A distributed photovoltaic cluster modeling method and system based on improved FCM algorithm

Abstract

The present invention belongs to the technical field of power electronic systems and discloses a distributed photovoltaic cluster modeling method and system based on an improved FCM algorithm to measure the photovoltaic unit outlet voltage V t and PCC point voltage V PCC The calculation formula of the equivalent electrical distance from the outlet of the photovoltaic unit to the PCC point under the radial structure is derived, and the equivalent electrical distance under the chain structure and complex network topology is calculated, and then the line impedance equivalent coefficient is calculated; the active power P, reactive power Q and port voltage V output of the photovoltaic unit are measured. t , proposed an index value to characterize the photovoltaic voltage support function; comprehensively considered the equivalent electrical distance, photovoltaic capacity and voltage support function, established a comprehensive clustering index that accurately characterizes the characteristics of distributed photovoltaic power generation systems; used the optimal solution obtained by the particle swarm clustering algorithm as the initial clustering center, used the FCM algorithm to calculate the optimal clustering result, and used the parameter equivalence to calculate the equivalent model.

Description

Technical Field

[0001] The present invention belongs to the technical field of power electronic systems, and in particular relates to a distributed photovoltaic cluster modeling method and system based on an improved FCM algorithm. Background Art

[0002] Vigorously developing distributed photovoltaics is a necessary way to help my country achieve a green and low-carbon transformation of its energy structure and achieve carbon peak and carbon neutrality. However, the increase in photovoltaic installed capacity causes current backflow, which in turn causes problems such as overvoltage, which will greatly restrict the absorption of distributed photovoltaics. In order to deeply study the voltage operating characteristics of the new distribution network and comprehensively optimize the voltage regulation strategy of photovoltaic inverters, it is necessary to establish a simulation model for the distributed photovoltaic system. However, due to the presence of dozens or even hundreds of grid-connected inverters in large-scale distributed photovoltaic power stations, detailed modeling will have problems such as long simulation time and large simulation scale, and may even cause "dimensionality curse". Existing simulation modeling methods cannot effectively take into account both the accuracy and speed of distributed photovoltaic simulation modeling.

[0003] The cluster equivalent modeling method clusters photovoltaic power generation units with similar dynamic characteristics into the same cluster, equating several photovoltaic units within the cluster to a single photovoltaic unit, thereby simplifying the model. Because it balances accuracy and speed, it provides a practical reference for distributed photovoltaic system simulation modeling. During the cluster equivalent modeling process, the choice of clustering indicators and clustering algorithms significantly influences the accuracy and speed of the clustering results. However, for distributed photovoltaic power generation systems, due to their varying line impedances, mixed voltage support functions, and varying photovoltaic capacities, traditional clustering indicators cannot fully characterize the output dynamic characteristics of distributed photovoltaic units, resulting in inaccurate clustering results and large errors between the equivalent model and the detailed model. Furthermore, traditional clustering algorithms are highly random and prone to local optimality, making it difficult to stably obtain the optimal clustering solution. This can lead to large errors in the clustering results, further compromising the effectiveness of distributed photovoltaic coordinated control.

[0004] To address the issue of clustering metric selection, some literature uses inverter control parameters as clustering indicators. By establishing an offline parameter sensitivity database, they online calculate the characteristic distance between these parameters and the measured information to assess their matching degree, and then perform clustering. Because the filter inductance of photovoltaic inverters is highly sensitive to various disturbances, some literature further introduces the filter inductance of photovoltaic inverters as a clustering indicator based on the inverter control parameters, thereby more fully reflecting the dynamic response characteristics of the inverter. However, photovoltaic inverter parameters are generally difficult to obtain. Some literature uses the active power, reactive power, output voltage, and current of the photovoltaic array and inverter as clustering indicators. The resulting clustering results can simultaneously reflect the static and dynamic characteristics of photovoltaic modules in a two-stage photovoltaic power station. Regarding the selection of indicators for fault conditions, some literature extracts the peak characteristic points of the active and reactive power response curves of the photovoltaic system under a three-phase short-circuit fault as clustering indicators. However, these clustering indicators do not consider the network structure characteristics of distributed photovoltaic systems and the voltage support function of photovoltaic inverters. Therefore, clustering indicators that comprehensively characterize the emerging characteristics of distributed photovoltaic systems are needed to achieve accurate equivalent modeling of distributed photovoltaic systems.

[0005] To address the issue of clustering algorithm selection, some literature has used the K-means clustering algorithm to calculate the Euclidean distance between samples and then divide photovoltaic power plants into clusters based on their distance. This method is low-complexity and easy to implement. However, the K-means clustering algorithm uses hard clustering, meaning each data point belongs to only one cluster center, making it sensitive to noise and outliers. Therefore, some literature has proposed using the FCM algorithm to mitigate this noise sensitivity. Unlike the K-means clustering algorithm, the FCM algorithm uses soft clustering, meaning each data point is assigned to multiple clusters with varying membership degrees, with the final clustering result determined by the membership degree. This algorithm is suitable for processing data with uncertainty and ambiguity, and its clustering results are more consistent with objective reality. However, the FCM algorithm randomly selects the initial cluster centers, and different initial cluster centers lead to completely different clustering results, making it prone to falling into local optima. Another literature has proposed introducing the Canopy algorithm before the FCM algorithm to find the initial cluster centers. However, this method is significantly affected by the algorithm parameter settings and is prone to falling into local optima when the parameters are not set correctly, resulting in low accuracy. Therefore, an accurate and efficient distributed photovoltaic cluster modeling method is needed to achieve accurate equivalent modeling of distributed photovoltaics in different scenarios.

[0006] The existing technologies in the clustering and equivalent modeling of distributed photovoltaic power generation systems have brought about the following technical problems in industrial applications:

[0007] 1. Insufficient selection of clustering indicators:

[0008] In existing technologies, clustering metrics primarily focus on PV inverter control parameters, filter inductance, and system characteristics such as active power, reactive power, output voltage, and current. While these metrics can reflect the static and dynamic characteristics of PV power plants, they fail to fully consider the network structure characteristics of distributed PV systems and the voltage support function of PV inverters. These new power system characteristics have a significant impact on the control, stability analysis, and fault handling of large-scale distributed PV systems. Clustering models that fail to incorporate these key characteristics can easily lead to inadequate reflection of the system's dynamic behavior during the equivalent modeling process, thereby impacting overall system operational efficiency and safety.

[0009] 2. Inverter parameters are difficult to obtain:

[0010] PV inverter parameters (such as filter inductance) are highly sensitive to system dynamics, but these parameters are difficult to obtain directly in real industrial applications. Without these parameters, relying on existing equivalent modeling techniques can lead to reduced accuracy, impacting the accuracy of PV power plant cluster modeling. This issue is particularly prominent in scenarios requiring precise control of inverter response, such as voltage support and dynamic response optimization.

[0011] 3. Limitations of existing clustering algorithms:

[0012] The K-means clustering algorithm, commonly used in existing technologies, has low computational complexity and is easy to implement. However, its hard clustering properties make it sensitive to noise and outliers, making it difficult to process complex data with uncertainty and ambiguity in real-world scenarios. Furthermore, while the FCM algorithm uses a soft clustering approach, which effectively handles ambiguity in data, its random selection of initial cluster centers makes it prone to falling into local optimal solutions, resulting in unstable clustering results. These limitations of clustering algorithms can easily lead to large errors in the equivalent modeling process. These errors can be amplified, particularly in large-scale distributed photovoltaic systems, impacting the reliability and stability of system control.

[0013] 4. Insufficient scene adaptability:

[0014] Existing clustering algorithms lack sufficient adaptability to diverse operating scenarios and fault conditions. For example, in special scenarios like three-phase short-circuit faults, existing indicator selection and clustering methods cannot fully reflect the system's characteristics. For large-scale distributed photovoltaic power generation systems, when the system encounters grid disturbances or faults, existing equivalent models struggle to accurately reflect the dynamic behavior of the entire system. This makes it difficult to ensure the dynamic stability and safety of the system as the scale of photovoltaic power generation systems continues to expand.

[0015] 5. Impact of local optimality problem:

[0016] Although some literature has attempted to optimize the initial cluster centers of FCM by introducing the Canopy algorithm, this method relies heavily on the algorithm parameter settings. If the parameters are not set properly, it will still fall into local optimality, affecting the accuracy of the clustering results and the reliability of the model. In scenarios with complex dynamic characteristics such as distributed photovoltaic power generation systems, this local optimality problem leads to deviations between system modeling and actual operation, ultimately affecting the system's scheduling, control, and operational efficiency.

[0017] In summary, the technical problem with existing technologies for equivalent modeling of distributed photovoltaic power generation systems lies in their failure to comprehensively consider the key characteristics of the system. Clustering algorithms are prone to local optima and large errors when processing complex data, which affects the modeling accuracy, stability, and adaptability of the entire system. These issues limit the potential application of distributed photovoltaic power generation systems in intelligent control, optimized operation, and fault handling. Summary of the Invention

[0018] In response to the problems existing in the existing technology, the present invention provides a distributed photovoltaic cluster modeling method based on an improved FCM algorithm, aiming to achieve efficient and accurate distributed photovoltaic cluster modeling with different electrical distances, mixed voltage support functions, and different photovoltaic capacities.

[0019] The present invention is implemented as follows: a distributed photovoltaic cluster modeling method based on an improved FCM algorithm, comprising:

[0020] 1) Measure the output voltage V of the photovoltaic unit t and PCC point voltage V PCC and the phase angle difference θ between the two, derive the calculation formula for the equivalent electrical distance from the photovoltaic unit outlet to the PCC point in the radial structure, and then calculate the equivalent electrical distance in the chain structure and complex network topology, and then calculate the line impedance equivalent coefficient;

[0021] 2) Measure the active power P, reactive power Q and port voltage V of the photovoltaic unit output t , an index value is proposed to characterize the photovoltaic voltage support function;

[0022] 3) Comprehensively consider equivalent electrical distance, photovoltaic capacity and voltage support function to establish a comprehensive clustering index that accurately characterizes the characteristics of distributed photovoltaic power generation systems;

[0023] 4) The optimal solution obtained by the particle swarm clustering algorithm is used as the initial cluster center, the FCM algorithm is used to calculate the optimal clustering result, and the parameter equivalence is used to calculate the equivalence model.

[0024] Furthermore, the calculation formula for the equivalent electrical distance from the distributed photovoltaic outlet to the PCC point under the radial structure is:

[0025]

[0026] Among them, A eq represents the equivalent line impedance from the PV unit outlet to the grid, B represents the line impedance from the PCC point to the grid, A represents the line impedance from the PV outlet to the PCC point, I d-eq , I q-eq The current from the PCC point to the grid, the subscripts "d" and "q" represent the d and q axis components respectively, I d , I q represents the current from the PV unit outlet to the PCC point, and T represents the transformation matrix of the PV units from their respective coordinate systems to the common coordinate system.

[0027] Furthermore, the calculation formula of the line impedance equivalent coefficient is:

[0028]

[0029] Among them LEC i A represents the line impedance equivalent coefficient of the i-th node, ieq A is the equivalent line impedance from the photovoltaic unit connection point of the i-th node to the grid. i represents the line impedance parameter from the ith PV unit to the PCC, and B represents the line impedance from the PCC to the grid.

[0030] Furthermore, the index value that characterizes the voltage support function of the photovoltaic unit inverter is:

[0031]

[0032] Among them, V represents the output voltage of the photovoltaic unit, P represents the output active power of the photovoltaic unit, Q represents the output reactive power of the photovoltaic unit, and C VP Indicates the correlation between the PV unit outlet voltage and active power, C VQ Indicates the correlation between the PV unit outlet voltage and reactive power;

[0033] Furthermore, the distributed photovoltaic clustering index that comprehensively considers the equivalent electrical distance, photovoltaic capacity and voltage support function is:

[0034] CCI=a×C VP +a×C VQ +(1-2a)×A eq

[0035] Among them, CCI represents the distributed photovoltaic comprehensive clustering index, a represents the normalized photovoltaic capacity, C VP and C VQ They represent the correlation between the PV unit outlet voltage and active power and reactive power, respectively. eq Indicates the equivalent electrical distance;

[0036] Further improvements to the FCM algorithm include:

[0037] The particle swarm algorithm searches for the optimal solution. Particles determine their flight direction based on their own optimal position and the global optimal position. The global optimal solution is obtained through convergence iteration and serves as the initial clustering center of the FCM algorithm.

[0038] The FCM algorithm updates the membership matrix and selects cluster centers based on the initial cluster centers obtained by the particle swarm algorithm, and it iterates to obtain the optimal clustering result.

[0039] Another object of the present invention is to provide a distributed photovoltaic cluster simplified model system based on the improved FCM algorithm, comprising:

[0040] Voltage measurement module, used to measure the photovoltaic unit outlet voltage V t and PCC point voltage V PCC ;

[0041] Phase angle measurement module, used to measure the phase angle difference θ between the photovoltaic unit outlet voltage and the PCC point voltage;

[0042] Power measurement module, used to measure the active power P and reactive power Q output by the photovoltaic unit;

[0043] Line impedance measurement module, used to measure line impedance Z s =R s +jX s ;

[0044] The module for deriving the equivalent electrical distance relationship is used to derive the expression of the equivalent electrical distance from the photovoltaic outlet to the PCC point under radial structure, chain structure and complex topology;

[0045] Line impedance equivalent coefficient calculation module, used to calculate the variation coefficient of equivalent line impedance;

[0046] Voltage support function index value calculation module, used to calculate the active power P, reactive power Q and port voltage V output by the photovoltaic unit t , calculate the index value that characterizes the photovoltaic voltage support function;

[0047] Comprehensive clustering index calculation module, used to calculate clustering index for clustering based on equivalent electrical distance, voltage support function and photovoltaic capacity;

[0048] Particle swarm optimization module, which is used to search for the global optimal solution and calculate the optimal cluster center using comprehensive clustering indicators and particle swarm optimization;

[0049] The fuzzy C-means clustering algorithm module is used to cluster several photovoltaic units using the optimal clustering center obtained by the particle swarm algorithm to obtain the optimal clustering result.

[0050] Another object of the present invention is to provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the distributed photovoltaic cluster modeling method based on the improved FCM algorithm.

[0051] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to execute the steps of the distributed photovoltaic cluster modeling method based on the improved FCM algorithm.

[0052] Another object of the present invention is to provide an information data processing terminal, which is used to implement the distributed photovoltaic equivalent model system based on the improved FCM algorithm.

[0053] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0054] First, we will address the technical problems existing in the above-mentioned prior art and provide some creative technical effects after solving the problems. The details are as follows:

[0055] (1) The equivalent electrical distance calculation method proposed in the present invention derives the equivalent line impedance from the outlet of the photovoltaic unit to the power grid under the radial structure, and then calculates the equivalent line impedance under the chain structure and complex network topology. It takes into account the equivalent electrical distance calculation of various network topologies rather than simply superimposing the line impedance.

[0056] (2) The line impedance equivalent coefficient calculation module proposed in the present invention quantifies the evolution characteristics of the equivalent line impedance by calculating the variation coefficient of the equivalent line impedance;

[0057] (3) The comprehensive clustering index proposed in this invention comprehensively considers the network topology, voltage support function and photovoltaic capacity of the distributed photovoltaic system, and further accurately reflects the dynamic response characteristics of the photovoltaic unit.

[0058] (4) The cluster modeling method proposed in the present invention can globally search for the optimal solution, avoiding the disadvantage of traditional algorithms that randomly select initial cluster centers and thus fall into local optimality, and can stably achieve accurate clustering equivalence.

[0059] (5) The cluster modeling method proposed in the present invention has strong applicability and can achieve accurate and rapid clustering for equivalent modeling of distributed photovoltaic systems under different complex network topologies and mixed voltage support functions.

[0060] Second, the present invention provides a distributed photovoltaic cluster modeling method and system based on an improved FCM algorithm. In view of the characteristics of distributed photovoltaic systems with different line impedances, mixed voltage support functions, and different photovoltaic capacities, the present invention measures the voltage at the photovoltaic unit outlet, PCC point, and grid, as well as the line impedance on each line, and derives and calculates the equivalent electrical distance under radial structure, chain structure, and complex network topology, and then calculates the line impedance equivalent coefficient. Then, the active power, reactive power, and voltage output of the photovoltaic unit are measured, and the index value representing the photovoltaic voltage support function is calculated. Then, the photovoltaic capacity is normalized and used as the weight of the equivalent electrical distance and voltage support function to establish a comprehensive clustering index that accurately represents the characteristics of the distributed photovoltaic system. Considering that traditional clustering algorithms randomly select initial cluster centers and are prone to falling into local optimality, the present invention introduces a particle swarm algorithm to perform a global search on the data samples, and uses the calculated optimal solution as the initial cluster center. The optimal clustering result is calculated using the FCM algorithm, and the equivalent model is calculated using parameter equivalence.

[0061] Third, as auxiliary evidence for the inventiveness of the claims of the present invention, it is also reflected in the following important aspects:

[0062] (1) The expected benefits and commercial value of the technical solution of the present invention after transformation are:

[0063] The cluster modeling method disclosed in the present invention can realize cluster modeling of distributed photovoltaic systems with mixed voltage support functions: measuring the photovoltaic units, PCC points and grid voltages and the impedance of each line, deriving and calculating the equivalent electrical distance under the radial structure, chain structure and complex network topology, and then calculating the line impedance equivalent coefficient, measuring the active power, reactive power and voltage output by the photovoltaic units, calculating the index value that characterizes the photovoltaic voltage support function, and then establishing a comprehensive clustering index that accurately characterizes the characteristics of the distributed photovoltaic system. The optimal solution obtained by the particle swarm clustering algorithm is used as the initial cluster center, the optimal clustering result is calculated using the FCM algorithm, and the equivalent model is calculated using parameter equivalence. This cluster modeling method can be widely used in the cluster modeling equivalent circuits of distributed photovoltaic systems with complex network topologies and mixed voltage support functions. Due to the comprehensive consideration of the equivalent electrical distance and photovoltaic voltage support characteristics, it can accurately reflect the dynamic characteristics of the photovoltaic units. At the same time, this cluster modeling method introduces the particle swarm algorithm to improve and overcome the shortcomings of the traditional FCM algorithm, greatly improving the accuracy of the clustering results.

[0064] (2) The technical solution of the present invention fills the technical gap in the industry at home and abroad:

[0065] At present, the cluster modeling methods at home and abroad use the output characteristics of photovoltaic units as clustering indicators, but this indicator is only applicable to centralized photovoltaic cluster modeling, and does not take into account the characteristics of different line impedances, mixed voltage support functions, and different photovoltaic capacities in distributed photovoltaic systems. When there are complex network topologies and different control modes such as Volt-Var, constant power factor, and constant reactive power in photovoltaic systems, existing studies only collect output characteristics such as voltage, active power, and reactive power of photovoltaic units as clustering indicators. These indicators cannot fully characterize the electrical characteristics of distributed photovoltaic systems, which affects the accuracy of clustering results. In addition, the existing clustering algorithm uses a random selection of initial cluster centers for subsequent clustering calculations. When the initial cluster centers are not selected properly, it is easy to fall into local optimality, which not only affects the rapidity of the clustering process, but also leads to inaccurate distributed photovoltaic clustering results. There is a large error between the equivalent model and the detailed model, which greatly limits its application in engineering. The cluster modeling method provided by the present invention fills this gap and provides a model basis for the clustering equivalence problem of distributed photovoltaic power generation systems.

[0066] Fourth, the distributed photovoltaic cluster modeling method based on the improved FCM algorithm of this invention solves several key technical problems in the existing technology in industrial applications and achieves significant technological progress. The following is a detailed analysis of the problems solved by this technical solution in industrial applications and its technological progress:

[0067] 1. Existing technical problems: low modeling accuracy and difficulty in handling complex photovoltaic cluster network structures

[0068] Existing photovoltaic cluster modeling methods often fail to accurately handle complex network topologies. This is particularly true in large-scale distributed photovoltaic systems, where the calculation of equivalent electrical distances is often inaccurate, leading to significant deviations in modeling results. Furthermore, existing technologies fail to fully consider key parameters such as the voltage support function of photovoltaic units and photovoltaic capacity, resulting in errors in the models' reflection of actual operating conditions.

[0069] This method significantly improves modeling accuracy by measuring the outlet voltage of the photovoltaic unit, the PCC point voltage, and the phase angle difference θ. This method combines the equivalent electrical distance calculation formulas for radial, chain, and complex network topologies to calculate the line impedance equivalent coefficient. Furthermore, a comprehensive clustering index is established by comprehensively considering three core parameters: equivalent electrical distance, photovoltaic capacity, and voltage support function. This further enhances modeling accuracy and accurately reflects the actual operating characteristics of photovoltaic clusters.

[0070] 2. Problems with existing technologies: Clustering algorithms are sensitive to initial conditions and have slow convergence speed

[0071] When dealing with distributed photovoltaic clusters, the traditional FCM (Fuzzy C-means) clustering algorithm is often limited by the choice of initial cluster centers, easily falling into local optima, and has a slow convergence rate. This results in low computational efficiency in practical applications, especially when dealing with large-scale photovoltaic clusters, making it difficult to meet the requirements of real-time modeling and rapid response.

[0072] This paper introduces a particle swarm optimization algorithm, using it to obtain the global optimal solution, which serves as the initial cluster centers for the FCM algorithm. This not only accelerates the convergence of the FCM algorithm but also avoids the risk of falling into local optima, thereby enabling rapid calculation of the optimal clustering results. This improvement significantly improves the efficiency of distributed photovoltaic cluster modeling, especially in large-scale application scenarios.

[0073] 3. Existing technical issues: Insufficient consideration of the voltage support capacity of photovoltaic units, affecting grid stability

[0074] Existing technologies often overlook the voltage support provided by photovoltaic cells in the power grid, resulting in a failure to fully consider their impact on grid voltage stability during modeling. This deficiency prevents distributed photovoltaic systems from fully utilizing their potential to support voltage during voltage regulation, leading to risks such as voltage fluctuations and grid instability.

[0075] This paper measures the active power, reactive power, and port voltage of PV cells to develop an indicator value that characterizes their voltage support capability. This innovative design accurately assesses the PV cell's ability to support grid voltage, allowing the actual impact of each PV cell to be fully considered during modeling. This optimizes voltage stability during grid operation and improves overall system reliability and safety.

[0076] 4. Existing technical issues: Failure to effectively consider the complexity of large-scale distributed photovoltaic systems

[0077] As the scale of distributed photovoltaic systems continues to expand, existing modeling technologies find it difficult to effectively handle the complex electrical relationships between large-scale photovoltaic unit clusters. In particular, when multiple photovoltaic units are connected in different topologies, existing simplified models often cannot accurately express their electrical characteristics.

[0078] This paper establishes a comprehensive set of clustering metrics by comprehensively considering the equivalent electrical distance, capacity differences, and voltage support capabilities between photovoltaic units. This approach effectively addresses complex photovoltaic system topologies. Combined with an improved FCM algorithm, this approach not only addresses the complexity of large-scale photovoltaic clusters but also generates accurate equivalent models, significantly improving the model's applicability and scalability. It is particularly well-suited for modeling and optimizing large-scale photovoltaic power generation systems in industry.

[0079] 5. Significant economic benefits and application value

[0080] The technical solution of this invention significantly improves the accuracy and efficiency of photovoltaic cluster modeling, thereby reducing the operating and maintenance costs of photovoltaic systems. Furthermore, the accuracy of the model enables more rational and precise grid scheduling, reduces the risks associated with voltage fluctuations, and improves the reliability of power supply. In large-scale applications, this improvement can bring significant economic and social benefits to grid operators and photovoltaic power generation companies, contributing to the promotion and popularization of distributed photovoltaic power generation.

[0081] This invention demonstrates significant technological advancement in addressing existing issues such as modeling accuracy, computational efficiency, voltage support capability assessment, and complex system processing. In industrial applications, it not only improves the management and scheduling efficiency of distributed photovoltaic clusters, but also enhances the safety and stability of power grid operations, providing strong support for the widespread adoption of photovoltaic power generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0083] Figure 1 This is a flow chart of a distributed photovoltaic cluster modeling method based on an improved FCM algorithm provided by an embodiment of the present invention;

[0084] Figure 2 Schematic diagram of the equivalent line impedance of distributed photovoltaics under a radial structure provided by an embodiment of the present invention;

[0085] Figure 3 is a phasor diagram of photovoltaic units converted from their respective coordinate systems to a common coordinate system according to an embodiment of the present invention;

[0086] Figure 4 Schematic diagram of the equivalent line impedance of distributed photovoltaics under a chain structure provided by an embodiment of the present invention;

[0087] Figure 5 Schematic diagram of the equivalent line impedance of distributed photovoltaics under the hybrid structure provided by an embodiment of the present invention;

[0088] Figure 6 This is an overall framework diagram of the improved FCM algorithm provided by an embodiment of the present invention;

[0089] Figure 7 This is a structural diagram of a distributed photovoltaic cluster modeling method based on an improved FCM algorithm provided by an embodiment of the present invention;

[0090] FIG8(a) shows the active power and reactive power curves at the PCC point of the model before decoupling and the equivalent model after decoupling under a radial structure according to an embodiment of the present invention; FIG8(b) shows the active power and reactive power curves at the PCC point of the model before decoupling and the equivalent model after decoupling under a chain structure according to an embodiment of the present invention; and FIG8(c) shows the active power and reactive power curves at the PCC point of the model before decoupling and the equivalent model after decoupling under a hybrid structure according to an embodiment of the present invention.

[0091] Figure 9 This is a decoupled radial power distribution network diagram of a distributed photovoltaic system containing 36 photovoltaic units provided by an embodiment of the present invention;

[0092] Figure 10 The reactive power curves at the PCC point of the equivalent model and the detailed model obtained by using the FCM algorithm and the improved FCM algorithm disclosed in the present invention provided in the embodiments of the present invention;

[0093] Figure 11 It is an active power curve at the PCC point of an equivalent model obtained by using the FCM algorithm and the improved FCM algorithm disclosed by the present invention, as well as a detailed model provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0094] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0095] The following are two industrial application examples of the distributed photovoltaic cluster modeling method based on the improved FCM algorithm:

[0096] Example 1: Distributed Photovoltaic Power Generation Management System in Smart Grid

[0097] In the operation of a large-scale smart grid, the grid includes multiple distributed photovoltaic power stations located in various urban and rural areas. Due to their varying locations and access points, these PV stations exhibit significant differences in power supply stability and grid voltage support capabilities. By introducing a photovoltaic cluster modeling method based on an improved FCM algorithm, power operators can accurately analyze the equivalent electrical distance between each photovoltaic unit and the PCC point, as well as the active and reactive power of each unit, and further measure its voltage support capability.

[0098] In practical application, the system first measures the voltage and power output of PV cells. It then uses a particle swarm optimization algorithm to optimize the initial cluster centers and employs an FCM algorithm to perform optimized cluster analysis on these cells. This process enables operators to classify PV cells in the grid into different categories, forming an equivalent model for real-time monitoring and regulation. The implementation of this system not only improves the grid's power dispatch efficiency but also enhances the stability of distributed PV generation and reduces the risk of failures caused by voltage fluctuations.

[0099] Example 2: Optimal Control of Distributed Photovoltaic Clusters in Large-Scale Industrial Parks

[0100] A large industrial park installed multiple photovoltaic power plants for independent power supply and connected them to the grid. Due to the wide distribution and uneven capacity of the photovoltaic units, voltage fluctuations within the park were frequent, impacting overall power quality. To optimize the park's photovoltaic power generation system, the park management employed a cluster modeling approach based on an improved FCM algorithm.

[0101] By measuring the voltage, output power, and electrical distance of each photovoltaic unit from the PCC point, the management system constructs comprehensive clustering indicators for the photovoltaic cluster within the park. Using a dual optimization approach using a particle swarm optimization algorithm and a FCM algorithm, the park's photovoltaic power plants are effectively grouped, and equivalent models are established for each group. Management utilizes these models to adjust and control voltage fluctuations within the park in real time, ensuring the stability and efficiency of photovoltaic power generation. Furthermore, optimized control reduces downtime and losses of industrial equipment caused by voltage instability, improving the continuity and economic benefits of industrial production.

[0102] In view of the problems existing in the prior art, the present invention provides a distributed photovoltaic cluster modeling method and an equivalent model system based on an improved FCM algorithm. The present invention is described in detail below with reference to the accompanying drawings.

[0103] like Figure 1 As shown, a distributed photovoltaic cluster modeling method based on the improved FCM algorithm includes the following steps:

[0104] 1) Measure the output voltage V of the photovoltaic unit t and PCC point voltage V PCC and the phase angle difference θ between the two, derive the calculation formula for the equivalent electrical distance from the photovoltaic unit outlet to the PCC point in the radial structure, and then calculate the equivalent electrical distance in the chain structure and complex network topology, and then calculate the line impedance equivalent coefficient;

[0105] 2) Measure the active power P, reactive power Q and port voltage V of the photovoltaic unit output t , an index value is proposed to characterize the photovoltaic voltage support function;

[0106] 3) Comprehensively consider equivalent electrical distance, photovoltaic capacity and voltage support function to establish a comprehensive clustering index that accurately characterizes the characteristics of distributed photovoltaic power generation systems;

[0107] 4) The optimal solution obtained by the particle swarm clustering algorithm is used as the initial cluster center, the FCM algorithm is used to calculate the optimal clustering result, and the parameter equivalence is used to calculate the equivalence model.

[0108] This distributed photovoltaic cluster modeling method, based on an improved FCM algorithm, achieves accurate photovoltaic power generation system modeling and analysis through multiple steps. First, in step 1, the equivalent electrical distance from the photovoltaic unit to the PCC (point of common coupling) is derived by measuring the voltage at the photovoltaic unit outlet and the voltage at the PCC (point of common coupling), as well as the phase angle difference θ between the two. This calculation formula was initially derived for simple radial structures and has been further extended to more complex chain structures and complex network topologies. The purpose of this step is to accurately calculate the electrical connection distance between photovoltaic units, laying the foundation for subsequent modeling.

[0109] In step 2, the system further measures the PV unit's output active power P, reactive power Q, and port voltage. Using these parameters, an index value is developed to characterize the PV voltage support function. This index value is introduced to measure the differences in voltage support capabilities among PV units, ensuring that the impact and support capabilities of each PV unit on the grid voltage are fully considered during the modeling process. Accurate characterization of the voltage support function is crucial for modeling PV clusters, as it reflects the voltage stability and overall performance of the cluster.

[0110] Next, step 3 comprehensively considers equivalent electrical distance, PV capacity, and voltage support capability to establish a comprehensive clustering index. This index more comprehensively characterizes the characteristics of distributed PV power generation systems. Especially in large-scale PV clusters, this comprehensive index effectively describes the characteristics of each PV unit, providing a more scientific and reasonable basis for subsequent cluster analysis. Using this clustering index, the system can group PV units, simplifying the modeling process for the entire PV cluster.

[0111] Finally, in step 4, the system first uses the particle swarm optimization (PSO) algorithm to obtain the optimal solution, which is used as the initial cluster center for the FCM (Fuzzy C-means) algorithm. Based on this, the FCM algorithm is used to optimally cluster the photovoltaic cells, resulting in the optimal clustering results. Based on these clustering results, parameter equivalence is performed and an equivalent model is generated. This process significantly improves the accuracy and efficiency of photovoltaic cluster modeling, ensuring that accurate equivalent models can be obtained in complex systems for subsequent analysis and control.

[0112] according to Figure 2The equivalent schematic diagram of the radiant structure photovoltaic system is shown in the figure. The grid-connected line matrix equation of the grid-side converter of the equivalent front circuit can be obtained as follows:

[0113]

[0114] in Represents the line impedance parameter matrix from the PV unit outlet to the PCC point, U td ,U tq They represent the d-axis and q-axis voltage components of the photovoltaic unit in its own coordinate system, U pccd ,U pccq Represent the voltage components of the PCC point on the d and q axes, I d ,I q They represent the d-axis and q-axis current components of the photovoltaic unit in its own coordinate system. Rearranging this formula yields:

[0115]

[0116] The subscripts 1 and 2 represent the electrical quantities of the first and second photovoltaic units, respectively. The grid-connected converters of each photovoltaic unit are vector-controlled in their own phase-locked coordinate system, and the grid-connected converters are aggregated through the collector network to form a new energy station. The voltage vector at the common connection point of each photovoltaic unit after being aggregated through the line impedance is used as the reference direction to establish a common coordinate system. The phasor diagram is shown in the figure below. Figure 3 As shown, the conversion matrix that converts the grid-side converter variables from their respective coordinate systems to the common coordinate system can be obtained:

[0117]

[0118] where x d-1 ,x q-1 ,x d-2 ,x q-2 Respectively represent the d-axis and q-axis components of the first photovoltaic unit and the second photovoltaic unit under the common coordinate axis, x 1d ,x 1q ,x 2d ,x 2q They represent the d-axis and q-axis components of the first photovoltaic unit and the second photovoltaic unit in their own coordinate systems, respectively. T1 and T2 represent the conversion matrices of the first photovoltaic unit and the second photovoltaic unit, respectively.

[0119] Arrange the above formulas and equate the d-axis and q-axis current components in their respective coordinate systems to the common coordinate system, and the current expression in the common coordinate system is obtained as follows:

[0120]

[0121] Among them I d-1 ,Id-1 ,I d-2 ,I q-2 They represent the d-axis and q-axis current components of the first and second photovoltaic units respectively in the common coordinate system. The current at the common connection point is equal to the sum of the output currents of each grid-side converter in the common coordinate system, and its expression is:

[0122]

[0123] Keeping the output voltage of the photovoltaic unit unchanged, that is, the difference between the output voltage and the grid voltage remains unchanged before and after the equivalence, the line matrix equation from the photovoltaic outlet to the PCC point is obtained as follows:

[0124]

[0125] Where B represents the line impedance parameter matrix from the PCC point to the power grid.

[0126] Keeping the consistency of the models before and after equivalence, the grid-connected line matrix equation of the grid-side converter of the equivalent circuit is obtained as follows:

[0127]

[0128] Arranging the above formulas yields:

[0129]

[0130] Among them A 1eq ,A 2eq Represent the equivalent electrical distances from the first PV unit and the second PV unit to the grid respectively.

[0131] For the chain structure, in order to ensure the consistency of the calculation method, the radial structure is regarded as a special form of the chain structure. The equivalent line impedance parameters of the grid-connected distributed photovoltaic under the chain structure can be calculated by multiple equivalences based on the radial structure. Figure 4 The equivalent schematic diagram of the chain-type photovoltaic system shown in the figure can be used to calculate the equivalent electrical distance. It is worth mentioning that for complex topological structures, they can be regarded as a topological reorganization of the radial structure and the chain structure, and the equivalent electrical distance can still be calculated using a similar method. Figure 5 The equivalent schematic diagram of the hybrid photovoltaic system shown can be used to calculate the equivalent electrical distance.

[0132] Furthermore, the calculation formula of the line impedance equivalent coefficient is:

[0133]

[0134] Among them LEC i A represents the line impedance equivalent coefficient of the i-th node, ieqA is the equivalent line impedance from the photovoltaic unit connection point of the i-th node to the grid. i represents the line impedance parameter from the ith PV unit to the PCC, and B represents the line impedance from the PCC to the grid.

[0135] Currently, the common voltage support functions are Volt-Var control, constant power factor control, and constant reactive power control. However, for distributed photovoltaic power generation systems, it is difficult to accurately and quickly know the voltage support function used by each photovoltaic inverter. Therefore, it is necessary to find suitable indicators to characterize the voltage support function of the photovoltaic inverter. The three control methods are all closely related to the active power and voltage or reactive power and voltage at the photovoltaic outlet. The three control methods can be completely distinguished based on the above correlation. Furthermore, it is proposed to use the correlation between voltage and active power and the correlation between voltage and reactive power as clustering indicators, and to calculate the correlation coefficient between samples as an estimate of the overall correlation coefficient to characterize the voltage support function of photovoltaics. The calculation formula of the correlation coefficient is as follows:

[0136]

[0137] where c ij For sample F i With sample F j The correlation coefficient between i ,F j ) is the sample F i With sample F j Covariance between Var(F i ) and Var(F j ) are sample F i and sample F j The variance of .

[0138] Furthermore, differences in PV capacity also have a certain impact on the clustering results. For PV units with larger capacity, their voltage support function has a stronger ability to regulate the distribution network voltage. In this case, the voltage support function is used as the more important clustering indicator, that is, the voltage support function has a greater weight, which is conducive to precise voltage regulation. For PV units with smaller capacity, in order to facilitate the distribution of centralized voltage regulation commands, PV units with closer equivalent electrical distances should be clustered together, and the equivalent electrical distance should be given a greater weight. In summary, using the normalized capacity value as the weight, the comprehensive clustering index (CCI) is proposed as follows:

[0139] CCI=a×C VP +a×C VQ +(1-2a)×A eq

[0140] Where a is the normalized capacity weight, C vp is the correlation coefficient between voltage and active power, C vq is the correlation coefficient between voltage and reactive power, A eq is the equivalent electrical distance.

[0141] Furthermore, the overall framework of the improved fuzzy C-means clustering algorithm is shown in the figure below: Figure 6 As shown. The objective function and constraints of the fuzzy C-means clustering algorithm are as follows:

[0142]

[0143] Among them, u ij is the sample point x i Relative to the cluster center v j The degree of membership of the cluster, m is the fuzzy index (m>1), d ij is the sample point x i With cluster center v j The Euclidean distance is obtained by iterative optimization of the objective function. In order to minimize the objective function J, the Lagrange multiplier method is used on the objective function under the condition of satisfying the constraints to obtain the membership matrix U and cluster center v. j .

[0144]

[0145] The specific description of the algorithm is as follows:

[0146]

[0147]

[0148] Because fuzzy C-clustering uses a random method to select initial cluster centers, the clustering results are highly dependent on the selection of the initial centers. If the initial centers are incorrectly selected, the subsequent clustering process will be significantly affected, and the optimal clustering results may be less than ideal. The number of clustering iterations will also increase. Randomly selected initial centers have a high degree of uncertainty, which directly affects the clustering effect. Therefore, we propose to first use a particle swarm optimization algorithm to globally optimize and select the optimal cluster centers, and then use the fuzzy C-means clustering algorithm to obtain the optimal clustering results.

[0149] The position of particle i in N-dimensional space is represented by vector X i =(x1,x2,…,x N ), the flight speed is represented by vector V i =(v1,v2,…,v N ). The particle updates its velocity and position using the following formula:

[0150] vi =w×v i +c1×rand()×(pbest i -x i )+c2×rand()×(gbest i -x i )

[0151] x i =x i +v i

[0152] Where w is the inertia factor, representing the trade-off between global optimization ability and local optimization ability, v i Refers to the speed of the particle, rand() is a random number between (0,1), x i is the current position of the particle, c1 and c2 are learning factors. The pseudo code of the particle swarm algorithm is as follows:

[0153]

[0154]

[0155] like Figure 7 As shown, the distributed photovoltaic cluster modeling method and equivalent model system based on the improved FCM algorithm provided by the embodiment of the present invention include:

[0156] Voltage measurement module, used to measure the photovoltaic unit outlet voltage V t and PCC point voltage V PCC ;

[0157] Phase angle measurement module, used to measure the phase angle difference θ between the photovoltaic unit outlet voltage and the PCC point voltage;

[0158] Power measurement module, used to measure the active power P and reactive power Q output by the photovoltaic unit;

[0159] Line impedance measurement module, used to measure line impedance Z s =R s +jX s ;

[0160] The module for deriving the equivalent electrical distance relationship is used to derive the expression of the equivalent electrical distance from the photovoltaic outlet to the PCC point under radial structure, chain structure and complex topology;

[0161] Line impedance equivalent coefficient calculation module, used to calculate the variation coefficient of equivalent line impedance;

[0162] Voltage support function index value calculation module, used to calculate the active power P, reactive power Q and port voltage V output by the photovoltaic unit t, calculate the index value that characterizes the photovoltaic voltage support function;

[0163] Comprehensive clustering index calculation module, used to calculate clustering index for clustering based on equivalent electrical distance, voltage support function and photovoltaic capacity;

[0164] Particle swarm optimization module, which is used to search for the global optimal solution and calculate the optimal cluster center using comprehensive clustering indicators and particle swarm optimization;

[0165] The fuzzy C-means clustering algorithm module is used to cluster several photovoltaic units using the optimal clustering center obtained by the particle swarm algorithm to obtain the optimal clustering result.

[0166] An application embodiment of the present invention provides a computer device, which includes a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the processor executes the steps of a distributed photovoltaic cluster modeling method based on an improved FCM algorithm.

[0167] An application embodiment of the present invention provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the steps of a distributed photovoltaic cluster modeling method based on an improved FCM algorithm.

[0168] An application embodiment of the present invention provides an information data processing terminal, which is used to implement a distributed photovoltaic cluster modeling method and an equivalent model system based on an improved FCM algorithm.

[0169] The present invention provides two embodiments of the invention: first, the equivalent electrical distance calculation method disclosed in the present invention is used to simplify the network structure into the form of "photovoltaic unit-line impedance-grid" under radial structure, chain structure and hybrid structure, and the simulation results are compared with those before simplification to verify its accuracy; second, in a large-scale distributed photovoltaic system, the clustering results of the improved FCM algorithm disclosed in the present invention are compared with those of the existing FCM algorithm to verify the effectiveness and advantages of the proposed clustering algorithm.

[0170] In a first embodiment, Figure 2 The radial structure shown, Figure 3 The chain structure shown and Figure 4 The line impedance parameters and line impedance equivalent coefficients before and after decoupling under the hybrid structure are shown in Tables 1, 2, and 3, respectively. The model before decoupling and the equivalent model after decoupling are simulated, and the simulation results are shown in Figure 8. The active power and reactive power curves of the PCC before and after decoupling are basically consistent, further proving the effectiveness of the equivalent electrical distance solution method disclosed in the present invention in decoupling distributed photovoltaics under radial structure, chain structure, and hybrid structure.

[0171] In the second embodiment, first Figure 9 The distributed photovoltaic system with 36 photovoltaic units shown in the figure is clustered and simulated in the decoupled radial distribution network using the improved FCM algorithm proposed in the present invention, where photovoltaic units of different colors represent Volt-Var control, constant reactive power control and constant power factor control, respectively. The depth of the color of the equivalent line impedance represents the numerical value. The larger the equivalent line impedance, the darker the color. The installed capacity of each photovoltaic unit is shown in Table 4, and the total photovoltaic installed capacity is 2640kW. The reactive power curve and active power curve of the equivalent model obtained by the FCM algorithm and the improved FCM algorithm disclosed by the present invention at the PCC point are shown as follows: Figure 10 and Figure 11 As shown, it can be seen that the active and reactive curves of the equivalent model obtained by using the improved FCM algorithm disclosed in the present invention are basically consistent with those of the detailed model at the PCC point, and the relative error of the reactive curve compared with the traditional FCM algorithm is reduced from 15.8% to 2.11%, which proves the effectiveness of the distributed photovoltaic cluster modeling method based on the improved FCM algorithm disclosed in the present invention.

[0172] Table 1 Line parameters before and after decoupling of the radial structure

[0173]

[0174] Table 2 Line parameters before and after chain structure decoupling

[0175]

[0176] Table 3 Line parameters before and after hybrid structure decoupling

[0177]

[0178] Table 4 Installed capacity of each photovoltaic unit

[0179]

[0180] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware portion can be implemented using dedicated logic; the software portion can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. Those skilled in the art will appreciate that the above-mentioned devices and methods can be implemented using computer-executable instructions and / or contained in processor control code, for example, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, or can be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0181] The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with this technical field within the technical scope disclosed by the present invention and within the spirit and principles of the present invention should be covered by the scope of protection of the present invention.

Claims

1. A distributed photovoltaic cluster modeling method based on an improved FCM algorithm, characterized in that: include: 1) Measure the output voltage V of the photovoltaic unit t and PCC point voltage V PCC and the phase angle difference θ between the two, derive the calculation formula for the equivalent electrical distance from the photovoltaic unit outlet to the PCC point in the radial structure, and then calculate the equivalent electrical distance in the chain structure and complex network topology, and then calculate the line impedance equivalent coefficient; 2) Measure the active power P, reactive power Q and port voltage V of the photovoltaic unit output t , an index value is proposed to characterize the photovoltaic voltage support function; 3) Comprehensively consider equivalent electrical distance, photovoltaic capacity and voltage support function to establish a comprehensive clustering index that accurately characterizes the characteristics of distributed photovoltaic power generation systems; 4) Using the optimal solution obtained by the particle swarm clustering algorithm as the initial cluster center, the FCM algorithm is used to calculate the optimal clustering result, and the equivalent model is calculated using the parameter equivalent; The calculation formula for the equivalent electrical distance from the photovoltaic unit outlet to the PCC point in the radial structure is: Among them, A eq represents the equivalent electrical distance, B represents the line impedance from the PCC point to the grid, A represents the line impedance from the photovoltaic outlet to the PCC point, I d-eq , I q-eq The current from the PCC point to the grid, the subscripts "d" and "q" represent the d-axis and q-axis components respectively, I d , I q represents the current from the PV unit outlet to the PCC point, T represents the transformation matrix of the PV units from their respective coordinate systems to the common coordinate system; The calculation formula of line impedance equivalent coefficient is: Among them LEC i A represents the line impedance equivalent coefficient of the i-th node, ieq is the equivalent line impedance from the photovoltaic unit connection point of the i-th node to the grid, A i represents the line impedance parameter from the ith PV unit to the PCC, and B represents the line impedance from the PCC to the grid.

2. The distributed photovoltaic cluster modeling method based on the improved FCM algorithm according to claim 1, characterized in that: The index value that characterizes the photovoltaic voltage support function is: In the formula, V represents the output voltage of the photovoltaic unit, P represents the output active power of the photovoltaic unit, Q represents the output reactive power of the photovoltaic unit, and c VP represents the correlation between the PV unit outlet voltage and active power, c VQ Indicates the correlation between the PV unit outlet voltage and reactive power.

3. The distributed photovoltaic cluster modeling method based on the improved FCM algorithm according to claim 1 is characterized in that: The comprehensive clustering index that accurately characterizes the characteristics of distributed photovoltaic power generation systems is: <h2 style=";text-align:left;direction:ltr">CCI = a×c<h2 style=";text-align:left;direction:ltr"> VP <h2 style=";text-align:left;direction:ltr"> +a×c<h2 style=";text-align:left;direction:ltr"> VQ <h2 style=";text-align:left;direction:ltr"> +(1-2a)×A<h2 style=";text-align:left;direction:ltr"> eq Among them, CCI represents the distributed photovoltaic comprehensive clustering index, a represents the normalized photovoltaic capacity, c VP and c VQ They represent the correlation between the PV unit outlet voltage and active power and reactive power, respectively. eq Indicates the equivalent electrical distance.

4. The distributed photovoltaic cluster modeling method based on the improved FCM algorithm according to claim 1, characterized in that: Improvements to the FCM algorithm include: The particle swarm algorithm searches for the optimal solution. Particles determine their flight direction based on their own optimal position and the global optimal position. The global optimal solution is obtained through convergence iteration and serves as the initial clustering center of the FCM algorithm. The FCM algorithm updates the membership matrix and selects cluster centers based on the initial cluster centers obtained by the particle swarm algorithm, and it iterates to obtain the optimal clustering result.

5. A distributed photovoltaic cluster modeling system based on an improved FCM algorithm that implements the distributed photovoltaic cluster modeling method based on the improved FCM algorithm as described in any one of claims 1 to 4, characterized in that: include: Photovoltaic measurement module, used to measure the PV unit outlet voltage, PCC point voltage and the phase angle difference θ between the two, and derive and calculate the equivalent electrical distance from the PV unit outlet to the PCC point in radial structure, chain structure and complex network topology; Line impedance equivalent coefficient calculation module, used to calculate the variation coefficient of equivalent line impedance; Power measurement module, used to measure the output active power P, reactive power Q and port voltage V of the photovoltaic unit t , and calculate the voltage support function index of the photovoltaic unit; Clustering index calculation module, which is used to comprehensively consider equivalent electrical distance, photovoltaic capacity and voltage support function to establish a comprehensive clustering index for distributed photovoltaic systems; The clustering calculation module is used to use the optimal solution of the particle swarm algorithm as the initial cluster center, calculate the optimal clustering result using the FCM algorithm, and generate an equivalent model through equivalent parameter calculation.

6. The distributed photovoltaic cluster modeling system based on the improved FCM algorithm according to claim 5, characterized in that: The photovoltaic measurement module further comprises: The calculation unit, based on the measured values of the PV unit outlet voltage and the PCC point voltage as well as the phase angle difference θ, uses the equivalent electrical distance calculation formula to derive and calculate the equivalent electrical distance under radial, chain and complex network topologies, and then calculates the line impedance equivalent coefficient.

7. The distributed photovoltaic cluster modeling system based on the improved FCM algorithm according to claim 5, characterized in that: The power measurement module further includes: The power calculation unit is used to calculate an index value representing the voltage support function of the photovoltaic unit inverter through a formula, and the index value is calculated based on the outlet voltage, output active power and reactive power of the photovoltaic unit.

8. The distributed photovoltaic cluster modeling system based on the improved FCM algorithm according to claim 5, characterized in that: The clustering index calculation module further includes: The index calculation unit is used to calculate the comprehensive clustering index of distributed photovoltaics. The comprehensive clustering index combines the photovoltaic capacity, equivalent electrical distance, and the correlation between the photovoltaic unit outlet voltage and the active power and reactive power.

9. The distributed photovoltaic cluster modeling system based on the improved FCM algorithm according to claim 5, characterized in that: The clustering calculation module includes: A particle swarm optimization unit is used to take the optimal solution of the particle swarm algorithm as the initial cluster center; The FCM algorithm unit is used to update the membership matrix and select cluster centers according to the initial cluster centers provided by the particle swarm optimization unit, iteratively calculate the optimal clustering results and generate an equivalent model.

Images