Distributed photovoltaic cluster voltage regulation and control method considering net load balance

Abstract

The invention relates to the technical field of distributed photovoltaic cluster voltage regulation and control, in particular to a distributed photovoltaic cluster voltage regulation and control method considering net load balancing. According to the method, multi-parameter monitoring equipment is installed at each node to collect data, the net load power is calculated, the space-time distribution characteristics of the net load power are analyzed, key nodes are determined, and a voltage threshold value is set. A comprehensive objective function is constructed by taking voltage stabilization and net load balancing as objectives, a multi-objective optimization model is established, a particle swarm optimization algorithm is adopted for solving, and a result is converted into an inverter regulation and control instruction. And the inverter executes the instruction and feeds back the instruction, the regulation and control effect is evaluated according to the feedback, and closed-loop control is realized. According to the method, the voltage stability of a power supply system is effectively improved, net load balancing is realized, the operation efficiency and reliability of a power system are improved, and technical support is provided for efficient application of a distributed photovoltaic cluster in a power distribution network.

Description

Technical Field

[0001] The present invention relates to the technical field of voltage regulation of distributed photovoltaic clusters, and in particular to a voltage regulation method for distributed photovoltaic clusters taking net load balancing into consideration. Background Art

[0002] With the rapid development of distributed photovoltaic power generation technology, a large number of distributed photovoltaic clusters are connected to the distribution network, which has changed the unidirectional flow characteristics of the traditional distribution network. There are many defects in the existing technology: Difficulty in voltage regulation: Distributed photovoltaic output is affected by factors such as light and weather, and has obvious intermittent and volatile characteristics. When the photovoltaic output changes, the voltage of the distribution network is prone to fluctuations or even exceeding the limit. Traditional voltage regulation methods are difficult to respond to such changes quickly and accurately, resulting in excessively high or low voltages at some nodes, affecting the normal operation and service life of power equipment and reducing the quality of power. For example, when the light intensity changes suddenly, the photovoltaic output changes dramatically, which may cause the voltage of nearby nodes to instantly exceed the allowable range, affecting the stability of power equipment.

[0003] Unbalanced net load: The load characteristics and photovoltaic distribution in different regions are different, resulting in an imbalanced net load between nodes in the distribution network. The net load in some areas is too high, while in other areas it is low, which not only increases the transmission loss of the power grid, but may also cause some lines to be overloaded, affecting the safe and stable operation of the power grid. Traditional load balancing methods do not fully consider the impact of distributed photovoltaics and cannot effectively solve the problem of unbalanced net load.

[0004] Insufficient data collection and analysis: Existing monitoring equipment and data collection methods are difficult to fully and accurately obtain the operating data of distributed photovoltaic clusters and distribution networks. The frequency of data collection is low and the accuracy is poor, which cannot reflect the dynamic changes of the system in real time. At the same time, there is a lack of effective analysis methods for the collected data, making it difficult to explore the temporal and spatial distribution characteristics behind the data, and unable to provide strong support for voltage regulation and load balancing.

[0005] Single control strategy: Existing distributed photovoltaic cluster output control strategies often focus on a single goal, such as voltage stability or load balancing, and ignore the mutual influence and coordinated optimization between multiple goals. This single-goal control strategy cannot achieve the overall optimal operation of the power system, reducing the system's operating efficiency and reliability.

[0006] To this end, we propose a distributed photovoltaic cluster voltage control method that considers net load balancing. Summary of the invention

[0007] In view of the above deficiencies in the prior art, the purpose of the present invention is to solve the problems of voltage regulation, net load imbalance, insufficient data collection and analysis, and single regulation strategy caused by distributed photovoltaic access in the prior art.

[0008] In order to achieve the above objectives, the present invention adopts the following technical solutions: The voltage control method of distributed photovoltaic cluster considering net load balancing includes the following steps: S1. Net load data collection and analysis: The net load data of each node is collected in real time through monitoring equipment, and the temporal and spatial distribution characteristics of the net load are analyzed using data mining algorithms to identify the peak and valley periods of the net load and the imbalance of spatial distribution; S2. Voltage monitoring and threshold setting: Based on the temporal and spatial distribution characteristics of the net load provided by S1, voltage sensors are installed at key nodes of the power supply system to monitor voltage changes in real time. The upper and lower voltage thresholds are set in combination with the temporal and spatial distribution characteristics of the net load to ensure that the voltage regulation objectives are clear and meet the actual operation requirements. S3. Formulate the output control strategy of distributed photovoltaic clusters. Using the results of S1 and S2, with the goal of voltage stability and net load balance of the power supply system, determine the output control target of the distributed photovoltaic cluster, and balance the net load of each node during the control. Establish a model to transform the output control problem into an optimization problem, search for the optimal output control solution, consider the constraints, obtain the best strategy and convert it into a control instruction to send to the inverter; S4. Control command execution and feedback. After receiving the control command, the inverter responds quickly and accurately adjusts the output power of the photovoltaic array to execute output control. It also has fault diagnosis and protection functions, monitors the working status in real time, and alarms and protects in case of abnormalities. After executing the control command, the voltage monitoring system and net load monitoring equipment are used to monitor the voltage changes of the power supply system and the net load distribution of each node in real time, and the voltage data and net load data after control are compared and analyzed with the control target to evaluate the control effect.

[0009] Furthermore, the monitoring equipment in S1 is a multi-parameter monitoring device installed at each node in the coverage area of ​​the distributed photovoltaic cluster according to the grid topology and node importance; on the photovoltaic power generation side, a photovoltaic array monitor with maximum power point tracking (MPPT) function is selected, which can accurately collect the real-time output power, voltage, current, ambient temperature, light intensity and other data of the photovoltaic components, providing a basis for analyzing the photovoltaic output characteristics; smart meters and power quality analyzers are used on the load side; the monitoring equipment is equipped with high-speed communication modules, supporting 4G / 5G, optical fiber or LoRa and other communication methods, to ensure that the data can be transmitted to the data processing center stably and in real time.

[0010] Furthermore, the net load data of the nodes in S1 is obtained by collecting active power data in real time through monitoring equipment, and then calculating the net load power of each node, and the calculation formula is: ; in, For Node Net load power; For Node Active power of load; For Node Active output of distributed photovoltaics.

[0011] Furthermore, the method for analyzing the spatiotemporal distribution characteristics of the net load using a data mining algorithm in S1 is as follows: The collected net load data is analyzed using the DBSCAN algorithm. The DBSCAN algorithm is used to identify the core points, boundary points and noise points in the data by setting the neighborhood radius and the minimum number of points, and then the data is divided into different clusters. In the spatial dimension, a spatial weight matrix is ​​constructed according to the geographical location and electrical connection relationship of the nodes, and the global Moran's index is used to judge the spatial distribution balance of the net load. The calculation formula of the global Moran's index is: ; is the global Moran index, is the total number of nodes, is the spatial weight matrix element, is the average value of the net load of all nodes; when When the value is greater than 0 and significant, it indicates that there is a spatial positive correlation in the net load, that is, high-value areas are adjacent to high-value areas, and low-value areas are adjacent to low-value areas, and there is an imbalance in spatial distribution; when When the value is less than 0 and significant, it indicates that there is a spatial negative correlation; When the value is equal to 0, it indicates that the spatial distribution is random.

[0012] Furthermore, when using the DBSCAN algorithm to analyze the net load data in S1, the method of setting the neighborhood radius and the minimum point is as follows: For the neighborhood radius, the net load data obtained in S1 are statistically analyzed, the Euclidean distance between each data point is calculated, the distribution of the Euclidean distance is obtained, and any percentile between the 85% and 90% quantiles in the distance distribution is selected as the initial value of the neighborhood radius, and fine-tuning is performed on this basis; The minimum number of points is determined by calculating the average density of the data points, first calculating the number of points in the neighborhood of each data point and then finding the average of these points; Furthermore, the DBSCAN algorithm is used to identify the core points, boundary points, and noise points in the data by setting the neighborhood radius and the minimum number of points, and then the data is divided into different clusters. The specific process is as follows: S101. Identification of core points, for a point in the data set ,by Draw a circle with the radius of the neighborhood as the center (a hypersphere in high-dimensional space), and this range is the point Neighborhood; if at point The number of points contained in the neighborhood of itself) is greater than or equal to the minimum number of points, then the point Determined to be a core point; S102. Determination of boundary points, if one The point is not a core point, but it falls on a core point In the neighborhood of , then this point It is the boundary point; although the number of points in the neighborhood of the boundary point does not reach the minimum number of points, it is associated with the core point; the boundary point is located at the edge of the data cluster, it plays the role of connecting different core point areas, and helps to determine the scope of the data cluster; S103. Determination of noise points: points that are neither core points nor boundary points are noise points. Noise points appear as isolated points in the data set, the number of points in their neighborhood is far less than the minimum number of points, and they are not associated with the neighborhood of any core point; S104. Data clustering: Starting from any core point, all core points and boundary points in its neighborhood are grouped into one cluster; since there are other core points in the neighborhood of the core point, and these core points have their own neighborhoods, by continuously expanding this neighborhood relationship, the interconnected core points and boundary points are gradually merged to form a complete data cluster; in the net load data, starting from the core point, the core points and boundary points in its neighborhood are grouped into one cluster, and then the related points in the neighborhood of these points are continued to be grouped until no new points can be added, thus determining a data cluster representing a specific net load distribution feature; repeating this process, processing other unclassified core points in the data set, thereby dividing the entire data set into different data clusters; each data cluster represents a group of nodes with similar net load spatiotemporal distribution characteristics, and by analyzing these clusters, the spatial distribution imbalance of the net load is identified, such as data clusters in some areas have a higher net load, while data clusters in other areas have a lower net load.

[0013] Furthermore, the key nodes in S2 include distributed photovoltaic centralized access points, load center nodes, voltage fluctuation sensitive nodes, etc.

[0014] Furthermore, the voltage sensor in S2 has real-time monitoring, data storage and fault self-diagnosis functions. It can accurately monitor voltage changes at a sampling frequency of 100 Hz, store data in a local cache, and upload it to the data processing center in real time through the communication network. When the sensor detects a fault in itself, it immediately issues an alarm and switches to a backup sensor to ensure the continuity of voltage monitoring.

[0015] Furthermore, the step of setting the voltage upper and lower thresholds in S2 includes: During peak load periods, considering that the system reactive power demand increases significantly and the voltage is prone to drop, the voltage lower limit threshold Set to: ; in is the system rated voltage, The allowable voltage drop ratio during the peak load period is determined based on historical data and system operation experience; During the load valley period, the voltage may rise due to the photovoltaic output. Set to: ; The voltage rise ratio allowed during the load valley period, usually between 0.03-0.07; In order to make the voltage threshold more in line with the actual operating conditions and to adapt to the dynamic changes of the power supply system, a dynamic adjustment mechanism of the voltage threshold is introduced, and a voltage change rate monitoring model is established to calculate the voltage change rate per unit time in real time. When the voltage change rate per unit time exceeds the set warning value, it indicates that the voltage is changing too fast. At this time, the voltage threshold is appropriately adjusted according to the voltage change trend. At the same time, the voltage threshold is fine-tuned in combination with the real-time changes in the net load to ensure the accuracy of voltage regulation.

[0016] Further, the S3 includes: S31. Determine the control target, take the voltage stability and net load balance of the power supply system as the goal, and construct a comprehensive objective function: ; ; ; in, represents the objective function; Indicates the net load balancing index; It is The actual voltage of each key node; Indicates the number of key nodes; Indicates voltage deviation index; is the number of key nodes; It is a key node The voltage weight, Determined according to the degree of influence of the node on the system stability; by setting different weights for different nodes, the influence of each node voltage on the overall stability of the system can be more accurately reflected; is the number of regions; For Region The node set within For Region The number of nodes in It is The net load of each node; Indicates the voltage deviation index weight coefficient; represents the weight coefficient of the net load balancing index, and , and The value is determined according to the actual operation requirements and system characteristics. Through multiple simulation tests, the system operation under different weight combinations is simulated. The optimal weight combination is selected based on the evaluation indicators such as system voltage deviation, net load balance and system stability. S32. Establish a model, take the active output and reactive output of each photovoltaic power station as decision variables, consider power balance constraints, photovoltaic power station output constraints, voltage constraints, line transmission power constraints and inverter switching frequency constraints, and use the comprehensive objective function In order to optimize the target, a model is established to transform the output regulation problem into an optimization problem: ; In S32: Power balance constraints: ; ; in, is the number of PV power stations in the distributed PV cluster; Indicates Active power output of each photovoltaic power station; Indicates Reactive power output of each photovoltaic power station; The active power input to the grid; Reactive power input to the grid; is the sum of active loads of all nodes, Indicates Active load of each node; is the sum of reactive loads of all nodes, Indicates Reactive load of each node; ; ; It is The maximum active output of a photovoltaic power station is determined by factors such as the rated power of the photovoltaic modules, light intensity and temperature; They are The minimum and maximum reactive power output of a photovoltaic power station is limited by the inverter capacity and power factor; , to ensure the output constraints of photovoltaic power stations at key nodes: The voltage is within the set threshold range; Line transmission power constraints: , It is a line The transmission power, It is a line The maximum permissible transmission power is to prevent line overload; consider the line resistance and reactance , , and They are respectively the active power and reactive power transmitted on the line, and their values ​​are determined by power flow calculation; Inverter switching frequency constraint: To avoid the impact of frequent inverter switching on equipment life and system stability, limit the number of inverter switching times within a certain time interval. ; Indicates that at a certain time interval The actual switching times of the inverter; The inverter is The maximum number of switching times allowed within a period is an upper limit value pre-set based on factors such as inverter equipment characteristics and system operation stability requirements; S33. Solve the model and use the particle swarm optimization algorithm (PSO) to search for the optimal output control solution; S34. Convert the optimized output values ​​of each photovoltaic power station into control instructions for the inverter.

[0017] Further, the S33 includes: In the PSO algorithm, each particle represents a photovoltaic power station output control scheme, and the position of the particle Indicates the output of each photovoltaic power station, Indicates The output control scheme of photovoltaic power station with particles Representative The particle corresponding to ( ) output of a photovoltaic power station; speed , Indicates The speed of a particle, Indicates The particle in Dimensions (corresponding to The speed determines the moving direction and step size of the particle in the search space. The particle is guided to find a better position in the search space by updating the speed. In each iteration, the particle updates the speed and position according to its own historical optimal position and the global optimal position. The update formula is: ; ; in is the inertia weight, , is the learning factor, , is a random number between [0,1], is the number of iterations; the optimal output control scheme is obtained through multiple iterative searches; Indicates The particle in The historical optimal position on the dimension; the global optimal position is in the corresponding The value of output adjustment of each photovoltaic power station; It represents the optimal solution found by the entire particle swarm in the previous iteration process in the corresponding The value of output adjustment of each photovoltaic power station; In order to improve the convergence speed and accuracy of the algorithm, the PSO algorithm is improved; the adaptive inertia weight is adopted , as the number of iterations increases, Gradually reduce from a larger value, so that the algorithm has a stronger global search ability in the early stage and a better local search ability in the later stage; at the same time, dynamically adjust the learning factor and , according to the fitness value of the particle, for particles with good fitness values, the learning factor is appropriately reduced to make them pay more attention to their own experience; for particles with poor fitness values, the learning factor is increased to encourage them to learn from excellent particles.

[0018] The S34 includes: converting the output value of each photovoltaic power station obtained by optimization into a control instruction of the inverter, and encoding the control instructions such as active output and reactive output into a data packet that complies with the protocol specification according to the communication protocol of the inverter; in the encoding process, verifying and encrypting the data to ensure the accuracy and security of the instruction transmission; sending the control instruction to the inverter of each photovoltaic power station through the communication network; to ensure that the instruction can be transmitted reliably, a redundant transmission and retransmission mechanism is adopted, and if the inverter does not receive the correct instruction within a certain period of time, it automatically requests retransmission; at the same time, a timestamp is added to the instruction so that the inverter can judge the timeliness of the instruction and avoid executing outdated instructions; Further, the S4 includes: S41. Command execution: After receiving the control command, the inverter responds quickly and accurately adjusts the output power of the photovoltaic array. The method based on model predictive control is combined with the real-time operating status and environmental parameters of the photovoltaic array to predict the photovoltaic output in the future, optimize the control strategy of the inverter in advance, and realize accurate adjustment of the output power of the photovoltaic array. At the same time, the inverter has multiple built-in protection mechanisms to monitor its own working status in real time, including temperature, current, voltage, power factor and other parameters; when an abnormal situation is detected, an alarm is immediately issued and protective measures are initiated; in addition, the inverter also has a fault diagnosis function, which can make a preliminary diagnosis of the fault and upload the fault information to the monitoring center, so that the operation and maintenance personnel can handle it in time; S42. Feedback evaluation: after executing the control command, the voltage change of the power supply system and the net load distribution of each node are monitored in real time through the voltage monitoring system and the net load monitoring equipment; the collection frequency (such as 1 minute / time) is set to obtain the real-time operation data of the system; the collected data is transmitted to the data processing center through the communication network; In the data processing center, the collected data after regulation is compared and analyzed with the regulation target; the voltage deviation index value and the net load balance index value are recalculated and compared with the voltage deviation index value and the net load balance index value before regulation and the regulation target value. If the voltage deviation index value after regulation is less than the voltage deviation index value before regulation, and the net load balance index value after regulation is less than the net load balance index value before regulation, it indicates that the regulation strategy is effective and the system voltage stability and net load balance are improved; if the expected effect is not achieved, the reasons are analyzed; according to the analysis results, the parameters of the optimization model are adjusted, such as re-determining the weight coefficient and , or improve the constraints; at the same time, check the data collection and analysis modules to ensure the accuracy and completeness of the data; then re-execute step S3 to generate a new control strategy until the control effect meets the requirements.

[0019] In summary, due to the adoption of the above technical solution, the beneficial technical effects of the invention are: The distributed photovoltaic cluster voltage control method considering net load balancing proposed in the present invention brings significant optimization effects to the power system as a whole through a series of closely coordinated steps, and effectively improves the comprehensive performance of the power system in the scenario of large-scale access of distributed photovoltaics; Enhance the stability of the power system. From the perspective of system stability, this method achieves multi-faceted collaborative optimization. In the data collection and analysis stage, the net load data is accurately obtained and its spatiotemporal distribution characteristics are deeply analyzed to provide a solid data foundation for subsequent regulation. Based on this, in the voltage monitoring and threshold setting link, a reasonable voltage threshold can be set in a targeted manner to provide a clear regulation target for voltage stability. The formulation of the output regulation strategy of the distributed photovoltaic cluster takes voltage stability and net load balance as the dual goals. By establishing and solving complex models, precise control of photovoltaic output is achieved. This series of operations cooperate with each other to effectively suppress the voltage instability caused by photovoltaic output fluctuations and load changes, maintain the system voltage in a reasonable range, reduce the impact of voltage anomalies on power equipment, reduce the risk of system failures caused by voltage problems, and enhance the stability of the entire power system. For example, when the light intensity changes rapidly, the system can adjust the photovoltaic output in time to maintain voltage stability and ensure the normal operation of the equipment.

[0020] Improve the operating efficiency of the power system. In terms of operating efficiency, the present invention has achieved remarkable results. In-depth analysis of net load data helps to accurately grasp the load change law, achieve net load balance by regulating the output of the photovoltaic cluster, and reduce power loss during grid transmission. At the same time, reasonable voltage regulation reduces the additional energy consumption of equipment caused by unreasonable voltage. In inverter control, precise power regulation enables the photovoltaic array to better track the maximum power point and improve the efficiency of photovoltaic power generation. Overall, this multi-link collaborative optimization method reduces the waste of energy during transmission and conversion, improves the energy utilization efficiency of the power system, and enables more reasonable allocation and utilization of power resources. Taking a regional power grid as an example, after applying this method, the system loss was significantly reduced and the energy utilization efficiency was significantly improved.

[0021] To ensure the reliability of the power system, from the perspective of reliability, the present invention has built a complete guarantee system. The monitoring equipment collects data in an all-round way and provides real-time feedback on the system operation status, providing a basis for timely discovery of potential problems. The voltage sensor continuously monitors the voltage of key nodes, and can quickly trigger the control mechanism once an abnormality occurs. The inverter has fault diagnosis and protection functions, and can respond in time when problems occur in the equipment to avoid the expansion of faults. The execution and feedback link of the control instruction ensures that the system is always in a stable operating state by real-time monitoring of the control effect and timely adjustment of the strategy. This all-round, multi-level guarantee mechanism greatly improves the ability of the power system to cope with various complex situations, ensures the reliability of power supply, reduces the occurrence of power outages, and improves the user's power experience. For example, when the load suddenly changes or the photovoltaic equipment fails, the system can adjust quickly to ensure uninterrupted power supply.

[0022] Adapting to the trend of distributed energy access, the present invention has important practical significance under the current trend of widespread access to distributed energy. It fully considers the characteristics of distributed photovoltaic clusters, and through innovative control methods, effectively solves the problems of voltage fluctuations and net load imbalance caused by distributed photovoltaic access. This enables the power system to better accept distributed photovoltaic energy, promote the efficient use of clean energy, and promote the transformation of the energy structure towards a green and sustainable direction. At the same time, the flexibility and scalability of the method enable it to adapt to the development trend of the increasing scale and types of distributed energy in the future, laying a solid foundation for building an intelligent and efficient modern power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a flow chart of the voltage control method of distributed photovoltaic cluster considering net load balancing; Figure 2 Flowchart for using DBSCAN algorithm to divide data into different clusters; Figure 3 Flowchart of a method for developing a distributed photovoltaic cluster output control strategy; Figure 4 Flowchart for regulating instruction execution and feedback. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the invention more clear, the invention is further described in detail below in combination with the embodiments. It should be understood that the specific embodiments described here are only used to explain the invention and are not used to limit the invention.

[0025] like Figure 1 As shown: The voltage control method of distributed photovoltaic cluster considering net load balancing includes the following steps: S1. Net load data collection and analysis: The net load data of each node is collected in real time through monitoring equipment, and the temporal and spatial distribution characteristics of the net load are analyzed using data mining algorithms to identify the peak and valley periods of the net load and the imbalance of spatial distribution; S2. Voltage monitoring and threshold setting: Based on the temporal and spatial distribution characteristics of the net load provided by S1, voltage sensors are installed at key nodes of the power supply system to monitor voltage changes in real time. The upper and lower voltage thresholds are set in combination with the temporal and spatial distribution characteristics of the net load to ensure that the voltage regulation objectives are clear and meet the actual operation requirements. S3. Formulate the output control strategy of distributed photovoltaic clusters. Using the results of S1 and S2, with the goal of voltage stability and net load balance of the power supply system, determine the output control target of the distributed photovoltaic cluster, and balance the net load of each node during the control. Establish a model to transform the output control problem into an optimization problem, search for the optimal output control solution, consider the constraints, obtain the best strategy and convert it into a control instruction to send to the inverter; S4. Control command execution and feedback. After receiving the control command, the inverter responds quickly and accurately adjusts the output power of the photovoltaic array to execute output control. It also has fault diagnosis and protection functions, monitors the working status in real time, and alarms and protects in case of abnormalities. After executing the control command, the voltage monitoring system and net load monitoring equipment are used to monitor the voltage changes of the power supply system and the net load distribution of each node in real time, and the voltage data and net load data after control are compared and analyzed with the control target to evaluate the control effect.

[0026] In S1, the monitoring equipment is a multi-parameter monitoring device installed at each node in the coverage area of ​​the distributed photovoltaic cluster according to the grid topology and node importance; on the photovoltaic power generation side, a photovoltaic array monitor with maximum power point tracking (MPPT) function is selected, which can accurately collect the real-time output power, voltage, current, ambient temperature, light intensity and other data of the photovoltaic components, providing a basis for analyzing the photovoltaic output characteristics; on the load side, smart meters and power quality analyzers are used, which can not only measure active power, reactive power, and apparent power, but also capture the harmonic components and three-phase imbalance of the load and other information, comprehensively reflecting the power consumption characteristics of the load; the monitoring equipment is equipped with a high-speed communication module, supporting 4G / 5G, optical fiber or LoRa and other communication methods, to ensure that the data can be transmitted to the data processing center stably and in real time.

[0027] The net load data of the nodes in S1 is obtained by collecting active power data in real time through monitoring equipment, and then calculating the net load power of each node. The calculation formula is: ; in, For Node Net load power; For Node Active power of load; For Node Active power output of distributed photovoltaics.

[0028] The method for analyzing the spatiotemporal distribution characteristics of the net load using a data mining algorithm in S1 is as follows: The DBSCAN algorithm is used to identify the core points, boundary points and noise points in the data by setting the neighborhood radius and the minimum number of points, and then the data is divided into different clusters. In the spatial dimension, a spatial weight matrix is ​​constructed according to the geographical location and electrical connection relationship of the nodes, and the global Moran's index is used to judge the spatial distribution balance of the net load. The calculation formula of the global Moran's index is: ; is the global Moran index, is the total number of nodes, is the spatial weight matrix element, is the average value of the net load of all nodes; when When the value is greater than 0 and significant, it indicates that there is a spatial positive correlation in the net load, that is, high-value areas are adjacent to high-value areas, and low-value areas are adjacent to low-value areas, and there is an imbalance in spatial distribution; when When the value is less than 0 and significant, it indicates that there is a spatial negative correlation; When the value is equal to 0, it indicates that the spatial distribution is random.

[0029] When using the DBSCAN algorithm to analyze the net load data in S1, the method of setting the neighborhood radius and the minimum point is: For the neighborhood radius, the net load data obtained in S1 is statistically analyzed, the Euclidean distance between each data point is calculated, the distribution of the Euclidean distance is obtained, and any percentile between the 85% quantile and the 90% quantile in the distance distribution is selected as the initial value of the neighborhood radius, and fine-tuning is performed on this basis; for example, the net load data distance between all nodes is calculated, and the 85% quantile of the distance value is taken as 0.25, and the neighborhood radius can be initially set to 0.25; The minimum number of points is determined by calculating the average density of the data points. First, the number of points in the neighborhood of each data point (the neighborhood is calculated with a larger initial neighborhood radius value) is calculated, and then the average of these points is calculated; if the average number of points is 6, the minimum number of points can be set to an integer close to the average, such as 5 or 7; like Figure 2 As shown in the figure, the DBSCAN algorithm is used to identify the core points, boundary points and noise points in the data by setting the neighborhood radius and the minimum number of points, and then the data is divided into different clusters. The specific process is as follows: S101. Identification of core points, for a point in the data set ,by Draw a circle with the radius of the neighborhood as the center (a hypersphere in high-dimensional space), and this range is the point Neighborhood; if at point The number of points contained in the neighborhood of itself) is greater than or equal to the minimum number of points, then the point Determined as a core point, in the net load data collected by S1, it is assumed that there is a node , its net load data is relatively stable over a period of time and is similar to the net load data of multiple surrounding nodes; after setting the neighborhood radius and the minimum number of points, if the node The number of nodes within the neighborhood radius of reaches or exceeds the minimum number of points, then the node The corresponding net load data point is the core point; this means that in terms of spatial distribution, the node There are enough similar net load data points around it, forming a relatively dense data area; S102. Determination of boundary points, if one The point is not a core point, but it falls on a core point In the neighborhood of , then this point is a boundary point; although the number of points in the boundary point's own neighborhood does not reach the minimum number of points, it is associated with the core point; for example, in the net load data set, the node The net load data of the spatial distribution has less points in its own neighborhood than the minimum number of points, but it is in the core point In the neighborhood of The corresponding net load data point is the boundary point; the boundary point is located at the edge of the data cluster, it plays the role of connecting different core point areas and helps to determine the scope of the data cluster; S103. Determination of noise points: points that are neither core points nor boundary points are noise points. Noise points are isolated points in the data set. The number of points in their neighborhood is far less than the minimum number of points, and they are not associated with the neighborhood of any core point. In the net load data, the net load data of individual nodes may be greatly different from the net load data of other nodes due to equipment failure, data collection errors, etc. Under the conditions of the set neighborhood radius and minimum number of points, the net load data point corresponding to the node does not belong to the neighborhood of any core point, and there are not enough points in its own neighborhood. In this case, this point is a noise point. S104. Data clustering: Starting from any core point, all core points and boundary points in its neighborhood are classified into one cluster; since there may be other core points in the neighborhood of the core point, and these core points have their own neighborhoods, by continuously expanding this neighborhood relationship, the interconnected core points and boundary points are gradually merged to form a complete data cluster; in the net load data, starting from the core point, the core points and boundary points in its neighborhood are included in one cluster, and then the related points in the neighborhood of these points are continued to be included until no new points can be added, so that a data cluster representing a specific net load distribution feature is determined; repeat this process, process other unclassified core points in the data set, so as to divide the entire data set into different data clusters; each data cluster represents a group of nodes with similar net load spatiotemporal distribution characteristics, and by analyzing these clusters, the spatial distribution imbalance of the net load is identified, such as the net load of data clusters in some areas is higher, while the net load of data clusters in other areas is lower.

[0030] The key nodes in S2 include distributed photovoltaic centralized access points, load center nodes, voltage fluctuation sensitive nodes, etc.

[0031] The voltage sensor in S2 has real-time monitoring, data storage and fault self-diagnosis functions. It can accurately monitor voltage changes at a sampling frequency of 100 Hz, store data in a local cache, and upload it to the data processing center in real time through a communication network. When the sensor detects a fault in itself, it immediately issues an alarm and switches to a backup sensor to ensure the continuity of voltage monitoring.

[0032] The step of setting the voltage upper and lower thresholds in S2 includes: During peak load periods, considering that the system reactive power demand increases significantly and the voltage is prone to drop, the voltage lower limit threshold Set to: ; in is the system rated voltage, The voltage drop ratio allowed during peak load hours is determined based on historical data and system operation experience, and is generally between 0.05 and 0.1. During the load valley period, the voltage may rise due to the photovoltaic output. Set to: ; The voltage rise ratio allowed during the load valley period, usually between 0.03-0.07; In order to make the voltage threshold more in line with the actual operating conditions and to adapt to the dynamic changes of the power supply system, a dynamic adjustment mechanism of the voltage threshold is introduced, and a voltage change rate monitoring model is established to calculate the voltage change rate per unit time in real time. When the voltage change rate per unit time exceeds the set warning value, it indicates that the voltage changes too fast. At this time, the voltage threshold is appropriately adjusted according to the voltage change trend. For example, if the voltage continues to rise and the voltage change rate per unit time is greater than 0.01 , the voltage upper limit threshold will be appropriately lowered to prevent damage to the equipment caused by excessive voltage; at the same time, combined with the real-time changes in net load, the voltage threshold will be fine-tuned to ensure the accuracy of voltage regulation.

[0033] like Figure 3 As shown, the S3 includes: S31. Determine the control target, take the voltage stability and net load balance of the power supply system as the goal, and construct a comprehensive objective function: ; ; ; in, represents the objective function; Indicates the net load balancing index; It is The actual voltage of each key node; Indicates the number of key nodes; Indicates voltage deviation index; is the number of key nodes; It is a key node The voltage weight, Determined according to the degree of influence of the node on the system stability; for example, the key node close to the power supply side and bearing important loads has a greater impact on the system stability. The value can be relatively large; while some secondary nodes The value is relatively small; by setting different weights for different nodes, the impact of each node voltage on the overall stability of the system can be more accurately reflected; is the number of regions; For Region The node set within For Region The number of nodes in It is The net load of each node; Indicates the voltage deviation index weight coefficient; represents the weight coefficient of the net load balancing index, and , and The value is determined according to the actual operation requirements and system characteristics. Through multiple simulation tests, the system operation under different weight combinations is simulated, and the optimal weight combination is selected based on the evaluation indicators such as system voltage deviation, net load balance and system stability. For example, in areas with high requirements for voltage stability, Set to 0.7, Set to 0.3; in areas with high requirements for load balancing, it can be appropriately increased The proportion of S32. Establish a model, take the active output and reactive output of each photovoltaic power station as decision variables, consider power balance constraints, photovoltaic power station output constraints, voltage constraints, line transmission power constraints and inverter switching frequency constraints, and use the comprehensive objective function In order to optimize the target, a model is established to transform the output regulation problem into an optimization problem: ; In the steps: Power balance constraints: ; ; in, is the number of PV power stations in the distributed PV cluster; Indicates Active power output of each photovoltaic power station; Indicates Reactive power output of each photovoltaic power station; The active power input to the grid; Reactive power input to the grid; is the sum of active loads of all nodes, Indicates Active load of each node; is the sum of reactive loads of all nodes, Indicates Reactive load of each node; Output constraints of photovoltaic power stations: ; ; It is The maximum active output of a photovoltaic power station is determined by factors such as the rated power of the photovoltaic modules, light intensity and temperature; They are The minimum and maximum reactive power output of a photovoltaic power station is limited by the inverter capacity and power factor; , ensuring that the voltage of key nodes is within the set threshold range; Line transmission power constraints: , It is a line The transmission power, It is a line The maximum permissible transmission power is to prevent line overload; consider the line resistance and reactance , , and They are respectively the active power and reactive power transmitted on the line, and their values ​​are determined by power flow calculation; Inverter switching frequency constraint: To avoid the impact of frequent inverter switching on equipment life and system stability, limit the number of inverter switching times within a certain time interval. ; Indicates that at a certain time interval The actual number of switching times of the inverter; "switching" here usually refers to the conversion operation of the inverter between different working modes (such as different power output states, active-reactive regulation mode switching, etc.); The inverter is The maximum number of switching times allowed within a power system is an upper limit value pre-set based on factors such as the characteristics of the inverter equipment and the requirements for system operation stability. It reflects the limitation on the switching operation frequency of the inverter to avoid damage to the life of the inverter equipment due to excessively frequent switching (such as accelerated aging of internal electronic components and increased wear of mechanical parts), and also prevents adverse effects on the stability of the entire power system (such as causing voltage fluctuations, increased harmonics, etc.); S33. Solve the model and use the particle swarm optimization algorithm (PSO) to search for the optimal output control solution; S34. Convert the optimized output values ​​of each photovoltaic power station into control instructions for the inverter.

[0034] The S33 includes: In the PSO algorithm, each particle represents a photovoltaic power station output control scheme, and the position of the particle Indicates the output of each photovoltaic power station, Indicates The output control scheme of photovoltaic power station with particles Representative The particle corresponding to The output of each photovoltaic power station; the position vector of each particle contains the output information of all photovoltaic power stations in the distributed photovoltaic cluster, and the optimal photovoltaic power station output control scheme is searched by adjusting the position of the particle; the speed , Indicates The speed of a particle, Indicates The particle in Dimensions (corresponding to Output adjustment of each photovoltaic power station, ), the speed determines the moving direction and step size of the particle in the search space, and the particle is guided to find a better position in the search space by updating the speed; in each iteration, the particle updates the speed and position according to its own historical optimal position and the global optimal position, and the update formula is: ; ; in is the inertia weight, , is the learning factor, , is a random number between [0,1], is the number of iterations; the optimal output control scheme is obtained through multiple iterative searches; Indicates The particle in The historical optimal position in dimension; the global optimal position in Dimension (corresponding to Output adjustment of each photovoltaic power station); It represents the optimal solution found by the entire particle swarm in the previous iteration. Dimension (corresponding to Output adjustment of each photovoltaic power station); In order to improve the convergence speed and accuracy of the algorithm, the PSO algorithm is improved; the adaptive inertia weight is adopted , as the number of iterations increases, Gradually reduce from a larger value, so that the algorithm has a stronger global search ability in the early stage and a better local search ability in the later stage; at the same time, dynamically adjust the learning factor and , according to the fitness value of the particle, for particles with good fitness values, the learning factor is appropriately reduced to make them pay more attention to their own experience; for particles with poor fitness values, the learning factor is increased to encourage them to learn from excellent particles.

[0035] The S34 includes: converting the optimized output value of each photovoltaic power station into a control instruction of the inverter, and encoding the control instructions such as active output and reactive output into a data packet that complies with the protocol specification according to the communication protocol of the inverter (such as ModbusTCP / IP protocol); in the encoding process, verifying and encrypting the data to ensure the accuracy and security of the instruction transmission; sending the control instructions to the inverters of each photovoltaic power station through a communication network (such as industrial Ethernet, 4G / 5G network, etc.); to ensure that the instructions can be transmitted reliably, a redundant transmission and retransmission mechanism is adopted. If the inverter does not receive the correct instruction within a certain period of time, it automatically requests retransmission; at the same time, a timestamp is added to the instruction so that the inverter can judge the timeliness of the instruction and avoid executing outdated instructions; like Figure 4 As shown, the S4 comprises: S41. Instruction execution. After receiving the control instruction, the inverter responds quickly and accurately adjusts the output power of the photovoltaic array. The method based on model predictive control (MPC) is used to combine the real-time operating status and environmental parameters (such as light intensity, temperature, etc.) of the photovoltaic array to predict the photovoltaic output in the future, optimize the control strategy of the inverter in advance, and realize the precise adjustment of the output power of the photovoltaic array. For example, according to the real-time light intensity and temperature data, the maximum power point of the photovoltaic array is predicted using the photovoltaic cell model. The inverter adjusts its own pulse width modulation (PWM) signal to make the photovoltaic array work near the maximum power point, thereby improving the power generation efficiency. At the same time, the inverter has multiple built-in protection mechanisms to monitor its own working status in real time, including parameters such as temperature, current, voltage, and power factor. When an abnormal situation is detected (such as overcurrent, overvoltage, overheating, and abnormal power factor), an alarm is immediately issued and protective measures are initiated, such as cutting off the circuit and reducing output power, to prevent equipment damage. In addition, the inverter also has a fault diagnosis function, which can make a preliminary diagnosis of the fault and upload the fault information to the monitoring center for timely processing by operation and maintenance personnel. For example, when the inverter temperature is detected to be too high, a high temperature alarm is immediately issued, and the output power is reduced to prevent the inverter from being damaged due to overheating, and the fault information (such as the specific value of the overheating and the time of occurrence) is uploaded to the monitoring center. S42. Feedback evaluation: after executing the control command, the voltage change of the power supply system and the net load distribution of each node are monitored in real time through the voltage monitoring system and the net load monitoring equipment; the collection frequency (such as 1 minute / time) is set to obtain the real-time operation data of the system; the collected data is transmitted to the data processing center through the communication network; At the data processing center, the collected data after regulation is compared and analyzed with the regulation target; the voltage deviation index value and the net load balance index value are recalculated and compared with the voltage deviation index value and the net load balance index value before regulation and the regulation target value. If the voltage deviation index value after regulation is less than the voltage deviation index value before regulation, and the net load balance index value after regulation is less than the net load balance index value before regulation, it indicates that the regulation strategy is effective and the system voltage stability and net load balance are improved; if the expected effect is not achieved, the reasons are analyzed, which may be inaccurate model parameters, unconsidered interference factors in the system, deviations in the regulation execution process, etc.; according to the analysis results, the parameters of the optimization model are adjusted, such as re-determining the weight coefficient and , or improve the constraints; at the same time, check the data acquisition and analysis modules to ensure the accuracy and completeness of the data; then re-execute step S3 to generate a new control strategy until the control effect meets the requirements. For example, if it is found that the voltage in a certain area is still beyond the threshold range and the net load balance is not as expected, it may be that the weight coefficient is set unreasonably through analysis. After re-adjusting the weight coefficient, control again until the system operation indicators meet the requirements.

[0036] The above description is a preferred embodiment of the invention and is not intended to limit the invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the invention should be included in the protection scope of the invention.

Claims

1. A distributed photovoltaic cluster voltage control method considering net load balancing, characterized in that: The following steps are involved: S1. Net load data collection and analysis: The net load data of each node is collected in real time through monitoring equipment, and the temporal and spatial distribution characteristics of the net load are analyzed using data mining algorithms to identify the peak and valley periods of the net load and the imbalance of spatial distribution; S2. Voltage monitoring and threshold setting: Based on the temporal and spatial distribution characteristics of the net load provided by S1, voltage sensors are installed at key nodes of the power supply system to monitor voltage changes in real time. The upper and lower voltage thresholds are set in combination with the temporal and spatial distribution characteristics of the net load to ensure that the voltage regulation objectives are clear and meet the actual operation requirements. S3. Formulate the output control strategy of distributed photovoltaic clusters. Using the results of S1 and S2, with the goal of voltage stability and net load balance of the power supply system, determine the output control target of the distributed photovoltaic cluster, and balance the net load of each node during the control. Establish a model to transform the output control problem into an optimization problem, search for the optimal output control solution, consider the constraints, obtain the best strategy and convert it into a control instruction to send to the inverter; S4. Control command execution and feedback. After receiving the control command, the inverter responds quickly and accurately adjusts the output power of the photovoltaic array to execute output control. It also has fault diagnosis and protection functions, monitors the working status in real time, and alarms and protects in case of abnormalities. After executing the control command, the voltage monitoring system and net load monitoring equipment are used to monitor the voltage changes of the power supply system and the net load distribution of each node in real time, and the voltage data and net load data after control are compared and analyzed with the control target to evaluate the control effect.

2. The distributed photovoltaic cluster voltage control method considering net load balancing according to claim 1 is characterized in that: The monitoring equipment in S1 is to install multi-parameter monitoring equipment at each node in the coverage area of ​​the distributed photovoltaic cluster according to the grid topology and node importance; on the photovoltaic power generation side, a photovoltaic array monitor with maximum power point tracking function is selected, which can accurately collect the real-time output power, voltage, current, ambient temperature and light intensity data of the photovoltaic components, providing a basis for analyzing the photovoltaic output characteristics; smart meters and power quality analyzers are used on the load side; the monitoring equipment is equipped with high-speed communication modules, supporting 4G / 5G, optical fiber or LoRa communication methods, to ensure that the data can be stably and in real time transmitted to the data processing center; The net load data of the nodes in S1 is obtained by collecting active power data in real time through monitoring equipment, and then calculating the net load power of each node. The calculation formula is: ； in, For Node Net load power; For Node Active power of load; For Node Active power output of distributed photovoltaics.

3. The distributed photovoltaic cluster voltage control method considering net load balancing according to claim 1 is characterized in that: The method for analyzing the spatiotemporal distribution characteristics of the net load using a data mining algorithm in S1 is as follows: The collected net load data is analyzed using the DBSCAN algorithm. The DBSCAN algorithm is used to identify the core points, boundary points and noise points in the data by setting the neighborhood radius and the minimum number of points, and then the data is divided into different clusters. In the spatial dimension, a spatial weight matrix is ​​constructed according to the geographical location and electrical connection relationship of the nodes, and the global Moran's index is used to judge the spatial distribution balance of the net load. The calculation formula of the global Moran's index is: ； is the global Moran index, is the total number of nodes, is the spatial weight matrix element, is the average value of the net load of all nodes; when When the value is greater than 0 and significant, it indicates that there is a spatial positive correlation in the net load, that is, high-value areas are adjacent to high-value areas, and low-value areas are adjacent to low-value areas, and there is an imbalance in spatial distribution; when When the value is less than 0 and significant, it indicates that there is a spatial negative correlation; When the value is equal to 0, it indicates that the spatial distribution is random.

4. The distributed photovoltaic cluster voltage control method considering net load balancing according to claim 3 is characterized in that: When using the DBSCAN algorithm to analyze the net load data in S1, the method of setting the neighborhood radius and the minimum number of points is as follows: For the neighborhood radius, the net load data obtained in S1 are statistically analyzed, the Euclidean distance between each data point is calculated, the distribution of the Euclidean distance is obtained, and any percentile between the 85% and 90% quantiles in the distance distribution is selected as the initial value of the neighborhood radius, and fine-tuning is performed on this basis; The minimum number of points is determined by calculating the average density of the data points, first calculating the number of points in the neighborhood of each data point, and then finding the average of these points.

5. The distributed photovoltaic cluster voltage control method considering net load balancing according to claim 3 is characterized in that: The DBSCAN algorithm is used to identify the core points, boundary points, and noise points in the data by setting the neighborhood radius and the minimum number of points, and then the data is divided into different clusters. The specific process is as follows: S101. Identification of core points, for a point in the data set ,by Draw a circle with the radius as the center and the neighborhood radius as the radius. This range is the point Neighborhood of; If at point If the number of points in the neighborhood of is greater than or equal to the minimum number of points, then point Determined to be a core point; S102. Determination of boundary points, if one The point is not a core point, but it falls on a core point In the neighborhood of , then this point It is the boundary point; S103. Determination of noise points: points that are neither core points nor boundary points are noise points. Noise points appear as isolated points in the data set, the number of points in their neighborhood is far less than the minimum number of points, and they are not associated with the neighborhood of any core point; S104. Data clustering: Starting from any core point, all core points and boundary points in its neighborhood are grouped into one cluster; since there are other core points in the neighborhood of the core point, and these core points have their own neighborhoods, by continuously expanding this neighborhood relationship, the interconnected core points and boundary points are gradually merged to form a complete data cluster; in the net load data, starting from the core point, the core points and boundary points in its neighborhood are grouped into one cluster, and then the related points in the neighborhood of these points are continued to be grouped until no new points can be added, thus determining a data cluster representing a specific net load distribution feature; repeating this process, processing other unclassified core points in the data set, thereby dividing the entire data set into different data clusters; each data cluster represents a group of nodes with similar net load spatiotemporal distribution characteristics, and by analyzing these clusters, the spatial distribution imbalance of the net load is identified.

6. The distributed photovoltaic cluster voltage control method considering net load balancing according to claim 1, characterized in that: In S2, the key nodes include distributed photovoltaic centralized access points, load center nodes, and voltage fluctuation sensitive nodes; the voltage sensor has real-time monitoring, data storage and fault self-diagnosis functions, and can accurately monitor voltage changes at a sampling frequency of 100Hz, and store the data in the local cache, and upload it to the data processing center in real time through the communication network. When the sensor detects its own fault, it immediately issues an alarm and switches to the backup sensor to ensure the continuity of voltage monitoring.

7. The distributed photovoltaic cluster voltage control method considering net load balancing according to claim 1, characterized in that: The step of setting the voltage upper and lower thresholds in S2 includes: During peak load periods, considering that the system reactive power demand increases significantly and the voltage is prone to drop, the voltage lower limit threshold Set to: ； in is the system rated voltage, The allowable voltage drop ratio during the peak load period is determined based on historical data and system operation experience; During the low load period, the voltage will rise due to the photovoltaic output, so the voltage upper limit threshold Set to: ； The voltage increase ratio allowed during the load valley period; In order to make the voltage threshold more in line with the actual operating conditions and to adapt to the dynamic changes of the power supply system, a dynamic adjustment mechanism of the voltage threshold is introduced, and a voltage change rate monitoring model is established to calculate the voltage change rate per unit time in real time. When the voltage change rate per unit time exceeds the set warning value, it indicates that the voltage is changing too fast. At this time, the voltage threshold is appropriately adjusted according to the voltage change trend. At the same time, the voltage threshold is fine-tuned in combination with the real-time changes in the net load to ensure the accuracy of voltage regulation.

8. The distributed photovoltaic cluster voltage control method considering net load balancing according to claim 1, characterized in that: The S3 includes: S31. Determine the control target, take the voltage stability and net load balance of the power supply system as the goal, and construct a comprehensive objective function: ； ； ； in, represents the objective function; Indicates the net load balancing index; It is The actual voltage of each key node; Indicates the number of key nodes; Indicates voltage deviation index; It is a key node The voltage weight, Determined based on the impact of the node on system stability; is the number of regions; For Region The node set within For Region The number of nodes in It is The net load of each node; Indicates the voltage deviation index weight coefficient; represents the weight coefficient of the net load balancing index, and , and The value is determined according to the actual operation requirements and system characteristics; S32. Establish a model, take the active output and reactive output of each photovoltaic power station as decision variables, consider power balance constraints, photovoltaic power station output constraints, voltage constraints, line transmission power constraints and inverter switching frequency constraints, and use the comprehensive objective function In order to optimize the target, a model is established to transform the output regulation problem into an optimization problem: ; In S32: Power balance constraints: ； ； in, is the number of PV power stations in the distributed PV cluster; Indicates Active power output of each photovoltaic power station; Indicates Reactive power output of each photovoltaic power station; The active power input to the grid; Reactive power input to the grid; is the sum of active loads of all nodes, Indicates Active load of each node; is the sum of reactive loads of all nodes, Indicates Reactive load of each node; Output constraints of photovoltaic power stations: ； ； It is The maximum active output of a photovoltaic power station is determined by the rated power of the photovoltaic modules, light intensity and temperature factors; They are The minimum and maximum reactive power output of a photovoltaic power station is limited by the inverter capacity and power factor; , ensuring that the voltage of key nodes is within the set threshold range; Line transmission power constraints: , It is a line The transmission power, It is a line The maximum permissible transmission power is to prevent line overload; consider the line resistance and reactance , , and They are respectively the active power and reactive power transmitted on the line, and their values ​​are determined by power flow calculation; Inverter switching frequency constraint: To avoid the impact of frequent inverter switching on equipment life and system stability, limit the number of inverter switching times within a certain time interval. ; Indicates that at a certain time interval The actual switching times of the inverter; The inverter is The maximum number of switching times allowed within a period; S33. Solve the model and use the particle swarm optimization algorithm PSO to search for the optimal output control solution; S34. Convert the optimized output values ​​of each photovoltaic power station into control instructions for the inverter.

9. The distributed photovoltaic cluster voltage control method considering net load balancing according to claim 8, characterized in that: The S33 includes: In the PSO algorithm, each particle represents a photovoltaic power station output control scheme, and the position of the particle Indicates the output of each photovoltaic power station, Indicates The output control scheme of photovoltaic power station with particles Representative The particle corresponding to The output of a photovoltaic power station; speed , Indicates The speed of a particle, Indicates The particle in the corresponding The speed of the output adjustment of each photovoltaic power station; in each iteration, the particle updates the speed and position according to its own historical optimal position and the global optimal position, and the update formula is: ； ； in is the inertia weight, , is the learning factor, , is a random number between [0,1], is the number of iterations; the optimal output control scheme is obtained through multiple iterative searches; Indicates The particle in The historical optimal position on the dimension; the global optimal position is in the corresponding The value of output adjustment of each photovoltaic power station; It represents the optimal solution found by the entire particle swarm in the previous iteration process in the corresponding The value of output adjustment of each photovoltaic power station; In order to improve the convergence speed and accuracy of the algorithm, the PSO algorithm is improved; adaptive inertia weight is adopted , as the number of iterations increases, Gradually reduce from a larger value, so that the algorithm has a stronger global search ability in the early stage and a better local search ability in the later stage; at the same time, dynamically adjust the learning factor and , according to the fitness value of the particle, for particles with good fitness values, the learning factor is appropriately reduced to make them pay more attention to their own experience; for particles with poor fitness values, the learning factor is increased to encourage them to learn from excellent particles.

10. The distributed photovoltaic cluster voltage control method considering net load balancing according to claim 8, characterized in that: The S34 includes: converting the output value of each photovoltaic power station obtained by optimization into a control instruction of the inverter, and encoding the active output and reactive output control instructions into data packets that comply with the protocol specifications according to the communication protocol of the inverter; in the encoding process, verifying and encrypting the data to ensure the accuracy and security of the instruction transmission; sending the control instructions to the inverters of each photovoltaic power station through the communication network; to ensure that the instructions can be transmitted reliably, a redundant transmission and retransmission mechanism is adopted, and if the inverter does not receive the correct instruction within a certain period of time, it automatically requests retransmission; at the same time, a timestamp is added to the instruction so that the inverter can judge the timeliness of the instruction and avoid executing outdated instructions.

11. The distributed photovoltaic cluster voltage control method considering net load balancing according to claim 1, characterized in that: The S4 includes: S41. Command execution: After receiving the control command, the inverter responds quickly and accurately adjusts the output power of the photovoltaic array. The method based on model predictive control is combined with the real-time operating status and environmental parameters of the photovoltaic array to predict the photovoltaic output in the future, optimize the control strategy of the inverter in advance, and realize accurate adjustment of the output power of the photovoltaic array. At the same time, the inverter has built-in multiple protection mechanisms to monitor its own working status in real time, including temperature, current, voltage, and power factor parameters; when an abnormal situation is detected, an alarm is immediately issued and protective measures are initiated; in addition, the inverter also has a fault diagnosis function, which can make a preliminary diagnosis of the fault and upload the fault information to the monitoring center, so that the operation and maintenance personnel can handle it in time; S42. Feedback evaluation: after executing the control command, the voltage change of the power supply system and the net load distribution of each node are monitored in real time through the voltage monitoring system and the net load monitoring equipment; the collection frequency is set to obtain the real-time operation data of the system; the collected data is transmitted to the data processing center through the communication network; At the data processing center, the collected data after regulation are compared and analyzed with the regulation target; the voltage deviation index and the net load balance index are recalculated and compared with the voltage deviation index value and the net load balance index value before regulation and the regulation target value. If the voltage deviation index value after regulation is less than the voltage deviation index value before regulation, and the net load balance index value after regulation is less than the net load balance index value before regulation, it indicates that the regulation strategy is effective and the system voltage stability and net load balance are improved; if the expected effect is not achieved, the cause is analyzed; according to the analysis results, the parameters of the optimization model are adjusted; at the same time, the data collection and analysis module is checked to ensure the accuracy and completeness of the data; then the S3 step is re-executed to generate a new regulation strategy until the regulation effect meets the requirements.

Images