

[0040] Based on a comprehensive evaluation of grid cohesion and inter-grid separation, a grid partitioning effectiveness index is defined. :

[0041]

[0042] The first term is the average profile coefficient. For grid The average dissimilarity of internal nodes For grid The average dissimilarity to the nearest neighbor grid node; the second term is a penalty term. To divide the total power transmitted by the severed tie line, The total transmission power of the system, This is the penalty coefficient;

[0043] Select to make The largest The value is used as the optimal number of grid cells.

[0044] Further, in step S5, a preliminary grid is generated according to the photovoltaic cluster grid partitioning scheme, and the grid boundary smoothing correction is performed on the preliminary grid, specifically including:

[0045] D1: Identify enclave nodes located at grid boundaries, i.e., nodes whose grids are inconsistent with the grids of their main neighbors in physical space.

[0046] D2: Define the boundary correction cost function Calculate the enclave node Spatiotemporal affinity gain after reclassification to its neighboring grid:

[0047]

[0048] like Then the node Ownership transferred to neighboring grid This eliminates geographically fragmented, tiny, isolated grids. and Representing nodes respectively The spatiotemporal affinity with neighboring grid m and the original grid.

[0049] Furthermore, the mesh reconstruction triggering mechanism in step S6 is achieved by calculating the mesh centroid drift rate. This is achieved, specifically as follows:

[0050]

[0051] in, yes Time of the first the centroid vector of the grid in the space-time feature space; only when greater than a preset drift threshold a significant change in the current weather flow field or power grid operation mode is determined, steps S2 to S5 are re-executed, otherwise the grid division result and the corresponding control strategy of the last time are used.

[0052] Beneficial effects: Compared with the prior art, the present application solves the problem of time and space output fluctuation mismatch and control difficulty of citywide photovoltaic clusters caused by cloud movement, can effectively balance the physical topological constraint and meteorological time delay characteristics, realizes accurate aggregation of photovoltaic clusters, and significantly improves the accommodation capacity and control flexibility of the distribution network to distributed energy. Specifically, the following four aspects are embodied.

[0053] 1. Breakthrough the limitation of traditional static division, significantly improve the cluster fluctuation suppression ability. The present application can accurately quantify the "time lag" characteristics of photovoltaic output fluctuation caused by cloud movement among different nodes by constructing a time-shift cross-correlation matrix that captures the movement characteristics of the weather system. By dividing the nodes with strong causal correlation but time dislocation into the same grid, the natural time and space smoothing effect within the cluster is effectively utilized, so that the power fluctuation rate of the cluster as a whole is significantly reduced, and the peak and frequency regulation pressure of the distribution network side is greatly reduced.

[0054] 2. Considering geographical and electrical dual constraints, ensure that the grid division conforms to the physical law of power grid operation. In view of the problem that the existing technology simply relies on geographical distance division to cause low electrical coupling degree, the present application constructs a heterogeneous space adjacency matrix that integrates geographical Euclidean distance and electrical connection impedance. This ensures that the divided grid is not only geographically adjacent, but also closely connected in electrical topology, avoiding the forced aggregation of physically adjacent nodes belonging to different substations or feeders, thereby effectively protecting the integrity of the power grid structure, facilitating the coordinated control of voltage and power flow.

[0055] 3. An adaptive dynamic reconstruction mechanism is established, and the robustness of the system to environmental changes is improved. The present application introduces a reconstruction triggering mechanism based on the drift rate of the grid centroid. The system can monitor the changes in the space-time feature space in real time, and when the weather flow field or power grid operation mode changes significantly, causing the drift to exceed the threshold, the grid reconstruction is automatically triggered. This dynamic response mechanism ensures that the grid division scheme is always optimally matched with the current meteorological environment and power grid state, overcoming the defect that the static division model is difficult to adapt to complex and changeable weather conditions.

[0056] 4. The shape of the grid boundary is optimized, and the practical feasibility of engineering application is improved. The present application identifies and processes enclave nodes by introducing a grid boundary smoothing correction strategy, eliminates small island grids that are broken in geography by using a boundary correction cost function, and optimizes the space of the grid. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1 is a flow chart of the method of the present application;

[0058] Figure 2 is a space-time trajectory graph of city area photovoltaic distribution and cloud movement;

[0059] Figure 3 is a space-time correlation analysis graph;

[0060] Figure 4 is a division result comparison graph;

[0061] Figure 5 is a cluster fluctuation suppression capability comparison graph under two division strategies. DETAILED DESCRIPTION

[0062] The present application will be further illustrated below in conjunction with the drawings and specific embodiments, and it should be understood that these embodiments are only used to illustrate the present application and not to limit the scope of the present application, and after reading the present application, various equivalent modifications of the present application by those skilled in the art fall within the scope defined by the appended claims.

[0063] Example 1:

[0064] As shown in the drawings, the present embodiment provides a city area photovoltaic cluster dynamic grid division method based on space-time coupling tensor, comprising the following steps: Figure 1

[0065] S1: Establish a city area photovoltaic monitoring database, and synchronously collect active power time series and corresponding geographic coordinates of all photovoltaic nodes in the city area within a preset historical sampling period;

[0066] S2: According to the data collected in step S1, construct a heterogeneous space adjacency matrix considering electrical topology constraints, and calculate the static space correlation degree between any two photovoltaic nodes by fusing geographic Euclidean distance and electrical connection impedance;

[0067] The construction method of the heterogeneous space adjacency matrix comprises:

[0068] A1: Construct a physical adjacency graph of photovoltaic nodes based on geographic coordinates by using Delaunay triangulation algorithm;

[0069] A2: For a pair of physically adjacent nodes , define its space distance metric as:

[0070] ​​​​​​

[0071] wherein, is the geographic Euclidean distance between nodes and ; is the electrical distance between two nodes in the power distribution network topology, i.e., the equivalent impedance modulus; and are the maximum geographic distance and the maximum electrical distance within the city area, respectively; is the geographic-electrical weight coefficient, taking a value in the range of ;

[0072] A3: Convert the distance metric into spatial similarity using the Gaussian kernel function to obtain the matrix element :

[0073]

[0074] wherein, is the spatial scale decay parameter; the heterogeneous spatial adjacency matrix is obtained according to the matrix element .

[0075] S3: According to the data collected in step S1, a time-shift cross-correlation matrix capturing the moving characteristics of the weather system is constructed, the maximum cross-correlation coefficient and the corresponding time lag between the photovoltaic output sequences are calculated through a sliding time window, and the dynamic time correlation degree between nodes is quantified;

[0076] The construction method of the time-shift cross-correlation matrix includes:

[0077] B1: The power normalized sequences and of any two photovoltaic nodes and are defined as the cross-correlation function , wherein, is the time lag, taking a value in the range of ;

[0078] B2: The maximum cross-correlation coefficient and the corresponding optimal lag time in the allowed lag range are calculated:

[0079]

[0080] B3: A dual time correlation degree containing fluctuation similarity and response synchronicity is constructed:

[0081]

[0082] wherein, is a time synchronization tolerance constant, the formula ensures that only nodes with similar fluctuation patterns and close occurrence time have high time correlation degree; according to the double time correlation degree obtaining the time-shift cross-correlation matrix .

[0083] S4: performing space-time feature tensor fusion, mapping the static spatial correlation degree and the dynamic time correlation degree into a global space-time affinity matrix through a nonlinear coupling function;

[0084] The space-time feature tensor fusion adopts a weighted coupling strategy based on Hadamard product, and the global space-time affinity matrix obtained is The element of the global space-time affinity matrix is calculated according to the following formula:

[0085]

[0086] wherein, is a coupling coefficient for adjusting the preference of space-time correlation; is a connectivity indication function, which is 1 only when nodes and have a physical connection path within the power supply range of the power transformation stations at each voltage level of the power grid, otherwise 0, so as to ensure that the divided grid does not split the physical structure of the power grid.

[0087] S5: based on the global space-time affinity matrix, the photovoltaic nodes are divided by using an improved spectral clustering algorithm, and the optimal grid number is adaptively determined by using a contour coefficient, and finally a photovoltaic cluster grid division scheme at this moment is generated;

[0088] The process of dividing the photovoltaic nodes by using the improved spectral clustering algorithm includes:

[0089] C1: calculating the degree matrix of the global space-time affinity matrix , wherein is a diagonal matrix, and the diagonal elements ;

[0090] C2: constructing a normalized Laplacian matrix :

[0091]

[0092] wherein, is an identity matrix;

[0093] C3: calculating the eigenvectors corresponding to the first smallest eigenvalues of the normalized Laplacian matrix , composing an eigenvector matrix , and performing dimensionality reduction on the row vector is normalized;

[0094] C4: Each row in the feature matrix is regarded as a sample point, and the K-means++ algorithm is used to cluster it, and the obtained cluster label is the grid number to which each photovoltaic node belongs.

[0095] The method for adaptively determining the optimal number of grids through the contour coefficient includes:

[0096] Based on the comprehensive evaluation of the grid cohesion and the grid separation, the grid division effectiveness index is defined :

[0097]

[0098] wherein the first term is the average contour coefficient, is the average dissimilarity of the nodes in the grid , the second term is the penalty term, is the average dissimilarity of the nodes in the grid and the nearest neighbor grid nodes; the second term is the total power transmission of the division cut-off tie line, is the total transmission power of the system, is the penalty coefficient;

[0099] The value of is selected as the optimal grid number.

[0100] According to the photovoltaic cluster grid division scheme, a preliminary grid is generated, and the grid boundary is smoothed and corrected, specifically including:

[0101] D1: Identify the enclave nodes on the grid boundary, that is, the nodes whose grid to which the node belongs is inconsistent with the grid to which the main neighbor node of the node in the physical space belongs;

[0102] D2: Define the boundary correction cost function , calculate the spatiotemporal affinity gain after reclassifying the enclave node to its neighbor grid:

[0103]

[0104] If , the ownership of the node is transferred to the neighbor grid , so as to eliminate the small island grid which is broken in geography; and respectively represent the spatiotemporal affinity of the node and the neighbor grid m and the original grid. ​​

[0105] S6: Set a grid drift threshold, when the spatial-temporal feature change of newly collected data causes the grid center offset to exceed the threshold, trigger grid reconstruction, otherwise keep the current partition scheme.

[0106] The triggering mechanism of grid reconstruction is achieved by calculating the grid centroid drift rate , specifically:

[0107]

[0108] Wherein, is the centroid vector of the th grid in the spatial-temporal feature space at time t; only when is greater than the preset drift threshold , it is determined that the current weather flow field or power grid operation mode has changed significantly, and steps S2 to S5 are re-executed, otherwise the grid partition result and the corresponding control strategy of the last time are used.

[0109] Embodiment 2:

[0110] In order to verify the effectiveness and superiority of the method of the present application, this embodiment constructs a municipal distribution network model containing 80 distributed photovoltaic nodes on the Matlab simulation platform for testing, as follows:

[0111] I. Simulation scenario and data basis

[0112] The experimental area is set to , 80 photovoltaic nodes are randomly distributed, and the electrical topology connection of the distribution network is simulated based on the minimum spanning tree (MST) algorithm. In order to truly restore the photovoltaic output fluctuation characteristics caused by the movement of the cloud cluster, the experiment introduces the Taylor frozen turbulence hypothesis, simulates a cloud cluster with Gaussian distribution characteristics to sweep the area at a wind speed of .

[0113] Figure 2 The space-time trajectory of the simulation scenario is shown. Figure 2 The black solid line in the figure represents the electrical connection topology of the distribution network, and the colored contour lines respectively show the spatial position and movement path of the cloud cluster at three sampling time points. It can be seen that the cloud cluster does not cover the entire area instantaneously, but presents a significant space-time evolution process, which provides the necessary physical scene for verifying the space-time coupling mechanism of the present application.

[0114] II. Analysis of the space-time correlation mechanism of photovoltaic fluctuation

[0115] Figure 3 Reveals the fluctuation hysteresis ignored by the prior art.

[0116] Figure 3 (a) selects two typical nodes, node A and node B, located upstream and downstream of the cloud moving path. It can be clearly observed that when the cloud passes, the upstream node A blue curve experiences power drop first, and the downstream node B orange curve experiences similar fluctuation pattern after a certain time delay. The two are strongly related in physics, but there is a misalignment in the time axis.

[0117] Figure 3 (b) further gives the time shift cross-correlation function analysis results of the two nodes. The coefficient obtained by traditional zero-lag correlation calculation is low, which cannot identify the correlation of the two; and the method of the present application accurately captures the optimal lag time , where the cross-correlation coefficient of the two reaches a peak. This strongly proves the necessity and scientificity of introducing the time shift cross-correlation matrix in the present application.

[0118] III. Comparison of grid division schemes

[0119] To intuitively evaluate the clustering effect of the present application, the number of grids , the present application method is compared with the traditional K-means clustering method based on geographic Euclidean distance, and the results are shown in Figure 4 .

[0120] Figure 4 (a) shows that the grid generated by the traditional geographic division method presents a regular block aggregation feature, such as circular distribution, which only considers the static geographic proximity. Since the moving direction of the cloud is ignored, nodes on the same cloud path but slightly far apart in geographic distance are forced to be divided into different grids.

[0121] Figure 4 (b) shows the division results of the present application method. It can be clearly seen that the generated grid shape presents a strip-shaped feature along the moving direction of the cloud. This division method successfully aggregates nodes with fluctuation causality and time sequence delay characteristics in the same cluster, realizes the deep coupling of physical space and meteorological space, and is the core advantage of the present application technical solution.

[0122] IV. Quantitative evaluation of cluster suppression performance

[0123] Figure 5 The average fluctuation rate of the aggregated power in each cluster under the two division strategies is compared, and the standard deviation of the power difference is used as the measure. Figure 5The column chart data in the figure shows that when the traditional geographical division (gray column) is used, due to the weak asynchronization of the fluctuation of the nodes in the cluster, the complementary smoothing effect is limited, and the fluctuation rate after aggregation is still high. When the space-time division method of the application (orange column) is used, due to the optimized combination of the nodes in the cluster according to the space-time correlation, the natural smoothing effect brought by the space-time dislocation is effectively utilized, so that the power fluctuation rate of the whole cluster is reduced by about 35%.

[0124] In summary, the simulation experiment fully proves that the method of the application can accurately capture the space-time drift characteristics of photovoltaic output, significantly improve the fluctuation smoothing capability of the photovoltaic cluster through dynamic grid division, and provide a more optimal regulation boundary for the stable operation of the power distribution network.

Claims

1. A method for dynamic grid partitioning of urban photovoltaic clusters based on spatiotemporal coupling tensors, characterized in that, Includes the following steps: S1: Establish a city-wide photovoltaic monitoring database and simultaneously collect the active power time series and corresponding geographical coordinates of all photovoltaic nodes within the city within a preset historical sampling period; S2: Based on the data collected in step S1, construct a heterogeneous spatial adjacency matrix that considers electrical topology constraints, and calculate the static spatial correlation degree between any two photovoltaic nodes by fusing geographical Euclidean distance and electrical connection impedance. S3: Based on the data collected in step S1, construct a time-shift cross-correlation matrix to capture the movement characteristics of the weather system, calculate the maximum cross-correlation coefficient between photovoltaic power output sequences and their corresponding time lag through a sliding time window, and quantify the dynamic time correlation between nodes. S4: Perform spatiotemporal feature tensor fusion, mapping the static spatial correlation and dynamic temporal correlation into a global spatiotemporal affinity matrix through a nonlinear coupling function; S5: Based on the global spatiotemporal affinity matrix, the photovoltaic nodes are divided using an improved spectral clustering algorithm, and the optimal number of grids is determined adaptively by the contour coefficient, thus generating the photovoltaic cluster grid division scheme at this moment. S6: Set a grid drift threshold. When the grid center offset caused by the spatiotemporal characteristics of newly acquired data exceeds the threshold, grid reconstruction is triggered; otherwise, the current partitioning scheme is maintained.

2. The method for dynamic grid partitioning of urban photovoltaic clusters based on spatiotemporal coupling tensors according to claim 1, characterized in that, The method for constructing the heterogeneous space adjacency matrix in step S2 includes: A1: Using the Delaunay triangulation algorithm, based on geographic coordinates Construct the physical adjacency graph of the photovoltaic nodes; A2: For physically adjacent node pairs Define its spatial distance metric for: ； in, For nodes and The geographical Euclidean distance; This represents the electrical distance between two nodes in the distribution network topology, i.e., the equivalent impedance magnitude. and These are the maximum geographical distance and the maximum electrical distance within the city limits, respectively. The geographical-electrical weighting coefficient has a range of values. ; A3: Use the Gaussian kernel function to transform the distance metric into spatial similarity, and obtain the matrix elements. : ； in, For spatial scale attenuation parameters; based on matrix elements Obtain the adjacency matrix of the heterogeneous space .

3. The method for dynamic grid partitioning of urban photovoltaic clusters based on spatiotemporal coupling tensors according to claim 2, characterized in that, The method for constructing the time-shift cross-correlation matrix in step S3 includes: B1: For any two photovoltaic nodes and Power normalized sequence and Define the cross-correlation function ,in, This refers to the time lag. B2: Calculate the maximum cross-correlation coefficient within the allowable lag range. and its corresponding optimal lag time : ； B3: Construct a dual time correlation degree that includes fluctuation similarity and response synchronization. : ； in, This is the time synchronization tolerance constant; based on the dual time correlation degree Obtain the time-shift cross-correlation matrix .

4. The method for dynamic grid partitioning of urban photovoltaic clusters based on spatiotemporal coupling tensors according to claim 3, characterized in that, In step S4, the spatiotemporal feature tensor fusion adopts a weighted coupling strategy based on Hadamard product to obtain a global spatiotemporal affinity matrix. elements The calculation formula is: ； in, To adjust the coupling coefficient for spatiotemporal correlation preference; For connectivity indicator functions, only if the node and The value is 1 if there is a physical connection path within the power supply range of substations at each voltage level in the distribution network, and 0 otherwise, to ensure that the divided grid does not physically break the power grid structure.

5. The method for dynamic grid partitioning of urban photovoltaic clusters based on spatiotemporal coupling tensors according to claim 4, characterized in that, The process of dividing photovoltaic nodes using the improved spectral clustering algorithm in step S5 includes: C1: Calculate the global spatiotemporal affinity matrix degree matrix ,in It is a diagonal matrix, with diagonal elements ; C2: Construct the normalized Laplace matrix : ； in, It is the identity matrix; C3: Calculate the normalized Laplacian matrix The former The eigenvectors corresponding to the smallest eigenvalues ​​form the feature matrix. and to The row vectors are normalized. C4: The feature matrix Each row in the array is considered a sample point. The K-means++ algorithm is used to cluster them, and the resulting cluster label is the grid number to which each photovoltaic node belongs.

6. The method for dynamic grid partitioning of urban photovoltaic clusters based on spatiotemporal coupling tensors according to claim 5, characterized in that, The method for adaptively determining the optimal number of meshes using contour coefficients in step S5 includes: Based on a comprehensive evaluation of grid cohesion and inter-grid separation, a grid partitioning effectiveness index is defined. : ； The first term is the average profile coefficient. For grid The average dissimilarity of internal nodes For grid The average dissimilarity to the nearest neighbor grid node; the second term is a penalty term. To divide the total power transmitted by the severed tie line, The total transmission power of the system, This is the penalty coefficient; Select to make The largest The value is used as the optimal number of grid cells.

7. The method for dynamic grid partitioning of urban photovoltaic clusters based on spatiotemporal coupling tensors according to claim 6, characterized in that, In step S5, a preliminary grid is generated according to the photovoltaic cluster grid partitioning scheme, and the grid boundary is smoothed and corrected. Specifically, this includes: D1: Identify enclave nodes located at grid boundaries, i.e., nodes whose grids are inconsistent with the grids of their main neighbors in physical space. D2: Define the boundary correction cost function Calculate the enclave node Spatiotemporal affinity gain after reclassification to its neighboring grid: ； like Then the node Ownership transferred to neighboring grid This eliminates geographically fragmented, tiny, isolated grids. and Representing nodes respectively The spatiotemporal affinity with neighboring grid m and the original grid.

8. The method for dynamic grid partitioning of urban photovoltaic clusters based on spatiotemporal coupling tensors according to claim 7, characterized in that, The mesh reconstruction triggering mechanism in step S6 is achieved by calculating the mesh centroid drift rate. This is achieved, specifically as follows: ； in, yes Time of the first The centroid vector of each grid in the spatiotemporal feature space; only when greater than the preset drift threshold If a significant change is found in the current weather flow field or power grid operation mode, steps S2 to S5 are re-executed; otherwise, the grid division results and corresponding control strategies from the previous moment are used.

Images