Method, system, device and storage medium for evaluating stability limits of a power transmission section

Abstract

The application discloses a method, system, device and storage medium for evaluating the stability limit of a power transmission section, comprising the following steps: pre-training a limit transmission capacity evaluation model of a power grid power transmission section by using a transfer learning method to obtain a TTC evaluation model based on transfer learning; performing active learning on the TTC evaluation model based on transfer learning to obtain a trained TTC evaluation model; and inputting each sample in an unlabeled sample pool into the trained TTC evaluation model to realize evaluation of the stability limit of the power transmission section. The application introduces transfer learning and active learning, and can reduce the time required for labeling new samples for updating a new scene model based on transfer learning. The application can evaluate the stability limit of the power transmission section, thereby improving the utilization efficiency of the power transmission channel, adapting to the working conditions of large-scale external transmission of new energy, and greatly reducing the demand for labeled samples and the time cost of applying the model to a new scene, and improving the efficiency.

Description

Technical Field

[0001] This invention belongs to the field of power system technology, specifically relating to a method, system, equipment, and storage medium for evaluating the stability limits of power transmission sections. Background Technology

[0002] High-proportion renewable energy power grids face numerous constraints and complex coupling mechanisms in their cascaded sections, resulting in low transmission section utilization. Factors hindering renewable energy transmission and the spatiotemporal distribution characteristics of various features are difficult to accurately identify using traditional deterministic calculation methods. High-proportion renewable energy power grids also suffer from numerous operating mode combinations and complex controllable strategies. Existing methods that implement fixed and stable quotas based on typical operating modes cannot maximize transmission channels, leading to a contradiction between renewable energy consumption demand and the efficient, safe, and stable operation of transmission channels. Furthermore, factors hindering renewable energy consumption are difficult to identify, and there is a lack of auxiliary decision-making support.

[0003] Traditionally, transmission sections are defined as several important interconnecting lines within a power system, identified by power system operation experts and dispatchers based on operational experience. However, with the continuous expansion of power grids and the strengthening of interconnections, the structure and operating conditions of power systems have become increasingly complex. Identifying sections solely based on interconnecting lines is no longer sufficient to meet the requirements for safe grid operation. Existing transmission section identification methods focus on power flow transfer, searching for a set of lines corresponding to the transferred power flow after a branch line is disconnected. This is visually represented by an increase in the current and power of the remaining lines at the section.

[0004] Most existing studies use the most severe, deterministic fault verification to determine a fixed stability limit based on typical operating conditions. However, this traditional method has the disadvantage of low transmission channel utilization efficiency and is not suitable for the large-scale transmission of new energy. Summary of the Invention

[0005] To overcome the problems in the prior art, the purpose of this invention is to provide a method, system, device and storage medium for evaluating the stability limit of a power transmission section. This method can evaluate the stability limit of a power transmission section, thereby improving the utilization efficiency of power transmission channels and adapting to the working conditions of large-scale transmission of new energy.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] A method for evaluating the stability limits of a power transmission section includes the following steps:

[0008] The transfer learning method was used to pre-train the ultimate transmission capacity assessment model of the power grid transmission section, and a transfer learning-based TTC assessment model was obtained.

[0009] The TTC evaluation model based on transfer learning is actively learned to obtain the trained TTC evaluation model;

[0010] Each sample in the unlabeled sample pool is input into the trained TTC evaluation model to achieve the stability limit assessment of the transmission section. The unlabeled sample pool is determined through the following process: under the new scenario, the power generation load is fluctuated and the power flow is calculated through Monte Carlo simulation, and the power flow constitutes the unlabeled sample pool.

[0011] Furthermore, the model for estimating the ultimate transmission capacity of the power grid transmission section is as follows:

[0012] P TTC =Max(ξ(c,s,z))

[0013] stG(c,s,z)=0,H(c,s,z)≤0

[0014] Among them, P TTC ξ represents the transmission capacity limit of the transmission section, c represents the power flow value on the section, and the control variables are the generator terminal voltage of the PV node, the active power of the generator, the reactive power of the generator at the PQ node, or the active power of the load; s represents the state variables, which are the line power flow, the reactive power of the generator at the PV node, or the generator terminal voltage at the PQ node; z represents the grid topology variables, G represents the power flow equation constraint, and H represents the safety constraint.

[0015] Furthermore, safety constraints include cross-sectional thermal stability constraints and transient power angle stability constraints.

[0016] Furthermore, the cross-sectional thermal stability constraint is as follows:

[0017]

[0018] in, For line I after a fault τ occurs k The current, I lkmax For Line I k Current limit.

[0019] Furthermore, the transient work angle stability constraint is:

[0020] |δ Gp (t)-δ Gq (t)|≤δ max

[0021] Where, δ Gp (t), δ Gq (t) represents the power angles of generators p and q at time t, and δ represents the power angles of generators p and q at time t. max This is the preset maximum power angle difference.

[0022] Furthermore, the objective function for the optimization of the power grid transmission section limit transmission capacity model is:

[0023]

[0024] Where h is the target domain D in the new scenario. t Source Domain D under historical operating scenarios s The model obtained by pre-training through sample co-transfer learning and These represent the empirical errors of model h in historical and new scenarios, respectively. This represents the empirical error weight value.

[0025] Furthermore, the error weight value Calculated using the following formula:

[0026]

[0027] Where d is the dimension of the hypothesis space of model h, A and β are constants, and m t =βm is the number of labeled samples in the target domain, and m is the size of the labeled sample set.

[0028] Furthermore, each sample in the unlabeled sample pool is represented as X. i =[P G Q G V G ,θ G ,P L Q L ,P AC ], where P G Q G V G ,θ G This represents the generator's active and reactive power, voltage, and phase angle, P. L Q L P represents the active and reactive power of the load. AC This indicates the initial active power flow on the AC line.

[0029] A system for evaluating the stability limits of a power transmission section includes:

[0030] The transfer learning module is used to transfer the ultimate transmission capacity assessment model of the power grid transmission section to obtain the TTC assessment model based on transfer learning.

[0031] The active learning module is used to actively learn the TTC evaluation model based on transfer learning to obtain the trained TTC evaluation model.

[0032] The evaluation module is used to input each sample in the unlabeled sample pool into the trained TTC evaluation model to achieve the stability limit evaluation of the transmission section. The unlabeled sample pool is determined through the following process: under the new scenario, the power generation load is fluctuated and the power flow is calculated through Monte Carlo simulation, and the power flow constitutes the unlabeled sample pool.

[0033] Furthermore, the model for evaluating the ultimate transmission capacity of the power grid transmission section is as follows:

[0034] P TTC =Max(ξ(c,s,z))

[0035] stG(c,s,z)=0,H(c,s,z)≤0

[0036] Among them, P TTC ξ represents the transmission capacity limit of the transmission section, c represents the power flow value on the section, and the control variables are the generator terminal voltage of the PV node, the active power of the generator, the reactive power of the generator at the PQ node, or the active power of the load; s represents the state variables, which are the line power flow, the reactive power of the generator at the PV node, or the generator terminal voltage at the PQ node; z represents the grid topology variables, G represents the power flow equation constraint, and H represents the safety constraint.

[0037] Furthermore, safety constraints include cross-sectional thermal stability constraints and transient power angle stability constraints.

[0038] Furthermore, the cross-sectional thermal stability constraint is as follows:

[0039]

[0040] in, For line I after a fault τ occurs k The current, I lkmax For Line I k Current limit.

[0041] A computer device includes a memory and a processor, the memory storing a computer program executable on the processor, the computer program, when executed by the processor, implementing the method for assessing the stability limits of a power transmission section as described above.

[0042] A computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the method for assessing the stability limits of a power transmission section as described above.

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

[0044] This invention provides a method for assessing the transmission limit of transmission sections by analyzing the correlation between correlation indicators of cascaded sections in a regional power grid with a high proportion of renewable energy in its spatiotemporal distribution. This method fully utilizes historical scenario data and reduces the computational cost of online updates. By introducing transfer learning pre-training, this invention derives the transfer generalization error bound and the optimal empirical error combination weights to guide the pre-training stage in obtaining a new scenario-based model with minimal generalization error. Furthermore, by introducing active learning and actively querying important samples based on TTC (Transfer-to-Concept) network sensitivity assessment, this significantly reduces the new sample labeling time required for updating the new scenario-based model. This invention can assess the stability limit of transmission sections, thereby improving the utilization efficiency of transmission channels, adapting to the conditions of large-scale renewable energy transmission, and significantly reducing the labeling sample requirements and time costs for applying the model to new scenarios, thus improving efficiency. Attached Figure Description

[0045] Figure 1 This is a schematic diagram illustrating the transfer learning method of the present invention.

[0046] Figure 2 This is a schematic diagram illustrating the active learning mechanism of the present invention.

[0047] Figure 3 This is a flowchart of the training process for the TTC evaluation network in a new scenario based on transfer learning and active learning.

[0048] Figure 4 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0049] The present invention will now be described in detail with reference to the accompanying drawings.

[0050] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0051] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.

[0052] This invention can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, elements, data structures, etc., that perform a specific task or implement a specific abstract data type. This invention can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0053] In this invention, terms such as "module," "device," and "system" refer to relevant entities applied to a computer, such as hardware, combinations of hardware and software, software, or software in execution. More specifically, for example, an element can be, but is not limited to, a process running on a processor, a processor, an object, an executable element, an execution thread, a program, and / or a computer. Furthermore, an application program or script running on a server, and the server itself, can also be an element. One or more elements may be in an execution process and / or thread, and elements may be localized on a single computer and / or distributed across two or more computers, and may be run on various computer-readable media. Elements can also communicate via local and / or remote processes based on signals having one or more data packets, for example, signals from data interacting with another element in a local system, a distributed system, and / or interacting with other systems via signals over a network of the Internet.

[0054] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising" or "including" include not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0055] This invention analyzes the correlation between correlation indicators of cascaded transmission channel sections based on the spatiotemporal distribution characteristics of new energy sources in regional power grids. Using existing power grid basic data, real-time operation data, power generation and consumption data, and combining transfer learning and active learning algorithms, it utilizes information from historical operation scenarios to assist model learning in new scenarios, even with a limited number of labeled samples. This significantly reduces computation time and improves model accuracy in new scenarios, providing a framework for the online application of artificial intelligence models in power grid dispatching.

[0056] See Figure 2The present invention provides a method for evaluating the stability limit of a power transmission section, comprising the following steps:

[0057] Step 1: Establish a model to evaluate the ultimate transmission capacity of the power grid transmission section:

[0058] The ultimate transmission capacity (Total Transfer, TTC) of a power grid transmission section is defined as the maximum power that the section can transmit under the premise of satisfying all anticipated safety constraints before and after a fault. The TTC evaluation problem of a transmission section can be described by the following constraints (the constraints are defined in the literature: Paensuwan N, Yokoyama A, Nakachi Y, and Verma S C. Improved risk-based TTC evaluation with system case partitioning. International Journal of Electrical Power & Energy Systems, 2013, 44(3): 530–539.). The model for evaluating the ultimate transmission capacity of a power grid transmission section is as follows:

[0059] P TTC =Max(ξ(c,s,z))

[0060] stG(c,s,z)=0,H(c,s,z)≤0

[0061] Among them, P TTC ξ represents the transmission capacity limit of the transmission section, c represents the power flow value on the section, and the control variables are the generator terminal voltage of the PV node, the active power of the generator, the reactive power of the generator at the PQ node, or the active power of the load; s represents the state variables, which are the line power flow, the reactive power of the generator at the PV node, or the generator terminal voltage at the PQ node; z represents the grid topology variables, G represents the power flow equation constraint, and H represents the safety constraint.

[0062] Safety constraints include cross-sectional thermal stability constraints and transient power angle stability constraints. The cross-sectional thermal stability constraints are as follows:

[0063]

[0064] in, For line I after a fault τ occurs k The current, I lkmax For Line I k Current limit.

[0065] The transient work angle stability constraint is:

[0066] |δ Gp(t)-δ Gq (t)|≤δ max

[0067] Where, δ Gp (t), δ Gq (t) represents the power angles of generators p and q at time t, and δ represents the power angles of generators p and q at time t. max This is the preset maximum power angle difference.

[0068] Step 2: Pre-train the Ultimate Transmission Capacity (TTC) assessment model for power grid transmission sections using transfer learning methods to obtain a TTC assessment model based on transfer learning. The specific process is as follows:

[0069] The transfer learning problem is defined as follows: given a source domain D s and the learning task T in the source domain s Target domain D t and the learning task T in the target domain t Based on source domain D s For learning task T s Use existing knowledge to assist the target domain D t For learning task T t The learning.

[0070] In this invention, D s Refers to the historical operating scenario, target domain D t Pointer to source domain D s A significantly different new scenario. Source Domain D s and target domain D τ The main difference lies in the operating mode, such as line breakage, shutdown, maintenance, and drastic changes in source load.

[0071] See Figure 1 To adapt to the TTC assessment problem under changing scenarios, a transfer learning training method is used to train the model for assessing the ultimate transmission capacity of the power grid transmission section: given historical operating scenario D s And new scene D t The feature space of the historical operation scenario is X. s The feature space of the new scene is X t They share the same feature space X s =X t The label space for historical operation scenarios is Y. s The label space for the new scene is Y. t Tag space Y s =Y t and conditional probability distribution Q s (y s |x s )=Q t (yt |x t ), Q s (y s |x s Let Q be the conditional probability distribution of the historical operating scenarios. t (y t |x t Let be the conditional probability distribution of the new scenario, where x represents the feature space elements of the historical scenarios. s ∈X s The tag space element y of the historical running scenario s ∈Y s The feature space element x of the new scene t ∈X t The tag space element y in the new scene t ∈Y t However, the data edge distribution is different, i.e., P s (x s )≠P t (x t ), P s (x s P represents the edge distribution of data from historical operational scenarios. t (x t () represents the edge distribution of data in the new scenario.

[0072] For the TTC assessment problem, the migration goal is to use historical operating scenario D. s The labeled data is used to learn a hypothesis that is as good as possible in the hypothesis space: x t →y t For the new scenario D t New scene tag space element y t ∈Y t Make predictions.

[0073] The starting point of transfer learning is to train a cross-sectional TTC depth evaluation network model applicable to new scenarios using a small number of newly labeled samples.

[0074] The objective function of the power grid transmission section limit transmission capacity model is defined as:

[0075]

[0076] Where h is the target domain D in the new scenario. t Source Domain D under historical operating scenarios s The model obtained by pre-training through sample co-transfer learning and These represent the empirical errors of model h in historical and new scenarios, respectively. The empirical error weights, set manually, are determined through an optimization method, ranging from 0 to 1. These error weights are then... By finding the optimal value, the error bound can be minimized, and the error weight value can be optimized. The solution result is:

[0077]

[0078] Where d is the dimension of the hypothesis space of model h, A and β are constants, and m t =βm is the number of labeled samples in the target domain, and m is the size of the labeled sample set, that is, m is the number of samples in the sample set that are from the target domain D. t The obtained samples and (1-β)m samples from the source domain D s The obtained sample.

[0079] Step 3: Actively learn the TTC evaluation model based on the sensitivity of the TTC evaluation network by transferring the TTC evaluation model to obtain the trained TTC evaluation model.

[0080] In the new scenario, Monte Carlo simulation is used to simulate the fluctuation of power generation load and calculate the power flow, which is then used to construct an unlabeled sample pool.

[0081] Each sample in the unlabeled sample pool is input into the trained TTC evaluation model to achieve the stability limit assessment of the transmission section. The specific process is as follows:

[0082] The core of active learning lies in the sample query function, which is used to evaluate sample importance and select important samples from unlabeled samples for subsequent learning. For existing metastable binary classification problems, the classifier outputs probability information along with the classification result, reflecting the uncertainty of the model's sample prediction, which can be used to evaluate the importance of the sample. When the new scene deviates significantly from the historical scene, the prediction error of the pre-trained model for the sample may increase; a high sample prediction error means that the sample has higher value for model learning. In this invention, active learning based on TTC to evaluate network sensitivity uses dimensionality-reduced sensitivity coordinates to measure the degree of deviation of each scene from the sample center, thus measuring the importance of the scene.

[0083] In the new scenario, Monte Carlo simulation is used to simulate the fluctuation of the generation load and calculate the power flow. The calculated power flow is used to form an unlabeled sample pool. For each sample in the unlabeled sample pool, a transfer learning-based TTC evaluation model is used to obtain the control variables and state variables of TTC relative to the generator active power output and terminal voltage.

[0084]

[0085]

[0086]

[0087]

[0088] in, and These are the initial values ​​of generator active power, generator terminal voltage, load active power, and load reactive power, respectively. and The generator active power, terminal voltage, and load active power are all random numbers between [-1, +1], and c is the generator active power. PG c VG and c PL The disturbance variables for generator active power, generator terminal voltage, and load active power are set to 0.3, 0.02, and 0.3, respectively.

[0089] Each sample in the unlabeled sample pool can be represented as X. i =[P G Q G V G ,θ G ,P L Q L ,P AC ], where P G Q G V G ,θ G The generator's active and reactive power, voltage, and phase angle are represented as state variables, P. L Q L P represents the active and reactive power of the load. AC This represents the initial active power flow on the AC line.

[0090]

[0091] With voltage V G and phase angle θ G Let x be the state variable, and P be the generator's active and reactive power. G Q G As control variables u, P L Q L As the dependent variable y, in the formula, f(x,u)=0 is the nodal power constraint equation, and y=y(x,u) is the relationship between power and voltage. The sensitivity vector is obtained by expanding the Taylor formula and taking the first term:

[0092]

[0093] Will Substituting, we get:

[0094]

[0095]

[0096] in

[0097]

[0098] In the formula, S xu S yu The sensitivity matrix, i.e., the sensitivity vector, is the result of a change in u causing changes in x and y respectively, where Δx and Δu are the changes between two different states.

[0099] To mitigate the weakening of sample diversity caused by high vector dimensionality, electrical quantities closely coupled with the cross-section are selected as inputs, thereby reducing the input dimensionality and enhancing the difference between samples. Specifically, in a new scenario based on optimizing the objective function combined with an active learning sample query method (Ben-David S, Blitzer J, Crammer K, et al. A theory of learning from different domains[J]. Machine Learning,2010,79(1-2):151-175.), the dimensionality of the transfer learning-based TTC evaluation model is reduced to obtain the trained TTC evaluation model, thus achieving evaluation.

[0100] The specific process for training the assessment model for the ultimate transmission capacity (Total Transfer, TTC) of power grid transmission sections is as follows:

[0101] See Figure 2 and Figure 3 Based on the transfer learning method and active learning method using the optimal empirical error weight combination, the trained TTC evaluation model is obtained. The specific process is as follows:

[0102] The first step is to combine historical operating scenarios D with weights based on empirical errors. s The sample set D consists of a sample of samples and a small number of randomly generated samples in the new scenario. t0 Combined with transfer learning pre-training, a relatively high-performance initial TTC evaluation model h0 is obtained;

[0103] The second step involves using the initial TTC evaluation model h0 obtained from the first step's pre-training to evaluate the sample set D. t0The sensitivity vector S0 is calculated using the method described above, and then PCA (Principal Component Analysis) is used to reduce the sensitivity vector S0 to 2 dimensions, resulting in the dimensionality-reduced matrix C0 and a 2-dimensional vector set. For a 2D vector set Calculate the Euclidean distance to the center point c to provide a basis for subsequent key sample queries;

[0104] Step 3: Generation of unlabeled sample pools and parameter settings:

[0105] Randomly generate unlabeled sample pool U in the new scenario t Set the labeled sample pool L t =D t0 Let i = 0, and set the target accuracy p, the maximum number of iterations n, and the number of new samples N per iteration; after iteration, obtain the model h. i When model h i If accuracy < p and i < n, proceed to step four;

[0106] The fourth step is to actively learn and proactively query key samples:

[0107] Using model h i For unlabeled sample pool U t For each sample in the dataset, calculate the sensitivity vector for each sample, and apply model h to the sensitivity vector of each sample. i Dimensionality reduction matrix C i Dimensionality is reduced to 2D. The distance from the 2D vector of each sample to the Euclidean distance center point c is calculated. The N samples with the largest distances are selected for simulation. TTC label samples are calculated and added to the labeled sample pool L. t Meanwhile, from the unlabeled sample pool U t TTC-labeled samples were removed from the sample list.

[0108] Step 5: Model fine-tuning and sample center update:

[0109] Because the new model can calculate the TTC-labeled samples and corresponding sensitivities more accurately than the previous model, it is necessary to recalculate the sensitivity vector on the initial random new scene sample set based on the new model, and then calculate the sample centers after dimensionality reduction. The labeled sample pool L is used. t Fine-tuning model h i Model h i+1 Using h i+1 For D t0 The sensitivity of the sample is calculated to obtain S. i+1 , will S i+1 Reducing to 2 dimensions yields the reduced-dimensional matrix C. i+1 and 2D vector set right Calculate the Euclidean distance from the center, i = i + 1;

[0110] Finally, after the iteration is complete, the final high-performance new scene model is output, namely the trained TTC evaluation model, such as... Figure 3 As shown.

[0111] This method comprises two stages: the first stage introduces transfer learning pre-training, deriving the transfer generalization error bound and the optimal empirical error combination weights to guide the pre-training stage in obtaining a new scene model with minimal generalization error; the second stage introduces active learning and model fine-tuning, actively querying important samples based on TTC to evaluate network sensitivity, significantly reducing the new sample labeling time required for model updates, and further improving the performance of the new scene model through model fine-tuning. This invention can significantly reduce the required labeled samples and time cost for applying the model to new scenes, improving the efficiency of model transfer.

[0112] This invention reduces the number of labeled samples required for training new scene models through active learning, thereby significantly reducing computation time.

[0113] The transfer learning and active learning intelligent algorithms used in this invention evaluate the stability limit of transmission sections. By utilizing information from historical operating scenarios to assist model learning in new scenarios and calculating the stability limit of transmission sections, the computation time can be significantly reduced. Through active learning and iterative fine-tuning, the accuracy of the TTC evaluation model in new scenarios can be further improved.

[0114] See Figure 4 A system for evaluating the stability limits of a power transmission section, comprising:

[0115] The transfer learning module is used to transfer the ultimate transmission capacity assessment model of the power grid transmission section to obtain the TTC assessment model based on transfer learning.

[0116] The active learning module is used to actively learn the TTC evaluation model based on transfer learning to obtain the trained TTC evaluation model.

[0117] The unlabeled sample pool acquisition module is used to simulate the fluctuation of power generation load and calculate the power flow in the new scenario through Monte Carlo simulation, and to construct an unlabeled sample pool of power flow.

[0118] The evaluation module is used to input each sample in the unlabeled sample pool into the trained TTC evaluation model to achieve the stability limit evaluation of the transmission section.

[0119] Furthermore, the model for evaluating the ultimate transmission capacity of the power grid transmission section is as follows:

[0120] P TTC =Max(ξ(c,s,z))

[0121] stG(c,s,z)=0,H(c,s,z)≤0

[0122] Among them, P TTC ξ represents the transmission capacity limit of the transmission section, c represents the power flow value on the section, and the control variables are the generator terminal voltage of the PV node, the active power of the generator, the reactive power of the generator at the PQ node, or the active power of the load; s represents the state variables, which are the line power flow, the reactive power of the generator at the PV node, or the generator terminal voltage at the PQ node; z represents the grid topology variables, G represents the power flow equation constraint, and H represents the safety constraint.

[0123] Furthermore, safety constraints include cross-sectional thermal stability constraints and transient power angle stability constraints.

[0124] Furthermore, the cross-sectional thermal stability constraint is as follows:

[0125]

[0126] in, For line I after a fault τ occurs k The current, I lkmax For Line I k Current limit.

[0127] A computer device includes a memory and a processor, the memory storing a computer program executable on the processor, the computer program, when executed by the processor, implementing the method for assessing the stability limits of a power transmission section as described above.

[0128] A computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the method for assessing the stability limits of a power transmission section as described above.

[0129] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0130] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0131] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0132] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for evaluating the stability limit of a power transmission section, characterized in that, Includes the following steps: The transfer learning method was used to pre-train the ultimate transmission capacity assessment model of the power grid transmission section, and a transfer learning-based TTC assessment model was obtained. The TTC evaluation model based on transfer learning is actively learned to obtain the trained TTC evaluation model. Each sample in the unlabeled sample pool is input into the trained TTC evaluation model to achieve the stability limit assessment of the transmission section. The unlabeled sample pool is determined through the following process: under the new scenario, Monte Carlo simulation is used to simulate the fluctuation of the power generation load and calculate the power flow, and the power flow constitutes the unlabeled sample pool. The model for evaluating the ultimate transmission capacity of power grid transmission sections is as follows: in, Represents the maximum transmission capacity of the transmission section. The current value represents the tidal current value on the cross section. Represents the control variables, which are the generator terminal voltage at the PV node, the generator active power, and the reactive power or load active power of the generator at the PQ node. Represents state variables, which are line power flow, reactive power of generators at PV nodes, or generator terminal voltage at PQ nodes. Represents the topology variables of the power grid. Represents the constraints of the power flow equations. Represents safety constraints; The objective function for the power grid transmission section limit transmission capacity model is: Where h is the target domain in the new scenario. Source domain in historical operating scenarios The model obtained by pre-training through sample co-transfer learning and These represent the empirical errors of model h in historical and new scenarios, respectively. These are empirical error weight values; The step of actively learning the TTC evaluation model based on transfer learning to obtain the trained TTC evaluation model specifically includes: Based on empirical error combination weights, historical operating scenarios are... The sample set consists of a small number of randomly generated samples in the new scenario and the original sample set. Combined with transfer learning pre-training, an initial TTC evaluation model is obtained. ; Using the pre-trained initial TTC evaluation model For the sample set Calculate the sensitivity vector and the sensitivity vector The PCA method is used to reduce the dimension to 2, resulting in the dimensionality-reduced matrix. and 2D vector set For a 2D vector set Calculate the Euclidean distance from the center point c; Randomly generate unlabeled sample pools in new scenarios Set up labeled sample pool Set i=0, target accuracy p, maximum number of iterations n, and number of new samples N per iteration; after iteration, obtain the model. When the model and When using a model For unlabeled sample pools For each sample in the dataset, calculate the sensitivity vector for each sample, and apply the model to the sensitivity vector of each sample. Dimensionality reduction matrix Dimensionality is reduced to 2D. The distance from the 2D vector of each sample to the Euclidean distance center point c is calculated. The N samples with the largest distances are selected for simulation. TTC label samples are calculated and added to the labeled sample pool. At the same time, from the unlabeled sample pool TTC-labeled samples are removed from the pool; each sample in the unlabeled sample pool is represented as... ,in, This indicates the generator's active and reactive power, voltage, and phase angle. This indicates the active and reactive power of the load. Indicates the initial active power flow on the AC line; Use labeled sample pools Fine-tuning model Get the model ,use right The sensitivity of the sample is calculated. ,Will Reduced to 2 dimensions to obtain the dimensionality reduction matrix and 2D vector set ,right Calculate the Euclidean distance from the center, i = i + 1; After the iteration is complete, the final trained TTC evaluation model is output.

2. The method for evaluating the stability limit of a power transmission section according to claim 1, characterized in that, Safety constraints include cross-sectional thermal stability constraints and transient power angle stability constraints.

3. The method for evaluating the stability limit of a power transmission section according to claim 2, characterized in that, The thermal stability constraint of the cross section is: in, For the occurrence of a fault Rear Line The current, For the line Current limit.

4. The method for evaluating the stability limit of a power transmission section according to claim 3, characterized in that, The transient work angle stability constraint is: in, , Let be the power angles of generators p and q at time t, respectively. This is the preset maximum power angle difference.

5. The method for evaluating the stability limit of a power transmission section according to claim 1, characterized in that, Error weight value Calculated using the following formula: Where d is the dimension of the hypothesis space of model h, and A,β are constants. denoted as the number of labeled samples in the target domain, and m as the size of the labeled sample set.

6. A system for evaluating the stability limits of a power transmission section, characterized in that, include: The transfer learning module is used to transfer the ultimate transmission capacity assessment model of the power grid transmission section to obtain the TTC assessment model based on transfer learning. The active learning module is used to actively learn the TTC evaluation model based on transfer learning to obtain the trained TTC evaluation model. The evaluation module is used to input each sample in the unlabeled sample pool into the trained TTC evaluation model to achieve the stability limit evaluation of the transmission section. The unlabeled sample pool is determined through the following process: under the new scenario, Monte Carlo simulation is used to fluctuate the generation load and calculate the power flow, and the power flow constitutes the unlabeled sample pool. The model for evaluating the ultimate transmission capacity of power grid transmission sections is as follows: in, Represents the maximum transmission capacity of the transmission section. The current value represents the tidal current value on the cross section. Represents the control variables, which are the generator terminal voltage at the PV node, the generator active power, and the reactive power or load active power of the generator at the PQ node. Represents state variables, which are line power flow, reactive power of generators at PV nodes, or generator terminal voltage at PQ nodes. Represents the topology variables of the power grid. Represents the constraints of the power flow equations. Represents safety constraints; The objective function for the power grid transmission section limit transmission capacity model is: Where h is the target domain in the new scenario. Source domain in historical operating scenarios The model obtained by pre-training through sample co-transfer learning and These represent the empirical errors of model h in historical and new scenarios, respectively. These are empirical error weight values; The step of actively learning the TTC evaluation model based on transfer learning to obtain the trained TTC evaluation model specifically includes: Based on empirical error combination weights, historical operating scenarios are... The sample set consists of a small number of randomly generated samples in the new scenario and the original sample set. Combined with transfer learning pre-training, an initial TTC evaluation model is obtained. ; Using the pre-trained initial TTC evaluation model For the sample set Calculate the sensitivity vector and the sensitivity vector The PCA method is used to reduce the dimension to 2, resulting in the dimensionality-reduced matrix. and 2D vector set For a 2D vector set Calculate the Euclidean distance from the center point c; Randomly generate unlabeled sample pools in new scenarios Set up labeled sample pool Set i=0, target accuracy p, maximum number of iterations n, and number of new samples N per iteration; after iteration, obtain the model. When the model and When using a model For unlabeled sample pools For each sample in the dataset, calculate the sensitivity vector for each sample, and apply the model to the sensitivity vector of each sample. Dimensionality reduction matrix Dimensionality is reduced to 2D. The distance from the 2D vector of each sample to the Euclidean distance center point c is calculated. The N samples with the largest distances are selected for simulation. TTC label samples are calculated and added to the labeled sample pool. At the same time, from the unlabeled sample pool TTC-labeled samples are removed from the pool; each sample in the unlabeled sample pool is represented as... ,in, This indicates the generator's active and reactive power, voltage, and phase angle. This indicates the active and reactive power of the load. Indicates the initial active power flow on the AC line; Use labeled sample pools Fine-tuning model Get the model ,use right The sensitivity of the sample is calculated. ,Will Reduced to 2 dimensions to obtain the dimensionality reduction matrix and 2D vector set ,right Calculate the Euclidean distance from the center, i = i + 1; After the iteration is complete, the final trained TTC evaluation model is output.

7. The system for evaluating the stability limits of a power transmission section according to claim 6, characterized in that, Safety constraints include cross-sectional thermal stability constraints and transient power angle stability constraints.

8. The system for evaluating the stability limits of a power transmission section according to claim 7, characterized in that, The thermal stability constraint of the cross section is: in, For the occurrence of a fault Rear Line The current, For the line Current limit.

9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing a computer program that can run on the processor, the computer program being executed by the processor to implement the method for assessing the stability limits of a power transmission section as described in any one of claims 1 to 5.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, causes the processor to perform the method for assessing the stability limits of a transmission section as described in any one of claims 1 to 5.

Images