A method for determining the transient stability of a power grid based on a critical confidence index

Abstract

This invention relates to the field of power system analysis, specifically to a method for determining the transient stability of a power grid based on a critical credibility index. The method includes: calculating the transient instability probability using real-time acquired transient response trajectories of electrical quantities at the initial stage of a fault and a trained transient stability discrimination model; predicting a transient stability label based on the transient instability probability; evaluating the predicted transient stability label to obtain the final transient stability discrimination result. This invention significantly improves the accuracy and generalization ability of the transient stability discrimination model by constructing an electromechanical transient simulation dataset covering multiple operating modes and fault types, and using transient response trajectories of electrical quantities as core feature inputs. The introduction of a critical credibility index quantifies the uncertainty of the model output results, reducing the possibility of instability conditions being misclassified as stable conditions, and improving the conservatism and accuracy of the discrimination results.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of power system analysis, in particular to a power grid transient stability discrimination method based on a critical confidence index. BACKGROUND

[0002] The transient stability of a power system is a key capability for maintaining synchronous operation after a large disturbance, and its analysis has irreplaceable value in preventing large-scale power outages and ensuring social power supply safety. Traditional analysis methods based on numerical simulation are time-consuming, making it difficult to meet the urgent needs of real-time analysis and rapid decision-making for complex power grids with a high proportion of new energy. Therefore, the introduction of artificial intelligence methods is crucial. Artificial intelligence technology can deeply mine massive historical and simulation data, build high-precision stability rapid evaluation and prediction models, and shorten the analysis time from minutes to milliseconds. This not only greatly improves the early warning and prevention and control capabilities of power grids for potential instability risks, supports dispatchers to take emergency control measures to avoid system collapse, but also provides a core computing tool for building a strong smart grid and promoting renewable energy consumption. Ultimately, transient stability analysis enabled by artificial intelligence is an important technical cornerstone for ensuring the stable operation of the national economy and social life, and has far-reaching significance for promoting energy transformation.

[0003] Based on the above analysis, the application of artificial intelligence to transient stability analysis has important value for the safe and stable operation of power systems and the development of the national economy. However, artificial intelligence methods currently still cannot achieve 100% accuracy in discrimination, and the tolerance of power systems for missed instability conditions is much lower than the tolerance for false positives. SUMMARY

[0004] (I) Invention purpose

[0005] The purpose of the present application is to provide a power grid transient stability discrimination method based on a critical confidence index, which evaluates the predicted transient stability, calculates the critical confidence index, and thus reduces the scenarios of missing instability conditions as stable conditions, improving the conservatism and accuracy of the discrimination results.

[0006] (II) Technical solutions

[0007] To solve the above problems, the present application provides a power grid transient stability discrimination method based on a critical confidence index, comprising:

[0008] Based on the preset operating mode and the corresponding combination of faults, electromechanical transient simulation is performed to obtain a simulation data set, which includes electrical quantity transient response trajectories and transient stability labels;

[0009] A transient stability discrimination model is constructed, with the electrical quantity transient response trajectories as input and the transient stability labels as output.

[0010] The transient stability discrimination model is trained using the simulation dataset to obtain the trained transient stability discrimination model;

[0011] Using the transient response trajectory of electrical quantities in the early stage of a fault acquired in real time and the trained transient stability discrimination model, the transient instability probability is calculated, and the transient stability label is predicted based on the transient instability probability.

[0012] The predicted transient stability labels are evaluated to obtain the final transient stability discrimination result.

[0013] In another aspect of the present invention, the transient response trajectory of electrical quantities in the simulation dataset is obtained by extracting the voltage amplitude and phase angle of each bus of the power grid, and the active power and reactive power transmitted by each branch of the power grid. The transient stability label in the simulation dataset is obtained by judging the generator power angle trajectory.

[0014] In another aspect of the present invention, the simulation dataset is divided into a training dataset and a reference dataset according to a preset ratio, and the training includes: training using the training dataset;

[0015] The transient stability discrimination model is based on a neural network with temporal feature processing capabilities, and the neural network includes at least one of a long short-term memory network, a gated recursive unit, and a Transformer network.

[0016] In another aspect of the present invention, evaluating the predicted transient stability label to obtain a final transient stability discrimination result includes:

[0017] Based on the trained transient stability discrimination model and the reference dataset, the instability probability prediction bias of each sample in the reference dataset is obtained;

[0018] Based on the instability probability prediction deviation, calculate the critical confidence index of the predicted transient stability label;

[0019] Based on the comparison between the critical confidence index and the preset critical threshold, if the critical confidence index is greater than the preset critical threshold, the predicted transient stability label is the final transient stability judgment result; otherwise, it is directly judged as unstable.

[0020] In another aspect of the present invention, the formula for calculating the transient instability probability is:

[0021] ;

[0022] in, This represents the transient response trajectory of electrical quantities. For corresponding The probability of transient instability, For transient stability discrimination model;

[0023] Based on the transient instability probability, predict the transient stability label, including:

[0024] If the transient instability probability is greater than or equal to the critical value for determining instability, instability is predicted.

[0025] If the transient instability probability is less than the critical value for determining instability, it is predicted to be stable;

[0026] The instability probability prediction bias is calculated using the following formula:

[0027] ;

[0028] in, For the corresponding electrical quantity transient response trajectory The instability probability prediction bias For corresponding The probability of transient instability, For corresponding The true transient stability label.

[0029] In another aspect of the invention, the transient stability label is predicted using the following formula:

[0030] ;

[0031] In the formula: To correspond to the transient response trajectory of electrical quantities The predicted transient stability label, This is an indicator function; if its input logic is true, it outputs label 1; otherwise, it outputs label 0. For corresponding The probability of transient instability, The critical value for determining instability.

[0032] In another aspect of the present invention, the critical confidence index is calculated using the following formula:

[0033] ;

[0034] in, This is the critical credibility index. This represents the cumulative probability for the corresponding condition. For the corresponding electrical quantity transient response trajectory The probability of transient instability, The critical value for determining instability.

[0035] In another aspect of the present invention, the cumulative probability of the corresponding condition is calculated using the following formula:

[0036] ;

[0037] ;

[0038] in, for The probability density function is characterized by a Gaussian distribution with Newman boundary conditions.

[0039] In another aspect of the present invention, the probability density function is expressed using the following formula:

[0040] ;

[0041] in, for The probability density function, for variance For statistical counting, for The mean.

[0042] In another aspect of the invention, the variance and mean are calculated using the following formula:

[0043] ;

[0044] in, for The mean, For transient stability discrimination model, This is the transient response trajectory of electrical quantities in the initial stage of a fault, acquired in real time. The preset number of neighborhood samples, For statistical counting, For reference dataset The first in the neighborhood For each sample, the distance metric within its neighborhood is Euclidean distance. Neighborhood samples are selected from the reference dataset. The smallest Euclidean distance One sample, To correspond to the transient response trajectory of electrical quantities The instability probability prediction bias.

[0045] (III) Beneficial Effects

[0046] The above-described technical solution of the present invention has the following beneficial technical effects:

[0047] This invention significantly improves the accuracy and generalization ability of the transient stability discrimination model by constructing an electromechanical transient simulation dataset covering various operating modes and fault types, and using the transient response trajectory of electrical quantities as the core feature input. A critical confidence index is introduced to quantify the uncertainty of the model output results. By measuring the model output results through the critical confidence index, scenarios where unstable conditions are missed as stable conditions are reduced, improving the conservatism and accuracy of the discrimination results. This enhances the accuracy of unstable scenario discrimination, ensures the conservatism of stability discrimination, and reduces the occurrence of serious accidents caused by missed unstable scenarios. Attached Figure Description

[0048] Figure 1 This is an overall flowchart of one embodiment of the present invention;

[0049] Figure 2 This is a power grid topology diagram according to an embodiment of the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments and the accompanying drawings. It should be understood that these descriptions are merely exemplary and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0051] Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0052] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0053] The invention will now be described in more detail with reference to the accompanying drawings. In the various drawings, the same elements are indicated by similar reference numerals. For clarity, the various parts in the drawings are not drawn to scale.

[0054] Example 1

[0055] A method for determining the transient stability of a power grid based on a critical confidence index. Figure 1 An overall flowchart of one embodiment of the present invention is shown, as follows: Figure 1 As shown, taking a real power grid in a certain region of China as an example, Figure 2 A power grid topology diagram according to an embodiment of the present invention is shown, such as Figure 2As shown, the power grid is the main grid of the provincial power system, which includes 15 synchronous generating units, 197 three-phase nodes, 24 new energy power plants and 4 DC transmission channels. The output of new energy can account for up to 50%.

[0056] The discrimination method includes:

[0057] Based on preset operating modes and corresponding fault combinations, electromechanical transient simulation is performed to obtain a simulation dataset. This dataset includes electrical quantity transient response trajectories and transient stability labels. In this embodiment, four typical operating modes and 3000 fault combinations are selected, including different N-1, N-2, and N-3 line combinations, different fault locations, and different clearing times. PSD-BPA software is used for transient simulation. The electrical quantity transient response trajectories in the simulation dataset are obtained by extracting the voltage amplitude and phase angle of each bus in the power grid, and the active and reactive power transmitted by each branch. Each electrical quantity transient response trajectory includes multiple curves. The voltage amplitude and phase angle of each bus, and the active and reactive power transmitted by each branch are each plotted with time as the X-axis and the corresponding value as the Y-axis, with one value per sampling point, forming multiple curves that serve as the electrical quantity transient response trajectories. These can be obtained by collecting the transient responses of these electrical quantities in the system after a fault. In this embodiment, there are a total of 881 electrical quantities, and each trajectory contains data from 30 time points before the fault, during the fault duration, and after the fault ends. The transient stability label in the simulation dataset is obtained by judging the generator power angle trajectory. The power angle has a true value in the simulation dataset, and the generator power angle trajectory is a curve with time on the X-axis and power angle on the Y-axis. In this embodiment, the transient stability label of the power system is determined by the generator power angle trajectory over an 8-second full-time domain. The transient stability label is denoted as […]. The range of values ​​is The value is denoted by 0, where 0 indicates transient stability and 1 indicates transient instability. The transient stability label uses the Transient Stability Index (TSI) to determine the stability of the system. The Transient Stability Index is defined as follows:

[0058] ;

[0059] In the formula: —An empirical threshold for judging transient stability; —The maximum power angle difference between any two generators during the transient process. When When the time is right, it is considered stable; otherwise, it is considered unstable.

[0060] The acquisition time of the electrical quantity transient response trajectory is the initial stage of the fault, i.e. the initial stage of the fault transient process. It includes the predetermined time of system operation before the fault, the duration of the fault, and the predetermined time after the fault is cleared. It can be set according to the actual situation. The predetermined time after the fault is cleared cannot exceed the decision time of safety control measures stipulated in the safety and stability guidelines.

[0061] A transient stability discrimination model is constructed. The input of the transient stability discrimination model is the transient response trajectory of the electrical quantity, and the output is the transient stability label. The transient stability discrimination model is based on a neural network with time-series feature processing capability. The neural network includes at least one of long short-term memory network, gated recursive unit and Transformer network.

[0062] The transient stability discrimination model is trained using the simulation dataset to obtain the trained transient stability discrimination model. The simulation dataset is divided into a training dataset and a reference dataset according to a preset ratio. The training includes training using the training dataset. The ratio of the training dataset to the reference dataset should satisfy the condition that the training dataset accounts for no less than 50% of the samples and the reference dataset accounts for no less than 20%. Further, the simulation dataset is divided into a training dataset, a reference dataset, and a test dataset according to a preset ratio. The simulation dataset is divided into a training dataset, a reference dataset, and a test set in a 6:2:2 ratio, where the test set is used to verify the final effect. In this embodiment, the transient stability discrimination model is constructed based on the GRU algorithm and trained using the training dataset.

[0063] Using the real-time acquired transient response trajectory of electrical quantities in the initial stage of a fault and the trained transient stability discrimination model, the transient instability probability is calculated, and the transient stability label is predicted based on the transient instability probability. In this embodiment, after the fault occurs, the transient response trajectory of electrical quantities is acquired through a phasor measurement unit, and the type of electrical quantity and the trajectory duration should be consistent. The formula for calculating the transient instability probability is:

[0064] ;

[0065] in, This represents the transient response trajectory of electrical quantities. For corresponding The probability of transient instability, This is a transient stability discrimination model.

[0066] Based on the transient instability probability, predict the transient stability label, including:

[0067] If the transient instability probability is greater than or equal to the critical value for determining instability, instability is predicted.

[0068] If the transient instability probability is less than the critical value for determining instability, the system is predicted to be stable; the critical value for instability can be obtained through empirical values ​​or simulation.

[0069] The transient stability label is predicted using the following formula:

[0070] ;

[0071] In the formula: To correspond to the transient response trajectory of electrical quantities The predicted transient stability label, This is an indicator function; if its input logic is true, it outputs label 1; otherwise, it outputs label 0. For corresponding The probability of transient instability, In this embodiment, the critical value for determining instability is 0.5.

[0072] The predicted transient stability labels are evaluated to obtain the final transient stability discrimination result, including:

[0073] Based on the trained transient stability discrimination model and the reference dataset, the instability probability prediction bias of each sample in the reference dataset is obtained; the instability probability prediction bias is calculated using the following formula:

[0074] ;

[0075] in, For the corresponding electrical quantity transient response trajectory The instability probability prediction bias For corresponding The probability of transient instability, For corresponding The true transient stability label.

[0076] Based on the instability probability prediction deviation, the critical confidence index of the predicted transient stability label is calculated; the critical confidence index is calculated using the following formula:

[0077] ;

[0078] in, This is the critical credibility index. This represents the cumulative probability for the corresponding condition. For the corresponding electrical quantity transient response trajectory The probability of transient instability, The critical value for determining instability.

[0079] The cumulative probability of the corresponding condition is calculated using the following formula:

[0080] ;

[0081] ;

[0082] in, for The probability density function is characterized by a Gaussian distribution with Newman boundary conditions.

[0083] The probability density function is expressed using the following formula:

[0084] ;

[0085] in, for The probability density function, for variance For statistical counting, for The mean.

[0086] The variance and mean are calculated using the following formula:

[0087] ;

[0088] in, for The mean, For transient stability discrimination model, This is the transient response trajectory of electrical quantities in the initial stage of a fault, acquired in real time. The preset number of neighborhood samples, For statistical counting, For reference dataset The first in the neighborhood For each sample, the distance metric within its neighborhood is Euclidean distance. Neighborhood samples are selected from the reference dataset. The smallest Euclidean distance One sample, To correspond to the transient response trajectory of electrical quantities The instability probability prediction bias.

[0089] Based on the comparison between the critical confidence index and a preset critical threshold, if the critical confidence index is greater than the preset critical threshold, the predicted transient stability label is the final transient stability determination result; otherwise, it is directly determined as instability. In this embodiment, the preset critical threshold is set to 0.8.

[0090] In this embodiment, the results of the power grid transient stability discrimination based on the critical confidence index are shown in Table 1.

[0091] Table 1. Results of power grid transient stability assessment using the GRU model and based on the critical confidence index.

[0092]

[0093] The test results recorded in Table 1 show that the method proposed in this embodiment can effectively improve the discrimination accuracy of unstable samples (from 96.54% to 99.81%), and the overall discrimination accuracy is improved from 97.21% to 97.82%.

[0094] It should be understood that the specific embodiments described above are merely illustrative or explanatory of the principles of the invention and do not constitute a limitation thereof. Therefore, any modifications, equivalent substitutions, improvements, etc., made without departing from the spirit and scope of the invention should be included within the protection scope of the invention. Furthermore, the appended claims are intended to cover all variations and modifications falling within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries.

[0095] The present invention has been described above with reference to embodiments thereof. However, these embodiments are merely illustrative and not intended to limit the scope of the invention. The scope of the invention is defined by the appended claims and their equivalents. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of the invention, and all such substitutions and modifications should fall within the scope of the invention.

[0096] Although embodiments of the present invention have been described in detail, it should be understood that various changes, substitutions, and modifications can be made to the embodiments of the present invention without departing from the spirit and scope of the invention.

[0097] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for power grid transient stability discrimination based on critical trust index, characterized in that, The method comprises the following steps: Based on the preset operation mode and the corresponding fault combination, electromechanical transient simulation is performed to obtain a simulation data set, which includes an electrical quantity transient response trajectory and a transient stability label; A transient stability discrimination model is constructed, the input of the transient stability discrimination model is the electrical quantity transient response trajectory, and the output is the transient stability label; The transient stability discrimination model is trained using the simulation data set to obtain a trained transient stability discrimination model; The transient instability probability is calculated using the real-time collected electrical quantity transient response trajectory at the initial stage of the fault and the trained transient stability discrimination model, and the transient stability label is predicted according to the transient instability probability; The predicted transient stability label is evaluated to obtain a final transient stability discrimination result, which comprises the following steps: Based on the trained transient stability discrimination model and a reference data set, the instability probability prediction deviation of each sample in the reference data set is obtained; According to the instability probability prediction deviation, the critical confidence index of the predicted transient stability label is calculated; The critical confidence index is compared with a preset critical threshold value, if the critical confidence index is greater than the preset critical threshold value, the predicted transient stability label is the final transient stability discrimination result, otherwise, it is directly discriminated as unstable; The critical confidence index is calculated by the following formula: ； in, This is the critical credibility index. This represents the cumulative probability for the corresponding condition. For the corresponding electrical quantity transient response trajectory The probability of transient instability, The critical value for determining instability.

2. The method for power system transient stability discrimination based on criticality confidence index according to claim 1, characterized in that, The electrical quantity transient response trajectory in the simulation data set is obtained by extracting the voltage amplitude and phase angle of each bus in the power grid and the active power and reactive power transmitted by each branch in the power grid, and the transient stability label in the simulation data set is obtained by judging the generator power angle trajectory.

3. The method for power system transient stability discrimination based on criticality confidence index according to claim 1, characterized in that, The simulation data set is divided into a training data set and a reference data set according to a preset proportion, and the training comprises training using the training data set; The transient stability discrimination model is based on a neural network with time sequence feature processing capability, and the neural network comprises at least one of a long short-term memory network, a gated recurrent unit and a Transformer network.

4. The power grid transient stability discrimination method based on the critical confidence index according to claim 3, wherein The transient instability probability calculation formula is: ； in, This represents the transient response trajectory of electrical quantities. For corresponding The probability of transient instability, For transient stability discrimination model; According to the transient instability probability, the transient stability label is predicted, which comprises the following steps: If the transient instability probability is greater than or equal to the critical value of discriminating instability, it is predicted to be unstable; If the transient instability probability is less than the critical value of discriminating instability, it is predicted to be stable; The instability probability prediction deviation is calculated by the following formula: ； wherein, is the instability probability prediction bias for the transient response trajectory of the corresponding electrical quantity, is the transient instability probability corresponding to is the true transient stability label corresponding to the transient response trajectory of the corresponding electrical quantity.​​ 5. The method for power system transient stability discrimination based on criticality confidence index according to claim 4, characterized in that, The transient stability label is predicted by the following formula: ； wherein: is a predicted transient stability label corresponding to the trajectory of the electrical quantities transient response, is an indicator function that outputs label 1 if its input logic is true, and label 0 otherwise, is a transient instability probability corresponding to is a threshold value that discriminates instability.​​ 6. The method of claim 5, wherein the criticality index is determined based on a ratio of a difference between the current and the reference value to the reference value. The cumulative probability of the corresponding condition is calculated by the following formula: ； ； wherein is a probability density function, characterized by a Gaussian distribution with Neumann boundary conditions.

7. The method of claim 6, wherein the criticality index is determined based on a ratio of a difference between the current and the previous value of the parameter and a difference between the previous value of the parameter and the value of the parameter at the previous time step. The probability density function is represented by the following formula: ； wherein, is the probability density function, is the variance, is the statistical count, is the mean.

8. The method of claim 7, wherein the criticality index is determined based on a ratio of a difference between the current and the previous value of the parameter and a difference between the previous value of the parameter and the value of the parameter at the previous time step. The variance and mean value are calculated by the following formula: ； wherein, is the mean value of, is a transient stability discrimination model, is a real-time collected electrical quantity transient response trajectory at the initial stage of a fault, is a preset number of neighborhood samples, is a statistical count, is the th sample in the neighborhood of the th sample in the reference data set, the distance measure in the neighborhood range is the Euclidean distance, and the neighborhood sample selection is the th sample in the reference data set with the minimum Euclidean distance, th sample in the reference data set with the minimum Euclidean distance, is a prediction deviation of the instability probability corresponding to the electrical quantity transient response trajectory .

Images