A new energy box transformer monitoring and control intelligent power distribution system and power distribution method

Abstract

This invention relates to the field of power grid control technology, specifically to an intelligent power distribution system and method for monitoring and controlling new energy transformer substations. The system includes: a data acquisition unit for acquiring multi-dimensional operating parameters of each transformer substation unit and status perception parameters of each monitoring node, wherein the multi-dimensional operating parameters include load fluctuation coefficient and reactive power compensation; and a transformer substation clustering unit for clustering all transformer substation units based on the deviation of multi-dimensional operating parameters between different units to determine the transformer substation cluster groups. This invention, through multi-dimensional parameter monitoring and analysis of each transformer substation unit, can classify and cluster transformer substation units according to different operating states and load fluctuation characteristics. This analysis method based on multi-dimensional data and power grid topology can achieve more accurate operating state identification, thereby optimizing the management and operation of the power distribution system and improving the overall reliability and stability of the power system.

Description

Technical Field

[0001] This invention relates to the field of power grid control technology, specifically to an intelligent power distribution system and method for monitoring and controlling new energy transformer substations. Background Technology

[0002] With the transformation of the global energy structure, traditional fossil fuels are gradually being replaced by renewable energy, especially the increasing proportion of new energy sources such as wind and solar power. The volatility and uncertainty of these new energy sources pose greater challenges to the management and dispatch of the power system. The output of new energy power generation is affected by factors such as weather and seasons, resulting in large fluctuations in power load. Therefore, more accurate and flexible monitoring and control systems are needed to ensure the stability of power supply.

[0003] Currently, traditional systems typically rely on manual monitoring and basic sensor data, which cannot perform comprehensive, real-time, multi-dimensional parameter monitoring. This makes it impossible to accurately identify the operating status and load fluctuations of each transformer unit, thus affecting the effectiveness of load regulation. Furthermore, in traditional systems, load regulation usually relies on static preset schemes and lacks flexible adaptive capabilities. When power demand changes drastically or load fluctuations are large, traditional systems have difficulty adjusting in real time, which may lead to energy waste or unstable power supply.

[0004] Furthermore, traditional systems cannot effectively assess and predict the stability of load regulation, and lack early warning mechanisms based on big data and algorithm analysis. When potential regulation risks occur, traditional systems cannot take measures to correct or optimize in advance, thereby increasing the probability of failure. Moreover, during the regulation process, they are easily affected by reactive power fluctuations, equipment failures, or changes in the external environment, resulting in low energy efficiency of the power system, and may even affect the reliability and stability of the entire system. Summary of the Invention

[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent power distribution system for monitoring and controlling new energy transformer substations, comprising:

[0006] The data acquisition unit is used to acquire multi-dimensional operating parameters of each transformer unit corresponding to the new energy transformer and the status perception parameters of each monitoring node. The multi-dimensional operating parameters include load fluctuation coefficient and reactive power compensation.

[0007] The transformer substation clustering unit is used to cluster all transformer substations based on the deviation of multi-dimensional operating parameters between different transformer substations, and to determine the transformer substation cluster group; based on the installation location of each transformer substation and the node distribution in the power grid topology, the electrical distance of each monitoring node in each transformer substation is determined.

[0008] The feature extraction unit is used to determine multiple operating state blocks for any given transformer substation unit; and to determine the state stability index of each operating state block based on the state perception parameters of the monitoring nodes and the fluctuation distribution of load characteristics in different operating state blocks.

[0009] The control and evaluation unit is used to screen load-affected blocks based on the distribution of monitoring nodes in the load fluctuation direction within the neighborhood of each operating state block; determine the load control quality of each load-affected block based on the electrical distance and power characteristics of the load-affected blocks and monitoring nodes; and determine the stability of the control effect of each load-affected block based on the distribution of similarity between the same areas of each load-affected block and other different transformer units in the corresponding transformer cluster, as well as the differences between the corresponding transformers.

[0010] The transformer substation screening unit is used to determine the overall operating quality of each transformer substation based on the state stability index, load control quality, and control effect stability of each load-affected block in each transformer substation unit; and to screen out high-quality operating transformer substation units based on the overall operating quality of all transformer substation units.

[0011] The power distribution control unit is used to perform intelligent power distribution control of the new energy transformer substation cluster based on the operating parameters and control strategies of the high-quality operating transformer substation unit.

[0012] Preferably, based on the deviation of multi-dimensional operating parameters among different transformer substation units, all transformer substation units are clustered to determine transformer substation cluster groups, including:

[0013] Select a preset number of benchmark reference units, and calculate the parameter deviation value based on the multi-dimensional operating parameters of each transformer substation unit and the benchmark reference units;

[0014] Obtain the load fluctuation coefficient deviation and reactive power compensation deviation of each transformer substation unit from the benchmark reference unit, as two dimensions of the parameter deviation value;

[0015] The sum of the reference calculated values ​​corresponding to the data in each dimension is minimized as the matrix optimization objective of the parameter deviation matrix, wherein the reference calculated value is the sum of the parameter deviation value of the corresponding dimension and the deviation adjustment term matched with the parameter deviation value, wherein the deviation adjustment term is the information entropy of the parameter deviation value;

[0016] Based on the matrix optimization objective, the parameter deviation value of each dimension is updated to obtain the target deviation value, and the parameter deviation matrix is ​​constructed based on all target deviation values;

[0017] Based on the parameter deviation matrix, all transformer substation units are clustered to determine the transformer substation cluster group.

[0018] Preferably, clustering all transformer substation units based on the parameter deviation matrix to determine transformer substation cluster groups includes:

[0019] Based on the parameter deviation matrix, a clustering objective function is constructed for each transformer substation unit.

[0020] Obtain the classification matrix and dimension weights of all transformer substation units. The classification matrix includes the probability that each transformer substation unit belongs to different cluster groups, and the dimension weights include the load fluctuation coefficient and the weight ratio of reactive power compensation in the cluster.

[0021] The clustering optimization objective is to minimize the sum of the classification bias and the dimensionality calculations. The classification bias is the deviation between the classification assignment matrix and the classification indicator matrix. The classification indicator matrix includes the preliminary assignment category of each transformer unit. The dimensionality calculations are the product of the dimensional weights and the corresponding dimensional clustering objective function, plus the sum of weight adjustment terms matched with the dimensional weights.

[0022] Based on the clustering optimization objective, the classification attribution matrix, classification indicator matrix, and dimension weights are updated, and the transformer substation cluster group is determined according to the updated classification indicator matrix.

[0023] Preferably, based on the parameter deviation matrix, a clustering objective function is constructed for each transformer substation unit, including:

[0024] Based on the parameter deviation matrix, construct the parameter difference matrix among all transformer substation units;

[0025] The parameter difference matrix is ​​combined with the classification matrix and dimension weights to construct a clustering objective function for each transformer substation unit. The clustering objective function is used to characterize the parameter consistency between the transformer substation unit and other units in its cluster group.

[0026] Preferably, the electrical distance between each monitoring node in each transformer substation is determined based on the installation location of each substation unit and the node distribution in the power grid topology, including:

[0027] Obtain the topology data of the target power grid, and select key nodes where power flow converges from the topology as reference nodes;

[0028] Obtain the power flow vector of the reference node over three consecutive operating cycles to determine the average direction of the power flow.

[0029] A two-dimensional cross-sectional plane is established with the reference node as the center and in the direction perpendicular to the direction of the tidal flow average, and this plane is used as the initial virtual cross-section;

[0030] Based on the number of transmission lines connected to the reference node and the impedance characteristics of the lines, the coverage of the initial virtual cross section is expanded, and the spatial boundary range of the virtual cross section is determined. The boundary range needs to cover the radiation area of ​​the monitoring nodes of all transformer substations connected to the reference node.

[0031] The real-time operating parameters of all monitoring nodes on any transformer substation unit are obtained, wherein the real-time operating parameters include node voltage amplitude, node current amplitude, and node active power.

[0032] Calculate the impedance value from each monitoring node to the virtual cross section, and confirm the impedance value as the electrical distance to the corresponding monitoring node;

[0033] Obtain the preset standard electrical distance at the reference node, wherein the standard electrical distance is an electrical distance reference corresponding to a known impedance value;

[0034] The electrical distance corresponding to the impedance value of each monitoring node is compared with the standard electrical distance, and the deviation between the two is calculated.

[0035] If the deviation value is less than the preset deviation threshold, the electrical distance of the monitoring node is confirmed to be valid; if the deviation value is greater than or equal to the preset deviation threshold, the real-time operating parameters of the monitoring node are reacquired, and the impedance value and electrical distance are recalculated.

[0036] Preferably, the impedance value from each monitoring node to the virtual cross-section is calculated, and the impedance value is confirmed as the electrical distance to the corresponding monitoring node, including:

[0037] Based on the valid sampling data of each monitoring node, extract the node voltage amplitude and node current amplitude at the same sampling time;

[0038] Based on the equivalent impedance reference value corresponding to the spatial boundary range of the virtual cross section, and combined with the voltage and current amplitudes of the monitoring node, the initial impedance value from the monitoring node to the virtual cross section is calculated.

[0039] Preferably, based on the state-aware parameters of the monitoring nodes and the fluctuation distribution of load characteristics in different operating state blocks, the state stability index of each operating state block is determined, including:

[0040] Determine the fluctuation characteristics of the state perception parameters of all monitoring nodes in each operating state block as the state fluctuation characteristics;

[0041] Determine the fluctuation characteristics of the parameter values ​​of all monitoring nodes in the power parameter space corresponding to the load rate dimension for each operating state block, and use these as load rate fluctuation characteristics;

[0042] The state fluctuation characteristics and load rate fluctuation characteristics of all operating state blocks are arranged according to the same sorting principle. Based on the difference in the sorting order of the state fluctuation characteristics and load rate fluctuation characteristics of each operating state block, the state stability index of each operating state block is determined. The difference in sorting order is negatively correlated with the state stability index.

[0043] Preferably, load-affected blocks are selected based on the distribution of monitoring nodes along the load fluctuation direction within the neighborhood of each operating state block, including:

[0044] Traverse the neighborhood range of each operating status block along the load fluctuation direction to determine multiple monitoring node sequences; if there are other monitoring nodes in the neighborhood range of the monitoring node sequence that are distributed before the monitoring nodes of the operating status block, then the corresponding monitoring node sequence is taken as the target monitoring node sequence, and the corresponding operating status block is taken as the load impact block.

[0045] Based on the load impact segments, the electrical distances to the monitoring nodes, and the power characteristics, the load regulation quality of each load impact segment is determined, including:

[0046] For any load impact segment, determine the ratio of the average electrical distance between the monitoring nodes before the load impact segment monitoring node in the target monitoring node sequence and the average electrical distance between the monitoring nodes of the load impact segment, and normalize it as the first load feature;

[0047] Determine the parameter value of the corresponding power dimension of each monitoring node in the power parameter space as a power feature. Determine the ratio of the average power feature between the load-affected block monitoring node in the target monitoring node sequence and the corresponding previous monitoring node, and normalize it as a second load feature.

[0048] The load control quality of each load influence block is determined based on the difference between the first load characteristic and the second load characteristic of different target monitoring node sequences in each load influence block. The difference between the first load characteristic and the second load characteristic is negatively correlated with the load control quality.

[0049] Preferably, the stability of the control effect of each load-affected block is determined based on the distribution of similarity between the same areas of each load-affected block and other different transformer units in the corresponding transformer cluster, as well as the differences between the corresponding transformers, including:

[0050] Determine the parametric similarity between the same area of ​​each load-affected block and each other substation unit in the corresponding substation cluster group, as the parametric similarity of each load-affected block relative to each other substation unit;

[0051] Determine the deviation of multi-dimensional operating parameters between the transformer substation unit and each other transformer substation unit, and perform negative correlation mapping as the operating similarity of the transformer substation unit relative to each other transformer substation unit;

[0052] The parameter similarity and operational similarity of each other transformer substation unit are arranged according to the same sorting principle. The mean square error of the arrangement order of each load influence block relative to the parameter similarity and operational similarity of all other transformer substation units is calculated to determine the stability of the control effect corresponding to each load influence block. The mean square error is negatively correlated with the stability of the control effect.

[0053] A smart power distribution method for monitoring and controlling new energy transformer substations, applicable to the aforementioned smart power distribution system for monitoring and controlling new energy transformer substations, comprising:

[0054] The system acquires multi-dimensional operating parameters for each transformer unit corresponding to the new energy transformer and status perception parameters for each monitoring node. The multi-dimensional operating parameters include load fluctuation coefficient and reactive power compensation.

[0055] Based on the deviation of multi-dimensional operating parameters between different transformer substations, all transformer substations are clustered to determine the transformer substation cluster group; based on the installation location of each transformer substation and the node distribution in the power grid topology, the electrical distance of each monitoring node in each transformer substation is determined.

[0056] For any given transformer substation unit, multiple operating status blocks are determined; based on the status perception parameters of the monitoring nodes and the fluctuation distribution of load characteristics in different operating status blocks, the status stability index of each operating status block is determined.

[0057] Based on the distribution of monitoring nodes in the load fluctuation direction within the neighborhood of each operating state block, load-affected blocks are selected; based on the electrical distance and power characteristics of the load-affected blocks and monitoring nodes, the load regulation quality of each load-affected block is determined; based on the distribution of similarity between the same areas of each load-affected block and other different transformer units in the corresponding transformer cluster, as well as the differences between the corresponding transformers, the stability of the regulation effect of each load-affected block is determined.

[0058] Based on the state stability index, load control quality, and control effect stability of each load-affected block in each transformer substation unit, the overall operating quality of each transformer substation unit is determined; based on the overall operating quality of all transformer substation units, high-quality operating transformer substation units are selected.

[0059] Based on the operating parameters and control strategies of the high-quality operating transformer substations, intelligent power distribution control is carried out on the new energy transformer substation cluster.

[0060] Compared with the prior art, the beneficial effects of the present invention are:

[0061] (1) By performing multi-dimensional parameter monitoring and analysis on each transformer substation, this invention can classify and cluster transformer substations according to different operating states and load fluctuation characteristics. This analysis method based on multi-dimensional data and power grid topology can achieve more accurate identification of operating states, thereby optimizing the management and operation of the power distribution system, improving the reliability and stability of the overall power system, and identifying key factors affecting the quality of load regulation through parameters such as load fluctuation, power characteristics and electrical distance. By accurately calculating the electrical distance and electrical characteristics of each monitoring node, load regulation can be carried out more efficiently, reducing energy loss and improving energy utilization efficiency.

[0062] (2) By evaluating the stability of the control effect of load impact segmentation, this invention can identify which areas of the power grid are more stable during the control process and which areas may have unstable control effects. Through analysis based on similarity distribution, it can provide early warning of possible stability problems in the power grid, helping the power system to take adjustment measures in advance. Furthermore, by real-time monitoring and calculation of electrical distance, the system can adaptively adjust the operation mode of the power grid to ensure that it can maintain efficient and stable operation even when the load fluctuates greatly or the power demand is uneven. This adaptive capability greatly enhances the flexibility and response speed of the power system. Attached Figure Description

[0063] Figure 1 This is a schematic diagram of the overall system architecture in one embodiment of the present invention;

[0064] Figure 2 This is a schematic flowchart of the overall method in one embodiment of the present invention.

[0065] In the diagram: 1. Data acquisition unit; 2. Substation clustering unit; 3. Feature extraction unit; 4. Control and evaluation unit; 5. Substation screening unit; 6. Power distribution control unit. Detailed Implementation

[0066] 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 determined by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0067] Example 1, please refer to Figure 1 This invention provides a technical solution: an intelligent power distribution system for monitoring and controlling new energy transformer substations, comprising:

[0068] Data acquisition unit 1 is used to acquire multi-dimensional operating parameters of each transformer unit corresponding to the new energy transformer and status perception parameters of each monitoring node. Among them, the multi-dimensional operating parameters include load fluctuation coefficient and reactive power compensation.

[0069] The transformer substation clustering unit 2 is used to cluster all transformer substations based on the deviation of multi-dimensional operating parameters between different transformer substations to determine the transformer substation cluster group; and to determine the electrical distance of each monitoring node in each transformer substation based on the installation location of each transformer substation and the node distribution in the power grid topology.

[0070] Feature extraction unit 3 is used to determine multiple operating status blocks for any given transformer substation unit; and to determine the state stability index of each operating status block based on the state perception parameters of the monitoring nodes and the fluctuation distribution of load characteristics in different operating status blocks.

[0071] The regulation and evaluation unit 4 is used to screen load-affected blocks based on the distribution of monitoring nodes in the load fluctuation direction within the neighborhood of each operating state block; determine the load regulation quality of each load-affected block based on the electrical distance and power characteristics of the load-affected blocks and monitoring nodes; and determine the stability of the regulation effect of each load-affected block based on the distribution of similarity between the same areas of each load-affected block and other different transformer units in the corresponding transformer cluster, as well as the differences between the corresponding transformers.

[0072] The transformer substation screening unit 5 is used to determine the overall operating quality of each transformer substation based on the state stability index, load control quality, and control effect stability of each load-affected block in each transformer substation; and to screen out high-quality operating transformer substations based on the overall operating quality of all transformer substations.

[0073] The power distribution control unit 6 is used to perform intelligent power distribution control of the new energy transformer substation cluster based on the operating parameters and control strategies of the high-quality operating transformer substation units.

[0074] It should be noted that the system collects multi-dimensional operating parameters of the new energy transformer substation and status sensing parameters of the monitoring nodes. These multi-dimensional operating parameters include load fluctuation coefficient and reactive power compensation. The load fluctuation coefficient refers to the degree of fluctuation in power load, used to measure the rate of load change. For example, in a commercial area, the load may be very high on weekdays, but will drop significantly on holidays; this fluctuation will affect the stability of the power distribution system. Reactive energy is the electrical energy used in the power system to maintain voltage stability and transmission efficiency; it does not directly perform work. Reactive power compensation equipment (such as capacitors, synchronous motors, etc.) can reduce reactive current on the lines and improve power quality. For example, assuming a new energy transformer substation is located in an urban commercial area, this unit will periodically collect data on load fluctuation and reactive power compensation changes. For instance, during peak evening hours, the load fluctuation coefficient may be 1.5, while the daytime reactive power compensation may be 5 kVAR (kilovar). This data will be recorded in real time and fed back to the system.

[0075] Cluster analysis is performed based on the deviation of multi-dimensional operating parameters among the transformer substations to determine the substation cluster groups, and the electrical distance of monitoring nodes in each substation unit is calculated. Based on parameters such as the load fluctuation coefficient and reactive power compensation of each substation unit, it is determined which substation units have similar operating states and group them into the same group. For example, substation units with similar load fluctuation coefficients may be grouped together for subsequent joint control. Electrical distance refers to the electrical distance between the substation unit and other equipment in the power grid (such as substations, other transformer substations, etc.), usually calculated using the power grid topology and line impedance. This helps determine the energy transmission and load balance situation. For example, suppose there are multiple transformer substation units in city A. Some substation units are located in industrial areas with high load fluctuation coefficients, while others are located in residential areas with low load fluctuation coefficients. The system will use a clustering algorithm to divide the substation units in these two areas into different groups and calculate the electrical distance between these substation units and other power distribution facilities (e.g., the distance from substation 1 to the substation is 3 kilometers).

[0076] For each transformer substation unit, multiple operating status blocks are identified, and the stability of the status is evaluated based on the status perception parameters of the monitoring nodes and the load characteristic fluctuation distribution. The operating status blocks are divided into multiple status blocks according to different operating conditions (such as high load, low load, normal state, etc.). The status stability index is used to measure the stability of different operating states. For example, whether the system can maintain voltage and frequency stability when the load fluctuates greatly. For example, suppose a transformer substation unit has a near full load and large fluctuations during the day in summer; and a lower and more stable load at night in summer. The system will divide these two different states into a "high load fluctuation" status block and a "low load stability" status block, and calculate their stability indices.

[0077] Based on the distribution of monitoring nodes along the load fluctuation direction within the neighborhood of each operating state block, load-affected blocks are selected, and the quality and stability of control are evaluated. A load-affected block refers to the area where load fluctuations impact the system under a specific state block; for example, some monitoring nodes may be significantly affected under high load fluctuation conditions. Control quality measures the effectiveness of system regulation under load fluctuation conditions; for example, reactive power compensation and load shifting can help the system remain stable under high load fluctuations. The effectiveness of control strategies is evaluated to ensure long-term effectiveness and prevent degradation due to changes in the external environment (such as weather changes or fluctuations in electricity demand). For example, under high load fluctuation conditions, if the system detects that the load characteristics of some monitoring nodes deviate from the normal range, the area where these nodes are located is selected as a "load-affected block," and the control quality of this area is evaluated. For example, increasing reactive power compensation can improve the power quality of this area, ensuring that load fluctuations do not affect system stability.

[0078] Based on the state stability index, load regulation quality, and regulation effect stability of each transformer substation unit, the overall operating quality of each transformer substation unit is evaluated, and high-quality operating transformer substation units are selected. For example, based on the evaluation results of each transformer substation unit, the system selects those transformer substation units that can efficiently regulate load fluctuations and have good stability. Suppose that in city A, after evaluation, transformer substation 1 has a more stable regulation effect and higher operating quality, then transformer substation 1 will be selected as a high-quality operating unit.

[0079] Based on the operating parameters and control strategies of high-performing transformer substations, intelligent power distribution control is implemented for the new energy transformer substation cluster. For example, if high-performing transformer substation 1 is selected, the system will refer to the operating parameters of this unit (such as load fluctuation coefficient, reactive power compensation, etc.) to intelligently adjust other transformer substations. For instance, if the control strategy of transformer substation 1 is to increase reactive power compensation to stabilize voltage, the system will adjust other transformer substations accordingly to achieve the best power distribution effect for the entire new energy transformer substation cluster.

[0080] In an optional embodiment, based on the deviation of multi-dimensional operating parameters among different transformer substation units, all transformer substation units are clustered to determine transformer substation cluster groups, including:

[0081] A preset number of benchmark reference units are selected, and the parameter deviation value is calculated based on the multi-dimensional operating parameters of each transformer substation unit and the benchmark reference units.

[0082] Obtain the load fluctuation coefficient deviation and reactive power compensation deviation of each transformer substation unit from the benchmark reference unit, as two dimensions of parameter deviation data.

[0083] The sum of the reference calculated values ​​corresponding to the data in each dimension is minimized as the matrix optimization objective of the parameter deviation matrix. The reference calculated value is the sum of the parameter deviation value of the corresponding dimension and the deviation adjustment term that matches the parameter deviation value. The deviation adjustment term is the information entropy of the parameter deviation value.

[0084] The parameter deviation value of each dimension is updated based on the matrix optimization objective to obtain the objective deviation value, and a parameter deviation matrix is ​​constructed based on all objective deviation values.

[0085] All transformer substation units are clustered based on the parameter deviation matrix to determine the transformer substation cluster group.

[0086] It should be noted that some substation units need to be selected as "benchmark" units. These benchmark units will be used as comparison objects in the subsequent analysis. Then, the "parameter deviation value" is calculated by comparing the multi-dimensional operating parameters (such as load fluctuation coefficient, reactive power compensation, etc.) of each substation unit with these benchmark units. These multi-dimensional operating parameters include the load fluctuation coefficient and reactive power compensation of each substation unit, which reflect the working status of the substation unit. The deviation of each substation unit and the benchmark units in these parameters is calculated. For example, if the load fluctuation coefficient of a certain substation unit is different from that of the benchmark unit, then this difference constitutes the parameter deviation value. For example, suppose three substation units are selected as benchmark units; the load fluctuation coefficient of a certain substation unit (A) is 1.2, while the load fluctuation coefficient of the benchmark unit is 1.0, then the load fluctuation coefficient deviation value of A is 0.2.

[0087] The deviation is divided into two dimensions: load fluctuation coefficient deviation and reactive power compensation deviation. By calculating the deviation values ​​of these two dimensions respectively, the performance differences of the transformer substation units can be understood more clearly. Load fluctuation coefficient deviation: indicates the degree of deviation of the transformer substation unit in load fluctuation. Reactive power compensation deviation: indicates the degree of deviation of the transformer substation unit in reactive current compensation. For example: assuming that the load fluctuation coefficient of transformer substation unit A is 1.2, the load fluctuation coefficient of the reference unit is 1.0, and the deviation is 0.2; at the same time, the reactive power compensation of A is 5kVAR, while that of the reference unit is 6kVAR, then the deviation of reactive power compensation is 1kVAR.

[0088] Minimizing the deviations in two dimensions (load fluctuation coefficient and reactive power compensation) is equivalent to finding a comprehensive objective by optimizing the sum of the parameter deviations in each dimension. The weighted sum of these deviations constitutes the "matrix optimization objective," which aims to make the performance of the transformer substation unit closer to the benchmark unit by minimizing these deviations. The parameter deviation value in each dimension is combined with the corresponding deviation adjustment term (such as information entropy) to obtain the reference calculated value. For example, assuming the deviation adjustment term is information entropy, the calculated deviation adjustment term for the load fluctuation coefficient is 0.3, and the deviation adjustment term for the reactive power compensation is 0.5. By minimizing their sum, an optimization objective is finally obtained, which is used to adjust and optimize the parameters of the transformer substation unit.

[0089] By optimizing the target, the parameter deviation value of each transformer substation is updated. Finally, based on these updated target deviation values, a "parameter deviation matrix" is constructed, which is the dataset of deviation values ​​for all transformer substations. For example, after matrix optimization, the load fluctuation coefficient deviation (from 0.2 to 0.1) and reactive power compensation deviation (from 1kVAR to 0.8kVAR) of transformer substation A are updated. These updated deviation values ​​constitute the new parameter deviation matrix.

[0090] Based on the constructed parameter deviation matrix, a clustering algorithm (such as K-means) is used to cluster all transformer substation units. Through clustering, the transformer substation units will be divided into multiple groups. The transformer substation units in these groups will have similar performance in terms of load fluctuation coefficient and reactive power compensation, which will facilitate subsequent control and management. For example, based on the parameter deviation matrix, transformer substation units A, B, and C may be clustered into one group because their load fluctuation coefficient and reactive power compensation deviation are similar within a certain range. At the same time, transformer substation units D and E may be clustered into another group because their parameter deviation values ​​are significantly different.

[0091] In an optional embodiment, clustering all transformer substation units based on the parameter deviation matrix to determine transformer substation cluster groups includes:

[0092] Based on the parameter deviation matrix, a clustering objective function is constructed for each transformer substation unit.

[0093] Obtain the classification matrix and dimensional weights of all transformer substation units. The classification matrix includes the probability that each transformer substation unit belongs to different cluster groups, and the dimensional weights include the load fluctuation coefficient and the weight ratio of reactive power compensation in the clusters.

[0094] The clustering optimization objective is to minimize the sum of classification bias and dimensionality calculations. Classification bias is the deviation between the classification assignment matrix and the classification indicator matrix. The classification indicator matrix includes the initial classification category of each transformer unit. Dimensionality calculations are the product of the weights of each dimension and the corresponding dimensional clustering objective function, plus the sum of weight adjustment terms matched with the dimensional weights.

[0095] Based on the clustering optimization objective, the classification attribution matrix, classification indicator matrix, and dimensional weights are updated, and the box transformer cluster group is determined according to the updated classification indicator matrix.

[0096] It should be noted that the purpose of the clustering objective function is to help evaluate the performance of each transformer substation unit in different clusters, and ultimately group similar transformer substation units into one cluster through optimization. The construction of the clustering objective function usually considers two aspects: intra-cluster variability: that is, the deviation of each transformer substation unit from other units in the cluster; inter-cluster variability: that is, the deviation between different clusters. For example, suppose there is a parameter deviation matrix, which contains the deviation values ​​of load fluctuation coefficient and reactive power compensation of multiple transformer substation units. Based on these deviation values, a clustering objective function is constructed with the goal of minimizing the difference in load fluctuation coefficient and reactive power compensation of each transformer substation unit within its own cluster, while maximizing the difference between groups.

[0097] During the clustering process, it is necessary to track the category to which each transformer substation unit belongs, as well as the weights of each dimension. This process involves: a classification matrix representing the probability that each transformer substation unit belongs to different cluster groups; typically, each row in the classification matrix represents a transformer substation unit, each column represents a cluster group, and each element value of the matrix represents the probability that the transformer substation unit belongs to that cluster group; the weight ratio of each dimension (such as load fluctuation coefficient and reactive power compensation) in the clustering; these weights determine the degree of influence of each dimension on the final clustering result.

[0098] The clustering optimization objective consists of two parts: classification bias, which represents the difference between the classification attribution matrix and the classification indicator matrix; the classification indicator matrix is ​​a preliminary matrix with only one non-zero element in each row, representing the unique cluster to which the transformer unit belongs; the smaller the classification bias, the higher the matching degree between the attribution matrix and the actual category; dimensionality calculation refers to the sum of the products of the weight of each dimension and the corresponding dimension's clustering objective function, further combined with a weight adjustment term; the weight adjustment term may be a factor matched with the dimension weights, adjusting the dimension's contribution to the objective function.

[0099] Based on the clustering optimization objective, the classification assignment matrix, classification indicator matrix, and dimension weights will be iteratively updated: Classification Assignment Matrix: The assignment matrix is ​​updated according to the probability of each transformer substation unit belonging to different cluster groups; Classification Indicator Matrix: The final category of each transformer substation unit is determined based on the current classification assignment matrix; Dimension Weights: The dimension weights are adjusted according to the current clustering results, so that dimensions that have a greater impact on the clustering results receive higher weights; Through iterative updates, the cluster group to which each transformer substation unit belongs is finally determined; The updated classification indicator matrix will reflect the final category of each transformer substation unit.

[0100] In an optional embodiment, based on the parameter deviation matrix, a clustering objective function is constructed for each transformer substation unit, including:

[0101] Based on the parameter deviation matrix, construct the parameter difference matrix among all box-type transformer units.

[0102] By combining the parameter difference matrix with the classification matrix and dimensional weights, a clustering objective function is constructed for each transformer substation unit. The clustering objective function is used to characterize the parameter consistency between the transformer substation unit and other units in its cluster group.

[0103] It should be noted that a parameter deviation matrix contains the deviation values ​​of each transformer substation unit in different dimensions; this matrix reflects the changes of each transformer substation unit in various parameters; for example, for each transformer substation unit, there may be values ​​in multiple dimensions such as load fluctuation coefficient and reactive power compensation; in order to evaluate the differences between transformer substation units, the parameter difference matrix can be calculated; each element of the parameter difference matrix represents the difference between two transformer substation units in various dimensions; this difference matrix helps to understand the relationship between each transformer substation unit and other transformer substation units, thus providing a basis for subsequent clustering.

[0104] To utilize this discrepancy information for clustering, a clustering objective function needs to be constructed by combining the parameter discrepancy matrix, the classification matrix, and the dimensional weights. The classification matrix represents the probability that each transformer substation unit belongs to each cluster group, with each element representing the probability of the substation unit belonging to a particular cluster group. The dimensional weights reflect the influence of each dimension (such as load fluctuation coefficient and reactive power compensation) on the clustering results. The clustering objective function can be used to measure the parameter consistency between each transformer substation unit and other units within its cluster group. The clustering objective function aims to minimize the discrepancy between the transformer substation unit and other units within the same group, taking into account the weights of different dimensions. The clustering objective function (e.g., by minimizing classification bias and dimensional calculation sums) will be used to optimize the clustering results. It will be iteratively updated: the probability of each transformer substation unit belonging to different cluster groups will be adjusted using optimization algorithms (such as K-means or fuzzy C-means); based on the clustering results, the dimensional weights will be adjusted so that dimensions with a greater impact on the clustering effect have higher weights.

[0105] In an optional embodiment, determining the electrical distance of each monitoring node in each transformer substation unit based on the installation location of each substation unit and the node distribution in the power grid topology includes:

[0106] Obtain the topology data of the target power grid, and select key nodes where power flow converges from the topology as reference nodes.

[0107] Obtain the power flow vector of the reference node over three consecutive operating cycles to determine the average direction of the power flow.

[0108] A two-dimensional cross-sectional plane is established with the reference node as the center and along the direction perpendicular to the direction of the tidal flow average, and this plane is used as the initial virtual cross-section.

[0109] Based on the number of transmission lines connected to the reference node and the line impedance characteristics, the coverage of the initial virtual cross section is expanded, and the spatial boundary range of the virtual cross section is determined. The boundary range must cover the monitoring node radiation area of ​​all transformer substations connected to the reference node.

[0110] Obtain the real-time operating parameters of all monitoring nodes on any transformer substation unit. The real-time operating parameters include node voltage amplitude, node current amplitude, and node active power.

[0111] Calculate the impedance value from each monitoring node to the virtual cross section, and confirm the impedance value as the electrical distance to the corresponding monitoring node.

[0112] Obtain the preset standard electrical distance at the reference node, where the standard electrical distance is the electrical distance reference corresponding to the known impedance value.

[0113] The electrical distance corresponding to the impedance value of each monitoring node is compared with the standard electrical distance, and the deviation between the two is calculated.

[0114] If the deviation value is less than the preset deviation threshold, the electrical distance of the monitoring node is confirmed to be valid; if the deviation value is greater than or equal to the preset deviation threshold, the real-time operating parameters of the monitoring node are reacquired, and the impedance value and electrical distance are recalculated.

[0115] It should be noted that data needs to be extracted from the topology of the target power grid. The topology describes the connection relationships of various nodes, transmission lines, transformers, and other equipment in the power grid. The topology data of the target power grid generally includes information such as the electrical connections of nodes, load distribution, and impedance of each line. By analyzing the topology, key nodes where power flows converge can be screened. These nodes are usually the core nodes for power flow and control in the power grid and are used as reference nodes. For example, suppose there is a power grid with multiple substations, lines, and load nodes. Key nodes where power flows converge can be identified by calculating the power flow data (such as voltage and power) of each node in the power grid. For example, at a certain moment, node A is the area with the most concentrated load in the power grid, and power converges here from multiple directions, so it is selected as the reference node.

[0116] The power flow vector of a reference node represents the direction and intensity of power flowing out of that node. To more accurately analyze the power flow characteristics of the power grid, it is necessary to obtain the power flow vector of the reference node over three consecutive operating cycles and determine the average direction of the power flow based on these data. For example, suppose we observe the reference node A and the power flow vectors measured over three consecutive operating cycles are as follows: First cycle: Power flow vector F1 = (1,3) (indicating that the vector direction is the positive x-axis and positive y-axis, and the current magnitude is 1 and 3); Second cycle: Power flow vector F2 = (2,2); Third cycle: Power flow vector F3 = (3,1); By calculating the average direction of these vectors, that is, averaging over each component: Average direction = ((1+2+3) / 3, (3+2+1) / 3) = (2,2), which indicates that the average direction of the power flow is along the vector (2,2), that is, starting from the reference node, flowing along the positive x and positive y directions.

[0117] Based on the average direction of the power flow at the reference node, a two-dimensional cross-sectional plane can be established around the reference node, perpendicular to the average direction of the power flow. This means that if the vector of the average direction is (2,2), then a vertical plane can be constructed based on this, and the range of the virtual cross-section can be defined with the reference node as the center. For example, assuming the calculated average direction of the power flow is (2,2), then the vertical direction can be (−2,2) (rotated 90 degrees), and a two-dimensional plane can be established around the reference node, perpendicular to the (2,2) direction. In this way, the power flow of the reference node and its surrounding area can be represented and analyzed more effectively.

[0118] The coverage area of ​​the virtual cross-section is expanded based on the number of transmission lines connected to the reference node and the impedance characteristics of these lines. That is, the spatial boundary of the virtual cross-section can be determined based on the number of transmission lines and the electrical characteristics (such as impedance) of each line. This boundary must cover the monitoring node radiation area of ​​all transformer substations connected to the reference node. For example, assuming reference node A connects to three transmission lines with impedances of Z1=0.1Ω for line 1, Z2=0.2Ω for line 2, and Z3=0.15Ω for line 3, the expansion range of the virtual cross-section can be calculated based on the impedance characteristics of these lines. For example, the electrical distance can be calculated based on the impedance of each line (e.g., using transmission line theory or the relationship between impedance and current). These calculations help determine the spatial boundary of the virtual cross-section, ensuring it covers all relevant monitoring node areas.

[0119] During the operation of the power grid, it is necessary to obtain the real-time operating parameters of all monitoring nodes on each transformer substation unit. These parameters typically include: node voltage amplitude (representing the voltage intensity of the node), node current amplitude (representing the current intensity of the node), and node active power (representing the active load or generation capacity of the node). These real-time parameters reflect the power status of the power grid under the current operating conditions. For example, suppose there is a transformer substation unit containing 3 monitoring nodes: Node 1: voltage amplitude V1=10V, current amplitude I1=5A, active power P1=50W; Node 2: voltage amplitude V2=11V, current amplitude I2=4.5A, active power P2=49W; Node 3: voltage amplitude V3=9.5V, current amplitude I3=5.2A, active power P3=51W.

[0120] Calculate the impedance value from each monitoring node to the virtual cross section; the impedance value can be calculated from the current and voltage of the node and the impedance characteristics of the line; these impedance values ​​will be converted into the corresponding electrical distance; for example: assuming that the voltage amplitude of node 1 is V1=10V, the current amplitude is I1=5A, and the line impedance is known to be Z=0.1Ω, then the electrical distance can be calculated by the following formula: electrical distance = V1 / I1 = 10 / 5 = 2m.

[0121] The standard electrical distance of the reference node is a preset reference value, usually obtained through historical data or calculation. The calculated electrical distance of each monitoring node is compared with the standard electrical distance of the reference node to obtain the deviation value. For example, assuming the standard electrical distance of the reference node is 2 meters, and the calculated electrical distance of node 1 is 2 meters, then the deviation value is: Deviation value = |2−2| = 0. If the deviation value is greater than the preset threshold, the parameters of the monitoring node need to be reacquired and the impedance and electrical distance need to be recalculated. If the deviation value is less than the preset threshold, the electrical distance of the monitoring node is considered valid; otherwise, the monitoring data needs to be reacquired to ensure the accuracy of the calculated electrical distance. For example, if the deviation value is greater than or equal to the preset threshold (e.g., 0.1 meters), it indicates that there is an error in the electrical distance calculation, and the monitoring parameters need to be reacquired.

[0122] In an optional embodiment, the impedance value from each monitoring node to the virtual cross-section is calculated, and the impedance value is used to determine the electrical distance to the corresponding monitoring node, including:

[0123] Based on the valid sampling data of each monitoring node, the node voltage amplitude and node current amplitude at the same sampling time are extracted.

[0124] Based on the equivalent impedance reference value corresponding to the spatial boundary range of the virtual cross section, and combined with the voltage and current amplitudes of the monitoring node, the initial impedance value from the monitoring node to the virtual cross section is calculated.

[0125] It should be noted that each monitoring node typically has a series of voltage and current data. By periodically sampling, the voltage amplitude and current amplitude at the same moment can be extracted. The voltage amplitude represents the intensity of the voltage (e.g., in volts), and the current amplitude represents the intensity of the current (e.g., in amperes). For example, suppose there is a monitoring node A with the following voltage and current data: voltage amplitude VA = 10V, current amplitude IA = 5A. These data were measured at the same moment (e.g., t = 10 seconds).

[0126] A virtual cross section is an imaginary surface or region established based on the power flow direction of the reference node and the structural characteristics of the power grid. Each cross section has an equivalent impedance reference value, which is usually calculated by the power grid model and reflects the ratio between current and voltage within the cross section. For example, suppose the equivalent impedance reference value Z of the virtual cross section is 0.2Ω, which is a preset impedance reference value based on the topology and electrical characteristics of the power grid.

[0127] By monitoring the voltage and current amplitudes of the node and the equivalent impedance reference value of the virtual cross section, the initial impedance value from the monitoring node to the virtual cross section can be calculated. Usually, the impedance value is calculated using Ohm's law, with the formula: Z=V / I.

[0128] Where Z is the impedance from the node to the virtual cross section, V is the voltage amplitude of the node, and I is the current amplitude of the node. For example, based on the above data, the voltage amplitude of the monitoring node A is VA = 10V, and the current amplitude is IA = 5A. Then the initial impedance from node A to the virtual cross section is: ZA = VA / IA = 10V / 5A = 2Ω.

[0129] In actual power grid analysis, it is also necessary to compare the calculated initial impedance value with the equivalent impedance reference value of the virtual cross section. If the calculated impedance value differs significantly from the reference value, it may be necessary to adjust the operating parameters of the power grid or further analyze the electrical characteristics of the node. For example, suppose the equivalent impedance reference value of the virtual cross section is Zvirtual = 0.2Ω, while the calculated initial impedance value of node A is ZA = 2Ω. Obviously, the difference between the two is significant, which may mean that there is a large deviation between the current and voltage of node A, and further analysis of its electrical characteristics is required.

[0130] In an optional embodiment, based on the state-aware parameters of monitoring nodes and the fluctuation distribution of load characteristics in different operating state blocks, a state stability index for each operating state block is determined, including:

[0131] The fluctuation characteristics of the state perception parameters of all monitoring nodes in each running state block are determined as state fluctuation characteristics.

[0132] The fluctuation characteristics of the parameter values ​​of all monitoring nodes in each operating state block corresponding to the load rate dimension in the power parameter space are determined as the load rate fluctuation characteristics.

[0133] The state fluctuation characteristics and load rate fluctuation characteristics of all operating state blocks are arranged according to the same sorting principle. Based on the difference in the sorting order of the state fluctuation characteristics and load rate fluctuation characteristics of each operating state block, the state stability index of each operating state block is determined. The difference in sorting order is negatively correlated with the state stability index.

[0134] It should be noted that during power grid operation, each monitoring node collects parameters such as voltage, current, and power at certain time intervals. State-aware parameters typically refer to the fluctuations of these power parameters (such as voltage amplitude, current amplitude, and power) at the monitoring nodes. Statistical analysis of these fluctuation characteristics yields state fluctuation characteristics, i.e., the fluctuation pattern of the monitoring node's state parameters within a specific time period. For example, suppose that during a certain period (e.g., 1 hour), voltage and current amplitudes are recorded at monitoring nodes A, B, and C respectively. Assume the voltage amplitude fluctuations are as follows: Node A: ±1V, Node B: ±2V, Node C: ±0.5V. Then, the state fluctuation characteristics are the statistical results of these voltage amplitude fluctuations. In this example, node B exhibits the largest voltage fluctuation, and its state fluctuation characteristics are also the most severe.

[0135] Load factor refers to the ratio of actual load to maximum load during power grid operation. The load factor of each monitoring node typically reflects the load usage of that node within a certain time period. Load factor fluctuation characteristics refer to the fluctuation pattern of load factor parameters (such as active power and apparent power) of all monitoring nodes in the power grid within a certain operating state segment. For example, suppose that the load factor changes of nodes A, B, and C within a certain time period are as follows: Node A: load factor fluctuation is 80% to 120%, Node B: load factor fluctuation is 60% to 150%, and Node C: load factor fluctuation is 90% to 110%. These load factor fluctuation characteristics indicate the magnitude of load changes at different nodes within that time period. Node B has the widest load fluctuation range, and its load factor fluctuation characteristics are the most significant.

[0136] The state fluctuation characteristics and load rate fluctuation characteristics of all operating state blocks are arranged according to the same sorting principle. This is usually a standardized sorting method based on the fluctuation amplitude or change of each monitoring node. For example, suppose that in a specific operating state block, the order of state fluctuation characteristics is as follows: Node A, Node B, Node C, while the order of load rate fluctuation characteristics is as follows: Node B, Node A, Node C. By comparing the order of state fluctuation characteristics and load rate fluctuation characteristics, the difference between the two can be calculated. This difference reflects the degree of matching between state fluctuation and load rate fluctuation of different monitoring nodes. If the order is very close, it means that the operating state block is relatively stable. Conversely, if the order difference is large, the stability of the block is poor. The state stability index is obtained by calculating the "difference" between the two sortings. Generally, the smaller the sorting difference, the higher the stability of the system. Conversely, the larger the difference, the more likely the system is to be in an unstable state.

[0137] For example, the state fluctuation characteristic ranking is: Node A (maximum fluctuation), Node B, Node C; the load rate fluctuation characteristic ranking is: Node B (maximum fluctuation), Node A, Node C. The "difference" between these rankings can be calculated. For instance, the difference between the state fluctuation ranking and the load rate ranking for Node A is 1, the difference for Node B is 0, and the difference for Node C is 0. Based on these differences, the state stability index of the entire operating state block can be calculated. If the sum of these differences is small (e.g., close to 0), it indicates system stability; if the sum is large, it indicates higher system instability.

[0138] In an optional embodiment, load-affected blocks are selected based on the distribution of monitoring nodes along the load fluctuation direction within the neighborhood of each operating state block, including:

[0139] Traverse the neighborhood range of each operating status block along the load fluctuation direction to determine multiple monitoring node sequences; if there are other monitoring nodes in the neighborhood range that are distributed before the monitoring nodes of the operating status block, then the corresponding monitoring node sequence is taken as the target monitoring node sequence, and the corresponding operating status block is taken as the load impact block.

[0140] Based on the load impact segments, the electrical distance to monitoring nodes, and power characteristics, the load regulation quality of each load impact segment is determined, including:

[0141] For any load impact segment, determine the ratio of the average electrical distance between the target monitoring node sequence and other monitoring nodes before the load impact segment monitoring node, and normalize it as the first load feature.

[0142] The parameter value of the corresponding power dimension of each monitoring node in the power parameter space is determined as the power feature. The ratio of the average power feature between the load-affected block monitoring node in the target monitoring node sequence and the corresponding previous monitoring node is determined and normalized as the second load feature.

[0143] The load control quality of each load influence block is determined based on the difference between the first load characteristic and the second load characteristic of different target monitoring node sequences in each load influence block. The difference between the first load characteristic and the second load characteristic is negatively correlated with the load control quality.

[0144] It should be noted that the process involves traversing the neighborhood of each operating state block along the load fluctuation direction: First, each operating state block in the power grid is identified, and the neighborhood of each state block is determined. The neighborhood typically refers to the area associated with that state block during load fluctuations. Next, multiple monitoring node sequences are determined: Along the load fluctuation direction, the neighborhood of each operating state block is traversed along the distribution direction of the monitoring nodes, resulting in multiple monitoring node sequences. These sequences reflect the load changes of different nodes in that direction. Finally, target monitoring node sequences are selected: If, among these monitoring node sequences, there exists a monitoring node whose electrical position precedes other nodes within its neighborhood (i.e., the node is located "upstream" of other nodes), then... If the location is specified, the sequence is identified as the "target monitoring node sequence", and the operating state block is labeled as the "load impact block". For example, suppose there is an operating state block A, and there are 3 monitoring nodes in the power grid: node X, node Y and node Z. Suppose that these nodes are distributed as follows in the load fluctuation direction: node X: small load fluctuation (low fluctuation), node Y: large load fluctuation (high fluctuation), node Z: moderate load fluctuation (medium fluctuation). When traversing the load fluctuation direction, if it is found that node Y is before node X and node Z is after node Y, then the monitoring node sequence from node Y to node Z is the target monitoring node sequence, and the operating state block A is the load impact block.

[0145] For each load impact block, firstly, the average electrical distance between the target monitoring node and all previous monitoring nodes in the load impact block monitoring sequence is calculated. Electrical distance typically refers to the actual distance of power transmission lines or the "virtual distance" along the power flow path. The ratio of these average electrical distances is calculated and normalized to obtain the first load characteristic. This characteristic reflects the "tightness" of the electrical distance between nodes, i.e., the efficiency of power flow. Each monitoring node has its corresponding power parameters, such as active power and reactive power. Within the load impact block, the ratio of the average power characteristics between the target monitoring node and all previous monitoring nodes in the load impact block monitoring sequence is calculated and normalized to obtain the second load characteristic. This reflects the power of each node. The magnitude of rate change and its response to load fluctuations; for example: Suppose there are the following monitoring nodes in a load-affected block B: Node 1: electrical distance = 100 meters, power = 10MW, Node 2: electrical distance = 150 meters, power = 12MW, Node 3: electrical distance = 200 meters, power = 15MW. Calculate: Average ratio of electrical distances: Assuming the electrical distance between nodes before Node 2 is 100 meters and the electrical distance between nodes before Node 3 is 150 meters; then, the first load characteristic is the ratio of the electrical distance between Node 3 and Node 2, assumed to be 1.5 (150 meters / 100 meters); Average ratio of power characteristics: Assuming the power before Node 2 is 10MW and the power before Node 3 is 12MW, then the second load characteristic is the ratio of 12MW to 10MW, i.e., 1.2.

[0146] Load control quality is determined by the difference between the first load characteristic and the second load characteristic. Assuming the first and second load characteristics have been calculated for multiple load influence blocks, for each load influence block, the difference between its first and second load characteristics needs to be compared. If the difference between these two characteristics is large, it indicates that the control quality of that load influence block is poor; conversely, it is better. For example, suppose the calculated load characteristics for two load influence blocks are as follows: Load influence block 1: First load characteristic = 1.5, Second load characteristic = 1.2, Difference = 0.3; Load influence block 2: First load characteristic = 2.0, Second load characteristic = 1.5, Difference = 0.5. By comparison, it can be concluded that the control quality of load influence block 1 is better than that of load influence block 2 because the difference between their characteristics is smaller.

[0147] In an optional embodiment, the stability of the control effect of each load-affected block is determined based on the distribution of similarity between the same areas of each load-affected block and other different transformer units in the corresponding transformer cluster, as well as the differences between the corresponding transformers. This includes:

[0148] Determine the parametric similarity between the same area of ​​each load-affected block and each other substation unit in the corresponding substation cluster group, as the parametric similarity of each load-affected block relative to each other substation unit.

[0149] Determine the deviation of multi-dimensional operating parameters between the transformer substation unit and each other transformer substation unit, and perform negative correlation mapping as the operating similarity of the transformer substation unit relative to each other transformer substation unit.

[0150] The parameter similarity and operational similarity of each other transformer substation unit are arranged according to the same sorting principle. The mean square error of the arrangement order of each load influence block relative to the parameter similarity and operational similarity of all other transformer substation units is calculated to determine the stability of the control effect corresponding to each load influence block. The mean square error is negatively correlated with the stability of the control effect.

[0151] It should be noted that the analysis focuses on the parameter similarity between each load-affected block and its associated prefabricated substation. Each prefabricated substation has multiple operating parameters, such as voltage, current, power, and load. The analysis requires comparing the parameters within the corresponding area of ​​each load-affected block with the parameters of other prefabricated substations in that area to calculate their similarity. Specifically, assuming there is a load-affected block A located in a region of the power grid, and this region has multiple prefabricated substations B, C, and D, the similarity between each prefabricated substation B, C, D, and A is calculated using the following method. Parameter similarity: Compare the voltage, current, and other parameters of transformer substation units B, C, and D with those of substation A, using Euclidean distance or correlation coefficient to measure their similarity. If the voltage and current of A and B are similar, their parameter similarity is high. Example: Assume that substation A has a voltage of 230V and a power of 10MW with transformer substation B, while transformer substation C has a voltage of 235V and a power of 9.8MW. The similarity of their voltage and power can be calculated. If they are very similar, then the parameter similarity between substation A and transformer substations B and C is high.

[0152] After calculating the parameter similarity between each transformer substation unit, the next step is to calculate the deviation of operating parameters between different transformer substation units. The deviation reflects the difference between the operating state of each transformer substation unit and other transformer substation units. The deviation can be measured by the differences between multi-dimensional operating parameters (such as voltage, current, power, etc.). After calculating the deviation, a negative correlation mapping is performed; that is, if the deviation between two transformer substation units is large (their parameter differences are large), then their operating similarity is low; conversely, if the deviation is small (the parameter differences are small), their operating similarity is high. Example: Suppose that the voltages of transformer substations B and D are 230V and 240V respectively, and their powers are 10MW and 8MW respectively; then the deviation between these two transformer substation units can be calculated, possibly using the difference of squares formula; the greater the difference in voltage and power, the higher the deviation and the lower the operating similarity.

[0153] The parameter similarity and operational similarity of each substation unit are arranged according to a certain sorting principle; for example, parameter similarity is sorted from high to low, and operational similarity is sorted from high to low. The mean square error (MSE) between the sorting of each load impact block A and all other substation units is calculated. The mean square error is a commonly used indicator to measure the difference between two sortings. By calculating the difference in sorting between A and other substation units in terms of parameter similarity and operational similarity, the control stability of the load impact block can be evaluated. The smaller the mean square error, the higher the consistency of A's sorting with other substation units during the control process, and the more stable the control effect. The larger the mean square error, the lower the consistency of A's sorting with other substation units during the control process, and the unstable control effect. Example: Suppose that for load impact block A, the following parameter similarity sorting is obtained: Parameter similarity sorting: B>C>D, Operational similarity sorting: C>B>D.

[0154] The mean square error calculation above shows the stability of the control effect of each load-affected block. The smaller the mean square error, the better the stability of the control effect. This is because a small mean square error means that the parameters and operation of each transformer unit are more consistent, indicating that the control process is more coordinated and more stable.

[0155] Example 2, please refer to Figure 2 This invention provides a technical solution: an intelligent power distribution method for monitoring and controlling new energy transformer substations, applicable to the aforementioned intelligent power distribution system for monitoring and controlling new energy transformer substations, comprising:

[0156] S1. Obtain multi-dimensional operating parameters for each transformer unit corresponding to the new energy transformer and status perception parameters for each monitoring node. Among them, the multi-dimensional operating parameters include load fluctuation coefficient and reactive power compensation.

[0157] S2. Based on the deviation of multi-dimensional operating parameters between different transformer substations, cluster all transformer substations to determine the transformer substation cluster group; based on the installation location of each transformer substation and the node distribution in the power grid topology, determine the electrical distance of each monitoring node in each transformer substation.

[0158] S3. For any transformer substation unit, determine multiple operating status blocks; based on the status perception parameters of the monitoring nodes and the fluctuation distribution of load characteristics in different operating status blocks, determine the status stability index of each operating status block.

[0159] S4. Based on the distribution of monitoring nodes in the load fluctuation direction within the neighborhood of each operating status block, select load-affected blocks; based on the electrical distance and power characteristics of the load-affected blocks and monitoring nodes, determine the load control quality of each load-affected block; based on the distribution of similarity between the same areas of each load-affected block and other different transformer units in the corresponding transformer cluster, as well as the differences between the corresponding transformers, determine the stability of the control effect of each load-affected block.

[0160] S5. Determine the overall operating quality of each transformer substation based on the state stability index, load control quality, and control effect stability of each load-affected block in each transformer substation unit; and select high-quality operating transformer substation units based on the overall operating quality of all transformer substation units.

[0161] S6. Based on the operating parameters and control strategies of high-quality operating transformer substations, intelligent power distribution control is carried out on the new energy transformer substation cluster.

[0162] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. An intelligent power distribution system for monitoring and controlling new energy transformer substations, characterized in that, include: The data acquisition unit is used to acquire multi-dimensional operating parameters of each transformer unit corresponding to the new energy transformer and status perception parameters of each monitoring node. The multi-dimensional operating parameters include load fluctuation coefficient and reactive power compensation. The transformer substation clustering unit is used to cluster all transformer substations based on the deviation of the multi-dimensional operating parameters between different transformer substations, and to determine the transformer substation cluster group; based on the installation location of each transformer substation and the node distribution in the power grid topology, the electrical distance of each monitoring node in each transformer substation is determined. The feature extraction unit is used to determine multiple operating state blocks for any given transformer substation unit; and to determine the state stability index of each operating state block based on the state perception parameters of the monitoring nodes and the fluctuation distribution of load characteristics in different operating state blocks. The control and evaluation unit is used to screen load-affected blocks based on the distribution of monitoring nodes in the load fluctuation direction within the neighborhood of each operating state block; determine the load control quality of each load-affected block based on the electrical distance and power characteristics of the load-affected blocks and monitoring nodes; and determine the stability of the control effect of each load-affected block based on the distribution of similarity between the same areas of each load-affected block and other different transformer units in the corresponding transformer cluster, as well as the differences between the corresponding transformers. The transformer substation screening unit is used to determine the overall operating quality of each transformer substation based on the state stability index, load control quality, and control effect stability of each load-affected block in each transformer substation unit; and to screen out high-quality operating transformer substation units based on the overall operating quality of all transformer substation units. The power distribution control unit is used to perform intelligent power distribution control of the new energy transformer substation cluster based on the operating parameters and control strategies of the high-quality operating transformer substation unit.

2. The intelligent power distribution system for monitoring and controlling new energy transformer substations according to claim 1, characterized in that, The process of clustering all transformer substation units based on the deviation of multi-dimensional operating parameters between different substation units to determine substation cluster groups includes: Select a preset number of benchmark reference units, and calculate the parameter deviation value based on the multi-dimensional operating parameters of each transformer substation unit and the benchmark reference units; Obtain the load fluctuation coefficient deviation and reactive power compensation deviation of each transformer substation unit from the benchmark reference unit, as two dimensions of the parameter deviation value; The sum of the reference calculated values ​​corresponding to the data in each dimension is minimized as the matrix optimization objective of the parameter deviation matrix, wherein the reference calculated value is the sum of the parameter deviation value of the corresponding dimension and the deviation adjustment term matched with the parameter deviation value, wherein the deviation adjustment term is the information entropy of the parameter deviation value; Based on the matrix optimization objective, the parameter deviation value of each dimension is updated to obtain the target deviation value, and the parameter deviation matrix is ​​constructed based on all target deviation values; Based on the parameter deviation matrix, all transformer substation units are clustered to determine the transformer substation cluster group.

3. The intelligent power distribution system for monitoring and controlling new energy transformer substations according to claim 2, characterized in that, Based on the parameter deviation matrix, all transformer substation units are clustered to determine the transformer substation cluster groups, including: Based on the parameter deviation matrix, a clustering objective function is constructed for each transformer substation unit. Obtain the classification matrix and dimension weights of all transformer substation units. The classification matrix includes the probability that each transformer substation unit belongs to different cluster groups, and the dimension weights include the load fluctuation coefficient and the weight ratio of reactive power compensation in the cluster. The clustering optimization objective is to minimize the sum of the classification bias and the dimensionality calculations. The classification bias is the deviation between the classification assignment matrix and the classification indicator matrix. The classification indicator matrix includes the preliminary assignment category of each transformer unit. The dimensionality calculations are the product of the dimensional weights and the corresponding dimensional clustering objective function, plus the sum of weight adjustment terms matched with the dimensional weights. Based on the clustering optimization objective, the classification attribution matrix, classification indicator matrix, and dimension weights are updated, and the transformer substation cluster group is determined according to the updated classification indicator matrix.

4. The intelligent power distribution system for monitoring and controlling new energy transformer substations according to claim 3, characterized in that, Based on the parameter deviation matrix, a clustering objective function is constructed for each transformer substation unit, including: Based on the parameter deviation matrix, construct the parameter difference matrix among all transformer substation units; The parameter difference matrix is ​​combined with the classification matrix and dimension weights to construct a clustering objective function for each transformer substation unit. The clustering objective function is used to characterize the parameter consistency between the transformer substation unit and other units in its cluster group.

5. The intelligent power distribution system for monitoring and controlling new energy transformer substations according to claim 4, characterized in that, Based on the installation location of each transformer substation unit and the node distribution in the power grid topology, the electrical distance to each monitoring node in each transformer substation unit is determined, including: Obtain the topology data of the target power grid, and select key nodes where power flow converges from the topology as reference nodes; Obtain the power flow vector of the reference node over three consecutive operating cycles to determine the average direction of the power flow. A two-dimensional cross-sectional plane is established with the reference node as the center and in the direction perpendicular to the direction of the tidal flow average, and this plane is used as the initial virtual cross-section; Based on the number of transmission lines connected to the reference node and the impedance characteristics of the lines, the coverage of the initial virtual cross section is expanded, and the spatial boundary range of the virtual cross section is determined. The boundary range needs to cover the radiation area of ​​the monitoring nodes of all transformer substations connected to the reference node. The real-time operating parameters of all monitoring nodes on any transformer substation unit are obtained, wherein the real-time operating parameters include node voltage amplitude, node current amplitude, and node active power. Calculate the impedance value from each monitoring node to the virtual cross section, and confirm the impedance value as the electrical distance to the corresponding monitoring node; Obtain the preset standard electrical distance at the reference node, wherein the standard electrical distance is an electrical distance reference corresponding to a known impedance value; The electrical distance corresponding to the impedance value of each monitoring node is compared with the standard electrical distance, and the deviation between the two is calculated. If the deviation value is less than the preset deviation threshold, the electrical distance of the monitoring node is confirmed to be valid; if the deviation value is greater than or equal to the preset deviation threshold, the real-time operating parameters of the monitoring node are reacquired, and the impedance value and electrical distance are recalculated.

6. The intelligent power distribution system for monitoring and controlling new energy transformer substations according to claim 5, characterized in that, Calculate the impedance value from each monitoring node to the virtual cross-section, and confirm the impedance value as the electrical distance to the corresponding monitoring node, including: Based on the valid sampling data of each monitoring node, extract the node voltage amplitude and node current amplitude at the same sampling time; Based on the equivalent impedance reference value corresponding to the spatial boundary range of the virtual cross section, and combined with the voltage and current amplitudes of the monitoring node, the initial impedance value from the monitoring node to the virtual cross section is calculated.

7. The intelligent power distribution system for monitoring and controlling new energy transformer substations according to claim 6, characterized in that, The determination of the state stability index for each operating state block based on the state perception parameters of the monitoring nodes and the fluctuation distribution of load characteristics in different operating state blocks includes: Determine the fluctuation characteristics of the state perception parameters of all monitoring nodes in each operating state block as the state fluctuation characteristics; Determine the fluctuation characteristics of the parameter values ​​of all monitoring nodes in the power parameter space corresponding to the load rate dimension for each operating state block, and use these as load rate fluctuation characteristics; The state fluctuation characteristics and load rate fluctuation characteristics of all operating state blocks are arranged according to the same sorting principle. Based on the difference in the sorting order of the state fluctuation characteristics and load rate fluctuation characteristics of each operating state block, the state stability index of each operating state block is determined. The difference in sorting order is negatively correlated with the state stability index.

8. The intelligent power distribution system for monitoring and controlling new energy transformer substations according to claim 7, characterized in that, The process of filtering load-affected blocks based on the distribution of monitoring nodes along the load fluctuation direction within the neighborhood of each operating state block includes: Traverse the neighborhood range of each operating status block along the load fluctuation direction to determine multiple monitoring node sequences; if there are other monitoring nodes in the neighborhood range of the monitoring node sequence that are distributed before the monitoring nodes of the operating status block, then the corresponding monitoring node sequence is taken as the target monitoring node sequence, and the corresponding operating status block is taken as the load impact block. Based on the load impact segments, the electrical distances to the monitoring nodes, and the power characteristics, the load regulation quality of each load impact segment is determined, including: For any load impact segment, determine the ratio of the average electrical distance between the monitoring nodes before the load impact segment monitoring node in the target monitoring node sequence and the average electrical distance between the monitoring nodes of the load impact segment, and normalize it as the first load feature; Determine the parameter value of the corresponding power dimension of each monitoring node in the power parameter space as a power feature. Determine the ratio of the average power feature between the load-affected block monitoring node in the target monitoring node sequence and the corresponding previous monitoring node, and normalize it as a second load feature. The load control quality of each load influence block is determined based on the difference between the first load characteristic and the second load characteristic of different target monitoring node sequences in each load influence block. The difference between the first load characteristic and the second load characteristic is negatively correlated with the load control quality.

9. The intelligent power distribution system for monitoring and controlling new energy transformer substations according to claim 8, characterized in that, The determination of the stability of the control effect of each load impact block based on the distribution of similarity between the same areas of other different transformer units in each load impact block and the corresponding transformer cluster, as well as the differences between the corresponding transformers, includes: Determine the parametric similarity between the same area of ​​each load-affected block and each other substation unit in the corresponding substation cluster group, as the parametric similarity of each load-affected block relative to each other substation unit; Determine the deviation of multi-dimensional operating parameters between the transformer substation unit and each other transformer substation unit, and perform negative correlation mapping as the operating similarity of the transformer substation unit relative to each other transformer substation unit; The parameter similarity and operational similarity of each other transformer substation unit are arranged according to the same sorting principle. The mean square error of the arrangement order of each load influence block relative to the parameter similarity and operational similarity of all other transformer substation units is calculated to determine the stability of the control effect corresponding to each load influence block. The mean square error is negatively correlated with the stability of the control effect.

10. A smart power distribution method for monitoring and controlling new energy transformer substations, applicable to the smart power distribution system for monitoring and controlling new energy transformer substations as described in any one of claims 1-9, characterized in that, include: The system acquires multi-dimensional operating parameters for each transformer unit corresponding to the new energy transformer and status perception parameters for each monitoring node. The multi-dimensional operating parameters include load fluctuation coefficient and reactive power compensation. Based on the deviation of multi-dimensional operating parameters between different transformer substations, all transformer substations are clustered to determine the transformer substation cluster group; based on the installation location of each transformer substation and the node distribution in the power grid topology, the electrical distance of each monitoring node in each transformer substation is determined. For any given transformer substation unit, multiple operating status blocks are determined; based on the status perception parameters of the monitoring nodes and the fluctuation distribution of load characteristics in different operating status blocks, the status stability index of each operating status block is determined. Based on the distribution of monitoring nodes in the load fluctuation direction within the neighborhood of each operating state block, load-affected blocks are selected; based on the electrical distance and power characteristics of the load-affected blocks and monitoring nodes, the load regulation quality of each load-affected block is determined; based on the distribution of similarity between the same areas of each load-affected block and other different transformer units in the corresponding transformer cluster, as well as the differences between the corresponding transformers, the stability of the regulation effect of each load-affected block is determined. Based on the state stability index, load control quality, and control effect stability of each load-affected block in each transformer substation unit, the overall operating quality of each transformer substation unit is determined; based on the overall operating quality of all transformer substation units, high-quality operating transformer substation units are selected. Based on the operating parameters and control strategies of the high-quality operating transformer substations, intelligent power distribution control is carried out on the new energy transformer substation cluster.

Images