Equivalent Modeling Method and System for Distributed Photovoltaic Clustering in Low-Voltage Systems
By combining generalized weather type clustering with spatial clustering, and selecting representative power stations through multi-index evaluation, an equivalent impedance model is constructed. This solves the problems of spatial correlation and meteorological differences in low-voltage distributed photovoltaic modeling, and improves the modeling accuracy and adaptability.
Patent Information
- Application Number
- CN202611024344.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-10
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2046-07-10
AI Technical Summary
Existing low-voltage distributed photovoltaic modeling methods do not effectively incorporate spatial correlation, lack consideration of meteorological and geographical location information, and lack efficient clustering strategies, making it difficult to meet application requirements.
A two-stage strategy combining generalized weather type clustering and spatial clustering is adopted. The improved dynamic time warping algorithm and adaptive spectral density clustering algorithm are used for clustering. The optimal representative power station is selected by combining multi-index comprehensive evaluation and an equivalent impedance model is constructed.
It significantly improves the accuracy and adaptability of low-voltage distributed photovoltaic equivalent modeling, solves the problems of large modeling errors and poor scene adaptability caused by ignoring meteorological differences and spatial correlations in traditional methods, and provides reliable model support.
Smart Images

Figure CN122532910B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of photovoltaic equivalent modeling technology, and in particular to a method and system for equivalent modeling of distributed photovoltaic clusters in low-voltage systems. Background Technology
[0002] At this critical juncture of the global energy structure transition towards cleaner and distributed energy, low-voltage distribution systems, as the main facilities connecting the distribution network and the user side, have become the core carrier for the large-scale integration of distributed photovoltaic (PV) systems. Compared to medium- and high-voltage systems, distributed PV systems integrated into low-voltage systems are characterized by a large number of units, small individual capacities, and scattered geographical distribution, with their output closely dependent on the spatial distribution of PV and meteorological conditions. Currently, research on distributed PV modeling technology focuses on the medium- and high-voltage domains, resulting in a mature system combining single-unit mechanism modeling and cluster equivalent modeling, which can meet the needs of system stability analysis and fault simulation. However, there are still technological gaps in distributed PV modeling for low-voltage systems, especially in cluster equivalent modeling oriented towards spatial correlation, and existing methods are difficult to adapt to the characteristics and engineering requirements of low-voltage systems.
[0003] Existing low-voltage photovoltaic (PV) modeling often follows the approach of scaling or idealizing parameters from medium- and high-voltage systems. Some schemes directly scale the parameters of medium- and high-voltage PV inverter models according to capacity and apply them to low-voltage scenarios, without considering the geographical differences in PV installation locations within low-voltage systems. For example, within the same transformer area, rooftop PV and ground-mounted PV output curves differ significantly due to different shading conditions. Similarly, PV in hilly areas and plains areas have different peak output times due to different solar radiation angles, yet they are included in the same model calculation. Other schemes use idealized models such as constant power sources and controlled current sources, ignoring the correlation between PV output and geographical location. This results in models that fail to reflect the output correlation of PV within the same transformer area, leading to significant discrepancies between simulation results and actual field measurements. Consequently, it is difficult to accurately characterize the actual characteristics of line power loss and transformer load rate fluctuations after PV integration.
[0004] Furthermore, photovoltaic power output is directly related to meteorological conditions, and existing methods do not consider weather type and geographical location in their modeling. On the one hand, they fail to classify generalized weather sample groups based on characteristics such as solar radiation intensity and cloud cover, and instead use a single model to adapt to all meteorological scenarios. This results in the model overestimating power output during sunny days and underestimating power output during cloudy and rainy days. On the other hand, even when weather types are classified, they are not further optimized in conjunction with geographical space. The use of the same attenuation coefficient amplifies the modeling error and fails to meet the requirements of low-pressure systems for refined adaptation to meteorological changes.
[0005] In summary, existing low-voltage distributed photovoltaic (PV) modeling methods fail to meet application requirements due to their lack of effective integration of spatial correlation, insufficient consideration of meteorological and geographical location information, and lack of efficient clustering strategies. Therefore, a clustering-based equivalent modeling method with spatial correlation as its core is needed to provide a solution to the modeling challenges of large-scale distributed PV in low-voltage systems, and to provide technical support for the safe operation of distribution networks and efficient PV consumption. Summary of the Invention
[0006] The purpose of this invention is to provide an equivalent modeling method and system for distributed photovoltaic clustering in low-voltage systems, aiming to solve the problems that traditional modeling methods fail to meet application requirements due to their ineffective integration of spatial correlation, lack of consideration for meteorological and geographical location information, and lack of efficient clustering strategies.
[0007] In a first aspect, the present invention provides a method for equivalent modeling of distributed photovoltaic clusters in low-voltage systems, the method comprising: Obtain the geographical location information and operation information of photovoltaics in the low-pressure area, and classify all photovoltaics in the low-pressure area into at least one generalized weather type group based on the geographical location information and operation information. An improved dynamic time warping algorithm and an adaptive spectral density clustering algorithm are used to spatially group each of the generalized weather type groups to obtain at least one photovoltaic cluster under each generalized weather type group. Calculate the average power of each photovoltaic cluster, and calculate the characteristic indicators of each photovoltaic unit in the photovoltaic cluster based on the average power. The characteristic indicators include Pearson correlation coefficient, dynamic time warping distance, and amplitude deviation index, so as to calculate the comprehensive representativeness score of each photovoltaic unit in any photovoltaic cluster based on all the characteristic indicators. The photovoltaic (PV) with the highest comprehensive representativeness score within the same PV cluster is selected as the target PV. The real-time active power of the target PV is collected, and the power values of other PVs are calculated based on the real-time active power of the target PV. An equivalent impedance model is then constructed based on the real-time active power and the power values.
[0008] Secondly, the present invention provides an equivalent modeling system for distributed photovoltaic clustering in low-voltage systems, the system comprising: The information acquisition module is used to acquire the geographical location information and operation information of photovoltaics in the low-pressure area, and to classify all photovoltaics in the low-pressure area into at least one generalized weather type group based on the geographical location information and operation information of photovoltaics. The cluster partitioning module is used to spatially partition each of the generalized weather type groups using an improved dynamic time warping algorithm and an adaptive spectral density clustering algorithm, so as to obtain at least one photovoltaic cluster under each generalized weather type group. The photovoltaic scoring module is used to calculate the average power of each photovoltaic cluster and to calculate the characteristic indicators of each photovoltaic unit in the photovoltaic cluster based on the average power. The characteristic indicators include Pearson correlation coefficient, dynamic time warping distance and amplitude deviation index, so as to calculate the comprehensive representativeness score of each photovoltaic unit in any photovoltaic cluster based on all the characteristic indicators. The model building module is used to select the photovoltaic with the highest comprehensive representativeness score within the same photovoltaic cluster as the target photovoltaic, collect the real-time active power of the target photovoltaic, calculate the power value of other photovoltaics based on the real-time active power of the target photovoltaic, and construct an equivalent impedance model based on the real-time active power and power value.
[0009] Thirdly, the present invention provides a storage medium that stores one or more programs, which, when executed by a processor, implement the above-described method for equivalent modeling of distributed photovoltaic clusters in low-voltage systems.
[0010] Fourthly, the present invention provides an electronic device, the electronic device comprising a memory and a processor, wherein: The memory is used to store computer programs; When the processor executes the computer program stored in the memory, it implements the above-mentioned low-voltage system distributed photovoltaic cluster equivalent modeling method.
[0011] Compared with the prior art, the present invention has the following advantages: This invention significantly improves the accuracy and adaptability of low-voltage distributed photovoltaic (PV) equivalent modeling by introducing a two-stage strategy combining generalized weather type clustering and spatial clustering, and integrating multi-index comprehensive evaluation to select the optimal representative power station. It effectively solves the problems of large modeling errors and poor scenario adaptability caused by neglecting meteorological differences and spatial correlations in traditional methods. Specifically, firstly, a weather type index is calculated based on the ratio of actual PV output to rated output, dividing all sampling times into generalized weather type groups such as sunny, cloudy, and rainy days. This allows the equivalent model to dynamically adjust the modeling benchmark according to different meteorological conditions, avoiding the problem of poor applicability of a single model across all weather scenarios. Based on this, for each generalized weather type group, a composite distance index integrating spatial Euclidean distance and electrical distance is further constructed. This is combined with a dynamic time warping algorithm to measure the shape similarity of PV power curves. Adaptive spectral density clustering automatically determines the number of clusters and cluster centers, achieving refined spatial clustering of geographically dispersed PV power plants with varying output patterns. This solves the problem of traditional clustering methods relying on a single distance index or manual setting. The invention addresses the issue of unreasonable clustering caused by threshold values. It then calculates the average power within each photovoltaic cluster and uses three characteristic indicators—Pearson correlation coefficient, dynamic time warping distance, and amplitude deviation index—to objectively assign a comprehensive representativeness score using the entropy weight method. The photovoltaic system with the highest score is selected as the representative power station, effectively solving the problem that a single indicator (such as only using the correlation coefficient or only using installed capacity) cannot fully reflect the matching degree of photovoltaic power output patterns. Finally, based on the real-time active power of the representative power station, the power values of other photovoltaic systems are calculated according to the proportion of installed capacity. A complete equivalent model is constructed by combining the equal power loss impedance models of radial and trunk topologies. This significantly reduces model complexity and simulation computation while preserving the consistency of power loss characteristics and transient response within the cluster. Through the joint optimization strategy of weather type-spatial clustering-multi-indicator representative selection, this invention can significantly improve the simulation accuracy and robustness of the low-voltage distributed photovoltaic equivalent model under different meteorological and geographical distribution scenarios, providing reliable model support for the planning, operation control, and fault analysis of low-voltage systems with large-scale distributed photovoltaic access. Attached Figure Description
[0012] Figure 1 This is a flowchart of an embodiment of the low-voltage system distributed photovoltaic clustering equivalent modeling method proposed in this invention; Figure 2 This is a schematic diagram of the structure of a detailed model of a low-voltage distributed photovoltaic system in this invention; Figure 3 This is a schematic diagram of the structure of the equivalent model of distributed photovoltaic systems in low-voltage systems in this invention; Figure 4 This is a comparison chart of the detailed model and the equivalent model voltages when Ufault = 50%Un in most rainy weather conditions in this invention. Figure 5 This is a power comparison diagram between the detailed model and the equivalent model when Ufault = 50%Un in most rainy days in this invention; Figure 6 This is a comparison chart of the detailed model and the equivalent model current when Ufault = 50%Un in most rainy weather in this invention; Figure 7 This is a comparison chart of the detailed model and the equivalent model voltages when Ufault = 50%Un on mostly cloudy days in this invention; Figure 8 This is a power comparison diagram between the detailed model and the equivalent model when Ufault = 50%Un on mostly cloudy days in this invention; Figure 9 This is a comparison chart of the detailed model and the equivalent model current when Ufault = 50%Un on mostly cloudy days in this invention; Figure 10 This is a comparison diagram of the detailed model and the equivalent model voltage when Ufault = 50%Un in most sunny days in this invention; Figure 11 This is a power comparison diagram between the detailed model and the equivalent model when Ufault = 50%Un in most sunny days in this invention; Figure 12 This is a comparison chart of the detailed model and the equivalent model current when Ufault = 50%Un in most sunny days in this invention; Figure 13 This is a comparison chart of the detailed model and the equivalent model voltage when Ufault = 80%Un in most rainy weather in this invention; Figure 14 This is a power comparison diagram between the detailed model and the equivalent model when Ufault = 80%Un in most rainy days in this invention; Figure 15 This is a comparison chart of the detailed model and the equivalent model current when Ufault = 80%Un in most rainy weather in this invention; Figure 16 This is a comparison chart of the detailed model and the equivalent model voltage when Ufault = 80%Un on mostly cloudy days in this invention; Figure 17 This is a power comparison diagram between the detailed model and the equivalent model when Ufault = 80%Un on mostly cloudy days in this invention; Figure 18 This is a comparison chart of the detailed model and the equivalent model current when Ufault = 80%Un on mostly cloudy days in this invention; Figure 19 This is a comparison diagram of the detailed model and the equivalent model voltage when Ufault = 80%Un in most sunny days in this invention; Figure 20 This is a power comparison diagram between the detailed model and the equivalent model when Ufault = 80%Un in most sunny days in this invention; Figure 21 This is a comparison chart of the detailed model and the equivalent model current when Ufault = 80%Un in most sunny days in this invention; Figure 22 This is a schematic diagram of the structure of a low-voltage system distributed photovoltaic clustering equivalent modeling system proposed in an embodiment of the present invention.
[0013] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation
[0014] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Unless otherwise defined, the technical or scientific terms used herein should have the ordinary meaning understood by those skilled in the art. The terms "comprising" and similar expressions used herein mean that the element or object preceding the word covers the element or object listed after the word and its equivalents, but does not exclude other elements or objects.
[0015] like Figure 1 As shown, an embodiment of the present invention provides an equivalent modeling method for distributed photovoltaic clustering in low-voltage systems. This method includes steps S101 to S104, wherein: Step S101: Obtain the photovoltaic geographic location information and photovoltaic operation information of the low-pressure area, and classify all photovoltaics in the low-pressure area into at least one generalized weather type group based on the photovoltaic geographic location information and photovoltaic operation information; It should be noted that this step classifies weather types based on the weather type index, forming a generalized weather type group. This reduces the computational dimensionality and improves the robustness of the grouping to meteorological changes. Specifically, the weather type index is first calculated using the following formula: ; in, Let i be the weather type index of the i-th photovoltaic cell within the c-th regional block at time t; The output power sample value of the i-th photovoltaic cell within the c-th regional block at time t; The rated output of the i-th photovoltaic cell within the c-th regional block; Regarding the formula for the aforementioned weather type index, it should be noted that this invention introduces a weather type index to map the ratio of photovoltaic power output to rated power output as a continuous index, and classifies photovoltaic power into generalized weather type groups such as sunny, cloudy, overcast, and showers based on a predefined index range. This exponential compression transforms complex meteorological states into low-dimensional discrete categories, reducing the computational complexity of subsequent spatial clustering. Simultaneously, photovoltaic power within the generalized group exhibits similar macroscopic power output characteristics, making the clustering results highly robust to short-term meteorological fluctuations. This avoids frequent inter-group re-partitioning caused by instantaneous power fluctuations, providing a stable input basis for subsequent spatial clustering.
[0016] Set a weather type index range corresponding to each weather type, and divide the weather type indices of all photovoltaics in all regional blocks according to the weather type index and the weather type index range to obtain at least one generalized weather type group.
[0017] For example, as shown in Table 1 below, there are multiple weather types and the range of weather type indices corresponding to each weather type.
[0018] Table 1. Weather Type Index Range and Weather Type Comparison
[0019] Step S102: Spatial clustering is performed on each of the generalized weather type groups using an improved dynamic time warping algorithm and an adaptive spectral density clustering algorithm to obtain at least one photovoltaic cluster under each generalized weather type group. It should be noted that, in the spatial clustering process, the spatial-electrical composite distance is first constructed according to the following formula: ; in, The spatial-electrical composite distance between the i-th photovoltaic (PV) and j-th PV units within the same generalized weather type group. The maximum spatial Euclidean distance between all photovoltaic pairs. To correspond to the maximum electrical distance between all photovoltaic pairs within the generalized weather type group, Spatial weighting coefficient, The electrical distance between the i-th photovoltaic (PV) and j-th PV grid connection points within the same generalized weather type group. The spatial Euclidean distance between the i-th photovoltaic and j-th photovoltaic units within the same generalized weather type group; In addition, weighting coefficients The determination method is to calculate based solely on spatial distance ( ) and based solely on electrical distance ( The silhouette coefficient of the grouping results is selected to maximize the silhouette coefficient. Value; if the difference in the contour coefficients of the two is less than 0.05, then take the value. The range of values is .
[0020] Regarding the formula for the spatial-electrical composite distance mentioned above, it should be noted that traditional clustering methods typically define the distance between photovoltaic (PV) units using only geographical latitude and longitude coordinates or relying solely on the electrical impedance of the transformer substations. Furthermore, the weights of spatial and electrical distances are usually set as fixed constants, leading to clustering results that ignore grid topology constraints or geographical clustering effects. To address this issue, this invention calculates the spatial Euclidean distance based on latitude and longitude and the electrical distance based on the magnitude of node self-impedance / mutual impedance, thereby constructing a composite distance formula. Electrical distance makes it easier for electrically coupled PV units to cluster together, while spatial distance ensures geographical proximity, providing a more comprehensive characterization of the intrinsic relationships between PV units. The weighting coefficients are the profile coefficients calculated for pure spatial distance clustering and pure electrical distance clustering, respectively. Dynamic selection ensures that the composite distance automatically favors dimensions with higher discriminative power under different transformer substation topologies, thereby maximizing inter-group separation and providing an optimal similarity measurement basis for subsequent clustering.
[0021] Furthermore, in some embodiments, the spatial distance between any two photovoltaic installations is calculated based on the latitude and longitude of the photovoltaic installation points; the electrical distance is calculated based on the transformer substation topology and line parameters, providing a quantifiable indicator for subsequent composite distance calculations and avoiding clustering based on a single indicator. ; in, Let be the spatial Euclidean distance between the i-th photovoltaic cell and the j-th photovoltaic cell; , These are the longitude and latitude coordinates of the i-th photovoltaic cell, respectively. , These are the longitude and latitude coordinates of the j-th photovoltaic cell, respectively. ; in, Let be the electrical distance between the i-th photovoltaic cell and the j-th photovoltaic cell grid connection point; , These are the self-impedance amplitudes of the i-th and j-th photovoltaic access points, respectively; This represents the mutual impedance magnitude between the two connection points.
[0022] Then, the dynamic time warping algorithm is used to obtain photovoltaic data. and The dynamic similarity of power is combined with the composite distance to construct a composite similarity matrix, which can simultaneously consider the static spatial structure and the dynamic power output process. Take the active power sequence of the i-th photovoltaic and the j-th photovoltaic. , , build Local distance matrix: ; in, The local distance matrix is the t-th row and the th... Column elements, Let be the power value of the i-th photovoltaic cell at time t. For the j-th other photovoltaic in Power value at time, This represents the total sampling time. Build Cumulative distance matrix ,initialization ,for ,make ;for ,make ; calculate recursively according to the following formula The remaining elements: ; in, Cumulative distance matrix The tth row of the middle Column elements; The dynamic time-warped distance can be obtained using the following formula: ; in, The dynamic time-normalized distance between the power curves of the i-th photovoltaic and the j-th photovoltaic; A composite similarity matrix is constructed based on the dynamic time warp distance and the spatial-electrical composite distance: ; in, Let be the element in the i-th row and j-th column of the composite similarity matrix S. The composite distance bandwidth parameter is defined as the composite distance between all photovoltaic pairs under the corresponding generalized weather type group. the median; The dynamic time-warped distance bandwidth parameter is defined as the dynamic time-warped distance between all photovoltaic pairs under the corresponding generalized weather type group. the median; Regarding the formula for the composite similarity matrix, it should be noted that conventional point-to-point Euclidean distance is highly sensitive to the translation and scaling of power waveforms. However, photovoltaic power output often experiences time delays due to cloud movement. Direct point-to-point comparisons may incorrectly classify two similar waveforms as dissimilar. This invention introduces a dynamic time warping algorithm. By constructing a local distance matrix and recursively accumulating it, the optimal time path between two power sequences is found, obtaining a dynamic time warped distance unaffected by phase shifts. This accurately reflects the overall morphological similarity of the two power output curves, ensuring that even if peaks and troughs are misaligned, photovoltaic systems with the same fluctuations can be correctly identified, improving clustering accuracy in dynamic scenarios.
[0023] Construct a weighted undirected graph using the composite similarity matrix as the edge weights. Here, the node set V contains all photovoltaic (PV) nodes, the edge set E contains all PV pairs, and the edge weights are... Calculate the normalized Laplacian matrix of the weighted undirected graph using the following formula: ; in, To normalize the Laplace matrix; It is the identity matrix; Let S be a degree matrix, and its i-th diagonal element is the sum of all elements in the i-th row of the composite similarity matrix S; right Perform feature decomposition, before extraction The eigenvectors corresponding to the smallest non-zero eigenvalues constitute the feature matrix. , , To normalize the order of the Laplacian matrix, while preserving clustering information, the computational cost of subsequent density peak clustering can be reduced, accelerating the clustering process and transforming the feature matrix... Each row is considered as a corresponding photovoltaic cell. The coordinates in the feature space are used to calculate the local density and relative distance of each photovoltaic (PV) unit. Decision values are generated based on the local density and relative distance, and all PV units under each generalized weather type group are clustered based on the decision values.
[0024] Regarding the construction of the weighted undirected graph described above, it should be noted that the weighted undirected graph is constructed using the composite similarity moment matrix as the edge weights. The normalized Laplacian matrix of the graph is calculated, and the eigenvectors corresponding to the first few smallest non-zero eigenvalues are extracted to form a low-dimensional feature matrix. Spectral decomposition transforms the originally complex graph cut problem into a simple clustering problem in a low-dimensional space. This significantly reduces the data dimensionality without losing key topological information, substantially reducing the computational cost of subsequent density peak clustering. Simultaneously, low-dimensional embedding allows the originally complex photovoltaic nodes in space to exhibit a clearer clustered distribution in the feature space, effectively amplifying inter-class differences and making subsequent clustering clearer and more efficient.
[0025] Furthermore, in some embodiments, the decision value is calculated according to the following formula: ; ; ; in, Let be the decision value for the i-th photovoltaic cell; For the first The relative distance between each photovoltaic unit; For the first Local density of photovoltaic cells in characteristic space The Euclidean distance between the i-th photovoltaic and the j-th photovoltaic in the feature space is the Euclidean distance between the vectors in the i-th row and the j-th row of F. The Euclidean distance bandwidth parameter is defined as the Euclidean distance between all photovoltaic pairs. The median of Indicates when When the maximum value is found among all photovoltaics, the maximum Euclidean distance from the i-th photovoltaic cell to the other photovoltaic cells is taken. Indicates when When the maximum value is not found in all photovoltaic systems, the density is higher in some local areas. Among all photovoltaics, take the minimum Euclidean distance from the i-th photovoltaic to all photovoltaics; Arrange the decision values in descending order, calculate the relative rate of change of adjacent decision values according to the following formula, and determine the number of clusters K and the initial cluster centers based on the K photovoltaic values with the largest relative rates of change: ; in, The k-th decision value after descending order and the k-th decision value The relative rate of change of each decision value; The value is the k-th decision value after sorting in descending order; n is the total number of photovoltaic cells. Regarding the formula for the decision values mentioned above, it should be noted that common clustering algorithms such as K-means require a preset number of clusters and random selection of initial centers, leading to inconsistent results in each run. This invention sorts the decision values in descending order, calculates the relative rate of change of adjacent decision values, and locks the K points with the largest relative rates of change as the initial cluster centers. When candidate values are not unique, a comprehensive screening is conducted using meteorological consistency indicators, root mean square error of power, and mean absolute error, selecting the value that best combines the three indicators to ensure that the number of clusters has a unique and physically meaningful solution. Finally, photovoltaic cells not at the initial centers are sorted by local density from high to low and sequentially assigned to the group containing the photovoltaic cells with a higher density and the closest Euclidean distance in the feature space. This avoids the randomness and local optima traps caused by the reliance on random initialization in K-means after traditional spectral clustering, resulting in stable and reproducible clustering results.
[0026] If the number of photovoltaic cells at the end of the top K photovoltaic cells with the largest relative change rates is not unique, then obtain the number of all possible candidate clusters and calculate each index according to the following formula: ; ; ; in, The variance of the weather type index for all photovoltaics in the c-th candidate cluster is given by the number of clusters, i.e., the meteorological consistency index. Let be the number of photovoltaic cells in all photovoltaic cells within the c-th candidate cluster. The average weather type index of all photovoltaics at time t among the c-th candidate clusters. The average root mean square error of power for all photovoltaic cells in the c-th candidate cluster is given. For the i-th photovoltaic in Average power over the time period This represents the average absolute error of the power of all photovoltaic cells in the c-th candidate cluster. Select to make Minimum corresponding The value is used as the final number of photovoltaic (PV) clusters. PVs included in these final clusters are used as initial cluster centers. PVs not in the initial cluster centers are sorted by their local density from high to low. Each PV is then assigned to the group containing the PV with the highest density and the closest Euclidean distance in the feature space, until all PVs are assigned. This completes the spatial clustering of PVs under this weather type, avoiding the randomness that may occur in K-means assignment after traditional spectral clustering.
[0027] Step S103: Calculate the average power of each photovoltaic cluster and calculate the characteristic index of each photovoltaic unit in the photovoltaic cluster based on the average power of the cluster. The characteristic index includes Pearson correlation coefficient, dynamic time warping distance and amplitude deviation index, so as to calculate the comprehensive representativeness score of each photovoltaic unit in any photovoltaic cluster based on all the characteristic indexes. In this step, based on the clustering results, a multi-index dynamic comprehensive evaluation method is used to determine the representative power stations of each photovoltaic cluster, forming a multi-cluster equivalent model. Specifically, the average power of the cluster is first calculated according to the following formula: ; in, Let be the average power of all photovoltaic cells in the c-th photovoltaic cluster at time t; This represents the total number of photovoltaic cells within the photovoltaic cluster. Let be the power of the i-th photovoltaic cell within the cluster at time t; The Pearson correlation coefficient, dynamic time warp distance, and amplitude deviation index are calculated using the following formulas: ; ; ; in, Let be the Pearson correlation coefficient between the i-th photovoltaic power and the cluster average power sequence; Let be the average power of the i-th photovoltaic cell within T; Cluster average power The mean within T, Let be the magnitude deviation index of the i-th photovoltaic power relative to the average power of the cluster; It is a non-zero positive number. Let be the dynamic time-normalized distance between the power of the i-th photovoltaic (PV) at time t and the average power of all PVs in the c-th PV cluster at time t. This dynamic time-normalized distance is calculated in essentially the same way as the dynamic time-normalized distance between the aforementioned PV power curves, except that it uses the active power of the i-th PV. The average power sequence of all photovoltaics within the c-th photovoltaic cluster , build The local distance matrix.
[0028] The overall representativeness score is calculated using the following formula: ; in, Score the overall representativeness of the i-th photovoltaic cell; These are the weighting coefficients for Pearson correlation coefficient, dynamic time-warped distance, and amplitude deviation index, respectively. for The value after positive normalization; for The value after negative normalization; for The value after negative normalization.
[0029] Regarding the representativeness scoring formula mentioned above, it should be noted that the Pearson correlation coefficient measures the linear correlation between photovoltaic power and the average power of the cluster, ensuring that the representative power station is highly consistent with the group in terms of trend fluctuations; the dynamic time warping distance allows for flexible matching in time series, identifies the global morphological similarity of the power waveform, and improves the shortcomings of the correlation coefficient in being insensitive to nonlinear scaling; the amplitude deviation index introduces non-zero positive numbers. As a stabilizing factor, it quantifies the relative difference between the average amplitude of individuals and the group, preventing equivalence errors caused by high correlation but vastly different amplitudes. This invention first performs positive normalization to unify the three factors to the same dimension, eliminating dimensional interference. The weighting coefficients can provide flexible adjustment space. In transient stability studies, the distance weight can be increased to highlight dynamic response, and in steady-state power flow analysis, the magnitude deviation weight can be increased to highlight power.
[0030] Step S104: Select the photovoltaic with the highest comprehensive representativeness score within the same photovoltaic cluster as the target photovoltaic, collect the real-time active power of the target photovoltaic, calculate the power value of other photovoltaics based on the real-time active power of the target photovoltaic, and construct an equivalent impedance model based on the real-time active power and power value.
[0031] In this step, the power values of other photovoltaic systems are calculated using the following formula: ; in, Let j be the power value of the j-th other photovoltaic power generation unit at time t. Let be the real-time active power of the target photovoltaic system at time t. To achieve the target installed capacity of photovoltaic power, Let be the installed capacity of the j-th other photovoltaic system; The low-voltage distribution area's collector lines are further divided into radial and trunk types. If it is a radial type, the equivalent impedance model is constructed according to the following formula: ; in, This is the equivalent impedance of a radial collector line. Let be the impedance of the i-th radial collector line. It is the sum of the active power of all photovoltaics flowing through the i-th radial collector line, where n is the total number of radial collector lines; If it is a trunk line type, then construct the equivalent impedance model according to the following formula: ; in, Let be the equivalent impedance after transformation of the nth trunk-type collector line. Let be the impedance of the nth collector line. Let be the active power flowing through the i-th photovoltaic cell in the n-th trunk collector line. This represents the total number of photovoltaic cells flowing through the nth trunk-type collector line.
[0032] Furthermore, the output results of the equivalent model and the detailed model under different weather conditions need to be compared to verify the effectiveness of this method. The effectiveness verification of the electromagnetic transient model under different weather conditions includes combinations of operating conditions with different voltage drop and voltage rise levels. The equivalent model and the detailed model are placed on the same electromagnetic transient simulation platform, with uniform simulation step size conditions set. Electrical quantities such as positive sequence voltage, active current, reactive current, active power, and reactive power at the grid connection point are synchronously collected for both models under various operating conditions. Based on the technical specifications for photovoltaic power plant modeling and parameter testing, the relative errors of each electrical quantity are calculated using the following formula. Error calculations are performed on the data for each operating condition to verify the effectiveness of the equivalent model. If the errors of all electrical quantities under all operating conditions meet the threshold requirements, the equivalent model is valid.
[0033] ; in, The relative error is N, where N is the number of sampling points. The electrical quantity value at the k-th sampling point in the equivalent model. This represents the electrical quantity value at the k-th sampling point in the detailed model.
[0034] For example, taking a region with a distributed photovoltaic low-voltage system as an example, the detailed model is as follows: Figure 2 As shown, the system includes a total of 17 distributed photovoltaic systems. Following the clustering equivalence method in steps 2 and 5, the equivalent model is obtained as follows: Figure 3 As shown, the equivalent model equates all photovoltaic (PV) systems to the 220kV side, encompassing a total of 7 distributed PV systems: 3 Class A distributed PV systems, 3 Class B distributed PV systems, and one distributed PV system that does not require equation. Based on the PSMODEL platform, the effectiveness of the equivalent electromagnetic transient model under different operating conditions was verified according to the operating conditions shown in Table 2.
[0035] Table 2 Test Conditions Table
[0036] Depend on Figures 4 to 6It can be seen that the photovoltaic output P = 0.3Pn, where Pn is the rated active power output of the photovoltaic system. Under the condition of a three-phase symmetrical fault, with the fault phase voltage Ufault = 50%Un, where Un is the rated voltage of the low-voltage distribution area, a comparison of voltage, power, and current between the detailed model and the equivalent model is shown. The average deviations of voltage, active power, reactive power, active current, and reactive current in the steady-state range between the detailed model and the equivalent model are 0, 0.004, 0.005, 0.004, and 0.005, respectively, and the average deviations in the transient range are 0, 0.003, 0.003, 0.003, and 0.003, respectively.
[0037] Depend on Figures 7 to 9 It can be seen that when the photovoltaic output P=0.6Pn, a three-phase symmetrical fault occurs, and the fault phase voltage Ufault=50%Un, the voltage, power, and current of the detailed model and the equivalent model are compared. The average deviations of voltage, active power, reactive power, active current, and reactive current of the detailed model and the equivalent model in the steady-state range are 0, 0.01, 0.009, 0.01, and 0.009, respectively, and the average deviations in the transient range are 0, 0.012, 0.005, 0.012, and 0.005, respectively.
[0038] Depend on Figures 10 to 12 It can be seen that when the photovoltaic output P=0.8Pn, a three-phase symmetrical fault occurs, and the fault phase voltage Ufault=50%Un, the voltage, power, and current of the detailed model and the equivalent model are compared. The average deviations of voltage, active power, reactive power, active current, and reactive current in the steady-state range between the detailed model and the equivalent model are 0.01, 0.012, 0.017, 0.012, and 0.017, respectively, and the average deviations in the transient range are 0.001, 0.01, 0.01, 0.01, 0.01, and 0.01, respectively.
[0039] Depend on Figures 13 to 15 It can be seen that when the photovoltaic output P=0.3Pn, a three-phase symmetrical fault occurs, and the fault phase voltage Ufault=80%Un, the voltage, power, and current of the detailed model and the equivalent model are compared. The average deviations of voltage, active power, reactive power, active current, and reactive current of the detailed model and the equivalent model in the steady-state range are 0, 0.003, 0.004, 0.003, and 0.004, respectively, and the average deviations in the transient range are 0, 0.001, 0.002, 0.001, and 0.002, respectively.
[0040] Depend on Figures 16 to 18It can be seen that when the photovoltaic output P = 0.6Pn, a three-phase symmetrical fault occurs, and the fault phase voltage Ufault = 80%Un, the voltage, power, and current of the detailed model and the equivalent model are compared. The average deviations of voltage, active power, reactive power, active current, and reactive current in the steady-state range between the detailed model and the equivalent model are 0, 0.003, 0.011, 0.003, and 0.011, respectively, and the average deviations in the transient range are 0, 0.002, 0.006, 0.002, and 0.006, respectively.
[0041] Depend on Figures 19 to 21 It can be seen that when the photovoltaic output P=0.8Pn, a three-phase symmetrical fault occurs, and the fault phase voltage Ufault=50%Un, the voltage, power, and current of the detailed model and the equivalent model are compared. The average deviations of voltage, active power, reactive power, active current, and reactive current in the steady-state range of the detailed model and the equivalent model are 0.001, 0.011, 0.019, 0.011, and 0.019, respectively, and the average deviations in the transient range are 0.011, 0.006, 0.012, 0.006, and 0.012, respectively. The above average deviation calculation results show that the verification results for operating conditions 1-6 all meet the requirements of "Q / GDW 10993-2017 Photovoltaic Power Station Modeling and Parameter Testing Procedure", verifying the effectiveness of the equivalent modeling method of this invention.
[0042] In summary, this invention significantly improves the accuracy and adaptability of low-voltage distributed photovoltaic (PV) equivalent modeling by introducing a two-stage strategy combining generalized weather type clustering and spatial clustering, and by integrating multiple indicators for comprehensive evaluation to select the optimal representative power station. This effectively solves the problems of large modeling errors and poor scenario adaptability caused by neglecting meteorological differences and spatial correlations in traditional methods. Specifically, firstly, a weather type index is calculated based on the ratio of actual PV output to rated output, dividing all sampling times into generalized weather type groups such as sunny, cloudy, and rainy days. This allows the equivalent model to dynamically adjust the modeling benchmark according to different meteorological conditions, avoiding the problem of poor applicability of a single model across all weather scenarios. Secondly, for each generalized weather type group, a composite distance index integrating spatial Euclidean distance and electrical distance is further constructed. This is combined with a dynamic time warping algorithm to measure the shape similarity of PV power curves. Adaptive spectral density clustering automatically determines the number of clusters and cluster centers, achieving refined spatial clustering of geographically dispersed PV power plants with varying output patterns. This solves the problem of traditional clustering methods relying on a single distance index or manual setting. The invention addresses the issue of unreasonable clustering caused by threshold values. It then calculates the average power within each photovoltaic cluster and uses three characteristic indicators—Pearson correlation coefficient, dynamic time warping distance, and amplitude deviation index—to objectively assign a comprehensive representativeness score using the entropy weight method. The photovoltaic system with the highest score is selected as the representative power station, effectively solving the problem that a single indicator (such as only using the correlation coefficient or only using installed capacity) cannot fully reflect the matching degree of photovoltaic power output patterns. Finally, based on the real-time active power of the representative power station, the power values of other photovoltaic systems are calculated according to the proportion of installed capacity. A complete equivalent model is constructed by combining the equal power loss impedance models of radial and trunk topologies. This significantly reduces model complexity and simulation computation while preserving the consistency of power loss characteristics and transient response within the cluster. Through the joint optimization strategy of weather type-spatial clustering-multi-indicator representative selection, this invention can significantly improve the simulation accuracy and robustness of the low-voltage distributed photovoltaic equivalent model under different meteorological and geographical distribution scenarios, providing reliable model support for the planning, operation control, and fault analysis of low-voltage systems with large-scale distributed photovoltaic access.
[0043] like Figure 22 As shown, one embodiment of the present invention proposes an equivalent modeling system for distributed photovoltaic clustering in low-voltage systems, the system comprising: The information acquisition module 10 is used to acquire the photovoltaic geographical location information and photovoltaic operation information of the low-pressure area, and to classify all photovoltaics in the low-pressure area into at least one generalized weather type group based on the photovoltaic geographical location information and photovoltaic operation information. Cluster partitioning module 20 is used to spatially partition each of the generalized weather type groups using an improved dynamic time warping algorithm and an adaptive spectral density clustering algorithm to obtain at least one photovoltaic cluster under each generalized weather type group. The photovoltaic scoring module 30 is used to calculate the average power of each photovoltaic cluster and to calculate the characteristic indicators of each photovoltaic in the photovoltaic cluster based on the average power of the cluster. The characteristic indicators include Pearson correlation coefficient, dynamic time warping distance and amplitude deviation index, so as to calculate the comprehensive representativeness score of each photovoltaic in any photovoltaic cluster based on all the characteristic indicators. The model building module 40 is used to select the photovoltaic with the highest comprehensive representativeness score in the same photovoltaic cluster as the target photovoltaic, collect the real-time active power of the target photovoltaic, calculate the power value of other photovoltaics based on the real-time active power of the target photovoltaic, and construct an equivalent impedance model based on the real-time active power and the power value.
[0044] In another aspect, the present invention also proposes a storage medium on which one or more programs are stored, which, when executed by a processor, implement the above-described method for equivalent modeling of distributed photovoltaic clusters in low-voltage systems.
[0045] In another aspect, the present invention also proposes an electronic device, including a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to realize the above-mentioned low-voltage system distributed photovoltaic cluster equivalent modeling method.
[0046] Those skilled in the art will understand that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can mean any means that can contain stored, communicated, propagated, or transmitted programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.
[0047] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.
[0048] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.
[0049] While embodiments of the present invention have been described in detail above, it will be apparent to those skilled in the art that various modifications and variations can be made to these embodiments. However, it should be understood that such modifications and variations fall within the scope and spirit of the invention as set forth in the claims. Furthermore, the invention described herein may have other embodiments and can be implemented or carried out in various ways.
Claims
1. A low-voltage system distributed photovoltaic sub-group equivalent modeling method, characterized in that, The method includes: Obtain the geographical location information and operation information of photovoltaics in the low-pressure area, and classify all photovoltaics in the low-pressure area into at least one generalized weather type group based on the geographical location information and operation information. An improved dynamic time warping algorithm and an adaptive spectral density clustering algorithm are used to spatially group each of the generalized weather type groups, resulting in at least one photovoltaic cluster under each generalized weather type group, specifically including: The spatial-electrical composite distance is constructed according to the following formula: ; wherein, is the space-electric compound distance between the ith photovoltaic and the jth photovoltaic within the same generalized weather type group, is the maximum space Euclidean distance among all photovoltaic pairs, is the maximum electric distance among all photovoltaic pairs within the corresponding generalized weather type group, is the space weight coefficient, is the electric distance between the ith photovoltaic and the jth photovoltaic point of common coupling within the same generalized weather type group, is the space Euclidean distance between the ith photovoltaic and the jth photovoltaic within the same generalized weather type group; Construct the active power sequence of the i-th photovoltaic and the j-th photovoltaic. Local distance matrix: ; in, The local distance matrix is the t-th row and the th... Column elements, Let be the power value of the i-th photovoltaic cell at time t. For the j-th other photovoltaic in Power value at time, This represents the total sampling time. Build Cumulative distance matrix ,initialization ,for ,make ;for ,make ; calculate recursively according to the following formula The remaining elements: ; in, Cumulative distance matrix The tth row of the middle Column elements; The dynamic time warping distance between the power curves of the i-th and j-th photovoltaic cells can be obtained using the following formula. : ; A composite similarity matrix S is constructed based on the dynamic time warp distance and the spatial-electrical composite distance; Construct a weighted undirected graph using the composite similarity matrix as edge weights, and calculate the normalized Laplacian matrix of the weighted undirected graph. ; right Perform feature decomposition, before extraction The eigenvectors corresponding to the smallest non-zero eigenvalues constitute the feature matrix. , the feature matrix Each row is considered as a corresponding photovoltaic cell. The coordinates in the feature space are used to calculate the local density and relative distance of each photovoltaic (PV) in the feature space. Decision values are generated based on the local density and relative distance. All PVs under each generalized weather type group are clustered based on the decision values. At least one PV cluster under each generalized weather type group is obtained based on the clustering results. Calculate the average power of each photovoltaic cluster, and calculate the characteristic indicators of each photovoltaic unit in the photovoltaic cluster based on the average power. The characteristic indicators include Pearson correlation coefficient, dynamic time warping distance, and amplitude deviation index, so as to calculate the comprehensive representativeness score of each photovoltaic unit in any photovoltaic cluster based on all the characteristic indicators. The photovoltaic (PV) with the highest comprehensive representativeness score within the same PV cluster is selected as the target PV. The real-time active power of the target PV is collected, and the power values of other PVs are calculated based on the real-time active power of the target PV. An equivalent impedance model is then constructed based on the real-time active power and the power values.
2. The method for equivalent modeling of distributed photovoltaic clusters in low-voltage systems according to claim 1, characterized in that, The step of obtaining the photovoltaic geographic location information and photovoltaic operation information of the low-pressure area, and classifying all photovoltaics in the low-pressure area into at least one generalized weather type group based on the photovoltaic geographic location information and photovoltaic operation information includes: Calculate the weather type index using the following formula: ; in, Let i be the weather type index of the i-th photovoltaic cell within the c-th regional block at time t; The output power sampled value of the i-th photovoltaic cell within the c-th regional block at time t; The rated output of the i-th photovoltaic cell within the c-th regional block; Set a weather type index range corresponding to each weather type, and divide the weather type indices of all photovoltaics in all regional blocks according to the weather type index and the weather type index range to obtain at least one generalized weather type group.
3. The method for equivalent modeling of distributed photovoltaic clusters in low-voltage systems according to claim 2, characterized in that, The step of spatially grouping each of the generalized weather type groups using an improved dynamic time warping algorithm and an adaptive spectral density clustering algorithm to obtain at least one photovoltaic cluster under each generalized weather type group includes: Construct a composite similarity matrix using the following formula: ; in, Let i be the element in the i-th row and j-th column of the composite similarity matrix S. The composite distance bandwidth parameter is defined as the composite distance between all photovoltaic pairs under the corresponding generalized weather type group. the median; The dynamic time-warped distance bandwidth parameter is defined as the dynamic time-warped distance between all photovoltaic pairs under the corresponding generalized weather type group. the median; Construct a weighted undirected graph using the composite similarity matrix as the edge weights. Here, the node set V contains all photovoltaic (PV) nodes, the edge set E contains all PV pairs, and the edge weights are... , Calculate the orthogonal Laplacian matrix of a weighted undirected graph using the following formula: ; in, To normalize the Laplace matrix; It is the identity matrix; Let S be a degree matrix, and its i-th diagonal element is the sum of all elements in the i-th row of the composite similarity matrix S.
4. The method for equivalent modeling of distributed photovoltaic clusters in low-voltage systems according to claim 3, characterized in that, The steps of calculating the local density and relative distance of each photovoltaic (PV) in the feature space, generating decision values based on the local density and relative distance, and clustering all PVs under each generalized weather type group based on the decision values include: Calculate the decision value using the following formula: ; ; ; in, Let be the decision value for the i-th photovoltaic cell; For the first The relative distance between each photovoltaic unit; For the first Local density of photovoltaic cells in characteristic space The Euclidean distance between the i-th photovoltaic and the j-th photovoltaic in the feature space is the Euclidean distance between the vectors in the i-th row and the j-th row of F. The Euclidean distance bandwidth parameter is defined as the Euclidean distance between all photovoltaic pairs. the median of Indicates when When the maximum value is found among all photovoltaics, the maximum Euclidean distance from the i-th photovoltaic cell to the other photovoltaic cells is taken. Indicates when When the maximum value is not found in all photovoltaic systems, the density is higher in some local areas. Among all photovoltaics, take the minimum Euclidean distance from the i-th photovoltaic to all photovoltaics; Arrange the decision values in descending order, calculate the relative rate of change of adjacent decision values according to the following formula, and determine the number of clusters K and the initial cluster centers based on the K photovoltaic values with the largest relative rates of change: ; in, The k-th decision value after descending order and the k-th decision value The relative rate of change of each decision value; The value is the k-th decision value after sorting in descending order; n is the total number of photovoltaic cells. If the number of photovoltaic cells at the end of the top K photovoltaic cells with the largest relative change rates is not unique, then obtain the number of all possible candidate clusters and calculate each index according to the following formula: ; ; in, Let be the number of photovoltaic cells in all photovoltaic cells within the c-th candidate cluster. The average root mean square error of power for all photovoltaic cells in the c-th candidate cluster is given. For the i-th photovoltaic in Average power over the time period This represents the average absolute error of the power of all photovoltaic cells in the c-th candidate cluster. Select to make Minimum corresponding The value is used as the final number of photovoltaic clusters, and the photovoltaic cells included in the final number of photovoltaic clusters are used as the initial cluster centers. The photovoltaic cells that are not in the initial cluster centers are sorted from high to low according to their local density. Each photovoltaic cell is then assigned to the group of the previously assigned photovoltaic cells with a higher density and the closest Euclidean distance in the feature space, until all photovoltaic cells are assigned. Let be the variance of the weather type index of all photovoltaics in the number of candidate clusters of type c, i.e., the meteorological consistency index.
5. The method for equivalent modeling of distributed photovoltaic clusters in low-voltage systems according to claim 4, characterized in that, The steps of calculating the average power of each photovoltaic cluster and calculating the characteristic indicators of each photovoltaic unit within the cluster based on the average power, including the Pearson correlation coefficient, dynamic time warping distance, and amplitude deviation index, to calculate the comprehensive representativeness score of each photovoltaic unit within any photovoltaic cluster based on all characteristic indicators include: The average power of the cluster is calculated using the following formula: ; in, Let be the average power of all photovoltaic cells in the c-th photovoltaic cluster at time t; This represents the total number of photovoltaic cells within the photovoltaic cluster. Let be the power of the i-th photovoltaic cell within the cluster at time t; The Pearson correlation coefficient, dynamic time warp distance, and amplitude deviation index are calculated using the following formulas: ; ; ; in, Let be the Pearson correlation coefficient between the i-th photovoltaic power and the cluster average power sequence; Let be the average power of the i-th photovoltaic cell within T; Cluster average power The mean within T, Let be the magnitude deviation index of the i-th photovoltaic power relative to the average power of the cluster; It is a non-zero positive number. Let be the dynamic time-warped distance between the power of the i-th photovoltaic cell at time t and the average power of all photovoltaic cells in the c-th photovoltaic cluster at time t. The overall representativeness score is calculated using the following formula: ; in, Score the overall representativeness of the i-th photovoltaic cell; These are the weighting coefficients for Pearson correlation coefficient, dynamic time-warped distance, and amplitude deviation index, respectively. for The value after positive normalization; for The value after negative normalization; for The value after negative normalization.
6. The method for equivalent modeling of distributed photovoltaic clusters in low-voltage systems according to claim 5, characterized in that, The steps of selecting the photovoltaic (PV) with the highest comprehensive representativeness score within the same PV cluster as the target PV, collecting the real-time active power of the target PV, calculating the power values of other PVs based on the real-time active power of the target PV, and constructing an equivalent impedance model based on the real-time active power and power values include: Calculate the power values of other photovoltaic systems using the following formula: ; in, Let j be the power value of the j-th other photovoltaic power source at time t. Let be the real-time active power of the target photovoltaic system at time t. For the target photovoltaic installed capacity, Let be the installed capacity of the j-th other photovoltaic system; The collector lines of the low-voltage distribution area are divided into radial and trunk types. If it is a radial type, the equivalent impedance model is constructed according to the following formula: ; in, This is the equivalent impedance of a radial collector line. Let be the impedance of the i-th radial collector line. It is the sum of the active power of all photovoltaics flowing through the i-th radial collector line, where n is the total number of radial collector lines; If it is a trunk line type, then construct the equivalent impedance model according to the following formula: ; in, Let be the equivalent impedance after transformation of the nth trunk-type collector line. Let be the impedance of the nth collector line. Let be the active power flowing through the i-th photovoltaic cell in the n-th trunk collector line. This represents the total number of photovoltaic cells flowing through the nth trunk-type collector line.
7. A low-voltage system distributed photovoltaic clustering equivalent modeling system, used to implement the low-voltage system distributed photovoltaic clustering equivalent modeling method as described in any one of claims 1-6, characterized in that, The system includes: The information acquisition module is used to acquire the geographical location information and operation information of photovoltaics in the low-pressure area, and to classify all photovoltaics in the low-pressure area into at least one generalized weather type group based on the geographical location information and operation information of photovoltaics. The cluster partitioning module is used to spatially partition each of the generalized weather type groups using an improved dynamic time warping algorithm and an adaptive spectral density clustering algorithm, so as to obtain at least one photovoltaic cluster under each generalized weather type group. The photovoltaic scoring module is used to calculate the average power of each photovoltaic cluster and to calculate the characteristic indicators of each photovoltaic unit in the photovoltaic cluster based on the average power. The characteristic indicators include Pearson correlation coefficient, dynamic time warping distance and amplitude deviation index, so as to calculate the comprehensive representativeness score of each photovoltaic unit in any photovoltaic cluster based on all the characteristic indicators. The model building module is used to select the photovoltaic with the highest comprehensive representativeness score within the same photovoltaic cluster as the target photovoltaic, collect the real-time active power of the target photovoltaic, calculate the power value of other photovoltaics based on the real-time active power of the target photovoltaic, and construct an equivalent impedance model based on the real-time active power and power value.
8. A storage medium, characterized in that, The storage medium stores one or more programs that, when executed by a processor, implement the low-voltage system distributed photovoltaic clustering equivalent modeling method as described in any one of claims 1-6.
9. An electronic device, characterized in that, The electronic device includes a memory and a processor, wherein: The memory is used to store computer programs; When the processor executes the computer program stored in the memory, it implements the low-voltage system distributed photovoltaic cluster equivalent modeling method as described in any one of claims 1-6.
Citation Information
Patent Citations
Online clustering equivalent modeling method of photovoltaic power plant
CN106451418A
Equivalent modeling method, device and equipment for photovoltaic power generation unit and storage medium
CN119475640A
