Power quality control method and system based on distributed power system

Abstract

The application discloses a power quality control method and system based on a distributed power system, relates to the technical field of power quality control, and comprises the following steps: acquiring an initial micro-grid topology structure and key features of nodes thereof, constructing a power quality index model and judging whether to perform topology control; solving conversion of the initial micro-grid topology structure into a simplest topology structure based on a Prim algorithm, and recording a remaining edge set; sorting the remaining edge set based on a neural network scoring model, and screening the remaining edge set according to system computing power; reconstructing the simplest topology structure based on the screened remaining edge set, and obtaining a to-be-controlled topology structure set; acquiring an optimal micro-grid topology structure in the topology structure set based on the power quality index model, comparing the optimal micro-grid topology structure with the initial micro-grid topology structure, obtaining a control instruction, and realizing efficient topology reconstruction control of the micro-grid of the distributed power system to regulate and control power quality.

Description

Technical Field

[0001] This invention relates to the field of power quality control technology, and more specifically, to a power quality control method and system based on distributed power systems. Background Technology

[0002] With the development of distributed energy systems, microgrids have gradually become an important component of power systems. However, due to problems such as unstable output of distributed power sources, frequent node failures, and severe load fluctuations, power quality is easily degraded, especially when microgrids operate in islanded mode, manifesting as voltage deviation, frequency fluctuations, and harmonic distortion. Existing control methods for microgrid topologies are mostly rigid. When power quality problems occur, traditional methods can only switch the microgrid topology through preset rules or manual operation, but they cannot dynamically assess which topology is most conducive to power quality, and may even cause unnecessary operations or failures during the switching process.

[0003] For example, the invention patent announcement CN108363306A discloses a method for determining parameters of a microgrid distributed controller based on linear quadratic optimization. This method establishes a small-signal model of the microgrid based on droop control to achieve reactive power sharing and average voltage recovery. It transforms the microgrid small-signal model into multiple single-input, single-output sub-models corresponding to each distributed power source. A distributed output feedback controller and a linear quadratic optimization objective function are designed. A stable feedback controller is selected, and the rate of change of the linear quadratic optimization objective function with respect to the feedback controller is calculated. Based on this rate of change, the feedback controller is improved so that the improved controller's quadratic optimization performance is superior to that of the original controller, thus obtaining a locally optimal distributed controller. This method designs a microgrid distributed controller based on a linear quadratic optimization strategy, achieving reactive power sharing and average voltage recovery in the microgrid, thereby improving the overall power quality of the microgrid.

[0004] For example, the invention patent announcement CN118054427A discloses a method for operating control of a unified power quality regulator. This invention provides a method for operating control of a unified power quality regulator, which includes: establishing an electrical value vector graph of the topology system; establishing a multi-objective optimization problem of the topology system based on the electrical value vector graph, and transforming the multi-objective optimization problem into a comprehensive single-objective optimization problem according to preset constraints; determining several sets of weight combinations, and obtaining the solution to the comprehensive single-objective optimization problem based on these weight combinations; determining a fitness function, selecting the optimal solution under the comprehensive single-objective optimization problem based on the fitness function, and determining the optimal weight coefficients under the current training conditions based on the optimal solution; selecting several training conditions as inputs, and the corresponding optimal weight coefficients as outputs, repeating the above steps for training to obtain a target surrogate model, and controlling the operation of the topology system in real time through the target surrogate model. The operating control method for a unified power quality regulator provided by this invention improves the stability and flexibility of system operation.

[0005] The above-disclosed technical solutions have at least the following technical problems: When faced with abnormal power quality or local faults, the system topology often relies on manual judgment or preset rule switching, lacking a data-driven adaptive optimization mechanism. This has little impact on the power quality of conventional power systems. However, in the microgrid islanding mode of distributed power systems, the power quality itself fluctuates greatly, which further exacerbates the problem of poor power quality in the islanding mode.

[0006] To address the above problems, this invention proposes a solution. Summary of the Invention

[0007] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a power quality control method and system based on distributed power systems. By constructing node state vectors, power quality index models, and topology optimization control mechanisms, dynamic evaluation and adaptive topology reconfiguration of microgrid power quality are achieved. This addresses the problems in existing technologies for microgrids in islanded mode, where topology switching relies on preset rules or manual judgment, response lags, and optimization effects are poor, making it difficult to ensure stable power quality in islanded mode.

[0008] To achieve the above objectives, the present invention provides the following technical solution: A power quality control method based on distributed power systems includes the following steps: obtaining the initial microgrid topology and key features of each node, constructing a power quality index model and determining whether topology control is required; converting the initial microgrid topology to its simplest form using the Prim algorithm and recording the remaining edge set; sorting the remaining edge set using a neural network scoring model and filtering it based on system computing power; reconstructing the simplest topology based on the filtered remaining edge set to obtain a set of topologies to be controlled; obtaining the optimal microgrid topology from the set of topologies based on the power quality index model and comparing it with the initial microgrid topology to obtain control commands.

[0009] In a preferred embodiment, the step of acquiring the initial microgrid topology and key features of each node, constructing a power quality index model, and determining whether to perform topology control specifically involves: acquiring the key features of each node in the initial microgrid topology; concatenating the key features of each node in the same order into a node state vector, and calculating the key feature deviation between the node state vector and a preset reference index; training the power quality index model based on the key feature deviation using a convolutional neural network, outputting the power quality deviation value, and determining whether to perform topology control based on the power quality deviation value.

[0010] In a preferred embodiment, the step of solving and transforming the initial microgrid topology into a simplified topology based on the Prim algorithm and recording the remaining edge set specifically involves: establishing state constraints, identifying nodes corresponding to key feature deviations that do not meet the state constraints as abnormal nodes, and removing them from the initial microgrid topology to obtain a first microgrid topology, which includes an edge set; solving the first microgrid topology based on the Prim algorithm to obtain a minimum spanning tree as the simplified topology; and recording the portion of the edge set in the first microgrid topology that does not belong to the simplified topology as the remaining edge set.

[0011] In a preferred embodiment, the sorting of the remaining edge set based on the neural network scoring model specifically involves: constructing edge feature vectors based on the key features of each node and the distance between edges between nodes; constructing an edge weight scoring model based on the neural network scoring model according to the historical edge feature vectors; outputting edge score values; and sorting and updating the remaining edge set.

[0012] In a preferred embodiment, the step of filtering the remaining edge set based on the system computing power specifically involves: obtaining system computing power status information and constructing a system computing power margin index based on a linear weighted model; obtaining the number of preferred edges based on the system computing power margin index, and constructing the filtered remaining edge set from the remaining edges corresponding to the number of preferred edges before filtering.

[0013] In a preferred embodiment, the step of reconstructing the simplest topology based on the filtered set of remaining edges to obtain a set of topologies to be controlled specifically involves: when the system simulates topology reconstruction, setting an upper limit on the number of remaining edges that can be added at one time as the maximum limit on the number of edges; and in each round of topology reconstruction, selecting edges from the filtered set of remaining edges that do not exceed the maximum limit on the number of edges, and combining them with the simplest topology to generate several topologies to be controlled, thus forming a set of topologies to be controlled.

[0014] In a preferred embodiment, obtaining the optimal microgrid topology from the set of topologies based on the power quality index model specifically involves: calculating the power quality deviation value of each topology to be controlled in the set of topologies to be controlled based on the power quality index model and sorting them to obtain the minimum power quality deviation value; and selecting the topology corresponding to the minimum power quality deviation value as the optimal topology.

[0015] In a preferred embodiment, the step of comparing the initial microgrid topology with the optimal topology to obtain control commands specifically involves: determining the set of edges to be disconnected and the set of edges to be closed by comparing the edge sets of the initial microgrid topology with those of the optimal topology; sorting the disconnection and closure operations based on load importance and edge priority, and generating corresponding disconnection and closure commands as control commands.

[0016] A system based on power quality control for distributed power systems includes: a state awareness module, which acquires the initial microgrid topology and key features of each node, constructs a power quality index model, and determines whether topology control should be performed; a simplified topology generation module, which uses the Prim algorithm to transform the initial microgrid topology into a simplified topology and records the remaining edge set; an edge scoring and filtering module, which sorts the remaining edge set based on a neural network scoring model and filters the remaining edge set according to the system's computing power; a topology construction module, which reconstructs the simplified topology based on the filtered remaining edge set to obtain a set of topologies to be controlled; and an optimal decision control module, which obtains the optimal microgrid topology from the set of topologies based on the power quality index model and compares it with the initial microgrid topology to obtain control commands.

[0017] The technical effects and advantages of the power quality control method for distributed power systems based on this invention are as follows: 1. This invention constructs a power quality index model and combines it with a simplified topology extraction and dynamic edge set update mechanism to achieve real-time perception and analysis of key characteristics of microgrids, thereby improving the accuracy of power quality anomaly identification and the scientific nature of topology optimization control decisions.

[0018] 2. This invention introduces a neural network scoring system and a system computing power margin for joint control, dynamically generating and selecting the optimal edge set to ensure that topology reconstruction not only improves power quality but also meets the requirements of computing power feasibility and system stability, thereby achieving an efficient and robust topology control strategy. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the power quality control method for distributed power systems according to the present invention.

[0020] Figure 2 This is a schematic diagram of the power quality control system based on a distributed power system according to the present invention.

[0021] Figure 3 This is a graph showing the variation of the total power quality deviation value of the distributed power system based on the present invention at different time periods. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0023] Example 1, Figure 1 The present invention provides a power quality control method based on a distributed power system, comprising the following steps: S1. Obtain the initial microgrid topology and key features of each node, construct a power quality index model and determine whether to perform topology control. The process of obtaining the initial microgrid topology and key features of each node, constructing a power quality index model, and determining whether to perform topology control specifically involves: Obtain the key characteristics of each node in the initial microgrid topology; The key features of each node are concatenated into a node state vector in the same order, and the deviation of the key features between the node state vector and the preset reference index is calculated. Based on the key feature deviation, a power quality index model is trained using a convolutional neural network, and the power quality deviation value is output. Based on the power quality deviation value, it is determined whether to perform topology control.

[0024] It should be noted that the key features include real-time voltage value, real-time frequency value, and real-time harmonic distortion rate; It should be noted that the node state vector is represented in the following form:

[0025] In the formula, Let be the node state vector corresponding to node i. Let be the voltage value at node i. Let i be the frequency value of node i. Let be the harmonic distortion rate of node i.

[0026] It should be noted that the preset reference parameters include a reference voltage of 220V, a reference frequency of 60Hz, and a reference harmonic distortion rate of 3%. It should be noted that the key characteristic deviations include voltage deviation, frequency deviation, and harmonic distortion rate deviation at each node. It should be noted that the following is a summary of the calculations in the above steps: The specific calculation formula for the power quality index model is as follows:

[0027] In the formula, This is the power quality deviation value. The number of nodes in a microgrid. Let be the voltage value at node i. For reference voltage, Let i be the frequency value of node i. For reference frequency, Let i be the harmonic distortion rate of node i. For reference harmonic distortion rate, For absolute voltage deviation, For absolute frequency deviation, For harmonic distortion rate deviation, For voltage deviation, For frequency deviation, As the weight of voltage deviation, As the weight of frequency deviation, This represents the weight of the harmonic distortion rate deviation.

[0028] It should be noted that, , and All settings are based on system priority and sensitivity to meet the requirements. Typically, these are empirical choices, optimized through simulation and historical data; It should be noted that the determination of whether to perform topology control based on power quality deviation value specifically refers to: When the power quality deviation value is greater than the preset power quality deviation threshold, a topology switch is initiated; otherwise, the initial microgrid topology remains unchanged, and the process ends.

[0029] It should be noted that the power quality deviation value is the output of the power quality index model and is used to evaluate the quality of power at nodes; the smaller the power quality deviation value, the better the power quality; the larger the power quality deviation value, the worse the power quality. It should be noted that the specific steps for obtaining the preset power quality deviation threshold are as follows: Based on experience, the maximum allowable voltage deviation is set at 7% (0.07), the maximum allowable frequency deviation is set at 0.2Hz, and the maximum allowable harmonic distortion (THD) deviation is set at 5% (0.05). The preset power quality deviation threshold is:

[0030] In the formula, To preset the power quality deviation threshold, As the weight of voltage deviation, As the weight of frequency deviation, The weight for the harmonic distortion rate deviation; It should be noted that, for the same node, a preset power quality deviation threshold is set. Deviation value from power quality The weights for voltage deviation, frequency deviation, and harmonic distortion deviation should be kept consistent.

[0031] S2, based on the Prim algorithm, solves the transformation of the initial microgrid topology to the simplest topology and records the remaining edge set; The process of transforming the initial microgrid topology to its simplest form using the Prim algorithm and recording the remaining edge set is as follows: State constraints are established, and nodes corresponding to key feature deviations that do not meet the state constraints are identified as abnormal nodes and removed from the initial microgrid topology to obtain the first microgrid topology, which includes an edge set. The Prim algorithm is used to solve the topology of the first microgrid, and the minimum spanning tree is obtained as the simplest topology. The portion of the edge set in the first microgrid topology that does not belong to the simplest topology is denoted as the remaining edge set.

[0032] It should be noted that the state constraints are:

[0033] In the formula, For voltage deviation, For frequency deviation, Let be the voltage value at node i. For reference voltage, Let i be the frequency value of node i. For reference frequency, The voltage tolerance is set empirically. , The allowable frequency deviation is set empirically. , Let be the harmonic distortion rate deviation at node i. The maximum harmonic distortion rate deviation is set to 5% based on experience; It should be noted that the obtained minimum spanning tree is the simplest topology structure, used to ensure the basic power supply connectivity of the microgrid; S3, sorts the remaining edge set based on the neural network scoring model, and filters the remaining edge set according to the system's computing power; The sorting of the remaining edge set based on the neural network scoring model is as follows: Construct edge feature vectors based on the key features of each node and the distances between edges between nodes; Based on historical edge feature vectors, an edge weight scoring model is constructed using a neural network scoring model. The edge score value is then output and the remaining edge set is updated in sorting order.

[0034] It should be noted that the specific steps for constructing the edge feature vector are as follows: Calculate the voltage difference between each node as the corresponding edge voltage value; Calculate the frequency difference between each node as the corresponding edge frequency value; The difference in total harmonic distortion (THD) of voltage between each node is calculated as the THD of the corresponding edge voltage. The edge feature vector is obtained by arranging the edge voltage value, edge frequency value, total harmonic distortion rate of edge voltage, and distance between each edge in sequence.

[0035] It should be noted that the following is a summary of the calculations in the above steps: The specific formula for the side voltage value is as follows:

[0036] The specific formula for the side frequency value is:

[0037] The specific formula for the total harmonic distortion rate of the side voltage is:

[0038] The edge feature vector is specifically represented as follows:

[0039] The edge weight scoring model is as follows:

[0040] In the formula, Let be the edge voltage value between node i and node j. Let be the voltage value at node i. Let be the edge frequency value between node i and node j. Let j be the frequency value of node j. Let be the total harmonic distortion rate of the edge voltage between node i and node j. Let be the total harmonic distortion of the voltage at node i. For edge feature vectors, Let be the line distance between node i and node j. The edge score is the score. It is a neural network scoring model.

[0041] It should be noted that, The edge score is the result output by the edge weight scoring model, where... A higher edge score indicates that the edge is more suitable for topology reconstruction. Furthermore, sorting the remaining edge set based on a neural network scoring model and filtering the remaining edge set according to the system's computing power has the following advantages: Achieving coordinated control of power quality optimization and system computing power utilization: The neural network scoring model is trained based on edge feature vectors, which can effectively identify edges that have a positive impact on power quality and retain them first; the system computing power margin index dynamically limits the number of edges in combination with the current computing power status, ensuring that the optimization control process operates efficiently within the system's processing range, achieving dual guarantees of performance and efficiency.

[0042] Improving the real-time performance and engineering applicability of topology reconfiguration strategies: By controlling the combination scale during the screening stage, the computational complexity of topology reconfiguration can be significantly reduced, avoiding computational delays caused by excessive combination, which helps to improve the response speed and deployment feasibility of reconfiguration strategies in actual microgrid systems.

[0043] Enhance the system's adaptability and robustness to complex operating scenarios: The joint mechanism can dynamically adjust the edge set selection strategy according to the system's operating status and external disturbances, adapting to typical scenarios such as load fluctuations, power supply changes, or local faults, thereby improving the stability and adaptability of topology optimization under changing operating conditions.

[0044] Suppressing the risk of combinatorial explosion and improving the efficiency of reconstruction decisions: In topology optimization, the number of candidate edge combinations grows exponentially. By using a dual mechanism of scoring and ranking and computing power screening, the solution space can be effectively reduced, ineffective calculations and resource waste can be avoided, and the speed and effectiveness of the decision-making process can be improved.

[0045] Promoting the intelligent and data-driven evolution of microgrid control: The scoring model has learning capabilities and can continuously optimize the side weight judgment logic as operating data accumulates, providing key support for building an intelligent microgrid control system with self-learning and self-adaptive capabilities.

[0046] The filtering of the remaining edge set based on the system's computing power specifically involves: Obtain system computing power status information and construct a system computing power margin index based on a linear weighted model; The number of preferred edges is obtained based on the system's computing power margin index, and the remaining edges corresponding to the number of preferred edges before filtering are used to construct the set of remaining edges after filtering.

[0047] It should be noted that the system computing power status information includes CPU idle rate, memory remaining rate, and the time spent on topology optimization in the previous cycle; It should be noted that the specific steps for obtaining the number of preferred edges are as follows: Define the relationship between the system's computing power margin index range and the number of optimal edges; Based on the specific value range of the system's computing power margin index, and based on the relationship between the system's computing power margin index range and the number of preferred edges, find the maximum number of preferred edges that the system can process within that range. It should be noted that the following is a summary of the calculations in the above steps: The specific formula for calculating the system's computing power margin index is as follows:

[0048] The specific relationship between the system's computing power margin index range and the number of optimal edges is as follows:

[0049] In the formula, This is a system computing power margin indicator. The system CPU idle rate, This represents the system's remaining memory percentage. The system time consumed in the previous topology optimization cycle. The empirical weight for CPU idle rate, The empirical weight of the system memory remaining rate, The empirical weights represent the system time consumed in the previous topology optimization cycle. This represents the optimal number of edges that can be added to the current system.

[0050] S4, reconstruct the simplest topology based on the filtered remaining edge set to obtain the set of topologies to be controlled; The process of reconstructing the simplest topology based on the filtered remaining edge set to obtain the set of topologies to be controlled is as follows: When simulating topology reconstruction in the system, a maximum limit is set on the number of remaining edges that can be added at one time, which serves as the maximum limit on the number of edges. In each round of simulation topology reconstruction, edges not exceeding the maximum limit number of edges are selected from the remaining edge set after filtering, and combined with the simplest topology to generate several topologies to be controlled, forming a set of topologies to be controlled.

[0051] It should be noted that the maximum number of edges is determined based on the upper limit of the number of remaining edges that can be added in a single operation. This upper limit is set based on the operational stability requirements during the microgrid topology reconfiguration process and can be comprehensively set based on the system's historical operating data, power quality fluctuation thresholds, and electrical safety margin factors.

[0052] Furthermore, when simulating topology reconstruction in the system, setting an upper limit on the number of remaining edges that can be added in a single step, as the maximum limit on the number of edges, has the following advantages: Controlling the magnitude of topology adjustments: By limiting the number of new edges added in each round of topology reconstruction, it is possible to effectively avoid excessive changes in the topology at one time, thereby reducing the risk of power quality fluctuations caused by structural abrupt changes. Ensuring system operational stability: Gradually introducing edge connections helps maintain the stability of key parameters such as voltage, current, and frequency during microgrid operation, and prevents abnormal phenomena such as local overload, harmonic accumulation, or islanding effect; Enhance the controllability and observability of the reconfiguration process: performing topology adjustments round by round and in stages makes it easier for the system to assess the trend of power quality changes at each stage, enabling the control system to intervene more accurately to optimize control decisions; Reduce system computational complexity: By limiting the search space of edge combinations, the number of candidate topologies in each round is effectively reduced, improving the system's ability to perform real-time optimization and control decisions under limited computing power; Facilitates linkage optimization with power quality assessment models: Only a small number of structural changes are introduced in each round, which helps to more clearly analyze the impact of specific edges on power quality indicators and improve the response accuracy and guidance value of the assessment model.

[0053] S5 obtains the optimal microgrid topology from the topology set based on the power quality index model and compares it with the initial microgrid topology to obtain control commands.

[0054] The process of obtaining the optimal microgrid topology from the set of topologies based on the power quality index model is as follows: Based on the power quality index model, calculate the power quality deviation value of each topology in the set of topologies to be controlled, sort them and obtain the minimum power quality deviation value. The topology corresponding to the minimum power quality deviation value is selected as the optimal topology.

[0055] The control commands obtained by comparing with the initial microgrid topology are as follows: By comparing the edge sets of the initial microgrid topology with those of the optimal topology, the sets of edges that need to be disconnected and the sets of edges that need to be closed are determined. The disconnect and close operations are sorted based on load importance and edge priority, and corresponding disconnect and close instructions are generated as control instructions.

[0056] It should be noted that the importance of the loads is classified according to the power supply guarantee requirements of the area. Areas with extremely high requirements for power supply continuity, such as medical institutions and data centers, are designated as high-level load areas, and their power supply must be guaranteed with priority. Commercial areas and office areas are designated as medium-level load areas, and their priority is lower than that of high-level loads when power resources are limited. Residential lighting and other similar areas are designated as low-level load areas, and a certain degree of power interruption or delay can be tolerated in extreme cases. It should be noted that the edge priority is divided according to the function type of the edge: the main edge connecting high-level load areas or important substation nodes is set as a high priority edge; the edge connecting ordinary load areas is set as a medium priority edge; and the backup or redundant edge that is only activated in case of fault, maintenance or emergency is set as a low priority edge. Figure 2 The system for the power quality control method based on distributed power systems according to the present invention includes: The state awareness module acquires the initial microgrid topology and key features of each node, constructs a power quality index model, and determines whether to perform topology control. The simplified topology generation module uses the Prim algorithm to transform the initial microgrid topology into the simplified topology and records the remaining edge set. The edge scoring and filtering module sorts the remaining edge set based on a neural network scoring model and filters the remaining edge set according to the system's computing power. The topology construction module reconstructs the simplest topology based on the filtered remaining edge set, thus obtaining a set of topologies to be controlled. The optimal decision control module obtains the optimal microgrid topology from the set of topologies based on the power quality index model and compares it with the initial microgrid topology to obtain control commands.

[0057] Figure 3 The curve trends of the total power quality deviation value of the power quality index model of the present invention for various time periods are given. The power quality index model includes voltage deviation, frequency deviation, THD deviation and total power quality deviation value, reflecting that the change of the total power quality deviation value curve is determined by the combined influence of voltage deviation, frequency deviation and THD deviation. The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0058] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0059] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0060] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0061] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0062] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A power quality control method based on distributed power systems, characterized in that, Includes the following steps: Obtain the initial microgrid topology and key characteristics of its nodes; The key features of each node are concatenated into a node state vector in the same order, and the key feature deviation is calculated based on the node state vector and the preset reference index. The key characteristic deviation is input into the power quality index model, and the power quality deviation value is output. Based on the power quality deviation value, it is determined whether to perform topology control. Specifically, when the power quality deviation value is greater than the preset power quality deviation threshold, a topology switch is determined; otherwise, the initial microgrid topology remains unchanged and the process ends. When it is determined that a topology switch is to be performed, the Prim algorithm is used to solve the problem of converting the initial microgrid topology to the simplest topology, and the remaining edge set is recorded. The remaining edge set is sorted based on a neural network scoring model, and then filtered based on the system's computing power. The filtering of the remaining edge set based on the system's computing power specifically involves: Obtain system computing power status information and construct a system computing power margin index based on a linear weighted model; The number of preferred edges is obtained based on the system's computing power margin index, and the remaining edges corresponding to the number of preferred edges before filtering are used to construct the set of remaining edges after filtering. The simplified topology is reconstructed based on the filtered remaining edge set, resulting in the set of topologies to be controlled, specifically: When simulating topology reconstruction in the system, a maximum limit is set on the number of remaining edges that can be added at one time, which serves as the maximum limit on the number of edges. In each round of topology reconstruction, edges not exceeding the maximum limit number of edges are selected from the remaining edge set after filtering, and combined with the simplest topology to generate several topology structures to be controlled, forming a set of topology structures to be controlled. The optimal microgrid topology is obtained from the set of topologies based on the power quality index model and compared with the initial microgrid topology to obtain control commands.

2. The power quality control method based on a distributed power system according to claim 1, characterized in that, The process of transforming the initial microgrid topology to its simplest form using the Prim algorithm and recording the remaining edge set is as follows: State constraints are established, and nodes corresponding to key feature deviations that do not meet the state constraints are identified as abnormal nodes and removed from the initial microgrid topology to obtain the first microgrid topology, which includes an edge set. The Prim algorithm is used to solve the topology of the first microgrid, and the minimum spanning tree is obtained as the simplest topology. The portion of the edge set in the first microgrid topology that does not belong to the simplest topology is denoted as the remaining edge set.

3. The power quality control method based on a distributed power system according to claim 1, characterized in that, The sorting of the remaining edge set based on the neural network scoring model is as follows: Construct edge feature vectors based on the key features of each node and the distances between edges between nodes; Based on historical edge feature vectors, an edge weight scoring model is constructed using a neural network scoring model. The edge score value is then output and the remaining edge set is updated in sorting order.

4. The power quality control method based on a distributed power system according to claim 1, characterized in that, The process of obtaining the optimal microgrid topology from the set of topologies based on the power quality index model is as follows: Based on the power quality index model, calculate the power quality deviation value of each topology in the set of topologies to be controlled, sort them and obtain the minimum power quality deviation value. The topology corresponding to the minimum power quality deviation value is selected as the optimal topology.

5. The power quality control method based on a distributed power system according to claim 1, characterized in that, The control commands are obtained by comparing the data with the initial microgrid topology, specifically: By comparing the edge sets of the initial microgrid topology with those of the optimal topology, the sets of edges that need to be disconnected and the sets of edges that need to be closed are determined. The disconnect and close operations are sorted based on load importance and edge priority, and corresponding disconnect and close instructions are generated as control instructions.

6. A system using the power quality control method for distributed power systems as described in any one of claims 1-5, characterized in that, include: The state awareness module acquires the initial microgrid topology and key features of each node, constructs a power quality index model, and determines whether to perform topology control. The simplified topology generation module uses the Prim algorithm to transform the initial microgrid topology into the simplified topology and records the remaining edge set. The edge scoring and filtering module sorts the remaining edge set based on a neural network scoring model and filters the remaining edge set according to the system's computing power. The topology construction module reconstructs the simplest topology based on the filtered remaining edge set, resulting in a set of topologies to be controlled. The optimal decision control module obtains the optimal microgrid topology from the set of topologies based on the power quality index model and compares it with the initial microgrid topology to obtain control commands.

Images