Stable voltage compensation control method applied to microgrid cluster

Abstract

The invention relates to the technical field of reactive power compensation of a power system, in particular to a stable voltage compensation control method applied to a micro-grid cluster, which specifically comprises the following steps of: analyzing principal components of all items of grid data of each node through a principal component analysis algorithm, extracting the first three principal components, and performing data weight distribution through an entropy weight method to obtain a data weight distribution result; constructing reactive compensation priorities of the nodes; the future reactive compensation priority of the node is predicted through the historical reactive compensation priority of the node, the fluctuation rate of prediction data is analyzed, and the dynamic compensation urgency of the node is constructed in combination with the clustering condition of the node at the current moment; a hierarchical resource allocation strategy is implemented according to the dynamic compensation urgency degree, the calculation process is updated in real time according to the reactive compensation feedback result, a self-optimization closed loop is formed, the problems that a traditional method is insufficient in quantization precision and poor in real-time performance are solved, the reactive demand matching precision and the voltage maintaining stable effect are improved, and the method is suitable for large-scale popularization and application. And the communication load and the calculation complexity are obviously reduced.

Description

Technical Field

[0001] The present application relates to the technical field of reactive power compensation in power systems, and in particular to a stable voltage compensation control method applied to a microgrid cluster. Background Art

[0002] With the large-scale grid connection of new energy, the popularization of direct current transmission and the increase in impact loads, traditional static compensation devices such as fixed capacitor banks and reactors are unable to meet the dynamic reactive power support needs due to problems such as slow response speed, low adjustment accuracy and harmonic amplification. As IGBT devices mature and control algorithms are optimized, the multi-level reactive power collaborative compensation system uses multiple SVG / STATCOM / reactive power compensation devices to be layered at different nodes in the microgrid cluster, combined with intelligent coordinated control strategies, to form a collaborative compensation network with complementary spatial distribution and temporal response.

[0003] In a multi-level reactive power coordinated compensation system, traditional centralized optimization methods need to incorporate massive variables such as node voltage, reactive power output, and line parameters of the entire network into a unified model. The optimization dimension grows exponentially with the scale of the power grid. Traditional methods face serious dimensional disasters. The huge amount of communication data and complex calculation process will significantly increase information transmission and processing time, causing communication delays. In addition, the exponential growth of the optimization dimension causes the algorithm convergence speed to drop sharply, making it difficult to generate dynamic control instructions in a timely manner. As a result, the reactive power compensation equipment cannot respond quickly to transient fluctuations in the power grid and cannot respond to changes in time, thereby causing the transient voltage stability of the power grid to deteriorate and triggering the risk of voltage collapse and instability. Summary of the Invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide a stable voltage compensation control method applied to a microgrid cluster. The technical solutions adopted are as follows: The present invention provides a method for controlling a stable voltage compensation applied to a microgrid cluster, the method comprising the following steps: Collect various grid data of each key node in the grid at each moment, and use the matrix composed of all grid data of all nodes at each moment as the grid data matrix at each moment; Obtaining the principal components of each node in the power grid data matrix through a principal component analysis algorithm; assigning weights to each type of data in the power grid data matrix; and analyzing the reactive power compensation priority of each node at each moment based on the principal components and the weights assigned to each type of data; Based on the reactive compensation priority of each node in the historical time period before the current moment, combined with a prediction algorithm, a predicted value of the reactive compensation priority of each node in a future preset time period is obtained; based on the data fluctuation of the predicted value of each node in the future preset time period, a predicted data fluctuation rate of each node at the current moment is constructed; the reactive compensation priority of all nodes at the current moment is clustered to obtain clusters, and based on the reactive compensation priority and predicted data fluctuation rate of each node at the current moment, combined with the data distribution in each cluster, the dynamic compensation urgency of each node at the current moment is constructed; Based on the clustering results of the dynamic compensation urgency of all nodes at the current moment, compensation resources are allocated to perform closed-loop control of the compensation system.

[0005] In one embodiment, the weight distribution for each type of data in the power grid data matrix is specifically as follows: The i-th row vector in the power grid data matrix consists of all the power grid data of the i-th node, and the j-th column vector consists of the j-th type of power grid data of all nodes; the power grid data matrix at each moment is used as the input of the weight distribution algorithm, and the output is the weight of each type of data in the power grid data matrix at each moment.

[0006] In one embodiment, the process of obtaining the reactive power compensation priority of each node at each moment is as follows: Obtain the variance contribution rate of each principal component of each node; extract the first preset number of principal components of each node in the power grid data matrix to construct a principal component score matrix; based on the principal component scores and corresponding variance contribution rates of the nodes in the principal component score matrix at each moment, combined with the power grid data values of each node at each moment and the weight assigned to each data, construct the reactive compensation priority of each node at each moment.

[0007] In one embodiment, the reactive power compensation priority is expressed as:

[0008] Where, is the reactive compensation priority of node i at the tth acquisition time; x is the number of principal components of each node extracted from any power grid data matrix; is the pth principal component score of node i in the principal component score matrix at the tth acquisition time; is the variance contribution rate of the pth principal component of node i in the principal component score matrix at the tth acquisition time; y is the number of power grid data types; is the weight assigned to the qth type of data in the power grid data matrix at the tth acquisition moment; is the value of the qth data of node i in the power grid data matrix at the tth acquisition time.

[0009] In one embodiment, the process of obtaining the predicted value of the reactive power compensation priority is as follows: Obtaining the reactive power compensation priority of each node at each moment in a preset historical period at the current moment, and using the sequence of the reactive power compensation priorities of each node at all moments in the preset historical period as the historical reactive power compensation priority sequence of each node at the current moment; The historical reactive power compensation priority sequence of each node is used as the input of the prediction model to obtain the predicted value of the reactive power compensation priority of each node within a preset time period in the future.

[0010] In one embodiment, the predicted data fluctuation rate of each node at the current moment is: the standard deviation of the reactive power compensation priority prediction values of each node at the current moment at all moments within a preset time period in the future.

[0011] In one embodiment, the process of obtaining the dynamic compensation urgency of each node at the current moment is as follows: Calculate the ratio between the mean value of the intra-cluster data of each cluster and the mean value of the reactive compensation priority of all nodes in the cluster, which is recorded as a first ratio; calculate the dynamic compensation urgency of each node at the current moment based on the first ratio, the reactive compensation priority of each node at the current moment and the predicted data volatility.

[0012] In one embodiment, the expression of the dynamic compensation urgency is: , where is the dynamic compensation urgency of node i at the current moment; is the first ratio of the cluster where node i is located at the current moment; is the reactive power compensation priority of node i at the current moment; is the mean value of the data in the cluster where node i is located at the current moment; is the predicted data volatility of node i at the current moment.

[0013] In one embodiment, the allocation of compensation resources based on the clustering result of the dynamic compensation urgency of all nodes at the current moment is specifically as follows: The dynamic compensation urgency of all nodes in the current network is used as the input of the clustering algorithm to obtain three clusters; Calculate the mean of the intra-cluster data for each cluster and record it as the mean dynamic compensation urgency. The cluster with the largest mean dynamic compensation urgency is designated as the high-demand cluster, the cluster with the smallest dynamic compensation urgency is designated as the low-demand cluster, and the remaining cluster is designated as the medium-demand cluster. Allocate the capacity of the reactive generator based on the grouping results, where the capacity allocated to the high-demand cluster > the capacity allocated to the medium-demand cluster > the capacity allocated to the low-demand cluster.

[0014] In one embodiment, the method for allocating weights to each type of data in the power grid data matrix is an entropy weight method.

[0015] The embodiments of the present application have at least the following beneficial effects: This application collects various grid data of each key node in the power grid at each moment, takes the matrix composed of all grid data of all nodes at each moment as the grid data matrix, obtains the principal components of each node in the grid data matrix through the principal component analysis algorithm, extracts the first three principal components, and assigns weights to each data through the entropy weight method, thereby constructing the reactive compensation priority of each node at each moment, retaining the core reactive driving characteristics and dynamically assigning parameter weights, accurately quantifying the real-time intensity of the node reactive demand, avoiding the parameter redundancy, dimensionality explosion and feature fuzziness problems caused by global data modeling in traditional methods, and solving the problems of insufficient quantification accuracy and poor real-time performance of traditional methods; through the reactive compensation of nodes in historical time periods Priority, predict the reactive compensation priority of nodes in future time periods, analyze the predicted data volatility of nodes at the current moment, and combine the clustering situation of nodes at the current moment to construct the dynamic compensation urgency of each node at the current moment. By integrating real-time demand intensity, cluster priority and future fluctuation risk, it dynamically reflects the spatiotemporal urgency of node reactive demand, avoiding the rigidity of static resource allocation strategy and the interference of communication delay on real-time control; implement a hierarchical resource allocation strategy according to the dynamic compensation urgency of each node, and update the calculation process in real time according to the power grid data matrix after reactive compensation feedback, forming a self-optimizing closed loop, improving the accuracy of reactive demand matching and voltage stability, and significantly reducing communication load and calculation complexity. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0017] Figure 1 A flowchart of a stable voltage compensation control method for a microgrid cluster provided in one embodiment of the present application; Figure 2 Schematic diagram of the process of obtaining reactive power compensation priority. DETAILED DESCRIPTION

[0018] To further illustrate the technical means and effects employed by this application to achieve the intended invention objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the stable voltage compensation control method for a microgrid cluster proposed in this application. In the following description, different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0019] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0020] The specific scheme of the stable voltage compensation control method applied to a microgrid cluster provided by the present application is described in detail below with reference to the accompanying drawings.

[0021] See also Figure 1 , which shows a flowchart of a stable voltage compensation control method for a microgrid cluster provided by an embodiment of the present application, the method comprising the following steps: Step S1 : collecting various grid data of each key node in the grid at each moment, and taking a matrix composed of all grid data of all nodes at each moment as a grid data matrix at each moment.

[0022] The physical architecture and hierarchical closed-loop control of the multi-level reactive power collaborative compensation system are constructed. Its core includes multi-level SVG / STATCOM / reactive power compensation devices (static var generators) hierarchically deployed at the nodes of the microgrid cluster. In this application, the nodes of the microgrid cluster refer to the transformers, junction boxes, new energy grid connection points, load centers, etc. in the microgrid cluster, which are in key positions in the microgrid cluster. Each device is equipped with an independent IGBT converter and a local controller. The microgrid cluster parameters of each node are collected in real time through various sensors, and low-latency data interaction is achieved by relying on 5G or fiber optic communication modules. The closed-loop control adopts a hierarchical structure, and the local control layer quickly responds to single-point reactive power demand. The coordinated control layer dynamically adjusts the global compensation strategy through real-time feedback of the microgrid cluster status, forming a "perception-decision-execution" closed loop to ensure the coordinated response of multi-level devices in space and time.

[0023] Assume that there are N nodes in the microgrid cluster. Current data at each node is collected through current transformers installed at each node of the microgrid cluster. Voltage data at each node is collected through voltage transformers installed at each node of the microgrid cluster. Real-time active and reactive power data at each node are collected through power sensors deployed at each node of the microgrid cluster. Impedance measurement units are installed at the beginning and end of the associated lines to collect equivalent line impedance data.

[0024] For the collection of various microgrid cluster data from each node, preferably, in the embodiment of the present application, the collection frequency of each type of data is set to 10 Hz. As other embodiments of the present application, the implementer can set the collection frequency according to actual conditions. All data collected by all nodes are subjected to overall normalization processing. Among them, overall normalization is a well-known technology, and the specific process is not repeated here.

[0025] At each acquisition moment, the current, voltage, active power, reactive power and line impedance data of all nodes are combined into a The matrix is denoted as the grid data matrix at the time of acquisition, where the i-th row vector in the matrix consists of all grid data from the i-th node, and the j-th column vector consists of the j-th type of grid data from all nodes. This matrix fully characterizes the dynamic operating state of the grid and provides a global data foundation for coordinated compensation.

[0026] Step S2, obtaining the principal components of each node in the power grid data matrix through the principal component analysis algorithm; assigning weights to each type of data in the power grid data matrix; and analyzing the reactive compensation priority of each node at each moment based on the principal components and the weights assigned to each type of data.

[0027] Since traditional multi-level reactive power coordinated compensation systems rely on global data modeling, the redundancy of massive power grid parameters and the significant coupling between parameters lead to data dimension explosion and key feature ambiguity, resulting in insufficient quantification accuracy of reactive power demand and low real-time control efficiency.

[0028] In the above-mentioned power grid data matrix at each acquisition moment, each node contains five-dimensional data. The power grid data matrix at each acquisition moment is used as the input of the PCA (principal component analysis) algorithm, and the first three principal components of each node in the matrix are extracted to obtain the principal component score matrix and the variance contribution rate of each principal component in the principal component score matrix. The size of the matrix is , the element in the i-th row and p-th column of the matrix is the p-th principal component score for the i-th node. The principal component score matrix eliminates redundancy through dimensionality reduction, extracts the core comprehensive characteristics of the grid nodes, and maps the original 5-dimensional parameters to a 3-dimensional space, characterizing the distribution characteristics of the nodes with respect to the key drivers of reactive power demand. The PCA algorithm is well known, and the specific process is not detailed here.

[0029] The grid data matrix at each acquisition moment is then used as input for the entropy weighting method. Weights are dynamically assigned by calculating the information entropy of each data type at different nodes. The output is the weight of each data type in the grid data matrix, recorded as the first weight. Weight assignment based on the entropy weighting method is well known to those skilled in the art and will not be elaborated on here.

[0030] It should be noted that for obtaining the weight of each data, this application only provides a method for calculating the weight. There are many existing methods for calculating weights. Implementers can also use other weight distribution algorithms to calculate the weight of each data. This application does not make specific restrictions.

[0031] Furthermore, based on the above analysis, taking node i as an example, the reactive power compensation priority of each node is constructed, and the expression is:

[0032] Where, is the reactive compensation priority of node i at the tth acquisition time; x is the number of principal components of each node extracted from any power grid data matrix, and in the embodiment of the present application, x=3; is the pth principal component score of node i in the principal component score matrix at the tth acquisition time; is the pth principal component variance contribution rate of node i in the principal component score matrix at the tth acquisition time; y is the number of power grid data types, and in the embodiment of the present application, y=5; is the first weight of the qth type of data in the power grid data matrix at the tth acquisition time; is the value of the qth data of node i in the power grid data matrix at the tth acquisition time.

[0033] is the projection value of the pth principal component obtained after PCA dimensionality reduction at the i-th node, reflecting the comprehensive characteristics of the node in terms of the key driving factors of reactive power demand. The variance contribution rate of the principal component p, that is, the proportion of the original data information retained by the principal component, is used as a weight and multiplied by the corresponding principal component score to sum up. It represents the comprehensive projection of the node on the core reactive driving factors and reflects the overall influence of node i on the global reactive demand of the power grid. The larger the value, the higher the priority of the reactive demand of node i in the global compensation. is the weight dynamically assigned to the qth grid data by the entropy weight method, reflecting the information value of this parameter in grid operation. Characterize the actual operating status of node i on the qth parameter. The sum of the product of the two reflects the comprehensive abnormality degree of node i on all key grid parameters. The larger the value, the more significant the deviation of the node's operating status from the normal range and the more urgent the demand for reactive power compensation.

[0034] The reactive power compensation priority of node i is comprehensively quantified by integrating the global characteristics of the principal component and the dynamic weight of the entropy weight method to measure the real-time intensity of the node's reactive power demand. The larger the value, the higher the priority of the node for collaborative compensation.

[0035] Step S3, based on the reactive compensation priority of each node in the historical time period before the current moment, combined with the prediction algorithm, obtain the predicted value of the reactive compensation priority of each node in the future preset time period; based on the data fluctuation of the predicted value of each node in the future preset time period, construct the predicted data fluctuation rate of each node at the current moment; cluster the reactive compensation priority of all nodes at the current moment to obtain each cluster cluster, based on the reactive compensation priority and predicted data fluctuation rate of each node at the current moment, combined with the data distribution in each cluster cluster, construct the dynamic compensation urgency of each node at the current moment.

[0036] Since traditional centralized optimization methods in multi-level reactive power coordinated compensation systems rely on global data synchronous transmission, they are prone to data delays or packet loss due to communication bandwidth limitations, resulting in control inaccuracy and even oscillation risks, and it is difficult to dynamically respond to the spatiotemporal changes in the reactive power demand of the power grid.

[0037] Therefore, based on the reactive compensation priority constructed above, the reactive compensation priorities of all nodes at the same moment are used as input to the K-means clustering algorithm. The number of cluster centers is set to K = 3 (high, medium, and low compensation demand groups). Three clusters are output, and the mean of the data within each cluster is calculated. The ratio of the mean of the data within each cluster to the mean of the reactive compensation priority of all nodes within the cluster is calculated and recorded as the first ratio. The K-means clustering algorithm is a well-known technology, and the specific process is not repeated here.

[0038] It should be noted that for the clustering of reactive compensation priorities of all nodes, this application only provides one clustering method. There are many existing clustering methods, and implementers can also use other clustering algorithms to cluster the reactive compensation priorities of all nodes. This application does not make specific restrictions.

[0039] The period within one hour before the current moment is taken as the observation period corresponding to the current moment, and the reactive compensation priority of each node at each moment in the observation period is calculated. The reactive compensation priority of each node at all moments in the observation period is taken as the historical reactive compensation priority of each node, and the historical reactive compensation priority of each node is sorted in ascending order of time to obtain the historical reactive compensation priority sequence of each node at the current moment.

[0040] Furthermore, taking the current moment node i as an example, the historical reactive power compensation priority sequence of node i is used as the input of the sliding window ARIMA (Autoregressive Integrated Moving Average Model) prediction model. Preferably, in the embodiment of the present application, the window size is set to As another embodiment of the present application, the implementer can set the window size according to the actual situation. The output is the predicted value of the reactive power compensation priority of node i in the next minute, that is, 600 data are predicted.

[0041] Furthermore, the predicted data fluctuation rate of node i is calculated by the predicted value of reactive power compensation priority of node i in the next minute at the current moment , predict the dynamic trend and uncertainty of the node's reactive power demand, and provide data support for optimizing the compensation strategy in advance. Preferably, in an embodiment of the present application, the standard deviation of the reactive power compensation priority prediction value of node i at all times in the next minute is used as the predicted data volatility of node i. Preferably, in other embodiments of the present application, the average absolute deviation of the reactive power compensation priority prediction value of node i at all times in the next minute can also be used as the predicted data volatility of node i.

[0042] Furthermore, based on the above analysis, taking node i as an example, the dynamic compensation urgency of each node is constructed, and the expression is:

[0043] Where, is the dynamic compensation urgency of node i at the current moment; is the first ratio of the cluster where node i is located at the current moment; is the reactive power compensation priority of node i at the current moment; is the mean value of the data in the cluster where node i is located at the current moment; is the predicted data volatility of node i at the current moment.

[0044] It reflects the deviation of the reactive power demand intensity of the cluster from the overall average level of the power grid. The larger the value, the higher the reactive power demand intensity of the cluster node is than the global average level of the power grid, and more compensation resources need to be allocated. Quantify the real-time intensity of node reactive power demand, is the mean value of the data within the cluster where node i is located, which is used to normalize the real-time influence of the node. Represents the priority adjustment factor within the cluster. The larger the value, the more prominent the compensation priority of the node in the cluster. is the predicted volatility of node i in the next minute, reflecting the uncertainty of reactive power demand. The larger its value is, the more likely the node may experience severe fluctuations in the future, and the system needs to give priority to dealing with the potential transient process of the node.

[0045] The greater the dynamic compensation urgency, the higher the spatiotemporal urgency of coordinated compensation for this node under the combined influence of the current real-time reactive power demand intensity, the global demand priority of the region, and the future fluctuation risk, and resources need to be allocated first to suppress the risk of voltage instability.

[0046] Step S4: allocating compensation resources based on the clustering results of the dynamic compensation urgency of all nodes at the current moment, and performing closed-loop control of the compensation system.

[0047] Based on the dynamic compensation urgency, the closed-loop control of the multi-level reactive power coordinated compensation system is optimized as follows: First, the dynamic compensation urgency of all nodes in the network at the current moment is used as the input of the K-means clustering algorithm, and the number of cluster centers K is set to 3, thus obtaining three clusters.

[0048] It should be noted that for the clustering of reactive compensation priorities of all nodes, this application only provides one clustering method. There are many existing clustering methods, and implementers can also use other clustering algorithms to cluster the reactive compensation priorities of all nodes. This application does not make specific restrictions.

[0049] Then, the mean value of the intra-cluster data of each cluster is calculated and recorded as the mean dynamic compensation urgency. The three clusters are grouped according to their size. The specific grouping method is to group the cluster with the largest mean dynamic compensation urgency as the high-demand cluster, the cluster with the smallest dynamic compensation urgency as the low-demand cluster, and the remaining cluster as the medium-demand cluster. Compensation resources are allocated hierarchically according to the grouping results. The specific allocation method is to deploy static VAR generators (SVG / STATCOM / reactive compensation devices) at the terminal nodes of the distribution network, connect them to the hierarchical closed-loop control system through local controllers and sensor networks, participate in real-time data acquisition and reactive compensation control, and dynamically allocate the capacity of the reactive generators through clustering. Specifically, the high-demand cluster is allocated 70% of the capacity, and the medium and low-demand clusters are allocated 25% and 5% of the capacity respectively, thereby significantly reducing the optimization dimension.

[0050] The above process collects data from multiple power grid nodes in real time, combines PCA dimensionality reduction with the entropy weight method to dynamically calculate the reactive compensation priority of each node, and uses K-means clustering and ARIMA forecasting to assess compensation urgency. Ultimately, compensation resources are allocated tiered by urgency. Reactive power compensation adjusts the supply and demand balance of reactive power, altering the power factor and impedance characteristics of the power grid, thereby generating a chain reaction of effects on voltage, current, active power, and reactive power to optimize the power system's operating state. The system monitors the grid data matrix after reactive compensation feedback and updates the calculation process in real time based on this data matrix, forming a self-optimizing closed loop. Based on the execution results, the system continuously adjusts its control parameters (such as weights and clustering groups) to ensure that the strategy always adapts to the real-time state of the power grid. This forms a closed-loop control cycle of "perception-decision-execution-optimization-feedback," achieving precise matching of reactive power demand and efficient voltage stability while significantly reducing communication load and computational complexity.

[0051] The process diagram of obtaining reactive power compensation priority is as follows: Figure 2 shown.

[0052] In summary, the embodiment of the present application provides a stable voltage compensation control method for a microgrid cluster, which collects various grid data of each key node in the grid at each moment, and uses the matrix composed of all grid data of all nodes at each moment as the grid data matrix. The principal component of each node in the grid data matrix is obtained by the principal component analysis algorithm, the first three principal components are extracted, and the weight of each data is assigned by the entropy weight method, thereby constructing the reactive compensation priority of each node at each moment, retaining the core reactive driving characteristics and dynamically assigning parameter weights, accurately quantifying the real-time intensity of the node reactive demand, avoiding the parameter redundancy, dimensionality explosion and feature fuzziness caused by global data modeling in traditional methods, and solving the problems of insufficient quantization accuracy and poor real-time performance of traditional methods. ; Through the reactive compensation priority of nodes in the historical time period, the reactive compensation priority of nodes in the future time period is predicted, the predicted data volatility of the nodes at the current moment is analyzed, and combined with the clustering situation of the nodes at the current moment, the dynamic compensation urgency of each node at the current moment is constructed. By integrating the real-time demand intensity, cluster priority and future fluctuation risk, the spatiotemporal urgency of the node reactive demand is dynamically reflected, avoiding the rigidity of the static resource allocation strategy and the interference of communication delay on real-time control; a hierarchical resource allocation strategy is implemented according to the dynamic compensation urgency of each node, and the calculation process is updated in real time according to the power grid data matrix after reactive compensation feedback, forming a self-optimizing closed loop, improving the accuracy of reactive demand matching and the voltage stability effect, and significantly reducing the communication load and calculation complexity.

[0053] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the above descriptions are of specific embodiments of the present application. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0054] The various embodiments in this application are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0055] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A stable voltage compensation control method applied to a microgrid cluster, characterized in that: The method comprises the following steps: Collect various grid data of each key node in the grid at each moment, and use the matrix composed of all grid data of all nodes at each moment as the grid data matrix at each moment; Obtaining the principal components of each node in the power grid data matrix through a principal component analysis algorithm; assigning weights to each type of data in the power grid data matrix; and analyzing the reactive power compensation priority of each node at each moment based on the principal components and the weights assigned to each type of data; Based on the reactive compensation priority of each node in the historical time period before the current moment, combined with a prediction algorithm, a predicted value of the reactive compensation priority of each node in a future preset time period is obtained; based on the data fluctuation of the predicted value of each node in the future preset time period, a predicted data fluctuation rate of each node at the current moment is constructed; the reactive compensation priority of all nodes at the current moment is clustered to obtain clusters, and based on the reactive compensation priority and predicted data fluctuation rate of each node at the current moment, combined with the data distribution in each cluster, the dynamic compensation urgency of each node at the current moment is constructed; Based on the clustering results of the dynamic compensation urgency of all nodes at the current moment, compensation resources are allocated to perform closed-loop control of the compensation system.

2. The stable voltage compensation control method applied to a microgrid cluster according to claim 1, characterized in that: The weight distribution of each data in the power grid data matrix is specifically as follows: The i-th row vector in the power grid data matrix consists of all the power grid data of the i-th node, and the j-th column vector consists of the j-th type of power grid data of all nodes; the power grid data matrix at each moment is used as the input of the weight distribution algorithm, and the output is the weight of each type of data in the power grid data matrix at each moment.

3. The stable voltage compensation control method applied to a microgrid cluster according to claim 1, characterized in that: The process of obtaining the reactive power compensation priority of each node at each moment is as follows: Obtain the variance contribution rate of each principal component of each node; extract the first preset number of principal components of each node in the power grid data matrix to construct a principal component score matrix; based on the principal component scores and corresponding variance contribution rates of the nodes in the principal component score matrix at each moment, combined with the power grid data values of each node at each moment and the weight assigned to each data, construct the reactive compensation priority of each node at each moment.

4. The stable voltage compensation control method applied to a microgrid cluster according to claim 3, characterized in that: The expression of reactive power compensation priority is: ； Where, is the reactive compensation priority of node i at the tth acquisition time; x is the number of principal components of each node extracted from any power grid data matrix; is the pth principal component score of node i in the principal component score matrix at the tth acquisition time; is the variance contribution rate of the pth principal component of node i in the principal component score matrix at the tth acquisition time; y is the number of power grid data types; is the weight assigned to the qth type of data in the power grid data matrix at the tth acquisition moment; is the value of the qth data of node i in the power grid data matrix at the tth acquisition time.

5. The stable voltage compensation control method applied to a microgrid cluster according to claim 1, characterized in that: The process of obtaining the predicted value of the reactive power compensation priority is as follows: Obtaining the reactive power compensation priority of each node at each moment in a preset historical period at the current moment, and using the sequence of the reactive power compensation priorities of each node at all moments in the preset historical period as the historical reactive power compensation priority sequence of each node at the current moment; The historical reactive power compensation priority sequence of each node is used as the input of the prediction model to obtain the predicted value of the reactive power compensation priority of each node within a preset time period in the future.

6. The stable voltage compensation control method applied to a microgrid cluster according to claim 1, characterized in that: The predicted data fluctuation rate of each node at the current moment is: the standard deviation of the reactive power compensation priority prediction values of each node at the current moment at all moments within a preset time period in the future.

7. The stable voltage compensation control method applied to a microgrid cluster according to claim 1, characterized in that: The process of obtaining the dynamic compensation urgency of each node at the current moment is as follows: Calculate the ratio between the mean value of the intra-cluster data of each cluster and the mean value of the reactive compensation priority of all nodes in the cluster, which is recorded as a first ratio; calculate the dynamic compensation urgency of each node at the current moment based on the first ratio, the reactive compensation priority of each node at the current moment and the predicted data volatility.

8. The stable voltage compensation control method applied to a microgrid cluster according to claim 7, characterized in that: The expression of the dynamic compensation urgency is: , where is the dynamic compensation urgency of node i at the current moment; is the first ratio of the cluster where node i is located at the current moment; is the reactive power compensation priority of node i at the current moment; is the mean value of the data in the cluster where node i is located at the current moment; is the predicted data volatility of node i at the current moment.

9. The stable voltage compensation control method applied to a microgrid cluster according to claim 1, characterized in that: The allocation of compensation resources based on the clustering results of the dynamic compensation urgency of all nodes at the current moment is specifically as follows: The dynamic compensation urgency of all nodes in the current network is used as the input of the clustering algorithm to obtain three clusters; Calculate the mean of the data within each cluster and record it as the mean of dynamic compensation urgency; The cluster with the largest mean value of dynamic compensation urgency is regarded as the high-demand cluster, the cluster with the smallest dynamic compensation urgency is regarded as the low-demand cluster, and the remaining cluster is regarded as the medium-demand cluster; The capacity of the reactive generator is allocated according to the grouping result, wherein the capacity allocated to the high-demand cluster>the capacity allocated to the medium-demand cluster>the capacity allocated to the low-demand cluster.

10. The stable voltage compensation control method applied to a microgrid cluster according to claim 1, characterized in that: The method for allocating weights to each type of data in the power grid data matrix is the entropy weight method.

Images