Power grid dispatching boundary operation scene generation method, device, equipment, medium and product

Abstract

The invention discloses a power grid dispatching boundary operation scene generation method, which comprises the following steps of: acquiring an ultra-short-term point prediction value of a target time period and a corresponding prediction error covariance matrix, and constructing conditional Gaussian distribution for each future time period; sampling the conditional multivariate Gaussian distribution to generate an initial scene set; inputting the initial scene set into a pre-trained scene risk discrimination model, and outputting a business index prediction value of each scene; based on the business index prediction value, screening out a target scene meeting a preset requirement from the initial scene set; according to the method, the quality and practicability of boundary scene generation can be remarkably improved, and the active defense capability of a dispatching system on extreme risks is enhanced, so that safe and stable operation of a power grid is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and in particular to a method, apparatus, equipment, medium and product for generating power grid dispatch boundary operation scenarios. Background Technology

[0002] In the field of power grid dispatch, the generation of ultra-short-term operating scenarios is a key technology to support real-time safety verification, market clearing, and risk warning. With the large-scale integration of new energy sources and the enhanced flexible interaction on the load side, the power grid operation status exhibits strong uncertainty, high time-varying characteristics, and complex coupling features. Traditional scenario generation methods based on historical data backtracking or deterministic prediction are no longer sufficient to meet the needs of dispatch decision-making. Especially at the ultra-short-term scale, it is necessary to quickly generate typical scenarios covering extreme boundary conditions to provide stress test inputs for the dispatch system, so as to verify the robustness of key indicators such as reserve capacity and cross-sectional margin, thereby ensuring the safe and stable operation of the power grid under disturbances such as new energy fluctuations and load mutations.

[0003] Existing scenario generation methods have multiple limitations, making it difficult to cover the risky operating conditions that dispatchers are truly concerned about. They cannot support dispatchers in predicting and responding to extreme operating conditions, thus affecting the safe and stable operation of the power grid. Summary of the Invention

[0004] This invention provides a method for generating boundary operation scenarios for power grid dispatching, which can significantly improve the quality and practicality of boundary scenario generation, enhance the dispatching system's proactive defense capability against extreme risks, and ensure the safe and stable operation of the power grid.

[0005] In a first aspect, embodiments of the present invention provide a method for generating power grid dispatch boundary operation scenarios, including: Obtain the ultrashort-term point prediction values ​​and corresponding prediction error covariance matrix for the target time period, and construct a conditional Gaussian distribution for each future time period; The conditional multivariate Gaussian distribution is sampled to generate an initial scene set; The initial set of scenarios is input into a pre-trained scenario risk discrimination model, which outputs the predicted business indicators for each scenario. Based on the predicted values ​​of the business metrics, target scenarios that meet the preset requirements are selected from the initial scenario set.

[0006] Furthermore, constructing a conditional Gaussian distribution for each future time period includes: For each future time period, a conditional multivariate Gaussian distribution is constructed with the predicted value of the ultra-short-term point as the mean and the prediction error covariance matrix as the variance.

[0007] Furthermore, the training process of the scenario risk discrimination model includes: The collected historical power grid operation data is used to construct a sample dataset; each sample dataset includes an operation status vector and a service label vector. The operation status vector includes the active power output of the generating units and the active power of the load nodes, and the service label vector includes the cross-sectional margin, positive reserve, renewable energy curtailment and congestion price. Set corresponding extreme scenario thresholds for each of the aforementioned business tag vectors, including cross-sectional margin threshold, positive reserve threshold, renewable energy curtailment threshold, and congestion price threshold; Based on the extreme scene threshold, scene labeling is performed on each sample data; A scenario risk discrimination model is constructed, and the model is trained using a labeled sample dataset.

[0008] Furthermore, the step of labeling each sample data based on the extreme scene threshold includes: Iterate through all the sample data mentioned above; When the cross-sectional margin in the sample data is less than or equal to the cross-sectional margin threshold, the sample data is marked as a low margin scenario. When the positive reserve in the sample data is greater than the positive reserve threshold, the sample data is marked as a high reserve scenario. When the amount of renewable energy curtailment in the sample data exceeds the renewable energy curtailment threshold, the sample data is marked as a high curtailment scenario. When the congestion price in the sample data is greater than the congestion price threshold, the sample data is marked as a high-price scenario.

[0009] Furthermore, after training the scenario risk discrimination model using the labeled sample dataset, the process includes: Calculate the average error of the predicted business indicators output by the scenario risk discrimination model; When the average error is less than the preset average error threshold, the scenario risk discrimination model is determined to be successfully trained.

[0010] Furthermore, the method also includes: The selected target scenarios are input into the SCED model and SCUC model for power system scheduling.

[0011] In a second aspect, embodiments of the present invention provide a power grid dispatch boundary operation scenario generation device, comprising: The data acquisition module is used to acquire the ultra-short-term point prediction values ​​and the corresponding prediction error covariance matrix for the target time period, and to construct a conditional Gaussian distribution for each future time period. The scene generation module is used to sample the conditional multivariate Gaussian distribution to generate an initial scene set; The risk discrimination module is used to input the initial set of scenarios into a pre-trained scenario risk discrimination model and output the predicted business indicator value for each scenario. The scenario filtering module is used to filter target scenarios that meet preset requirements from the initial scenario set based on the predicted values ​​of the business indicators.

[0012] Thirdly, embodiments of the present invention provide an electronic device, comprising: Memory, used to store computer programs; A processor for executing the computer program; Wherein, when the processor executes the computer program, it implements the power grid dispatch boundary operation scenario generation method described in any of the first aspects above.

[0013] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program, which, when executed, implements the power grid dispatch boundary operation scenario generation method described in any of the first aspects above.

[0014] Fifthly, embodiments of the present invention provide a computer program product, including computer instructions, which, when executed by a processor, implement the power grid dispatch boundary operation scenario generation method described in any of the first aspects above.

[0015] Compared with existing technologies, the power grid dispatch boundary operation scenario generation method provided by this invention has the following advantages: It obtains the ultra-short-term point prediction values ​​and corresponding prediction error covariance matrices for the target time period, and constructs a conditional Gaussian distribution for each future time period; it samples the conditional multivariate Gaussian distribution to generate an initial scenario set; it inputs the initial scenario set into a pre-trained scenario risk discrimination model to output the business indicator prediction values ​​for each scenario; based on the business indicator prediction values, it selects target scenarios that meet preset requirements from the initial scenario set. This invention can significantly improve the quality and practicality of boundary scenario generation, enhance the dispatch system's proactive defense capability against extreme risks, and ensure the safe and stable operation of the power grid. Attached Figure Description

[0016] To more clearly illustrate the technical features of the embodiments of the present invention, the drawings used in the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a method for generating power grid dispatch boundary operation scenarios according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a power grid dispatch boundary operation scenario generation device provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0018] 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.

[0019] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein is for the purpose of describing embodiments of the invention only and is not intended to limit the invention.

[0021] In a first aspect, embodiments of the present invention provide a method for generating power grid dispatch boundary operation scenarios, see [link to relevant documentation]. Figure 1 This is a flowchart illustrating an embodiment of a method for generating power grid dispatch boundary operation scenarios provided by the present invention.

[0022] like Figure 1 As shown, the method includes the following steps: S1: Obtain the predicted value of the ultra-short-term point for the target time period and the corresponding prediction error covariance matrix, and construct a conditional Gaussian distribution for each future time period; S2: Sample the conditional multivariate Gaussian distribution to generate an initial scene set; S3: Input the initial set of scenarios into the pre-trained scenario risk discrimination model and output the predicted business indicator value for each scenario; S4: Based on the predicted values ​​of the business indicators, select target scenarios that meet the preset requirements from the initial scenario set.

[0023] In practice, the predicted value of the ultra-short-term point of the target period and the corresponding prediction error covariance matrix are obtained, and a conditional multivariate Gaussian distribution is constructed for each future period. The conditional multivariate Gaussian distribution of each period is randomly sampled and combined to form a large number of future operation scenarios, generating an initial scenario set. For each generated initial scenario, a pre-trained scenario risk discrimination model is used to judge the risk and output the predicted value of business indicators. According to the preset business requirements, the target scenario that meets the requirements is selected from the initial scenario set.

[0024] In summary, this invention utilizes a pre-trained scenario risk discrimination model to rapidly evaluate business indicators for scenarios, accurately identifying high-risk typical scenarios, significantly improving the generation quality and practicality of boundary scenarios, enhancing the scheduling system's proactive defense capability against extreme risks, and ensuring the safe and stable operation of the power grid.

[0025] In one optional implementation, constructing a conditional Gaussian distribution for each future time period includes: For each future time period, a conditional multivariate Gaussian distribution is constructed with the predicted value of the ultra-short-term point as the mean and the prediction error covariance matrix as the variance.

[0026] Specifically, for each future time period, a conditional multivariate Gaussian distribution is constructed with the mean of the ultra-short-term point prediction value and the variance of the prediction error covariance matrix.

[0027] In one optional implementation, the training process of the scenario risk discrimination model includes: The collected historical power grid operation data is used to construct a sample dataset; each sample dataset includes an operation status vector and a service label vector. The operation status vector includes the active power output of the generating units and the active power of the load nodes, and the service label vector includes the cross-sectional margin, positive reserve, renewable energy curtailment and congestion price. Set corresponding extreme scenario thresholds for each of the aforementioned business tag vectors, including cross-sectional margin threshold, positive reserve threshold, renewable energy curtailment threshold, and congestion price threshold; Based on the extreme scene threshold, scene labeling is performed on each sample data; A scenario risk discrimination model is constructed, and the model is trained using a labeled sample dataset.

[0028] Specifically, historical power grid operation data within a preset time period in the target area is first collected and used to construct a sample dataset. Each sample data point contains two sets of core information: an operation status vector and a business label vector. The operation status vector serves as the input feature of the model, and only active power data strongly correlated with intraday clearing is retained, i.e., the active power output of the generating units. and active power of load nodes Running state vector It is represented as follows: ; in, This refers to the number of units in operation, typically ≤2000. This represents the number of load nodes, typically ≤8000.

[0029] The business label vector serves as the model's learning objective, directly corresponding to the four risk indicators most relevant to dispatchers: cross-sectional margin (whether lines are nearing overload), reserve (whether the system's reserve power is sufficient), renewable energy curtailment (whether wind / solar power is being wasted), and congestion pricing (whether electricity prices in a certain area are abnormally high). It is represented as follows: ; in, For cross-sectional margin, For backup purposes. For the abandoned electricity of new energy sources, To block electricity prices.

[0030] For each business tag vector, set corresponding extreme scenario thresholds, including cross-sectional margin threshold, positive reserve threshold, renewable energy curtailment threshold, and congestion price threshold. For example, the cross-sectional margin threshold can be set to 10%. When the cross-sectional margin is less than this threshold, it indicates that the line load rate has reached 90%, approaching the overload limit, and there is a risk of low margin. The positive reserve threshold can be set to 18% of the total system charge. When the positive reserve is greater than or equal to this threshold, it indicates that there is too much reserve power, the generation cost is too high, and there is a risk of high reserve cost. The renewable energy curtailment threshold can be set to 20% of the predicted renewable energy output. When the renewable energy curtailment is greater than or equal to this threshold, it indicates that wind / solar power waste exceeds 20%, the absorption capacity is insufficient, and there is a high risk of curtailment. The congestion price threshold can be set to 150% of the average node price of the entire network. When the congestion price is greater than or equal to this threshold, it indicates that the electricity price in a certain area has skyrocketed, there is a line congestion problem, and there is a risk of high electricity price. Based on the set extreme scenario thresholds, scenario labeling is performed on each sample data.

[0031] To construct a scenario risk discrimination model, a lightweight model such as a 2-layer MLP (Multilayer Perceptron) can be selected. The input layer dimension of the model is "number of operating units + number of load nodes", the hidden layer has 128 neurons, and the output layer is 4-dimensional, corresponding to the predicted values ​​of 4 types of risk indicators. The loss function of the model is MSE (mean squared error). The optimization objective is to make the risk indicator values ​​predicted by the model as close as possible to the actual label values ​​of the samples. An early stopping mechanism is adopted to verify the model accuracy in real time during training. If the accuracy no longer improves, training is stopped to avoid overfitting.

[0032] The scene risk discrimination model is trained using the labeled sample dataset to obtain the trained scene risk discrimination model.

[0033] This embodiment constructs a sample set containing operation status vectors and multi-dimensional business label vectors from historical power grid operation data, and sets extreme scenario thresholds for key indicators such as cross-sectional margin, positive reserve, renewable energy curtailment, and congestion pricing. This enables accurate risk classification and labeling of historical scenarios. The scenario risk discrimination model trained based on this can automatically identify potential extreme risks in the operation status, avoiding physical infeasibility of clearing results due to missing scenarios.

[0034] In one optional implementation, the step of labeling each sample data based on the extreme scene threshold includes: Iterate through all the sample data mentioned above; When the cross-sectional margin in the sample data is less than or equal to the cross-sectional margin threshold, the sample data is marked as a low margin scenario. When the positive reserve in the sample data is greater than the positive reserve threshold, the sample data is marked as a high reserve scenario. When the amount of renewable energy curtailment in the sample data exceeds the renewable energy curtailment threshold, the sample data is marked as a high curtailment scenario. When the congestion price in the sample data is greater than the congestion price threshold, the sample data is marked as a high-price scenario.

[0035] Specifically, the well-organized sample dataset is read, and for each sample, the four core indicators in its business label vector are individually verified. As long as the threshold conditions are met, the corresponding scenario is labeled. When the cross-sectional margin in the sample data is less than or equal to the cross-sectional margin threshold, the sample data is labeled as a low margin scenario. When the positive reserve in the sample data is greater than the positive reserve threshold, the sample data is labeled as a high reserve scenario. When the renewable energy curtailment in the sample data is greater than the renewable energy curtailment threshold, the sample data is labeled as a high curtailment scenario. When the congestion price in the sample data is greater than the congestion price threshold, the sample data is labeled as a high price scenario. After labeling, the sample data and all the labeled data are bound and stored.

[0036] It should be noted that the labeling of the four scenarios is independently judged and does not affect each other. Therefore, a single sample may be labeled with multiple labels. For example, sample 1 may be labeled with "low margin, high reserve, and high electricity price" at the same time, or it may be unlabeled (none of the four indicators meet the threshold). This multi-label labeling can accurately cover the complex risk scenarios in the actual operation of the power grid and avoid missing the complex risks that dispatchers care about.

[0037] This embodiment performs refined scene annotation on sample data based on preset extreme thresholds, providing a structured and highly recognizable labeling system for model training.

[0038] In one optional implementation, after training the scene risk discrimination model using the labeled sample dataset, the process includes: Calculate the average error of the predicted business indicators output by the scenario risk discrimination model; When the average error is less than the preset average error threshold, the scenario risk discrimination model is determined to be successfully trained.

[0039] Specifically, after training the scenario risk discrimination model, it is necessary to conduct an evaluation and verification, calculate the average error of the predicted business indicators output by the model, and only when the average error is less than the preset average error threshold can the training be considered successful and the model can be put into practical use.

[0040] This embodiment quantifies the model training effect, which can objectively and accurately measure the model's prediction accuracy and ensure that the model used in practical applications has a highly reliable risk prediction capability.

[0041] In an optional implementation, the method further includes: The selected target scenarios are input into the SCED model and SCUC model for power system scheduling.

[0042] Specifically, the selected target scenarios are input into the SCED (Security-Constrained Economic Dispatch) model and the SCUC (Security-Constrained Unit Commitment) model for safety verification and dispatch decisions. Based on the output of the two models, the dispatch center issues dispatch instructions to each power plant and substation to ensure the stable operation of the power system.

[0043] Secondly, embodiments of the present invention provide a power grid dispatch boundary operation scenario generation device, see [link to relevant documentation]. Figure 2 This is a schematic diagram of an embodiment of a power grid dispatch boundary operation scenario generation device provided by the present invention.

[0044] like Figure 2 As shown, the device includes: The data acquisition module 21 is used to acquire the ultra-short-term point prediction values ​​and the corresponding prediction error covariance matrix for the target time period, and to construct a conditional Gaussian distribution for each future time period. Scene generation module 22 is used to sample the conditional multivariate Gaussian distribution to generate an initial scene set; Risk discrimination module 23 is used to input the initial scenario set into the pre-trained scenario risk discrimination model and output the business indicator prediction value for each scenario; The scenario filtering module 24 is used to filter out target scenarios that meet preset requirements from the initial scenario set based on the predicted values ​​of the business indicators.

[0045] In one optional implementation, constructing a conditional Gaussian distribution for each future time period includes: For each future time period, a conditional multivariate Gaussian distribution is constructed with the predicted value of the ultra-short-term point as the mean and the prediction error covariance matrix as the variance.

[0046] In one optional implementation, the training process of the scenario risk discrimination model includes: The collected historical power grid operation data is used to construct a sample dataset; each sample dataset includes an operation status vector and a service label vector. The operation status vector includes the active power output of the generating units and the active power of the load nodes, and the service label vector includes the cross-sectional margin, positive reserve, renewable energy curtailment and congestion price. Set corresponding extreme scenario thresholds for each of the aforementioned business tag vectors, including cross-sectional margin threshold, positive reserve threshold, renewable energy curtailment threshold, and congestion price threshold; Based on the extreme scene threshold, scene labeling is performed on each sample data; A scenario risk discrimination model is constructed, and the model is trained using a labeled sample dataset.

[0047] In one optional implementation, the step of labeling each sample data based on the extreme scene threshold includes: Iterate through all the sample data mentioned above; When the cross-sectional margin in the sample data is less than or equal to the cross-sectional margin threshold, the sample data is marked as a low margin scenario. When the positive reserve in the sample data is greater than the positive reserve threshold, the sample data is marked as a high reserve scenario. When the amount of renewable energy curtailment in the sample data exceeds the renewable energy curtailment threshold, the sample data is marked as a high curtailment scenario. When the congestion price in the sample data is greater than the congestion price threshold, the sample data is marked as a high-price scenario.

[0048] In one optional implementation, after training the scene risk discrimination model using the labeled sample dataset, the process includes: Calculate the average error of the predicted business indicators output by the scenario risk discrimination model; When the average error is less than the preset average error threshold, the scenario risk discrimination model is determined to be successfully trained.

[0049] In an optional embodiment, the device is further configured to: The selected target scenarios are input into the SCED model and SCUC model for power system scheduling.

[0050] Thirdly, embodiments of the present invention provide an electronic device, see [link to previous document]. Figure 3 The diagram shown is a structural schematic of an electronic device provided in an embodiment of the present invention.

[0051] like Figure 3 As shown, the device includes: Memory 31 is used to store computer programs; Processor 32 is used to execute the computer program; When the processor 32 executes the computer program, it implements the power grid dispatch boundary operation scenario generation method as described in any of the above embodiments.

[0052] For example, the computer program may be divided into one or more modules / units, which are stored in the memory 31 and executed by the processor 32 to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0053] The processor 32 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0054] The memory 31 can be used to store the computer programs and / or modules. The processor 32 implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory 31 and calling the data stored in the memory 31. The memory 31 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory 31 may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital card (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0055] It should be noted that the aforementioned electronic devices include, but are not limited to, processors and memory, as will be understood by those skilled in the art. Figure 3 The structural diagram is merely an example of the electronic device described above and does not constitute a limitation on the electronic device. It may include more components than shown in the diagram, or combine certain components, or use different components.

[0056] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a computer program, which, when executed, implements the power grid dispatch boundary operation scenario generation method described in any of the above embodiments.

[0057] It should be understood that the present invention can implement all or part of the processes in the above-described method for generating power grid dispatch boundary operation scenarios, or it can be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the above-described method for generating power grid dispatch boundary operation scenarios. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0058] Fifthly, embodiments of this application also provide a computer program product, which is stored in a storage medium and executed by at least one processor to implement the power grid dispatch boundary operation scenario generation method described in any of the above embodiments.

[0059] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. It should be noted that, for those skilled in the art, several equivalent obvious modifications and / or equivalent substitutions can be made without departing from the technical principles of the present invention, and these obvious modifications and / or equivalent substitutions should also be considered within the scope of protection of the present invention.

Claims

1. A method for generating power grid dispatch boundary operation scenarios, characterized in that, The method comprises the following steps: obtaining an ultra-short-term point prediction value and a corresponding prediction error covariance matrix of a target period, and constructing a conditional Gaussian distribution for each future period; sampling the conditional multivariate Gaussian distribution to generate an initial scenario set; inputting the initial scenario set into a pre-trained scenario risk discrimination model to output a business index prediction value of each scenario; based on the business index prediction value, screening a target scenario meeting a preset requirement from the initial scenario set.

2. The power grid dispatch boundary operating scenario generation method of claim 1, wherein, The step of constructing a conditional Gaussian distribution for each future period comprises the following steps: for each future period, constructing a conditional multivariate Gaussian distribution with the ultra-short-term point prediction value as the mean value and the prediction error covariance matrix as the variance.

3. The method of claim 1, wherein, The training process of the scenario risk discrimination model comprises the following steps: constructing a sample data set from collected power grid historical operation data; wherein each sample data comprises an operation state vector and a business label vector, the operation state vector comprises active power output of an on-line unit and active power of a load node, and the business label vector comprises a section margin, a positive reserve, a new energy curtailment and a congestion price; setting a corresponding extreme scenario threshold for each business label vector, including a section margin threshold, a positive reserve threshold, a new energy curtailment threshold and a congestion price threshold; based on the extreme scenario threshold, scenario labeling is performed on each sample data; constructing a scenario risk discrimination model, and training the scenario risk discrimination model using the labeled sample data set.

4. The power grid dispatch boundary operating scenario generation method of claim 3, wherein, The step of performing scenario labeling on each sample data based on the extreme scenario threshold comprises the following steps: traversing all the sample data; when the section margin in the sample data is less than or equal to the section margin threshold, labeling the sample data as a low margin scenario; when the positive reserve in the sample data is greater than the positive reserve threshold, labeling the sample data as a high reserve scenario; when the new energy curtailment in the sample data is greater than the new energy curtailment threshold, labeling the sample data as a high curtailment scenario; when the congestion price in the sample data is greater than the congestion price threshold, labeling the sample data as a high price scenario.

5. The method of claim 3, wherein, After the scenario risk discrimination model is trained using the labeled sample data set, the following steps are performed: calculating the average error of the business index prediction value output by the scenario risk discrimination model; when the average error is less than a preset average error threshold, determining that the scenario risk discrimination model is qualified for training.

6. The power grid dispatch boundary operating scenario generation method of claim 1, wherein, The method further comprises the following steps: inputting the screened target scenario into an SCED model and an SCUC model to perform power system dispatching.

7. A power grid dispatch boundary operating scenario generation apparatus characterized by, The method comprises the following steps: a data acquisition module, configured to obtain an ultra-short-term point prediction value and a corresponding prediction error covariance matrix of a target period, and construct a conditional Gaussian distribution for each future period; a scenario generation module, configured to sample the conditional multivariate Gaussian distribution to generate an initial scenario set; a risk discrimination module, configured to input the initial scenario set into a pre-trained scenario risk discrimination model to output a business index prediction value of each scenario; A scene screening module is configured to screen a target scene meeting preset requirements from the initial scene set based on the service index prediction value.

8. An electronic device, comprising: The method comprises the following steps: A memory is configured to store a computer program; A processor is configured to execute the computer program; When the processor executes the computer program, the power grid scheduling boundary operation scene generation method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed to implement the power grid scheduling boundary operation scene generation method according to any one of claims 1 to 6.

10. A computer program product, characterised in that, The computer readable storage medium stores a computer program, and the computer program is executed to implement the power grid scheduling boundary operation scene generation method according to any one of claims 1 to 6.

Images