A power system coherent unit grouping method and system

Abstract

The application discloses a power system coherent unit grouping method and system, comprising the following steps: establishing unit power angle data of a generator unit according to simulation data and historical data of actual operation of a power system; performing dimension reduction processing on the unit power angle data, determining an initial optimal clustering number by using a first CH index, performing clustering analysis on all units, and generating an initial coherent unit grouping result; modifying the first CH index by using a penalty function considering internal connectivity of a coherent unit group, determining a target optimal clustering number by using a generated second CH index, and iteratively clustering the initial coherent unit grouping result by using the target optimal clustering number until the penalty function reaches a preset value, and generating a target coherent unit grouping result. The application fully considers internal unit connectivity of a post-isolation subsystem, and the clustering result is tested and modified by using the penalty function, so that the post-isolation island phenomenon in the internal unit of the post-isolation subsystem is effectively avoided, and the engineering practicability of the coherent unit grouping technology is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system stability analysis, and particularly relates to a power system coherent unit grouping method and system. BACKGROUND

[0002] The interconnected power system has obvious advantages in integrating resources, improving power consumption economy and power supply reliability, etc. In actual operation, such a huge power grid will inevitably be disturbed by various factors, and the complex operation mode and high coupling of the modern power system will significantly increase the probability of regional failure spreading to cause large-scale power outages. When an extreme disaster causes a large system failure, blocking the propagation of the failure, avoiding cascading reactions and preventing large-scale power outages are the primary considerations in emergency strategies.

[0003] Active splitting as a reliable guarantee for the last line of defense of the power grid plays an important role in effectively preventing the spread of system failures. Coherent unit identification as a key technology for active splitting plays a fundamental role in determining the allocation of units in the post-split subsystem and formulating splitting decisions. Reasonable coherent unit grouping can ensure the stable operation of each subsystem after splitting. However, the number of generator units in the system is too large, and directly calculating the operation data of all generator units is time-consuming and inefficient, which cannot meet the real-time and online requirements of active splitting. On the other hand, the clustering algorithm commonly used for coherent grouping only considers the operation information of the units without considering the connectivity between the units in the clustering results, which cannot meet the requirements of actual engineering. SUMMARY

[0004] The present application aims to provide a power system coherent unit grouping method and system to solve the problems of large amount of calculation, low grouping efficiency and failure to consider the connectivity between the units in the clustering results in the existing coherent unit grouping method, which cannot meet the requirements of actual engineering.

[0005] To achieve the above-mentioned purpose, the present application provides a power system coherent unit grouping method, comprising:

[0006] According to the simulation data and the actual operation history data of the power system, the unit power angle data of the generator units is established;

[0007] The unit power angle data is processed by dimension reduction, the initial optimal clustering number is determined by using the first CH index, and clustering analysis is performed on all units to generate an initial coherent unit grouping result;

[0008] The first CH index is modified by using a penalty function considering the connectivity of the coherent unit group to obtain a second CH index;

[0009] The second CH index is used to determine a target optimal cluster number, and the initial coherent generator group clustering result is iteratively clustered by using the target optimal cluster number until the penalty function reaches a preset value, and a target coherent generator group clustering result is generated.

[0010] Further, the first CH index is used to determine an initial optimal cluster number, and all generators are subjected to cluster analysis, including:

[0011] The maximum value in the first CH index value is taken as the optimal cluster number, and all generators are divided into k groups by using k-means clustering; wherein,

[0012] The first CH index is:

[0013]

[0014] In the formula, k is the number of clusters, n is the number of all data samples, tr(B k ) and tr(W k ) are the traces of the inter-class difference matrix and the intra-class difference matrix, respectively, and the values are the sums of all elements on the main diagonal lines of the respective matrices, B k and W k are the inter-class distance and the intra-class distance, and satisfy:

[0015]

[0016] In the formula, S j is a set of all data samples x allocated to the jth cluster, n j is the number of data samples in the jth cluster, μ j represents the jth cluster center, and μ is the center of all data samples.

[0017] Further, the first CH index is modified by using a penalty function considering the internal connectivity of the coherent generator group to obtain a second CH index, including:

[0018] CH'(k) = CH(k) - α(β1+β2+…+β k );

[0019] In the formula, α(β1+β2+…+β k ) is the penalty function, α is the penalty coefficient, β k is the state coefficient, and the value is 0 or 1, which is determined according to the connectivity of all generators in the kth group after clustering.

[0020] Further, the initial coherent generator group clustering result is iteratively clustered by using the target optimal cluster number until the penalty function reaches a preset value, and a target coherent generator group clustering result is generated, including:

[0021] The initial coherent unit clustering result is subjected to connectivity test and target optimal cluster number update by k-means clustering, and the target coherent unit clustering result is generated after iterative clustering until the state coefficient values of all unit groups are 0.

[0022] Further, the unit power angle data is replaced by phase-locked angle data:

[0023] x w (kT)=δ pll (kT)-δ pll (0)k=0,1,2,…;

[0024] In the formula, x w (kT) is the wind turbine power angle increment, δ pll (kT) is the phase-locked angle data collected in the kth time interval.

[0025] Further, before the dimensionality reduction processing of the unit power angle data, further comprising:

[0026] The unit power angle data is subjected to data denoising, abnormal data cleaning, missing data completion and difference calculation processing.

[0027] Further, the dimensionality reduction processing of the unit power angle data comprises:

[0028] The unit power angle data is subjected to time-frequency analysis to determine a unit power angle time-frequency feature data set;

[0029] The unit power angle time-frequency feature data set is subjected to dimensionality reduction processing by principal component analysis.

[0030] Further, the time-frequency analysis of the unit power angle data comprises:

[0031] The unit power angle data is subjected to discrete S transform to form a time-frequency feature modulus matrix:

[0032]

[0033] In the formula, is a time-frequency feature modulus matrix, N is the number of samples, is the frequency, is the Fourier transform function of the signal x(kT).

[0034] Further, the power system simulation data is steady-state calculation simulation data generated according to various analysis software capable of power system steady-state flow calculation;

[0035] The actual operation history data is historical steady-state flow data recorded by actual operation under multiple grid structure and multiple operation modes; wherein,

[0036] The plurality of grid structures include normal, maintenance, newly-operated line and main transformer, and all grid structures under the operation scheduling considering topology control;

[0037] The plurality of operation modes include a plurality of power flow operation modes under the parameters of conventional generator units, new energy units, loads and direct current transmission power.

[0038] The application also provides a power system homologous unit group division system, comprising:

[0039] A sample data construction unit is configured to establish unit power angle data of generator units according to power system simulation data and historical data of actual operation;

[0040] A dimension reduction processing unit is configured to perform dimension reduction processing on the unit power angle data, determine an initial optimal cluster number by using a first CH index, and perform cluster analysis on all units to generate an initial homologous unit group division result;

[0041] An index correction unit is configured to modify the first CH index by using a penalty function considering the internal connectivity of the homologous unit group to obtain a second CH index;

[0042] A unit group division unit is configured to determine a target optimal cluster number by using the second CH index, and iteratively cluster the initial homologous unit group division result by using the target optimal cluster number until the penalty function reaches a preset value to generate a target homologous unit group division result.

[0043] Compared with the prior art, the application has the following beneficial effects:

[0044] The power system homologous unit group division method provided by the application can effectively reduce the calculation amount of homologous unit group identification by performing dimension reduction operation on sample data. By considering the requirement that the units allocated to the same group in the clustering result need to satisfy the connectivity, a penalty function is added to test and modify the clustering result, effectively avoiding the existence of islands in the internal subsystem after clustering, and improving the engineering practicability of the homologous unit group division technology. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the application, the following will briefly introduce the drawings needed in the embodiments. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0046] Figure 1 is a flowchart of the power system homologous unit group division method provided by an embodiment of the application;

[0047] Figure 2 is a step procedure block diagram of a power system coherent generating unit grouping method provided by some embodiments of the present application;

[0048] Figure 3 is a schematic diagram of an IEEE39 node system connection provided by some embodiments of the present application;

[0049] Figure 4 is a schematic diagram of a system connection in which only generator node connection relationships are reserved provided by some embodiments of the present application;

[0050] Figure 5 is a schematic diagram of a generator group power angle increment curve after a PSDEdit platform simulation system fault provided by some embodiments of the present application;

[0051] Figure 6 is a schematic diagram of a coherent generating unit grouping result obtained by clustering in an initial state provided by some embodiments of the present application;

[0052] Figure 7 is a schematic diagram of a coherent generating unit grouping result obtained after modifying generator connection relationships provided by some embodiments of the present application;

[0053] Figure 8 is a schematic diagram of a structure of a power system coherent generating unit grouping system provided by some embodiments of the present application. DETAILED DESCRIPTION

[0054] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work under the premise that the embodiments in the present application are within the scope of protection of the present application.

[0055] It should be understood that the step numbers used herein are only for the convenience of description, and are not limited to the execution sequence of the steps.

[0056] It should be understood that the terms used in the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the present application specification and the appended claims, unless otherwise clear from the context, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0057] The terms "comprise" and "include" indicate the presence of described features, integers, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or sets thereof.

[0058] The term "and / or" means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0059] Please refer to Figure 1 The embodiment of the present application provides a power system synchronous unit grouping method. As shown in the figure, the power system synchronous unit grouping method comprises steps S10 to S40. Each step is as follows: Figure 1

[0060] S10, according to power system simulation data and actual operation history data, establishing the unit power angle data of the generator set.

[0061] In this step, first, the target power grid structure and the target power grid operation mode data are used, combined with the simulation data generated by the power system analysis and calculation model and the actual operation history data of the power grid, to form the unit power angle data.

[0062] In a specific embodiment, the plurality of grid structures includes normal, maintenance, newly-operated lines and main transformers, and all grid structures under the operation scheduling condition considering topology control; the plurality of operation modes includes a plurality of power flow operation modes under the parameters of conventional generator sets, new energy units, loads and direct current transmission power, such as winter large, winter small, summer large, summer small, early year mode, etc.

[0063] The power system simulation data is steady-state calculation simulation data generated according to various analysis software capable of performing power system steady-state power flow calculation, such as simulation data of steady-state calculation by PSASP, PSD-BPA, PSS / E and the like;

[0064] The actual operation history data is the historical steady-state power flow data recorded under the actual operation of the plurality of grid structures and the plurality of operation modes.

[0065] As a preferred embodiment, after obtaining the unit power angle data, in order to improve the data quality, the data is also subjected to preprocessing operation, including data denoising, abnormal data cleaning, missing data completion and difference calculation processing of the unit power angle data. The specific data denoising method, abnormal data cleaning method, missing data completion method and difference calculation processing method can be selected according to the actual situation, and the present embodiment is not limited in any way.

[0066] Specifically, the unit power angle data is:

[0067] x i (kT)=δ i (kT)-δ i (0)k=0,1,2,…;

[0068] In the formula, T is the sampling time interval, x​i (kT) is the i-th unit power angle increment data, δ i (kT) represents the value of the i-th unit power angle collected in the k-th time interval.

[0069] As a preferred embodiment, the unit power angle data is equivalently replaced by the phase-locked angle data to obtain:

[0070] x w (kT) = δ pll (kT) - δ pll (k) k = 0, 1, 2, …;

[0071] In the formula, x w (kT) is the wind turbine power angle increment, δ pll (kT) is the phase-locked angle data collected in the k-th time interval. It should be noted that the difference between the phase-locked angle and the phase angle of the voltage at the grid connection point of the wind farm reflects the degree of coherence between the wind turbine and the AC power grid, and has the same physical meaning as the conventional unit power angle.

[0072] S20, dimensionally process the unit power angle data, determine the initial optimal cluster number using the first CH index, and perform cluster analysis on all units to generate an initial coherent unit grouping result.

[0073] In a specific embodiment, the dimensionally processing the unit power angle data comprises the following steps:

[0074] 1) Perform time-frequency analysis on the unit power angle data to determine the unit power angle time-frequency feature data set.

[0075] In this step, when performing time-frequency analysis, specifically, discrete S transform is performed on the unit power angle data to form a time-frequency feature modulus value matrix:

[0076]

[0077] In the formula, is the time-frequency feature modulus value matrix, N is the number of samples, is the frequency, is the Fourier transform function of the signal x(kT).

[0078] Then, the unit power angle time-frequency feature data set is determined according to the time-frequency feature modulus value matrix.

[0079] 2) Dimensionally process the unit power angle time-frequency feature data set using principal component analysis.

[0080] In this step, the principal component analysis (PCA) is used for dimension reduction processing, specifically, after the unit time-frequency feature data set is processed by the steps of decentering, calculating the covariance, calculating the eigenvalues and eigenvectors of the covariance matrix, etc., the eigenvalues are sorted from large to small, and the information possession ratio is determined according to the needs:

[0081]

[0082] Then the first k largest eigenvalues are selected, where λ i is the i-th eigenvalue of the covariance matrix, and the corresponding k eigenvectors are respectively taken as row vectors to obtain the eigenvector matrix P, and the new k-dimensional data set is X' = PX.

[0083] Further, the first CH index is used to determine the initial optimal cluster number, and clustering analysis is performed on all units, including:

[0084] The maximum value in the first CH index value is taken as the optimal cluster number, and all units are divided into k groups by k-means clustering; wherein,

[0085] The first CH index is:

[0086]

[0087] In the formula, k is the number of clusters, n is the number of all data samples, tr(B k ) and tr(W k ) are the traces of the inter-class difference matrix and the intra-class difference matrix respectively, and the values are the sums of all elements on the main diagonal of the respective matrices, B k and W k are the inter-class distance and the intra-class distance, and satisfy:

[0088]

[0089] In the formula, S j is a set of all data samples x allocated to the j-th group of clusters, n j is the number of data samples in the j-th group of clusters, μ j represents the j-th cluster center, and μ is the center of all data samples.

[0090] S30, the first CH index is modified by using a penalty function considering the internal connectivity of the coherent group to obtain a second CH index.

[0091] Specifically, the second CH index in this embodiment is:

[0092] CH'(k) = CH(k) - α(β1+β2+…+β k );

[0093] In the formula, a (β1+β2+…+β k ) is a penalty function, a is a penalty coefficient, and the value is given by a person; β k is a state coefficient with a value of 0 or 1, and the value is determined according to the connectivity of all generator groups in the kth cluster after clustering. If there is a generator in the cluster that is not connected to other generators in the cluster, β k takes a value of 1, otherwise β k takes a value of 0.

[0094] S40, determining the target optimal clustering number by using the second CH index, and iteratively clustering the initial coherent generator group clustering result by using the target optimal clustering number until the value of the penalty function reaches a preset value, to generate a target coherent generator group clustering result.

[0095] Specifically, in this step, the target optimal clustering number is determined according to the second CH index, the connectivity of the initial coherent generator group clustering result is tested by using the k-means clustering, and the target optimal clustering number is updated by using the initial optimal clustering number. After the above steps are repeated, that is, iteratively clustered, until all the state coefficients β k take a value of 0, at which time each cluster has connectivity, and the coherent generator group clustering is completed, to generate a target coherent generator group clustering result.

[0096] Please refer to Figure 2 In a specific embodiment, the entire implementation process of the above-mentioned power system coherent generator group clustering method is given, including first collecting all generator angle information and calculating the angle increment, then performing time-frequency analysis on the angle increment, and using principal component analysis for dimension reduction, then calculating the CH index in the first clustering algorithm to determine the optimal clustering number K. After determining the optimal clustering number K, the k-means clustering algorithm is used to perform initial coherent generator group clustering, and then it is judged whether each cluster satisfies the connectivity. If not, go to the loop step, that is, calculate the value of the penalty function, to recalculate the CH index. Then return to execute the step of determining the optimal clustering number K, until each cluster satisfies the connectivity, to determine the final clustering result, and end the process.

[0097] In summary, the power system coherent generator group clustering method provided by the present application can effectively reduce the coherent cluster recognition calculation amount by performing dimension reduction operation on the sample data. By considering the requirement that the connection between each generator in the same group in the clustering result needs to satisfy the connectivity, a penalty function is added to test and modify the clustering result, effectively avoiding the existence of isolated islands in the subsystem after clustering, and improving the engineering practicability of the coherent generator group clustering technology.

[0098] For example, in order to verify the effect of the clustering method provided by the present application, a specific numerical value is substituted in an embodiment, including the following steps:

[0099] Step one: Collect the power angle information of each unit in real time, and obtain the power angle increment data by differential calculation. The IEEE 39-bus system is used for simulation analysis, and the power flow and stability calculation of the IEEE 39-bus system is carried out through the PSDEdit platform. It is assumed that the three-phase short circuit of the line connecting the 3rd node and the 18th node occurs at the moment when the fault is equal to 0s, and the fault disappears after 0.14s. Take the 30th generator node as the reference node, and output the power angle data of the remaining 9 generators within 4s from the occurrence of the fault. Among them, Figure 3 is the connection diagram of the IEEE 39-bus system, Figure 4 is the system connection diagram with only the connection relationship of the generator nodes retained; and Figure 5 is the power angle increment curve of the generator units after the simulation system fault by the PSDEdit platform.

[0100] Step two: Discrete S transform is performed on the unit power angle increment data to obtain the unit time-frequency characteristic modulus value matrix. Principal component analysis is performed on the obtained unit time-frequency characteristic modulus value matrix to remove the influence of secondary characteristics and noise and improve the calculation efficiency.

[0101] Step three: The optimal clustering number is determined by means of the CH index, and on this basis, k-means clustering is performed on all the generator units to determine the initial coherent unit grouping result. The CH clustering evaluation index is used to compare different clustering numbers, and the CH index scores of each clustering number are shown in Table 1:

[0102] Table 1 Clustering number and its CH index

[0103]

[0104] According to Table 1, the CH index takes the maximum value when k=3, so the optimal clustering number is 3. The clustering result is shown in Figure 6 . The 31st, 32nd and 39th generators are in the first group, the 33rd, 34th, 35th and 36th generators are in the second group, and the 37th and 38th generators are in the third group.

[0105] Step four: The values of the CH index under different original clustering numbers are modified by means of the penalty function, and the optimal clustering number is re-determined. All the load nodes in the system are integrated into the generator nodes, so that only the generator nodes are left in the system, and the simplified system connection diagram as shown in Figure 4 is obtained. Based on this, the connection relationship between each unit is determined. It can be seen that the unit groups obtained in step three meet the requirement of connectivity.

[0106] Step five: Iterative clustering according to the new optimal cluster number until the penalty function takes the value of 0 to ensure that all the generators in each coherent cluster meet the connectivity requirement after the system is decomposed into several subsystems.

[0107] As a comparison, the connection between generator No. 39 and generator No. 31 is disconnected. According to the previous clustering result, generator Nos. 31, 32 and 39 are divided into the same cluster, but after modifying the connection relationship, generator No. 39 has no connection with generator Nos. 31 and 32, so the connectivity checking link gets a penalty function which is not 0. After subtracting the penalty function, the CH value becomes smaller than the original one. By comparing the index scores under different cluster numbers again, the new optimal cluster number is 4, and the new clustering result is shown in Table 1. Figure 7 As can be seen, after a new round of clustering, generator No. 39 is divided into a cluster alone due to the requirement of connectivity.

[0108] The improved clustering algorithm proposed in the application is compared with the traditional clustering algorithm from the aspects of the coherent degree of the clustered cluster and the connectivity. The coherent clustering results obtained by using different feature extraction methods and clustering methods are shown in Table 2.

[0109] Table 2 Comparison of generator clustering results under different feature extraction and clustering methods

[0110]

[0111]

[0112] Using the coherent criterion The coherent clustering results of the generators obtained by the four different feature extraction and clustering methods in Table 2 are verified and compared, wherein Δδ i (t) and Δδ j (t) are the power angle increments of generators i and j respectively, T is the observation time, k s is the proportional coefficient.

[0113] The criterion value obtained by the method proposed in the embodiment is 0.8644, and the criterion values obtained by the other three methods are all 1.7918. It can be seen that the clustering result obtained by using the above method has good in-cluster coherence.

[0114] In addition, the Figure 3To compare the connectivity. The grouping result obtained by the method provided in the application shows obvious regional characteristics on the wiring diagram, and the No. 1, 2 and 3 machine groups all meet the requirement of connectivity. In the grouping result obtained by the PCA&k-means and ICA&k-means methods, the No. 38 and 39 generators lose the connected line with the No. 31 and 32 generators in the same group after being divided from the machine group, and cannot meet the engineering practice.

[0115] Referring to Figure 8 The power system coherent machine grouping system provided in some embodiments of the application further comprises:

[0116] The sample data construction unit 01 is configured to establish the machine group power angle data of the generators according to the simulation data and the historical data of actual operation of the power system;

[0117] The dimension reduction processing unit 02 is configured to perform dimension reduction processing on the machine group power angle data, determine the initial optimal cluster number by using the first CH index, and perform cluster analysis on all the machines to generate an initial coherent machine grouping result.

[0118] The index correction unit 03 is configured to modify the first CH index by using a penalty function considering the internal connectivity of the coherent machine group to obtain a second CH index.

[0119] The machine grouping unit 04 is configured to determine the target optimal cluster number by using the second CH index, and iteratively cluster the initial coherent machine grouping result by using the target optimal cluster number until the penalty function reaches a preset value to generate a target coherent machine grouping result.

[0120] It can be understood that the power system coherent machine grouping system provided in the embodiments is used to execute the power system coherent machine grouping method described in any of the above embodiments and achieve the same effect, which will not be further described here.

[0121] In the several embodiments provided in the application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There can be another division way for the actual implementation, for example, a plurality of units or pages can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual coupling or direct coupling or communication connection between the units can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0122] The units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed on multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0123] In addition, the functional units in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of hardware plus software functional unit.

[0124] The integrated unit realized in the form of software functional unit can be stored in a computer readable storage medium. The software functional unit stored in a storage medium includes a plurality of instructions for making a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor execute part of the steps of the method described in each embodiment of the present application. The foregoing storage medium includes a variety of program code storage media such as a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0125] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solution deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for grouping synchronized generator units in a power system, characterized in that, include: Based on power system simulation data and historical data from actual operation, the generator unit power angle data is established; The power angle data of the units is subjected to dimensionality reduction processing. The first CH index is used to determine the initial optimal number of clusters, and cluster analysis is performed on all units to generate the initial coherent unit clustering results. The first CH index is modified by using a penalty function that considers the internal connectivity of the coherent group to obtain the second CH index; The target optimal number of clusters is determined using the second CH index. The initial coherent group clustering results are iteratively clustered using the target optimal number of clusters until the penalty function reaches a preset value, at which point the target coherent group clustering results are generated. Specifically, k-means clustering is used to perform connectivity checks on the initial coherent group clustering results, and the initial optimal number of clusters is updated according to the target optimal number of clusters. The initial coherent group clustering results are then iteratively clustered until all group state coefficients are 0, at which point the target coherent group clustering results are generated.

2. The method for grouping synchronized generator units in a power system according to claim 1, characterized in that, The process of determining the initial optimal cluster number using the first CH index and performing cluster analysis on all units includes: The maximum value among the first CH index values ​​is taken as the optimal number of clusters, and all machine groups are divided using k-means clustering. A fleet of machines; among them The first CH indicator is: ； In the formula, The number of clusters, The total number of data samples. and are the traces of the between-class deviation matrix and the trace of the within-class deviation matrix, respectively, and their values ​​are the sum of all elements on the main diagonal of their respective matrices. and Let the distances be between classes and within classes, and satisfy the following: ； In the formula, To be assigned to the All data samples for group clustering The set constituted For the first The number of data samples in a cluster. Indicates the first Cluster centers, It serves as the center of all data samples.

3. The method for grouping synchronized generator units in a power system according to claim 2, characterized in that, The modification of the first CH index using a penalty function that considers the internal connectivity of the coherent machine group to obtain the second CH index includes: ； In the formula, For the penalty function, It is the penalty coefficient. These are state coefficients, taking values ​​of 0 or 1, depending on the order of clustering. The connectivity of all generator sets in a single generator group is determined.

4. The method for grouping synchronized generator units in a power system according to claim 1, characterized in that, The unit power angle data is equivalently replaced by phase-locked angle data: ； In the formula, This represents the power angle increment of the wind turbine. For the first Phase-locked angle data collected at time intervals.

5. The method for grouping synchronized generator units in a power system according to claim 1, characterized in that, Before performing dimensionality reduction processing on the unit power angle data, the method further includes: The unit power angle data is processed by denoising, cleaning abnormal data, filling in missing data, and differential calculation.

6. The method for grouping synchronized generator units in a power system according to claim 1, characterized in that, The dimensionality reduction processing of the unit power angle data includes: Perform time-frequency analysis on the unit's power angle data to determine the time-frequency characteristic dataset of the unit's power angle; Principal component analysis was used to reduce the dimensionality of the unit's power angle time-frequency characteristic dataset.

7. The method for grouping synchronized generator units in a power system according to claim 1, characterized in that, The power system simulation data is steady-state calculation simulation data generated by various analysis software capable of performing power system steady-state power flow calculations; The historical data from actual operation refers to historical steady-state power flow data recorded under various grid structures and operating modes; among which... The various grid structures include normal, maintenance, and newly commissioned lines and main transformers, as well as all grid structures under the operation and scheduling scenario considering topology control; The various operating modes include conventional generator sets, new energy generator sets, and various power flow operating modes under load and DC transmission power parameters.

8. A power system coordinating generator grouping system, characterized in that, include: The sample data construction unit is used to establish the generator set power angle data based on power system simulation data and historical data from actual operation; The dimensionality reduction processing unit is used to perform dimensionality reduction processing on the unit power angle data, determine the initial optimal cluster number using the first CH index, and perform cluster analysis on all units to generate the initial coherent unit clustering results. The index correction unit is used to modify the first CH index using a penalty function that takes into account the internal connectivity of the coherent group, so as to obtain the second CH index. The clustering unit is used to determine the target optimal number of clusters using the second CH index, and to iteratively cluster the initial coherent clustering results using the target optimal number of clusters until the penalty function reaches a preset value, thereby generating the target coherent clustering results. Specifically, k-means clustering is used to perform connectivity checks on the initial coherent clustering results, and the initial optimal number of clusters is updated according to the target optimal number of clusters to iteratively cluster the initial coherent clustering results until all cluster state coefficients are 0, thereby generating the target coherent clustering results.

Images