Distributed photovoltaic secondary equivalence modeling method based on parameter identification
By employing a parameter identification-based secondary equivalent modeling method for distributed photovoltaic inverters, and using Canopy-FCM clustering and multi-convolutional layer neural network for grouping and identification, the problem of unequal transient response characteristics between class A and class B photovoltaic inverters in existing technologies is solved, thereby improving the representativeness and accuracy of the model. This method is suitable for rapid simulation and analysis of large-scale low-voltage power distribution systems.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-03
- Publication Date
- 2026-03-13
AI Technical Summary
Existing equivalent modeling methods for distributed photovoltaics are insufficient in terms of control characteristic equivalence, and cannot accurately reflect the transient response characteristics of Class A and Class B distributed photovoltaics. Furthermore, the integration of parameter identification technology is weak, making it difficult to meet high standards in terms of simulation accuracy and reliability.
A secondary equivalent modeling method for distributed photovoltaic inverters based on parameter identification is adopted. By calculating the grid-connected equivalent impedance, the Canopy-FCM clustering algorithm is used to group inverters into Class A and Class B. Parameter identification is performed by combining a multi-convolutional layer hybrid convolutional neural network to construct a multi-machine equivalent model and perform secondary equivalent modeling to form Class A and Class B equivalent single-machine inverters.
It achieves accurate clustering of distributed photovoltaic clusters, improves the representativeness and accuracy of the equivalent model, ensures the realistic reproduction of dynamic characteristics during voltage drop and recovery, reduces computational complexity, and is suitable for rapid simulation and analysis of large-scale low-voltage power distribution systems.
Smart Images
Figure CN121659871A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of power system modeling and simulation technology, specifically relating to a distributed photovoltaic secondary equivalent modeling method based on parameter identification. Background Technology
[0002] With the continuous expansion of distributed photovoltaic (PV) grid integration, its impact on power system operation and stability is becoming increasingly significant. In actual power grid simulations, detailed modeling of each distributed PV unit leads to high model complexity, high computational resource consumption, and low simulation efficiency. Therefore, conducting research on equivalent modeling of distributed PV systems is of great significance for improving power grid simulation efficiency and supporting system planning and operation.
[0003] Currently, mainstream methods for equivalent modeling of distributed photovoltaic systems mainly revolve around the following ideas: 1) Capacity-weighted equivalent method: This method aggregates multiple distributed photovoltaic (PV) units into a single equivalent unit, whose capacity is the sum of the cluster's capacities. Electrical parameters are then weighted and averaged according to capacity. While simple and efficient, this method ignores the differences in the electrical location of distributed PV units within the grid and fails to reflect the impact of varying voltage levels within the cluster on the aggregated output, resulting in significant errors during transient processes.
[0004] 2) Impedance Aggregation Equivalent Method: This method attempts to preserve the grid connection of distributed photovoltaic clusters by calculating their equivalent grid-connected impedance. While it improves the equivalence of electrical connections, its core principle remains the linear aggregation of circuit parameters.
[0005] 3) Cluster Analysis Equivalence Method: In recent years, clustering algorithms (such as K-means) based on grid connection impedance and geographical location of distributed photovoltaic systems have become a research focus. This method improves the rationality of clustering to a certain extent.
[0006] However, existing methods suffer from two prominent common shortcomings: First, insufficient equivalence of control systems. Current methods primarily focus on the aggregation of primary electrical systems (such as impedance and capacity), while offering limited equivalence studies on the core of distributed photovoltaic (PV) transient behavior—the inverter control system. In particular, they fail to rigorously distinguish the fundamental differences in low-voltage ride-through control strategies between Type A and Type B PV systems. Type A PV systems possess active support capabilities for low-voltage ride-through control, while Type B PV systems exhibit "damping-and-gradual-recovery" characteristics. Mixing these two types of PV systems with vastly different control characteristics in equivalence leads to severe distortion in the equivalent model's response during voltage dips and recovery, failing to accurately simulate the dynamic characteristics of the real system. Second, weak integration with parameter identification techniques. Traditional equivalent model parameters are mostly based on manufacturer nameplate data or theoretical calculations. In reality, due to equipment aging, environmental factors, and the complexity of their own control strategies, the actual operating characteristics of distributed PV systems deviate from the theoretical model, making it difficult to meet high-standard simulation requirements in terms of accuracy and reliability. Summary of the Invention
[0007] This invention aims to address the shortcomings of existing distributed photovoltaic (PV) equivalent modeling methods in terms of control characteristic equivalence. It proposes a secondary equivalent modeling method for PV based on parameter identification, which is intended to more accurately reflect the transient response characteristics of Class A and Class B distributed PV systems. This will enable the construction of an equivalent model that simplifies the model and reproduces the real transient response of the population, providing a reliable tool for the safety and stability analysis and operation control of the power grid.
[0008] The present invention adopts the following technical solution: The present invention provides a method for secondary equivalent modeling of distributed photovoltaic inverters based on parameter identification, characterized by the following steps: Step 1: Calculate the grid-connected equivalent impedance of each distributed photovoltaic inverter based on the grid topology; Step 2: According to the different control strategies of distributed photovoltaic inverters under low voltage, p+q Distributed photovoltaic inverters are divided into Class A distributed photovoltaic inverters with low-voltage ride-through capability and Class B distributed photovoltaic inverters with low-voltage blocking capability. Step 3: Using the grid-connected equivalent impedance of each distributed photovoltaic inverter as the clustering index, the Canopy-FCM clustering algorithm is used to cluster Class A distributed photovoltaic inverters with low-voltage ride-through capability and Class B distributed photovoltaic inverters with low-voltage blocking capability, respectively, to obtain... c A One Class A distributed photovoltaic cluster, c B One Class B distributed photovoltaic cluster; Step 4: Use parameter identification methods to respectively... cA A Class A distributed photovoltaic cluster and c B Identify Class B distributed photovoltaic clusters and obtain the primary equivalent control parameters for each distributed photovoltaic cluster; Step 5: Calculate the equivalent line impedance of each distributed photovoltaic cluster, and construct a multi-machine equivalent model for each distributed photovoltaic cluster by combining the primary equivalent control parameters. Step 6: Apply system-level disturbances to the Type A multi-machine equivalent model and the Type B multi-machine equivalent model respectively. Record the electrical quantities at the output of the Type A multi-machine equivalent model and the output of the Type B multi-machine equivalent model respectively. Then, input them into a multi-convolutional layer hybrid convolutional neural network for processing to obtain the secondary equivalent control parameters of the Type A multi-machine equivalent model and the secondary equivalent control parameters of the Type B multi-machine equivalent model respectively. Step 7, containing c A The Class A multi-machine equivalent models of several Class A equivalent machines are aggregated into a single Class A equivalent machine, and c A The total capacity of distributed photovoltaic inverters in a Class A distributed photovoltaic cluster is taken as the equivalent capacity of a Class A equivalent single unit, and is obtained through equation (5). c A The equivalent line impedance of a Class A distributed photovoltaic cluster is used as the equivalent impedance of a Class A single unit. Will include c B The equivalent models of multiple equivalent B-type machines are aggregated into a single equivalent B-type machine, and... c B The total capacity of distributed photovoltaic inverters in a Class B distributed photovoltaic cluster is taken as the equivalent capacity of a Class B equivalent single unit, and is obtained through equation (5). c B The equivalent line impedance of a Class B distributed photovoltaic cluster is used as the equivalent impedance of a Class B equivalent single unit; thus, a secondary equivalent model containing one Class A equivalent single unit and one Class B equivalent single unit is constructed.
[0009] The characteristic of the distributed photovoltaic inverter secondary equivalent modeling method based on parameter identification described in this invention is that step 1 utilizes equation (1) a The grid-connected equivalent impedance of a distributed photovoltaic inverter : (1) In formula (1): the 1st to the 2nd p The first distributed photovoltaic inverter is radially connected, the second... p +1~ p+q Each distributed photovoltaic inverter is connected via a trunk line. , These represent the total current and total impedance of the outer layer circuitry, respectively. , The first a Current and impedance of the lines connected to a distributed photovoltaic inverter; For the first j Line current of the lines connected to a distributed photovoltaic inverter; For the first i The impedance of the lines connected to each distributed photovoltaic inverter.
[0010] Furthermore, step 3 includes the following steps: Step 3.1: Perform the first stage of Canopy coarse clustering on the Class A distributed photovoltaic inverters to obtain the initial cluster centers; Step 3.2, for c The initial cluster centers are then subjected to a second stage of fuzzy C-means fine clustering to obtain the final cluster centers. V A and membership matrix U A ; Step 3.3, according to V A and U A Step 3.1.1 D A Each data point is assigned to the cluster center with the highest membership degree, thus obtaining c A One Class A distributed photovoltaic cluster; Step 3.4: Following the process in Steps 3.1-3.3, perform the first stage of Canopy coarse clustering and the second stage of fuzzy C-means fine clustering on the Class B distributed photovoltaic inverters to obtain... c B A Class B distributed photovoltaic cluster.
[0011] Furthermore, step 3.1 includes the following steps: Step 3.1.1: Take the grid-connected equivalent impedance of any distributed photovoltaic inverter in the Class A distributed photovoltaic inverter as a data point b, and take the real part resistance of data point b. and imaginary reactance As a two-dimensional data point Thus, a two-dimensional data point set for a type A distributed photovoltaic inverter is constructed. D A ; Select the first distance threshold T 1 and second distance threshold T 2, and T 1> T 2; Step 3.1.2, from D A Randomly select a data point P As the center point; and calculate D A All other data points in P Euclidean distance, for all distances less than T All data points of 1 were added to P In the set of center points, the center point P Add to the set of center points Mc In the middle; all distances less than T 2 data points from D A Remove from, thus updating D A The updated two-dimensional data point set is obtained. D’ A ; Step 3.1.3, will D’ A Assign to D A Then, return to step 3.1.2 until... D’ A Until the set of centers is empty, the final set of centers is obtained. Mc and will Mc of c Each center point is used as c The initial cluster centers.
[0012] Furthermore, step 5 includes the following steps: Step 5.1: Calculate the first step using equation (2). c In a distributed photovoltaic cluster, the first f Voltage difference between a distributed photovoltaic inverter and the point of common coupling : (2) In formula (2): , The first c In a distributed photovoltaic cluster, the first f The impedance of the line connected to the first distributed photovoltaic inverter and the first f The active power output of a distributed photovoltaic inverter; The voltage at the point of common coupling. c 1=1,2,…, c A + c B ; Step 5.2: Calculate the equivalent first step using equations (3) and (4) respectively. cWeighted average voltage difference between distributed photovoltaic inverters and the point of common coupling in a distributed photovoltaic cluster And the voltage difference between the equivalent distributed photovoltaic inverter and the point of common coupling. : (3) (4) In equations (3) and (4): N 1 is the first c The number of distributed photovoltaic inverters in a distributed photovoltaic cluster; For the first c In a distributed photovoltaic cluster, the first f Equivalent line impedance of a distributed photovoltaic inverter; Step 5.3, let Using equation (5), we obtain the first... c Equivalent line impedance of a distributed photovoltaic cluster : (5) Step 5.4: Following the process of steps 5.1-5.3, we obtain... c A A Class A distributed photovoltaic cluster and c B The equivalent line impedance of a Class B distributed photovoltaic cluster is used to construct a system containing... c A A Class A equivalent machine, a Class A multi-machine equivalent model, and including c B A Class B multi-machine equivalent model of a Class B equivalent machine.
[0013] The present invention provides an electronic device, including a memory and a processor, characterized in that the memory is used to store a program supporting the processor in performing the method described therein, and the processor is configured to execute the program stored in the memory.
[0014] The present invention discloses a computer-readable storage medium storing a computer program, characterized in that the computer program is executed by a processor to perform the steps of the method described thereon.
[0015] Compared with existing technologies, the beneficial effects of this invention are reflected in: 1. This invention calculates the grid-connected equivalent impedance of each photovoltaic inverter and employs the Canopy-FCM hybrid clustering algorithm to finely group Class A photovoltaic systems with low-voltage ride-through capability and Class B photovoltaic systems with low-voltage blocking capability. This method not only fully considers the essential differences in control strategies between Class A and Class B photovoltaic systems but also reflects their electrical connection relationships in the power grid through impedance characteristics. This ensures that the equivalent model can realistically reproduce the group dynamic characteristics of the cluster during transient processes such as voltage dips and recovery, thus improving the representativeness of the equivalent model.
[0016] 2. This invention employs multi-convolutional layer hybrid convolutional neural network and other parameter identification techniques to automatically identify the PI control parameters, low-voltage ride-through control parameters, and sealing characteristic parameters of the equivalent model. This method can extract equivalent parameters that truly reflect the dynamic characteristics of the system from massive simulation data, overcoming the limitations of traditional methods that rely on ideal parameters, and improving the model accuracy and its consistency with field measured data.
[0017] 3. This invention proposes a secondary equivalent model strategy. First, a multi-machine equivalent model based on multiple clusters is constructed to retain key dynamic characteristics. Then, through secondary parameter identification, it is further aggregated into one Class A equivalent machine and one Class B equivalent machine. This stepwise simplification method, while preserving the system's key dynamic behaviors, reduces the number of model nodes and computational complexity, achieving a balance between simulation accuracy and computational efficiency. It is suitable for rapid simulation and analysis of large-scale low-voltage power distribution systems.
[0018] 4. By calculating the relative errors of various electrical quantities and comparing them with industry standards, this invention systematically verifies that the response performance of the equivalent model under various disturbances conforms to the actual system behavior, and the error is strictly controlled within the allowable range of engineering, thus ensuring the credibility of the equivalent model. Attached Figure Description
[0019] Figure 1 This is a schematic diagram of the distributed photovoltaic secondary equivalent modeling process of the present invention; Figure 2 A schematic diagram of the radial structure of a photovoltaic power generation unit; Figure 3 This is a schematic diagram of the equivalent structure of a photovoltaic power generation unit trunk line. Figure 4 Detailed model of a distributed photovoltaic system; Figure 5 A comparison chart of power simulation results between the detailed model and the single-unit equivalent model of a distributed photovoltaic system; Figure 6 A comparison of power simulation results between the detailed model and the quadratic equivalent model of a distributed photovoltaic system. Detailed Implementation
[0020] In this embodiment, as Figure 1 As shown, a distributed photovoltaic (PV) secondary equivalent modeling method based on parameter identification is a high-precision modeling method that can simultaneously consider the realism of electrical connections and the differences in control characteristics, and can extract equivalent parameters from data using parameter identification technology. This method, through a technical route of "impedance calculation - classification and clustering - parameter identification - multi-level equivalence," aims to achieve the grouping and equivalence of Class A and Class B distributed PV systems. This addresses a critical technical bottleneck that urgently needs to be solved for accurate system simulation in the context of high-proportion distributed PV grid integration. Specifically, the method includes the following steps: Step 1: Collect data such as line impedance parameters, transformer parameters, rated capacity of distributed photovoltaic inverters, grid connection level and inverter type (Class A or Class B) of the distributed photovoltaic area to be equivalently calculated, and calculate the grid-connected equivalent impedance of each distributed photovoltaic inverter. Step 1.1 Collect the resistance data of all relevant 220kV, 110kV, 35kV, and 10kV lines in this area. R and reactance X Parameters (unit: Ω / km) include the short-circuit impedance, no-load loss, rated capacity, and turns ratio of the distribution transformer; the installation location of each distributed photovoltaic inverter, its corresponding grid node, rated capacity (kVA), grid connection type (clearly distinguishing between Class A and Class B distributed photovoltaic inverters), and the inverter's factory control parameters. All line parameters are reduced to the same voltage level (220kV side), and a detailed simulation model of this area is built in PSASP for subsequent result comparison.
[0021] Step 1.2: Based on the power grid topology, calculate the grid-connected equivalent impedance of each distributed photovoltaic inverter using equation (1): (1) In formula (1): For the first a The grid-connected equivalent impedance of a distributed photovoltaic inverter, where, for example Figure 2 The first to the second shown p Each distributed photovoltaic inverter is connected radially, such as Figure 3 The first one shown p +1~ p+q Each distributed photovoltaic inverter is connected via a trunk line. , These represent the total current and total impedance of the outer layer circuitry, respectively. , The first a Current and impedance of the lines connected to a distributed photovoltaic inverter; For the first j The line current of the lines connected to each distributed photovoltaic unit; For the first iThe impedance of the lines connected to each distributed photovoltaic inverter.
[0022] Step 2: According to the different control strategies of distributed photovoltaic inverters under low voltage, p+q Distributed photovoltaic inverters are divided into Class A distributed photovoltaic inverters with low-voltage ride-through capability and Class B distributed photovoltaic inverters with low-voltage blocking capability. Step 3: Using the grid-connected equivalent impedance of each distributed photovoltaic inverter as the clustering index, the Canopy-FCM clustering algorithm is used to cluster Class A distributed photovoltaic inverters with low-voltage ride-through capability and Class B distributed photovoltaic inverters with low-voltage blocking capability, respectively, to obtain... c A One Class A distributed photovoltaic cluster, c B One Class B distributed photovoltaic cluster; The Canopy-FCM clustering algorithm combines the coarse-scanning, fast pre-classification of the Canopy algorithm with the fine-grained fuzzy clustering of the Fuzzy C-means (FCM) algorithm. It leverages the high efficiency of the Canopy algorithm, which can quickly determine the number of clusters and approximate centroids, and the high accuracy of the FCM algorithm, which can handle fuzzy cases where data points belong to multiple clusters. The first stage uses Canopy coarse clustering to quickly and roughly divide the data, initially determining the number and range of cluster centers. This provides optimized initial cluster centers for the FCM algorithm, preventing it from getting trapped in local optima due to improper initial center selection and significantly reducing the data complexity and number of iterations required by the subsequent FCM algorithm. The second stage uses Fuzzy C-means (FCM) fine clustering. Building upon the initial data provided by Canopy, it performs fine-grained fuzzy clustering, considering the continuity of impedance distribution and the fuzziness of boundary points. This better handles distributed photovoltaic units with impedance characteristics located in boundary regions, making the clustering results more consistent with the electrical continuity characteristics of the actual power grid.
[0023] Step 3.1: Perform the first stage of Canopy coarse clustering on the Class A distributed photovoltaic inverters to obtain the initial cluster centers: Step 3.1.1: Take the grid-connected equivalent impedance of any distributed photovoltaic inverter in the Class A distributed photovoltaic inverter as a data point b, and take the real part resistance of data point b. and imaginary reactance As a two-dimensional data point Thus, a two-dimensional data point set for a type A distributed photovoltaic inverter is constructed. D A ; Select the first distance threshold T 1 and second distance threshold T 2, and T 1> T2. For dataset A, set a first distance threshold. T 1A =0.5pu, second distance threshold T 2A =0.3pu; For the B-class dataset, set the first distance threshold. T 1B =0.5pu, second distance threshold T 2B =0.3pu.
[0024] Step 3.1.2, from D A Randomly select a data point P As the center point; and calculate D A All other data points in P Euclidean distance, for all distances less than T All data points of 1 were added to P In the set of centers, the center point will be... P Add to the set of center points Mc In the middle; all distances less than T 2 data points from D A Remove from, thus updating D A The updated two-dimensional data point set is obtained. D’ A ; Step 3.1.3, will D’ A Assign to D A Then, return to step 3.1.2 until... D’ A Until it becomes empty, thus obtaining the final result. Mc Set, and Mc set c The number of clusters whose initial cluster centers are each centroid; Step 3.2: Based on the number of clusters c A second stage of fuzzy C-means fine clustering is performed on the initial cluster centers to obtain the final cluster centers. V A and membership matrix U A .
[0025] Step 3.3, according to V A and U A ,Will D AEach data point is assigned to the cluster center with the highest membership degree, thus obtaining c A One Class A distributed photovoltaic cluster; Step 3.4: Following the process in Steps 3.1-3.3, perform the first stage of Canopy coarse clustering and the second stage of fuzzy C-means fine clustering on the Class B distributed photovoltaic inverters to obtain... c B A Class B distributed photovoltaic cluster.
[0026] Step 4: Use parameter identification methods to respectively... c A A Class A distributed photovoltaic cluster and c B Identify Class B distributed photovoltaic clusters and obtain the primary equivalent control parameters for each distributed photovoltaic cluster; Step 4.1: For each cluster obtained in Step 3, build a detailed simulation model in PSASP software, apply voltage drop disturbance, set 3 sets of operating conditions to instantly drop the grid connection point voltage to 0.2pu, 0.5pu, and 0.8pu, and restore it after 0.15 seconds, and obtain transient response data of voltage, current, active power, and reactive power at the cluster output. Step 4.2: In the convolutional layer convolution operation type of the convolutional neural network algorithm, select the multi-convolutional layer hybrid convolution mode using adaptive edge-filled Full convolution and small-kernel Valid convolution. Use this multi-convolutional layer hybrid convolutional neural network algorithm to identify the PI parameters, low-voltage ride-through control parameters, and low-voltage blocking characteristic parameters of each cluster equivalent model. The objective function is to minimize the error between the detailed simulation data and the equivalent simulation data of each cluster. Finally, the PI controller parameters and low-voltage ride-through control parameters of the Class A equivalent machine, and the blocking duration of the Class B equivalent machine are obtained. T Active current recovery slope K p reactive current recovery slope K q .
[0027] Step 5: Calculate the equivalent line impedance of each distributed photovoltaic cluster, and construct a multi-machine equivalent model for each distributed photovoltaic cluster by combining the primary equivalent control parameters. Step 5.1: Calculate the first step using equation (2). c In a distributed photovoltaic cluster, the first f Voltage difference between a distributed photovoltaic inverter and the point of common coupling : (2) In formula (2): , The firstc In a distributed photovoltaic cluster, the first f The impedance of the line connected to the first distributed photovoltaic inverter and the first f The active power output of a distributed photovoltaic inverter; The voltage at the point of common coupling. c 1=1,2,…, c A + c B ; Step 5.2: Calculate the equivalent first step using equations (3) and (4) respectively. c Weighted average voltage difference between distributed photovoltaic inverters and the point of common coupling in a distributed photovoltaic cluster And the voltage difference between the equivalent distributed photovoltaic inverter and the point of common coupling. : (3) (4) In equations (3) and (4): N 1 is the first c The number of distributed photovoltaic inverters in a distributed photovoltaic cluster; For the first c In a distributed photovoltaic cluster, the first f Equivalent line impedance of a distributed photovoltaic inverter.
[0028] Step 5.3, let Using equation (5), we obtain the first... c Equivalent line impedance of a distributed photovoltaic cluster : (5) Step 5.4: Following the process of steps 5.1-5.3, we obtain... c A A Class A distributed photovoltaic cluster and c B The equivalent line impedance of a Class B distributed photovoltaic cluster is used to construct a system containing... c A A Class A equivalent machine, a Class A multi-machine equivalent model, and including c B A Class B multi-machine equivalent model of a Class B equivalent machine.
[0029] Step 6: Based on the multi-machine equivalent model, apply system-level disturbances to the Type A multi-machine equivalent model and the Type B multi-machine equivalent model respectively, record the electrical quantities at the output of the Type A multi-machine equivalent model and the output of the Type B multi-machine equivalent model respectively, and input them into a multi-convolutional layer hybrid convolutional neural network for processing to obtain the secondary equivalent control parameters of the Type A multi-machine equivalent model and the secondary equivalent control parameters of the Type B multi-machine equivalent model respectively. Step 7, containing c A The Class A multi-machine equivalent models of several Class A equivalent machines are aggregated into a single Class A equivalent machine, and c A The total capacity of distributed photovoltaic inverters in a Class A distributed photovoltaic cluster is taken as the equivalent capacity of a single Class A unit. c A The equivalent line impedance of a Class A distributed photovoltaic cluster is obtained by formula (5). Will include c B The equivalent models of multiple equivalent B-type machines are aggregated into a single equivalent B-type machine, and... c B The total capacity of distributed photovoltaic inverters in a Class B distributed photovoltaic cluster is taken as the equivalent capacity of a single Class B unit. c B The equivalent line impedance of a Class B distributed photovoltaic cluster is obtained by formula (5) to obtain the equivalent impedance of a Class A equivalent single unit; thus, a secondary equivalent model containing a Class A equivalent single unit and a Class B equivalent single unit is constructed.
[0030] Step 8: Set up various voltage drop disturbance operating conditions, compare the dynamic response of the secondary equivalent model with the detailed model and the single-machine equivalent model in terms of voltage, power, frequency, etc., and calculate the weighted average deviation with reference to standards such as GB / T 32892-2016. If the error index under all verification conditions is within the allowable range of engineering, the secondary equivalent modeling method is determined to be accurate and effective and can be used to replace the detailed model for system-level simulation analysis. The practicality of the secondary equivalent model for common single-machine equivalent models is also compared.
[0031] In this embodiment, an electronic device includes a memory and a processor. The memory stores a program that supports the processor in executing the above-described method, and the processor is configured to execute the program stored in the memory.
[0032] In this embodiment, a computer-readable storage medium stores a computer program, which is executed by a processor to perform the steps of the above method.
[0033] Example: 1. Collect the resistance data of all relevant 220kV, 110kV, 35kV, and 10kV lines in this area. R and reactance X Parameters (unit: Ω / km), including the short-circuit impedance, no-load loss, rated capacity, and turns ratio of the distribution transformer; the grid node corresponding to the installation location of each distributed photovoltaic inverter, its rated capacity (kVA), grid connection type (clearly distinguishing between Class A or Class B distributed photovoltaic inverters), and the inverter's factory control parameters. All line parameters are reduced to the same voltage level (220kV side), and a detailed simulation model of this area is built in PSASP. Photovoltaic power generation unit parameters are shown in Table 1, transformer parameters in Table 2, and collector line parameters in Table 3. A detailed model of the distributed photovoltaic system is available in [Table missing]. Figure 4 .
[0034] Table 1 Photovoltaic power generation unit parameters
[0035] Table 2 Transformer Parameters
[0036] Table 3 Parameters of the collector line
[0037] 2. Calculate the grid-connected equivalent impedance of each distributed photovoltaic inverter according to step 2. The calculation results are shown in Table 4.
[0038] Table 4 Equivalent Impedance of Distributed Photovoltaic Inverters Connected to the Grid
[0039] 3. Following step 3, the Canopy-FCM clustering algorithm was used to group the distributed photovoltaic inverters of categories A and B respectively. The grouping results are shown in Table 5.
[0040] Table 5. Results of Distributed Photovoltaic Inverter Clustering
[0041] 4. Following step 4, for each cluster obtained in step 3, build a detailed simulation model in PSASP software, apply voltage dip disturbances, and set three operating conditions to momentarily drop the grid connection point voltage to 0.2 pu, 0.5 pu, and 0.8 pu, lasting for 0.15 seconds before recovering. Obtain transient response data such as voltage, current, active power, and reactive power at the cluster outlet. Use a multi-convolutional layer hybrid convolutional neural network algorithm to identify the control parameters of each cluster's equivalent model. Using minimizing the error between the detailed simulation data and the equivalent simulation data of each cluster as the objective function, finally obtain the PI controller parameters and low-voltage ride-through control parameters of the Class A equivalent machine, and the wave blocking duration of the Class B equivalent machine. TActive current recovery slope K p reactive current recovery slope K q The identification results are shown in Table 6.
[0042] Table 6 Identification results of control parameters for each cluster
[0043] 5. Calculate the equivalent line impedance for each cluster according to step 5, and construct a multi-machine equivalent model, including... c A + c B The equivalent line impedance calculation results for each cluster are shown in Table 7.
[0044] Table 7 Calculation results of equivalent line impedance for each cluster
[0045] 6. Construct multi-machine equivalent models. Apply system-level disturbances to the Type A and Type B multi-machine equivalent models constructed in step 5, record the electrical quantities at the output of the equivalent machines, and use a multi-convolutional layer hybrid convolutional neural network algorithm to perform secondary parameter identification on the simulation results of the Type A and Type B distributed photovoltaic multi-machine equivalent models, respectively, to obtain a set of Type A single-machine equivalent control parameters and Type B single-machine equivalent control parameters for the final simplified model, and calculate their secondary equivalent impedance.
[0046] 7. Construct a simplified model containing only one Class A and one Class B equivalent machine, and verify its accuracy by comparing it with the detailed model and the single-machine equivalent model. The power comparison results between the detailed model and the single-machine equivalent model are shown below. Figure 5 For detailed power comparison results between the model and the quadratic equivalent model, see [link to relevant documentation]. Figure 6 According to the "G-BT 29319-2024 Technical Regulations for Photovoltaic Power Generation Systems Access to Distribution Networks", the error is within the allowable range, verifying the accuracy of the equivalent results.
Claims
1. A method for secondary equivalent modeling of distributed photovoltaic inverters based on parameter identification, characterized in that, Includes the following steps: Step 1: Calculate the grid-connected equivalent impedance of each distributed photovoltaic inverter based on the grid topology; Step 2: According to the different control strategies of distributed photovoltaic inverters under low voltage, p+q Distributed photovoltaic inverters are divided into Class A distributed photovoltaic inverters with low-voltage ride-through capability and Class B distributed photovoltaic inverters with low-voltage blocking capability. Step 3: Using the grid-connected equivalent impedance of each distributed photovoltaic inverter as the clustering index, the Canopy-FCM clustering algorithm is used to cluster Class A distributed photovoltaic inverters with low-voltage ride-through capability and Class B distributed photovoltaic inverters with low-voltage blocking capability, respectively, to obtain... c A One Class A distributed photovoltaic cluster c B One Class B distributed photovoltaic cluster; Step 4: Use parameter identification methods to respectively... c A A Class A distributed photovoltaic cluster and c B Identify Class B distributed photovoltaic clusters and obtain the primary equivalent control parameters for each distributed photovoltaic cluster; Step 5: Calculate the equivalent line impedance of each distributed photovoltaic cluster, and construct a multi-machine equivalent model for each distributed photovoltaic cluster by combining the primary equivalent control parameters. Step 6: Apply system-level disturbances to the Type A multi-machine equivalent model and the Type B multi-machine equivalent model respectively. Record the electrical quantities at the output of the Type A multi-machine equivalent model and the output of the Type B multi-machine equivalent model respectively. Then, input them into a multi-convolutional layer hybrid convolutional neural network for processing to obtain the secondary equivalent control parameters of the Type A multi-machine equivalent model and the secondary equivalent control parameters of the Type B multi-machine equivalent model respectively. Step 7, containing c A The Class A multi-machine equivalent models of several Class A equivalent machines are aggregated into a single Class A equivalent machine, and c A The total capacity of distributed photovoltaic inverters in a Class A distributed photovoltaic cluster is taken as the equivalent capacity of a Class A equivalent single unit, and is obtained through equation (5). c A The equivalent line impedance of a Class A distributed photovoltaic cluster is used as the equivalent impedance of a Class A single unit. Will include c B The equivalent models of multiple equivalent B-type machines are aggregated into a single equivalent B-type machine, and... c B The total capacity of distributed photovoltaic inverters in a Class B distributed photovoltaic cluster is taken as the equivalent capacity of a Class B equivalent single unit, and is obtained through equation (5). c B The equivalent line impedance of a Class B distributed photovoltaic cluster is used as the equivalent impedance of a Class B equivalent single unit; thus, a secondary equivalent model containing one Class A equivalent single unit and one Class B equivalent single unit is constructed.
2. The method for secondary equivalent modeling of distributed photovoltaic inverters based on parameter identification according to claim 1, characterized in that, In step 1, the equation (1) is used. a The grid-connected equivalent impedance of a distributed photovoltaic inverter : (1) In formula (1): the 1st to the 2nd p The first distributed photovoltaic inverter is radially connected, the second... p +1~ p+q Each distributed photovoltaic inverter is connected via a trunk line. , These represent the total current and total impedance of the outer layer circuitry, respectively. , The first a Current and impedance of the lines connected to a distributed photovoltaic inverter; For the first j Line current of the lines connected to a distributed photovoltaic inverter; For the first i The impedance of the lines connected to each distributed photovoltaic inverter.
3. The method for secondary equivalent modeling of distributed photovoltaic inverters based on parameter identification according to claim 2, characterized in that, Step 3 includes the following steps: Step 3.1: Perform the first stage of Canopy coarse clustering on the Class A distributed photovoltaic inverters to obtain the initial cluster centers; Step 3.2, for c The initial cluster centers are then subjected to a second stage of fuzzy C-means fine clustering to obtain the final cluster centers. V A and membership matrix U A ; Step 3.3, according to V A and U A Step 3.1.1 D A Each data point is assigned to the cluster center with the highest membership degree, thus obtaining c A One Class A distributed photovoltaic cluster; Step 3.4: Following the process in Steps 3.1-3.3, perform the first stage of Canopy coarse clustering and the second stage of fuzzy C-means fine clustering on the Class B distributed photovoltaic inverters to obtain... c B A Class B distributed photovoltaic cluster.
4. The method for secondary equivalent modeling of distributed photovoltaic inverters based on parameter identification according to claim 3, characterized in that, Step 3.1 includes the following steps: Step 3.1.1: Take the grid-connected equivalent impedance of any distributed photovoltaic inverter in the Class A distributed photovoltaic inverter as a data point b, and take the real part resistance of data point b. and imaginary reactance As a two-dimensional data point Thus, a two-dimensional data point set for a type A distributed photovoltaic inverter is constructed. D A ; Select the first distance threshold T 1 and second distance threshold T 2, and T 1> T 2; Step 3.1.2, from D A Randomly select a data point P As the center point; and calculate D A All other data points in P Euclidean distance, for all distances less than T All data points of 1 were added to P In the set of center points, the center point P Add to the set of center points Mc In the middle; all distances less than T 2 data points from D A Remove from, thus updating D A The updated two-dimensional data point set is obtained. D’ A ; Step 3.1.3, will D’ A Assign to D A Then, return to step 3.1.2 until... D’ A Until the set of centers is empty, the final set of centers is obtained. Mc and will Mc of c Each center point is used as c The initial cluster centers.
5. The method for secondary equivalent modeling of distributed photovoltaic inverters based on parameter identification according to claim 4, characterized in that, Step 5 includes the following steps: Step 5.1: Calculate the first step using equation (2). c In a distributed photovoltaic cluster, the first f Voltage difference between a distributed photovoltaic inverter and the point of common coupling : (2) In formula (2): , The first c In a distributed photovoltaic cluster, the first f The impedance of the line connected to the first distributed photovoltaic inverter and the first f The active power output of a distributed photovoltaic inverter; The voltage at the point of common coupling. c 1=1,2,…, c A + c B ; Step 5.2: Calculate the equivalent first step using equations (3) and (4) respectively. c Weighted average voltage difference between distributed photovoltaic inverters and the point of common coupling in a distributed photovoltaic cluster And the voltage difference between the equivalent distributed photovoltaic inverter and the point of common coupling. : (3) (4) In equations (3) and (4): N 1 is the first c The number of distributed photovoltaic inverters in a distributed photovoltaic cluster; For the first c In a distributed photovoltaic cluster, the first f Equivalent line impedance of a distributed photovoltaic inverter; Step 5.3, let Using equation (5), we obtain the first... c Equivalent line impedance of a distributed photovoltaic cluster : (5) Step 5.4: Following the process of steps 5.1-5.3, we obtain... c A A Class A distributed photovoltaic cluster and c B The equivalent line impedance of a Class B distributed photovoltaic cluster is used to construct a system containing... c A A Class A equivalent machine, a Class A multi-machine equivalent model, and including c B A Class B multi-machine equivalent model of a Class B equivalent machine.
6. An electronic device, comprising a memory and a processor, characterized in that, The memory is used to store a program that supports a processor in executing the method of any one of claims 1-5, the processor being configured to execute the program stored in the memory.
7. A computer-readable storage medium storing a computer program thereon, characterized in that, The computer program is executed by the processor to perform the steps of the method according to any one of claims 1-5.