Transient stability emergency control key feature quantity mining method and device and storage medium

Abstract

The application discloses a transient stability emergency control key feature quantity mining method and device and a storage medium, and comprises the following steps: step 1: adopting different operation modes under a set expected power grid fault, electromechanical transient time domain simulation is carried out; step 2: according to the simulation result, the transient stability result is calculated, the transient mode sample data including transient stability and transient instability are formed, and an initial feature quantity set is selected; step 3: for each transient mode sample of transient instability, the leading group and the key section are combined to form a model-driven emergency control feature quantity set; step 4: the redundant and negative influence feature quantity in the initial feature quantity set is removed to obtain a data-driven emergency control feature quantity set; and step 5: the model-driven emergency control feature quantity and the data-driven emergency control feature quantity set are combined to obtain an emergency control key feature quantity set. The application is more efficient and has more complete data.

Description

Technical Field

[0001] This invention relates to the field of power system automation technology, and in particular to a method, device and storage medium for mining key feature quantities of transient stability emergency control. Background Technology

[0002] In formulating emergency control strategies for Level 2 defense standard faults as stipulated in the guidelines for power system safety and stability, it is necessary to determine key emergency control characteristics, including control sections, mode words, and critical generating units. Existing methods for identifying these key emergency control characteristics primarily rely on the experience of system professionals or model mechanisms, offering strong intuitiveness, mechanistic understanding, and interpretability.

[0003] With the construction of new power systems, power electronic equipment, mainly based on new energy power generation and ultra-high voltage direct current (UHVDC), is widely connected to the power grid, resulting in greater uncertainty, complexity, and nonlinearity in the power grid. The analytical efficiency and adaptability of traditional emergency control key feature mining methods are increasingly inadequate for the needs of power grid development, potentially leading to the omission of feature information and control strategy mismatch. Summary of the Invention

[0004] The purpose of this invention is to address the problems existing in the prior art by providing a more efficient and comprehensive method, device, and storage medium for mining key features of transient stable emergency control.

[0005] Specifically, the present invention is implemented using the following technical solutions.

[0006] In a first aspect, the present invention provides a method for mining key feature quantities of transient stable emergency control, including:

[0007] Step 1: Perform electromechanical transient time-domain simulations under different operating modes under the pre-set grid fault conditions;

[0008] Step 2: Calculate the transient stability results based on the simulation results, form transient mode sample data including transient stability and transient instability, and select feature quantities from the transient mode sample data to form an initial feature quantity set;

[0009] Step 3: For each transient mode sample of transient instability, merge its leading group of machines and key sections to form a set of model-driven emergency control features.

[0010] Step 4: Remove redundant and negative impact features from the initial feature set to obtain the data-driven emergency control feature set;

[0011] Step 5: Using the support vector machine ten-fold cross-validation method, calculate the evaluation index of the set after adding the elements of the data-driven emergency control feature set to the model-driven emergency control feature set in sequence, and take the set with the largest evaluation index as the emergency control key feature set.

[0012] Preferably, step 1 specifically includes:

[0013] Step 1.1: Based on the obtained power grid model and parameters, determine the anticipated power grid faults;

[0014] Step 1.2: Generalize the power supply and load of the basic operating mode to obtain multiple operating modes;

[0015] Step 1.3: Perform electromechanical transient time-domain simulations using different operating modes under the assumed grid fault conditions.

[0016] Preferably, step 2 specifically includes:

[0017] Step 2.1: Obtain the generator dynamic response trajectory based on the simulation results;

[0018] Step 2.2: Identify the unit's coherence based on the generator's dynamic response trajectory and divide the generator into M coherent generator groups;

[0019] Step 2.3: Divide the M coherent machine groups into leading group groups S according to different methods. k And the remaining group of machines A k The cluster set {Z} under different partitioning methods is obtained. k =(A k ,S k |k=1,2,…,2 M -1}, where A k +S k =Ω G , k represents the k-th partitioning method, Z k Ω represents the cluster under the k-th partitioning method. G Indicates M coherent machine groups;

[0020] Step 2.4: Assign cluster Z to each partitioning method k As a transient mode sample, the corresponding transient power angle stability margin η(Z) is calculated according to the Extended Equal Area Criterion (EEAC) method. k If η(Z) k If Z is less than 0, then determine Z. k For transient instability mode, the label is set to 0. If η(Z) k If Z is greater than 0, then determine Z. k To obtain transient stable mode, the label is set to 1, thus obtaining transient mode sample data;

[0021] Step 2.5: Select features from transient pattern sample data, normalize them, and use them as the initial feature set F.

[0022] Preferably, step 3 specifically includes:

[0023] Step 3.1: For transient mode samples that are transiently unstable in the transient mode sample data, obtain the characteristic quantities of their leading group of machines to form a characteristic quantity set A1;

[0024] Step 3.2: Calculate the synchronization force coefficient of each branch during the transient process after a power grid fault according to the following formula.

[0025]

[0026] In the formula: λ l (t) is the synchronizing force coefficient of branch l at time t, U l,1 (t) and U l,2 (t) represents the per-unit voltage values ​​of the beginning and end of branch l at time t; θ is the set of branches in the power grid; l (t) represents the phase angle difference between the voltages at the beginning and end of branch l at time t; X l P is the per-unit reactance of branch l; l (0) represents the active power of branch l before the fault occurred;

[0027] Step 3.3: For each transient mode sample of transient instability, select its strongly correlated branches to form the corresponding set of strongly correlated branches Ω. all ;

[0028] Step 3.4: Based on the set of strongly correlated branches Ω all The key section branches of each transient mode sample with transient instability are obtained, and the feature quantities of all key section branches are formed into a feature quantity set A2.

[0029] Step 3.5: Merge feature set A1 and feature set A2 to obtain the model-driven emergency control feature set.

[0030] Preferably, step 3.3 specifically includes:

[0031] Step 3.3.1: For each transient mode sample that experiences transient instability, obtain all branches that separate the leading group and the remaining groups, forming a branch set Ω`. all ;

[0032] Step 3.3.2: Calculate the angle difference Δδ between the center of inertia of the leading group and the center of inertia of the remaining groups at each moment according to the following formula. coi :

[0033]

[0034] Where, δ S δ A , respectively, are the inertia center angles of the leading group and the remaining group; S and A represent the sets of the leading group and the remaining group, respectively; M i M j δ represents the inertia of the i-th group of the leading group and the j-th group of the remaining groups; i δ j This represents the power angle of the i-th machine group in the leading group and the j-th machine group in the remaining group;

[0035] Step 3.3.3: From the branch set Ω` all Branches that satisfy the following conditions are selected to form the set Ω of strongly correlated branches of the transient mode sample of this transient instability. all :

[0036]

[0037] In the formula: t coi,max For Δδ coi The maximum moment; t λl,min For the transient process λ l (t) is the moment of minimum; λ l,min For λ l The minimum value of (t); λ l (t) is the synchronous force coefficient of branch l; t ε and λ ε This is a preset threshold.

[0038] Preferably, step 3.4 specifically includes:

[0039] Step 3.4.1: From the set of strongly correlated branches Ω all In the search process, the branch L with the smallest synchronization force coefficient is added to the critical section branch set Ω. int And branch L from Ω all Remove;

[0040] Step 3.4.2: Determine Ω int Check if a cut set has been formed; if so, proceed to step 3.4.4; otherwise, execute step 3.4.3.

[0041] Step 3.4.3: From Ω all The search finds the branch P that has not yet been traversed and has the smallest synchronization force coefficient during the transient process, and determines whether it is a parallel branch with branch L; if so, branch P is added to Ω. int If the condition is met, proceed to step 3.4.2; otherwise, repeat step 3.4.3 until Ω has been traversed. all If none of the branches satisfy the condition of forming a parallel branch with branch L, the process fails and exits.

[0042] Step 3.4.4: Determine whether the faulty branch and branch L are parallel transmission lines, and determine whether to add the faulty branch to Ω. int Does this result in a cut set? If both are true, then add the faulty branch to Ω. int If yes, proceed to step 3.4.5; otherwise, proceed directly to step 3.4.5.

[0043] Step 3.4.5: Ω int The set of branches corresponding to the key sections of the instability mode.

[0044] Preferably, step 4 specifically includes:

[0045] Step 4.1: Determine whether there is an intersection between the initial feature set and the model-driven emergency control feature set. If so, delete the intersection from the initial feature set to form the first set.

[0046] Step 4.2: Calculate the Pearson correlation coefficient of each element in the first set. For elements with a Pearson correlation coefficient of 1, randomly keep one and delete the rest to obtain the second set.

[0047] Step 4.3: Calculate the weight coefficients of each element in the second set according to the following formula, and delete the negative impact elements with weight coefficients less than 0 to obtain the set of data-driven emergency control features:

[0048]

[0049] In the formula, δ r Ω represents the weight coefficient of feature r in the second set Θ, μ is the weight factor, and Ω is the weight coefficient. S and Ω U S represents the set of transient mode sample data for transient instability and transient stability, where Ω is... S The number of samples, U is Ω U The number of samples, x u Represents Ω S The u-th sample, x u,nh x represents u In Ω S The nearest neighbor in x u,nm x represents u In Ω U The nearest neighbor in x v Represents Ω U The v-th sample, x v,nh x represents v In Ω U The nearest neighbor in x v,nm x represents v In Ω S The nearest neighbor, diff, is calculated using the following formula:

[0050]

[0051] In the formula, x represents a The value of the characteristic quantity r, x represents a,b The value of the feature quantity r, and K represents the vector composed of the feature quantities of all samples in the second set Θ.

[0052] Preferably, step 5 specifically includes:

[0053] Step 5.1: Add the features in the data-driven emergency control feature set to the model-driven emergency control feature set in descending order of their weight coefficients. Each addition of a feature forms a candidate set until the data-driven emergency control feature set is empty, resulting in several candidate sets.

[0054] Step 5.2: Perform 10-fold cross-validation based on support vector machine for each candidate set, calculate the F2 evaluation index based on the confusion matrix, and determine the candidate set with the largest F2 evaluation index as the final set of emergency control features.

[0055] Secondly, the present invention provides a transient stable emergency control key feature quantity mining device, including a processor and an executable program stored in a memory and capable of running on the processor, wherein the processor implements the above-mentioned method when executing the executable program.

[0056] Thirdly, the present invention provides a storage medium containing a computer-executable program, which, when executed by a computer processor, is used to perform the above-described method.

[0057] Compared with the prior art, the beneficial effects of the present invention are as follows: In response to the problem of power grid transient stability, the present invention is based on the idea of ​​joint data and model driving, giving full play to the advantages of the strong data mining capabilities of machine learning algorithms, supplementing and improving the traditional method of determining emergency control feature quantities based on models and mechanisms, avoiding the omission of key feature quantity information, and improving the efficiency and completeness of key feature quantity mining. Attached Figure Description

[0058] Figure 1 This is a flowchart illustrating the method for mining key features of transient stable emergency control provided by the present invention.

[0059] Figure 2 This is a schematic diagram of the structure of the transient stability emergency control key feature quantity mining device provided by the present invention. Detailed Implementation

[0060] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. The reference to "embodiment" herein means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0062] Example 1

[0063] This embodiment provides a method for mining key feature quantities of transient stable emergency control, such as... Figure 1 As shown, it includes the following steps 1 to 5.

[0064] Step 1: Perform electromechanical transient time-domain simulations under different operating modes under the assumed power grid fault conditions.

[0065] In practice, step 1 specifically includes:

[0066] Step 1.1: Based on the obtained grid generator model and parameters, determine the anticipated grid faults; specifically, the anticipated grid faults can be determined based on the second-level safety and stability standards of the power system safety and stability guidelines.

[0067] Step 1.2: Generalize the power supply and load of the basic operating mode to obtain multiple operating modes; specifically, the Monte Carlo sampling method can be used for generalization;

[0068] Step 1.3: Perform electromechanical transient time-domain simulations using different operating modes under the assumed grid fault conditions.

[0069] Step 2: Calculate the transient stability results based on the simulation results, form transient mode sample data including transient stability and transient instability, and select feature quantities from the transient mode sample data to form an initial feature quantity set.

[0070] In practice, step 2 specifically includes:

[0071] Step 2.1: Obtain the generator dynamic response trajectory based on the simulation results;

[0072] Step 2.2: Identify the unit's coherence based on the generator's dynamic response trajectory and divide the generator into M coherent generator groups;

[0073] Step 2.3: Divide the M coherent machine groups into leading group groups S according to different methods. k And the remaining group of machines A k The cluster set {Z} under different partitioning methods is obtained. k =(A k ,S k |k=1,2,…,2 M -1}, where A k +S k =Ω G , k represents the k-th partitioning method, Z k Ω represents the cluster under the k-th partitioning method. G Represents M coherent machine groups; there are 2 ways to partition them. M -1 type, therefore Z k There are also 2 M -1 item;

[0074] Step 2.4: Assign cluster Z to each partitioning method k As a transient mode sample, the corresponding transient power angle stability margin η(Z) is calculated according to the Extended Equal Area Criterion (EEAC) method. k If η(Z) k If Z is less than 0, then determine Z. k For transient instability mode, the label is set to 0. If η(Z) k If Z is greater than 0, then determine Z. k To obtain transient stable mode, the label is set to 1, thus obtaining transient mode sample data;

[0075] Step 2.5: Select feature quantities from the transient mode sample data, and normalize them to form the initial feature quantity set F. The selection of feature quantities can be based on scheduling operation and expert experience. Considering the different dimensions of various electrical quantities, to facilitate subsequent data processing, dimensionless and normalized processing is performed on each electrical quantity in the initial feature quantity set F, as shown in the following formula:

[0076]

[0077] Among them, y X Let X represent the normalized electrical quantity, and max and min represent the minimum and maximum values ​​in the dataset, respectively.

[0078] Step 3: For each transient mode sample of transient instability, merge the characteristic quantities of its leading group of machines and the characteristic quantities of key section branches to form a set of model-driven emergency control characteristic quantities.

[0079] In practice, step 3 specifically includes:

[0080] Step 3.1: For transient mode samples that are transiently unstable in the transient mode sample data, obtain the characteristic quantities of their leading group of machines to form a characteristic quantity set A1;

[0081] Step 3.2: Calculate the synchronization force coefficient of each branch during the transient process after a power grid fault according to the following formula.

[0082]

[0083] In the formula: λ l (t) is the synchronizing force coefficient of branch l at time t, U l,1 (t) and U l,2 (t) represents the per-unit voltage values ​​of the beginning and end of branch l at time t; For the set of branches in the power grid; μ l (t) represents the active power amplitude of branch l during the dynamic process; θ l (t) represents the phase angle difference between the voltages at the beginning and end of branch l at time t; X l P is the per-unit reactance of branch l; l (0) represents the active power of branch l before the fault occurred;

[0084] Step 3.3: For each transient mode sample of transient instability, select its strongly correlated branches to form the corresponding set of strongly correlated branches Ω. all ;

[0085] Step 3.4: Based on the set of strongly correlated branches Ω all The key section branches of each transient mode sample with transient instability are obtained, and the feature quantities of all key section branches are formed into a feature quantity set A2.

[0086] Step 3.5: Merge feature set A1 and feature set A2 to obtain the model-driven emergency control feature set. To ensure data consistency, the merged data also needs to be normalized.

[0087] In practice, step 3.3 specifically includes:

[0088] Step 3.3.1: For each transient mode sample that experiences transient instability, obtain all branches that separate the leading group and the remaining groups, forming a branch set Ω`. all ;

[0089] Step 3.3.2: Calculate the angle difference Δδ between the center of inertia of the leading group and the center of inertia of the remaining groups at each moment according to the following formula. coi :

[0090]

[0091] Where, δ S δ A , respectively, are the inertia center angles of the leading group and the remaining group; S and A represent the sets of the leading group and the remaining group, respectively; M i M j δ represents the inertia of the i-th group of the leading group and the j-th group of the remaining groups; i δ j This represents the power angle of the i-th machine group in the leading group and the j-th machine group in the remaining group;

[0092] Step 3.3.3: From the branch set Ω` all Branches that satisfy the following conditions are selected to form the set Ω of strongly correlated branches of the transient mode sample of this transient instability. all :

[0093]

[0094] In the formula: t coi,max For Δδ coi The maximum moment; t λl,min For the transient process λ l (t) is the moment of minimum; λ l,min For λ l The minimum value of (t); λ l (t) is the synchronous force coefficient of branch l; t ε and λ ε This is a preset threshold.

[0095] In practice, step 3.4 specifically includes:

[0096] Step 3.4.1: From the set of strongly correlated branches Ω all In the search process, the branch L with the smallest synchronization force coefficient is added to the critical section branch set Ω. int And branch L from Ω all Remove;

[0097] Step 3.4.2: Determine Ω int Check if a cut set has been formed; if so, proceed to step 3.4.4; otherwise, execute step 3.4.3.

[0098] Step 3.4.3: From Ω all The search finds the branch P that has not yet been traversed and has the smallest synchronization force coefficient during the transient process, and determines whether it is a parallel branch with branch L; if so, branch P is added to Ω. int If the condition is met, proceed to step 3.4.2; otherwise, repeat step 3.4.3 until Ω has been traversed. all If none of the branches satisfy the condition of forming a parallel branch with branch L, the process fails and exits.

[0099] Step 3.4.4: Determine whether the faulty branch and branch L are parallel transmission lines, and determine whether to add the faulty branch to Ω. int Does this result in a cut set? If both are true, then add the faulty branch to Ω. int If yes, proceed to step 3.4.5; otherwise, proceed directly to step 3.4.5.

[0100] Step 3.4.5: Ω int The set of branches corresponding to the key sections of the instability mode.

[0101] Step 4: Remove redundant and negative impact features from the initial feature set to obtain the data-driven emergency control feature set.

[0102] In practice, step 4 specifically includes:

[0103] Step 4.1: Determine whether there is an intersection between the initial feature set F and the model-driven emergency control feature set A. If so, then delete the intersection of the initial feature set to form the first set F1 = FF∩A;

[0104] Step 4.2: Calculate the Pearson correlation coefficient of each element in the first set. For elements with a Pearson correlation coefficient of 1, randomly keep one and delete the rest to obtain the second set F2 = F1 - F1', where F1' is the deleted element with a Pearson correlation coefficient of 1.

[0105] The formula for calculating the Pearson correlation coefficient is as follows:

[0106]

[0107] Among them, (x j ,y j Let m be m samples of random variables X and Y. This is the sample mean.

[0108] Step 4.3: Calculate the weight coefficients of each element in the second set according to the following formula, and delete the negative impact elements with weight coefficients less than 0 to obtain the set of data-driven emergency control features:

[0109]

[0110] In the formula, δ r Ω represents the weight coefficient of feature r in the second set Θ, μ is the weight factor, and Ω is the weight coefficient. S and Ω U S represents the set of transient mode sample data for transient instability and transient stability, where Ω is... S The number of samples, U is Ω UThe number of samples, x u Represents Ω S The u-th sample, x u,nh x represents u In Ω S The nearest neighbor in x u,nm x represents u In Ω U The nearest neighbor in x v Represents Ω U The v-th sample, x v,nh x represents v In Ω U The nearest neighbor in x v,nm x represents v In Ω S The nearest neighbor, diff, is calculated using the following formula:

[0111]

[0112] In the formula, x represents a The value of the characteristic quantity r, x represents a,b The value of the feature quantity r, and K represents the vector composed of the feature quantities of all samples in the second set Θ.

[0113] The larger the positive weight coefficient of a feature, the more critical its impact on classification; if the weight coefficient of a feature is less than 0, it indicates that the feature has a negative impact on classification and should be removed.

[0114] Step 5: Calculate the evaluation index of the set after adding the elements of the model-driven emergency control feature set to the data-driven emergency control feature set based on the support vector machine, and take the set with the largest evaluation index as the emergency control key feature set.

[0115] In specific implementation, step 5 includes:

[0116] Step 5.1: Add the features in the data-driven emergency control feature set to the model-driven emergency control feature set in descending order of their weight coefficients. Each addition of a feature forms a candidate set until the data-driven emergency control feature set is empty, resulting in several candidate sets.

[0117] Step 5.2: Perform 10-fold cross-validation based on support vector machine (SVM) on each candidate set, calculate the F2 evaluation index based on the confusion matrix, and determine the candidate set with the largest F2 evaluation index as the final set of emergency control features.

[0118] The F2 metric, based on the confusion matrix, is shown in Table 1.

[0119] Table 1 Confusion Matrix

[0120]

[0121]

[0122] Based on the confusion matrix, the formulas for calculating precision, recall, and F-measure are as follows:

[0123]

[0124]

[0125]

[0126] F β The index is a combination of precision and recall. β measures the relative importance of precision and recall. When β = 1, it is the standard F1 index.

[0127] Example 2

[0128] Figure 2 This is a schematic diagram of the structure of a device provided in Embodiment 3 of the present invention. This embodiment provides services for implementing the method of Embodiment 1 of the present invention. Figure 2 As shown, the device may include: a memory 301 storing a computer-executable program; a processor 302 coupled to the memory 301; the processor 302 calls the computer-executable program stored in the memory 301 to perform the steps in the method described in Embodiment 1.

[0129] Memory 301 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory. The device may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, memory 301 may be used to read and write non-removable, non-volatile magnetic media (commonly referred to as a "hard disk drive"). A program / utility having a set (at least one) of program modules may be stored, for example, in memory 301. Such program modules include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. The computer-executable program of the program modules typically performs the functions and / or methods described in the embodiments of the present invention.

[0130] The code for performing the operations of the present invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages.

[0131] The processor 302 executes various functional applications and data processing by running programs stored in the memory 301, such as implementing the method provided in Embodiment 1 of the present invention.

[0132] Example 3

[0133] This invention provides a storage medium containing a computer-executable program, which, when executed by a computer processor, is used to perform the method of Embodiment 1.

[0134] The storage medium of embodiments of the present invention may be any combination of one or more computer-readable media. A computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0135] The code for a computer-executable program that performs the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0136] Of course, the computer-executable program provided in the embodiments of the present invention is not limited to the above-described method operations, but can also perform related operations in the methods provided in any embodiment of the present invention.

Claims

1. A method for mining key feature quantities of transient stable emergency control, characterized in that, include: Step 1: Perform electromechanical transient time-domain simulations under different operating modes under the pre-set grid fault conditions; Step 2: Calculate the transient stability results based on the simulation results, form transient mode sample data including transient stability and transient instability, and select feature quantities from the transient mode sample data to form an initial feature quantity set; Step 3: For each transient mode sample of transient instability, merge its leading group of machines and key sections to form a set of model-driven emergency control features. Step 4: Remove redundant and negative impact features from the initial feature set to obtain the data-driven emergency control feature set; Step 5: Using the support vector machine ten-fold cross-validation method, calculate the evaluation index of the set after adding the elements of the data-driven emergency control feature set to the model-driven emergency control feature set in sequence, and take the set with the largest evaluation index as the emergency control key feature set. Step 3 specifically includes: Step 3.1: For transient mode samples that are transiently unstable in the transient mode sample data, obtain the characteristic quantities of their leading group of machines to form a characteristic quantity set A1; Step 3.2: Calculate the synchronization force coefficient of each branch during the transient process after a power grid fault according to the following formula. , In the formula: Let be the synchronization force coefficient of branch l at time t. and Let be the per-unit values ​​of the voltages at the beginning and end of branch l at time t; A collection of branches in a power grid The phase angle difference between the voltages at the beginning and end of branch l at time t; The per-unit reactance value of branch l; The active power of branch l before the fault occurred; Step 3.3: For each transient mode sample of transient instability, select its strongly correlated branches to form the corresponding set of strongly correlated branches Ω. all ; Step 3.4: Based on the set of strongly correlated branches Ω all The key section branches of each transient mode sample with transient instability are obtained, and the feature quantities of all key section branches are formed into a feature quantity set A2. Step 3.5: Merge feature set A1 and feature set A2 to obtain the model-driven emergency control feature set; Step 3.3 specifically includes: Step 3.3.1: For each transient mode sample that experiences transient instability, obtain all branches that separate the leading group and the remaining groups, forming a branch set Ω`. all ; Step 3.3.2: Calculate the angle difference between the center of inertia of the leading group and the center of inertia of the remaining groups at each moment according to the following formula. : ， ， , in, , , respectively, are the inertia center angles of the leading group and the remaining group; S and A represent the sets of the leading group and the remaining group, respectively; , This represents the inertia of the i-th machine group in the leading group and the j-th machine group in the remaining group; , This represents the power angle of the i-th machine group in the leading group and the j-th machine group in the remaining group; Step 3.3.3: From the branch set Ω` all Branches that satisfy the following conditions are selected to form the set Ω of strongly correlated branches of the transient mode sample of this transient instability. all : , In the formula: for The biggest moment; During the transient process The minimum moment; for The minimum value; Let be the synchronizing force coefficient of branch l at time t; and This is a preset threshold.

2. The method for mining key features of transient stable emergency control according to claim 1, characterized in that: Step 1 specifically includes: Step 1.1: Based on the obtained power grid model and parameters, determine the anticipated power grid faults; Step 1.2: Generalize the power supply and load of the basic operating mode to obtain multiple operating modes; Step 1.3: Perform electromechanical transient time-domain simulations using different operating modes under the assumed grid fault conditions.

3. The method for mining key features of transient stable emergency control according to claim 1, characterized in that: Step 2 specifically includes: Step 2.1: Obtain the generator dynamic response trajectory based on the simulation results; Step 2.2: Identify the unit's coherence based on the generator's dynamic response trajectory and divide the generator into M coherent generator groups; Step 2.3: Divide the M coherent machine groups into leading group groups S according to different methods. k And the remaining group of machines A k The cluster set {Z} under different partitioning methods is obtained. k =(A k ,S k | k=1,2,…,2 M -1}, where A k +S k =Ω G , k represents the k-th partitioning method, Z k Ω represents the cluster under the k-th partitioning method. G Indicates M coherent machine groups; Step 2.4: Assign cluster Z to each partitioning method k As a transient mode sample, the corresponding transient power angle stability margin η(Z) is calculated according to the Extended Equal Area Criterion (EEAC) method. k If η(Z) k If Z is less than 0, then determine Z. k For transient instability mode, the label is set to 0. If η(Z) k If Z is greater than 0, then determine Z. k To obtain transient stable mode, the label is set to 1, thus obtaining transient mode sample data; Step 2.5: Select features from transient pattern sample data, normalize them, and use them as the initial feature set F.

4. The method for mining key features of transient stable emergency control according to claim 1, characterized in that: Step 3.4 specifically includes: Step 3.4.1: From the set of strongly correlated branches Ω all In the search process, the branch L with the smallest synchronization force coefficient is added to the critical section branch set Ω. int And branch L from Ω all Remove; Step 3.4.2: Determine Ω int Check if a cut set has been formed; if so, proceed to step 3.4.4; otherwise, execute step 3.4.

3. Step 3.4.3: From Ω all The search finds the branch P that has not yet been traversed and has the smallest synchronization force coefficient during the transient process, and determines whether it is a parallel branch with branch L; if so, branch P is added to Ω. int If the condition is met, proceed to step 3.4.2; otherwise, repeat step 3.4.3 until Ω has been traversed. all If none of the branches satisfy the condition of forming a parallel branch with branch L, the process fails and exits. Step 3.4.4: Determine whether the faulty branch and branch L are parallel transmission lines, and determine whether to add the faulty branch to Ω. int Does this result in a cut set? If both are true, then add the faulty branch to Ω. int If yes, proceed to step 3.4.5; otherwise, proceed directly to step 3.4.

5. Step 3.4.5: Ω int The set of branches corresponding to the key sections of the instability mode.

5. The method for mining key features of transient stable emergency control according to claim 1, characterized in that: Step 4 specifically includes: Step 4.1: Determine whether there is an intersection between the initial feature set and the model-driven emergency control feature set. If so, delete the intersection from the initial feature set to form the first set; Step 4.2: Calculate the Pearson correlation coefficient of each element in the first set. For elements with a Pearson correlation coefficient of 1, randomly keep one and delete the rest to obtain the second set. Step 4.3: Calculate the weight coefficients of each element in the second set according to the following formula, and delete the negative impact elements with weight coefficients less than 0 to obtain the set of data-driven emergency control features: , In the formula, Represents the second set The weighting coefficients of the feature quantity r in the middle. As a weighting factor, and S represents the set of transient mode sample data representing transient instability and transient stability. The number of samples, U is The number of samples, express The u-th sample, express exist The nearest neighbor in express exist The nearest neighbor in express The v-th sample, express exist The nearest neighbor in express exist The nearest neighbor in The calculation formula is: , In the formula, express The value of the characteristic quantity r, express The value of the characteristic quantity r, Represents the second set A vector composed of the feature values ​​of all samples.

6. The method for mining key features of transient stability emergency control according to claim 5, characterized in that: Step 5 specifically includes: Step 5.1: Add the features in the data-driven emergency control feature set to the model-driven emergency control feature set in descending order of their weight coefficients. Each addition of a feature forms a candidate set until the data-driven emergency control feature set is empty, resulting in several candidate sets. Step 5.2: Perform 10-fold cross-validation based on support vector machine for each candidate set, calculate the F2 evaluation index based on the confusion matrix, and determine the candidate set with the largest F2 evaluation index as the final set of emergency control features.

7. A device for mining key characteristic quantities for transient stable emergency control, characterized in that, The invention includes a processor and an executable program stored in a memory and executable on the processor, characterized in that: when the processor executes the executable program, it implements the method as described in any one of claims 1-6.

8. A storage medium containing a computer-executable program, characterized in that, The computer executable program, when executed by a computer processor, is used to perform the method as described in any one of claims 1-6.

Images