Method and system for judging transient stability of power system after fault

Abstract

The invention provides a method and a system for judging transient stability of a power system after a fault. The method comprises the following steps: acquiring an initial power system operation state parameter; determining a PEBS exit point and a critical energy function value on the basis, determining a dominant unstable equilibrium point and calculating a transient critical energy threshold value; based on the two energy values, obtaining a difference value, and if the difference value is smaller than or equal to a preset difference value threshold value, obtaining a sample set and establishing an initial neural network model; updating the initial model based on the data and a preset weight matrix until convergence, and obtaining a neural network model; and acquiring a corresponding Lyapunov function value based on the data and the neural network model, if the Lyapunov function value is smaller than a preset stability threshold value, determining stability, otherwise, determining instability, and realizing judgment of transient stability after the power system has a fault. According to the method for judging the transient stability of the power system after the fault, the problems of high parameter dependence and low prediction precision are avoided, and accurate analysis of steady-state analysis and judgment of the power system is realized.

Description

Technical Field

[0001] This invention relates to the field of power system analysis and control, and in particular to a method and system for determining transient stability after a power system fault. Background Technology

[0002] With the increasing global demand for clean energy, large-scale grid connection of new energy sources such as wind power and photovoltaics, as well as the widespread integration of distributed power sources, microgrids, and flexible loads into the power system, the new power system exhibits a multi-element coordinated operation of "source, grid, load, and storage," with dynamic changes in its topology and significantly increased uncertainty in its operating state. Traditional power system steady-state operation analysis mainly relies on precise physical models, solving the power system's steady-state equations to obtain the system's operating state. This method can accurately analyze steady-state operation in traditional power systems. However, in new power systems, with the large-scale integration of new energy sources and the dynamic changes in the grid topology, traditional methods face significant challenges. New energy output is affected by natural conditions, exhibiting intermittency and randomness; power load changes rapidly due to user behavior, seasonal variations, and other factors; and the grid topology is frequently adjusted due to the switching on and off of distributed power sources and line maintenance. These factors make it difficult for steady-state operation analysis methods based on traditional physical models to accurately and quickly adapt to changes, leading to decreased accuracy and reliability of the analysis results, failing to meet the requirements of online reliable steady-state operation analysis for new power systems.

[0003] Under the current technological background, steady-state operation analysis of power systems still mainly relies on traditional technical approaches, while some preliminary data-driven explorations exist. Traditional steady-state analysis methods based on physical models are currently the most widely used technology in power system dispatch and operation. Their core relies on accurate physical models of components and topology, obtaining the system's steady-state operating state by establishing and solving power flow equations. In new power systems, although this type of method supplements the modeling of new energy units, such as equating wind power and photovoltaics to PQ nodes or PV nodes, its core still depends on fixed physical formulas and parameter inputs. With the development of artificial intelligence technology, some studies have attempted to use data-driven algorithms to replace or assist physical model calculations. These methods attempt to circumvent the dependence of physical models on parameter accuracy by collecting historical operating data, constructing a mapping relationship between "input features and steady-state output," and directly predicting steady-state indicators by inputting the current system state after training the model. However, most are limited to single uncertainties and have not systematically studied the adaptability under dynamic topology changes. The intermittent nature of new energy output and the randomness of load lead to frequent changes in initial operating conditions. To cover these uncertainties, traditional methods require extensive scenario enumeration for analysis, resulting in a dramatic increase in computational load. New power systems demand online analysis with response times ranging from seconds to minutes to meet the needs of real-time scheduling and operational decision-making. The computational efficiency of traditional methods is clearly insufficient to meet this requirement, making it difficult to provide timely and effective analysis results when faced with rapidly changing power system operating states. Traditional physical models rely on fixed node-branch correlation matrices to describe the power grid topology. When the grid topology changes, the power flow equations need to be reconstructed and the dimensions of the Jacobian matrix adjusted. The entire model reconstruction process requires manual intervention or pre-set rules, resulting in excessively long response times. This makes it impossible to cope with the "high-frequency dynamic" characteristics of topologies in new power systems, easily leading to a disconnect between steady-state analysis results and actual operating states, thus failing to provide accurate evidence for the safe and stable operation of the power system. Existing data-driven steady-state analysis methods rely on time-domain simulation and lack quantitative evaluation capabilities. Training data is mostly derived from time-consuming time-domain simulations, which can only output qualitative results of "stable / unstable" and cannot quantitatively provide key indicators such as stability margin and critical fault clearing time. This makes it difficult to support dispatchers in formulating precise prevention and control strategies. Moreover, the time-consuming nature of time-domain simulation itself will offset the rapid analysis advantage of data-driven methods. Data-driven models do not incorporate basic physical constraints of power systems (such as Kirchhoff's laws and Lyapunov stability conditions), making them "black box" models. When the operating point is outside the range of training data, the prediction results may violate physical laws, resulting in a sharp drop in accuracy and an inability to interpret the physical meaning of the prediction results. Furthermore, the impact of random fluctuations in wind power output on the unstable equilibrium point (UEP) of the system has not been systematically studied, and no suitable mathematical models and data mapping relationships have been constructed for new topologies such as wind power virtual synchronous generators (VSGs). This makes it difficult to cope with new power systems dominated by power electronic devices. Summary of the Invention

[0004] The present invention aims to provide a method and system for determining transient stability after a power system fault, so as to solve the above-mentioned technical problems, avoid the problems of strong dependence on parameters and low prediction accuracy, and achieve accurate analysis and determination of power system steady state.

[0005] To address the aforementioned technical problems, this invention provides a method for determining transient stability after a power system fault, comprising:

[0006] Obtain initial power system operating status parameters;

[0007] Based on the initial power system operating state parameters, determine the PEBS exit point and critical energy function value;

[0008] Based on the PEBS exit point, the dominant unstable equilibrium point is determined and the transient critical energy threshold is calculated;

[0009] Based on the critical energy function value and the transient critical energy threshold, the difference is obtained. If the difference is less than or equal to the preset difference threshold, the sample set is obtained and the initial neural network model is established.

[0010] The initial neural network model is iteratively updated based on the sample set, the initial neural network model, and the preset weight matrix until the preset convergence condition is met, thereby obtaining the neural network model.

[0011] Based on the initial power system operating state parameters and the neural network model, the Lyapunov function value corresponding to the initial power system operating state parameters is obtained. If the Lyapunov function value is less than the preset stability threshold, the power system is determined to be stable. If the Lyapunov function value is greater than or equal to the preset stability threshold, the power system is determined to be unstable, thus realizing the determination of transient stability after a power system fault.

[0012] In the above scheme, the PEBS exit point and critical energy function value are initially determined using initial power system operating state parameters, laying the groundwork for more accurate calculation of the unstable equilibrium point. Next, the PEBS exit point allows for more precise identification of the dominant unstable equilibrium point of the power system, yielding a transient critical energy threshold that better reflects actual fault scenarios and improving the reliability of the critical energy index. Then, by calculating the difference between the critical energy function value and the transient critical energy threshold, if the difference is less than or equal to a preset difference threshold, a sample set is acquired and an initial neural network model is established. This ensures data validity before acquiring the samples and initial model framework required for subsequent neural network model training, providing a data and model foundation. Subsequently, the neural network model parameters are continuously optimized using the sample set, and a preset weight matrix ensures the rationality of the training direction, enabling the final neural network model to meet the preset convergence conditions and satisfy the model requirements for transient stability determination. Finally, the corresponding Lyapunov function value is obtained using the initial power system operating state parameters and the neural network model. This value is then compared with the preset stability threshold to determine whether the power system is transiently stable after a fault, achieving accurate determination of the transient stability state after a power system fault. The above steps can avoid the problems of strong dependence on parameters and low prediction accuracy, and achieve accurate analysis and judgment of power system steady state.

[0013] Furthermore, it also includes:

[0014] Based on the critical energy function value and the transient critical energy threshold, the difference is obtained. If the difference is greater than the preset difference threshold, the dominant unstable equilibrium point is determined again based on the PEBS exit point, and the transient critical energy threshold is calculated.

[0015] In the above scheme, if the difference is greater than the preset difference threshold, the dominant unstable equilibrium point is re-determined based on the PEBS exit point and the transient critical energy threshold is recalculated. This avoids distortion of subsequent analysis results due to single calculation deviations. By recalculating and correcting the deviations, the reliability of the transient critical energy threshold is ensured, providing accurate data for subsequent steps.

[0016] Furthermore, the determination of several PEBS exit points and critical energy function values ​​based on initial power system operating state parameters includes:

[0017] Based on the initial power system operating state parameters, the generator motion equations are established;

[0018] The transient energy function is obtained based on the generator's equation of motion.

[0019] Based on the transient energy function and the preset fault trajectory, several PEBS exit points and critical energy function values ​​are determined.

[0020] In the above scheme, generator motion equations are established using initial power system operating state parameters, constructing a dynamic mathematical model that conforms to the physical mechanism of the power system. This provides the underlying physical basis for subsequent transient energy function derivation and fault trajectory analysis, ensuring that subsequent calculations are consistent with the actual dynamic characteristics of the power system. Next, the transient energy function is obtained through the generator motion equations, transforming the dynamic behavior of the power system into a quantitative indicator of energy change, providing a mathematical tool for PEBS exit point determination and critical energy calculation. Subsequently, the PEBS exit point is located using the transient energy function and a preset fault trajectory. The PEBS exit point is then substituted into the transient energy function to obtain the critical energy function value, yielding several sets of results and improving the data reliability and coverage for subsequent calculations of the dominant unstable equilibrium point.

[0021] Furthermore, the step of determining the dominant unstable equilibrium point and calculating the transient critical energy threshold based on several PEBS exit points includes:

[0022] Integrate for any PEBS exit point, obtain the corresponding integral value, and determine the PEBS exit point corresponding to the integral value that meets the preset target conditions as the initial value;

[0023] Based on the initial values, solve for the dominant unstable equilibrium point;

[0024] The transient critical energy threshold is calculated based on the dominant unstable equilibrium point.

[0025] In the above scheme, by obtaining the integral value corresponding to any PEBS exit point, and then selecting the integral value that meets the preset target conditions, the initial value is determined. This provides an initial iteration point for the subsequent accurate solution of the dominant unstable equilibrium point, avoiding iteration divergence or result deviation due to improper selection of the initial value, thus improving the solution efficiency and accuracy. Next, the dominant unstable equilibrium point is determined using the initial value. Compared with directly using the PEBS exit point as the unstable equilibrium point, this method can further correct errors, making the obtained dominant unstable equilibrium point more consistent with the actual dynamic characteristics of the power system after a fault, and providing reliable parameters for subsequent transient critical energy calculation. Subsequently, by calculating the transient critical energy threshold of the dominant unstable equilibrium point, the energy boundary of the power system's transient stability is quantified, providing a key energy benchmark for subsequent comparison with the critical energy function value, neural network training, and final stability determination.

[0026] Further, the step of obtaining the difference based on the critical energy function value and the transient critical energy threshold, and if the difference is less than or equal to a preset difference threshold, then obtaining a sample set and establishing an initial neural network model; includes:

[0027] Based on the critical energy function value and the transient critical energy threshold, the difference is obtained. If the difference is less than or equal to the preset difference threshold, sampling is performed in the preset feasible region to obtain a preset number of samples.

[0028] Obtain a sample set based on a preset number of samples;

[0029] Establish an initial neural network model.

[0030] In the above scheme, sampling is conducted within a predefined feasible region to ensure that the collected samples cover possible operating states of the system. This avoids insufficient generalization ability of the subsequent model due to limited sample distribution, while also ensuring that the samples conform to the physical characteristics of the power system, laying the foundation for building a reliable sample set. Then, by integrating a predefined number of samples to obtain a sample set, the scattered sampling data is transformed into structured training data, providing standardized input for the initial neural network model training and ensuring that the model can learn mapping relationships based on comprehensive power system information.

[0031] Further, the iterative update of the initial neural network model based on the sample set, the initial neural network model, and the preset weight matrix until a preset convergence condition is met to obtain the neural network model includes:

[0032] Based on the sample set and the initial neural network model, obtain the first Lyapunov function value;

[0033] Based on the first Lyapunov function value and the preset weight matrix, a risk function is constructed;

[0034] The initial neural network model is iteratively updated based on the risk function until the preset convergence condition is met, thus obtaining the neural network model.

[0035] In the above scheme, the first Lyapunov function value is obtained through a sample set and an initial neural network model. The physical parameters of the power system in the sample set are transformed into a quantitative expression of the Lyapunov function, providing a computational basis for subsequent risk function construction and model optimization, ensuring that model learning revolves around Lyapunov stability determination. Next, a risk function is constructed using the first Lyapunov function value and a preset weight matrix, transforming the Lyapunov stability condition into an optimizable mathematical index while preventing model overfitting and providing a clear optimization objective for the iterative update of the initial neural network model. Finally, the initial neural network model is iteratively updated based on the risk function, continuously optimizing the model to accurately learn functional relationships that conform to the Lyapunov stability condition until a preset convergence condition is reached, ensuring that the output neural network model can reliably output Lyapunov function values ​​for transient stability determination.

[0036] This invention provides a system for determining transient stability after a power system fault, comprising a parameter acquisition module, an energy calculation module, an energy determination module, a neural network construction module, an iterative optimization module, and a transient determination module, specifically:

[0037] The parameter acquisition module is used to acquire initial power system operating status parameters;

[0038] The energy calculation module is used to determine several PEBS exit points and critical energy function values ​​based on the initial power system operating state parameters.

[0039] The energy determination module is used to determine the dominant unstable equilibrium point and calculate the transient critical energy threshold based on several PEBS exit points.

[0040] The neural network construction module is used to obtain the difference based on the critical energy function value and the transient critical energy threshold. If the difference is less than or equal to the preset difference threshold, then a sample set is obtained and an initial neural network model is established.

[0041] The iterative optimization module is used to iteratively update the initial neural network model based on the sample set, the initial neural network model and the preset weight matrix until the preset convergence condition is met, thereby obtaining the neural network model.

[0042] The transient determination module is used to obtain the Lyapunov function value corresponding to the initial power system operating state parameters and the neural network model. If the Lyapunov function value is less than the preset stability threshold, the power system is determined to be stable. If the Lyapunov function value is greater than or equal to the preset stability threshold, the power system is determined to be unstable, thereby realizing the determination of transient stability after a power system fault.

[0043] This invention provides a system for determining transient stability after a power system fault. In practical applications, it only requires a parameter acquisition module to initially determine the PEBS exit point and critical energy function value using initial power system operating state parameters, laying the groundwork for subsequent, more accurate calculations of the unstable equilibrium point. Next, an energy determination module, using the PEBS exit point, can more accurately pinpoint the dominant unstable equilibrium point of the power system, obtaining a transient critical energy threshold that better reflects the actual fault scenario and improving the reliability of the critical energy index. Then, a neural network construction module calculates the difference between the critical energy function value and the transient critical energy threshold. If the difference is less than or equal to a preset difference threshold, a sample set is acquired and an initial neural network model is built, ensuring data validity before acquiring the samples and initial model framework required for subsequent neural network model training, providing a data and model foundation. Finally, an iterative optimization module continuously optimizes the neural network model parameters using the sample set, combined with a preset weight matrix to ensure the rationality of the training direction, so that the final neural network model reaches the preset convergence condition, meeting the model requirements for transient stability determination. Finally, a transient determination module is employed. This module obtains the corresponding Lyapunov function value using the initial power system operating state parameters and the neural network model. This value is then compared to a preset stability threshold to determine whether the power system is transiently stable after a fault, thus achieving accurate determination of the transient stability state after a power system fault. These steps avoid the problems of strong parameter dependence and low prediction accuracy, enabling accurate analysis and determination of the power system's steady-state state.

[0044] Furthermore, the neural network building module is also used for:

[0045] Based on the critical energy function value and the transient critical energy threshold, the difference is obtained. If the difference is greater than the preset difference threshold, the dominant unstable equilibrium point is determined again based on the PEBS exit point, and the transient critical energy threshold is calculated.

[0046] In the above scheme, if the difference is greater than the preset difference threshold, the dominant unstable equilibrium point is re-determined based on the PEBS exit point and the transient critical energy threshold is recalculated. This avoids distortion of subsequent analysis results due to single calculation deviations. By recalculating and correcting the deviations, the reliability of the transient critical energy threshold is ensured, providing accurate data for subsequent steps.

[0047] Furthermore, the energy calculation module is used to determine several PEBS exit points and critical energy function values ​​based on initial power system operating state parameters; including:

[0048] Based on the initial power system operating state parameters, the generator motion equations are established;

[0049] The transient energy function is obtained based on the generator's equation of motion.

[0050] Based on the transient energy function and the preset fault trajectory, several PEBS exit points and critical energy function values ​​are determined.

[0051] In the above scheme, generator motion equations are established using initial power system operating state parameters, constructing a dynamic mathematical model that conforms to the physical mechanism of the power system. This provides the underlying physical basis for subsequent transient energy function derivation and fault trajectory analysis, ensuring that subsequent calculations are consistent with the actual dynamic characteristics of the power system. Next, the transient energy function is obtained through the generator motion equations, transforming the dynamic behavior of the power system into a quantitative indicator of energy change, providing a mathematical tool for PEBS exit point determination and critical energy calculation. Subsequently, the PEBS exit point is located using the transient energy function and a preset fault trajectory. The PEBS exit point is then substituted into the transient energy function to obtain the critical energy function value, yielding several sets of results and improving the data reliability and coverage for subsequent calculations of the dominant unstable equilibrium point.

[0052] Furthermore, the energy determination module is used to determine the dominant unstable equilibrium point and calculate the transient critical energy threshold based on several PEBS exit points; including:

[0053] Integrate for any PEBS exit point, obtain the corresponding integral value, and determine the PEBS exit point corresponding to the integral value that meets the preset target conditions as the initial value;

[0054] Based on the initial values, solve for the dominant unstable equilibrium point;

[0055] The transient critical energy threshold is calculated based on the dominant unstable equilibrium point.

[0056] In the above scheme, by obtaining the integral value corresponding to any PEBS exit point, and then selecting the integral value that meets the preset target conditions, the initial value is determined. This provides an initial iteration point for the subsequent accurate solution of the dominant unstable equilibrium point, avoiding iteration divergence or result deviation due to improper selection of the initial value, thus improving the solution efficiency and accuracy. Next, the dominant unstable equilibrium point is determined using the initial value. Compared with directly using the PEBS exit point as the unstable equilibrium point, this method can further correct errors, making the obtained dominant unstable equilibrium point more consistent with the actual dynamic characteristics of the power system after a fault, and providing reliable parameters for subsequent transient critical energy calculation. Subsequently, by calculating the transient critical energy threshold of the dominant unstable equilibrium point, the energy boundary of the power system's transient stability is quantified, providing a key energy benchmark for subsequent comparison with the critical energy function value, neural network training, and final stability determination. Attached Figure Description

[0057] Figure 1 A flowchart illustrating a method for determining transient stability after a power system fault, provided in an embodiment of the present invention;

[0058] Figure 2This is an architecture diagram of a power system transient stability determination system provided in an embodiment of the present invention. Detailed Implementation

[0059] 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.

[0060] This embodiment addresses the steady-state analysis challenges brought about by the high proportion of new energy access, dynamic topology changes, and the dominance of power electronics in new power systems. It adopts a dual-path collaborative architecture of "data-model hybrid drive" and "pure data drive".

[0061] This embodiment provides a method for determining transient stability after a power system fault. Please refer to the flowchart below. Figure 1 ,include:

[0062] Step S1: Obtain initial power system operating status parameters;

[0063] Step S2: Based on the initial power system operating state parameters, determine the PEBS exit point and critical energy function value;

[0064] Step S3: Based on the PEBS exit point, determine the dominant unstable equilibrium point and calculate the transient critical energy threshold;

[0065] Step S4: Based on the critical energy function value and the transient critical energy threshold, obtain the difference. If the difference is less than or equal to the preset difference threshold, obtain the sample set and establish the initial neural network model.

[0066] Step S5: Iteratively update the initial neural network model based on the sample set, the initial neural network model, and the preset weight matrix until the preset convergence condition is met to obtain the neural network model;

[0067] Step S6: Based on the initial power system operating state parameters and the neural network model, obtain the Lyapunov function value corresponding to the initial power system operating state parameters. If the Lyapunov function value is less than the preset stability threshold, the power system is determined to be stable. If the Lyapunov function value is greater than or equal to the preset stability threshold, the power system is determined to be unstable, thus realizing the determination of transient stability after a power system fault.

[0068] In this embodiment, the PEBS exit point and critical energy function value are initially determined using initial power system operating state parameters, laying the groundwork for more accurate calculation of the unstable equilibrium point. Next, the PEBS exit point allows for more precise identification of the dominant unstable equilibrium point of the power system, yielding a transient critical energy threshold that better reflects actual fault scenarios and improving the reliability of the critical energy index. Then, by calculating the difference between the critical energy function value and the transient critical energy threshold, if the difference is less than or equal to a preset difference threshold, a sample set is acquired and an initial neural network model is established to ensure data validity before acquiring the samples and initial model framework required for subsequent neural network model training, providing a data and model foundation. Subsequently, the neural network model parameters are continuously optimized using the sample set, and a preset weight matrix is ​​used to ensure the rationality of the training direction, enabling the final neural network model to reach the preset convergence condition and meet the model requirements for transient stability determination. Finally, the corresponding Lyapunov function value is obtained using the initial power system operating state parameters and the neural network model, and this value is compared with the preset stability threshold c. * The magnitude of the value is used to determine whether the power system is transiently stable after a fault, thus achieving accurate determination of the transient stability state after a power system fault. The preset stability threshold c is used. * The specific process for obtaining the threshold is as follows: Candidate points randomly generated within the preset feasible region are input into the neural network model, and the corresponding Lyapunov function is calculated. Through numerical optimization, possible state combinations are traversed, and the maximum value is selected as the preset stability threshold c. * After a power system fault, the initial power system operating state parameters are input into a neural network model to calculate the corresponding Lyapunov function value V. If V <c * If V≥c, then the power system is in a stable region and the power system is transiently stable; * If the power system exceeds its stability region, it will become transiently unstable. The above steps avoid the problems of strong parameter dependence and low prediction accuracy, achieving accurate analysis and judgment of power system steady-state conditions. The Lyapunov function construction scheme based on neural networks in this embodiment is relatively conservative. However, on the one hand, conservatism is unavoidable in the fields of direct methods and classical transient stability analysis; on the other hand, transient synchronous stability judgment needs to rely on relatively conservative results—this not only provides a safety margin for fault identification and elimination, but also avoids the situation where the actual power system instability is not detected, leading to the failure of protection actions, thereby preventing the further expansion of the fault's impact.

[0069] Furthermore, it also includes:

[0070] Based on the critical energy function value and the transient critical energy threshold, the difference is obtained. If the difference is greater than the preset difference threshold, the dominant unstable equilibrium point is determined again based on the PEBS exit point, and the transient critical energy threshold is calculated.

[0071] In this embodiment, if the difference is greater than the preset difference threshold, the dominant unstable equilibrium point is re-determined based on the PEBS exit point and the transient critical energy threshold is recalculated. This avoids distortion of subsequent analysis results due to single calculation deviations. By recalculating and correcting the deviations, the reliability of the transient critical energy threshold is ensured, providing accurate data for subsequent steps.

[0072] Furthermore, the determination of several PEBS exit points and critical energy function values ​​based on initial power system operating state parameters includes:

[0073] Based on the initial power system operating state parameters, the generator motion equations are established;

[0074] The transient energy function is obtained based on the generator's equation of motion.

[0075] Based on the transient energy function and the preset fault trajectory, several PEBS exit points and critical energy function values ​​are determined.

[0076] In this embodiment, the generator motion equations are established using the initial power system operating state parameters. For a multi-machine system, the following assumptions are made: the n generators adopt a classical second-order model, the load adopts a constant impedance model, the network transfer conductance is ignored, and the network nodes are shrunk to nodes within the generators. Then, the generator motion equations can be expressed as follows:

[0077]

[0078] In the formula, Let ω represent the rotor angle of the i-th generator. i Let P represent the rotor angular velocity of the i-th generator. mi M represents the mechanical power of the i-th generator. i Represents the unit inertia constant E of the i-th generator. i B represents the amplitude of the internal electromotive force of the i-th generator. ij The transfer susceptance from node i to node j is sin(δ). i -δ j The sine value of the rotor angle difference is represented by δ. All parameters mentioned above are obtained from the initial power system operating state parameters. The nth generator is taken as the reference generator. (n-1) relative rotor angles and (n-1) relative angular velocities are defined as follows: δ m =δ i -δ n ,ω m =ω i -ω n If i = 1, 2, ..., n-1, then the system equations after the fault can be simplified as follows:

[0079]

[0080] The equilibrium point of the power system is the solution where ω = 0 and f(δ) = 0. In the formula, δ and ω are both (n-1)-dimensional vectors; f is an (n-1)-dimensional function vector with the following components:

[0081]

[0082] In the formula, P mn M represents the mechanical power of the nth generator. n The inertial constant of the nth generator, E n B represents the amplitude of the internal electromotive force of the nth generator. nj The transfer susceptance from node n to node j is sin(δ). in -δ jn ) represents the sine value of the rotor angular difference, sinδ nj This represents the sine value of the rotor angle difference between the nth reference machine and the jth generator. By constructing a dynamic mathematical model that conforms to the physical mechanism of the power system, a fundamental physical basis is provided for subsequent derivation of the transient energy function and fault trajectory analysis, ensuring that subsequent calculations are consistent with the actual dynamic characteristics of the power system. Next, the transient energy function is obtained through the generator's equations of motion:

[0083]

[0084] In the formula, V(δ,ω) represents the transient energy function, V KE (ω) represents the transient kinetic energy, V PE (δ) represents transient potential energy, used to describe the "potential energy" stored in a power system due to rotor angular deviation and electromagnetic power interaction. It is a core physical quantity for judging transient stability and consists of two parts: mechanical power potential energy and electromagnetic power potential energy. Electromagnetic power potential energy ω represents the steady-state rotor angle of the i-th generator; i M represents the rotor angular velocity of the i-th generator, which is the difference between the actual angular velocity and the synchronous angular velocity ω0 (a preset value), reflecting the change in the generator rotor speed; i E represents the moment of inertia coefficient of the i-th generator, used to characterize the rotor's ability to store kinetic energy; i E j Let represent the internal electromotive force amplitudes of the i-th and j-th generators, respectively; This represents the steady-state rotor angular difference. Based on the transient energy function expression, the generator's equations of motion can be rewritten as follows:

[0085]

[0086] By transforming the dynamic behavior of the power system into a quantitative indicator of energy change, a mathematical tool is provided for determining the PEBS exit point and calculating the critical energy. Subsequently, the PEBS exit point is located using a transient energy function and a pre-defined fault trajectory. Specifically, the PEBS (Potential Energy Boundary Surface) is formed by the trajectory connecting the points of maximum potential energy along any direction from the equilibrium point. Taking the time derivative of the transient energy function and expanding the derivation, the following result is obtained: Inside the PEBS, V... PE >0; outside of PEBS, V PE <0; and above PEBS, V PE =0. At this point, the directional derivative is equal to zero, that is:

[0087]

[0088] m represents the participating transient energy (V) PE The number of generators analyzed. Therefore, when V PE When (δ) = 0, this point is the PEBS exit point. Substituting several PEBS exit points into the transient energy function yields the critical energy function value, obtaining several sets of results to improve the data reliability and coverage of subsequent calculations of dominant unstable equilibrium points. Finally, based on maxV... PE (δ)=V cr Determine the critical energy function value.

[0089] Furthermore, the step of determining the dominant unstable equilibrium point and calculating the transient critical energy threshold based on several PEBS exit points includes:

[0090] Integrate for any PEBS exit point, obtain the corresponding integral value, and determine the PEBS exit point corresponding to the integral value that meets the preset target conditions as the initial value;

[0091] Based on the initial values, solve for the dominant unstable equilibrium point;

[0092] The transient critical energy threshold is calculated based on the dominant unstable equilibrium point.

[0093] In this embodiment, the dominant unstable equilibrium point is solved using the BCU method. The BCU method is developed based on the PEBS method and is used to calculate unstable equilibrium points more accurately. This is achieved by obtaining the integral value corresponding to any PEBS exit point: Then, select the integral values ​​that meet the preset target conditions (the first time the minimum value is reached; the specific value of this minimum value needs to be obtained through professional analysis and calculation based on the actual situation of the power system) and determine them as the initial values. This provides an initial iteration point for accurate subsequent solving of the dominant unstable equilibrium point, avoiding iteration divergence or result deviation due to improper initial value selection, and improving solution efficiency and accuracy. Next, the equation f(δ) = 0 is solved using the initial value and Newton's iteration method to determine the dominant unstable equilibrium point δ. UEP Compared to directly using the PEBS exit point as the unstable equilibrium point, this method can further correct errors, making the obtained dominant unstable equilibrium point more consistent with the actual dynamic characteristics of the power system after a fault, thus providing reliable parameters for subsequent transient critical energy calculations. Subsequently, δ... UEP Substitute V into the preset MSVR prediction model PE (δ) Calculate the transient critical energy threshold: V cr =V PE (δ UEP )=V(δ UP By calculating the transient critical energy threshold, the energy boundary of transient stability of the power system is quantified. Furthermore, compared with the PEBS method, the BCU method can find a more accurate unstable equilibrium point, providing a key energy benchmark for subsequent comparison with the critical energy function value, neural network training, and final stability determination.

[0094] Further, the step of obtaining the difference based on the critical energy function value and the transient critical energy threshold, and if the difference is less than or equal to a preset difference threshold, then obtaining a sample set and establishing an initial neural network model; includes:

[0095] Based on the critical energy function value and the transient critical energy threshold, the difference is obtained. If the difference is less than or equal to the preset difference threshold, sampling is performed in the preset feasible region to obtain a preset number of samples.

[0096] Obtain a sample set based on a preset number of samples;

[0097] Establish an initial neural network model.

[0098] In this embodiment, random sampling is performed within a preset feasible region D, and each sample contains power system state variables and associated parameters (such as M). i P mi P ei (electromagnetic power P) ei It is the power transmitted by the generator through the power grid, and its formula is: P ei =∑ j≠i E i E j B ij sinδ ijThis process ensures that the collected samples cover possible operating states of the system, avoiding insufficient generalization ability of subsequent models due to limited sample distribution, while also ensuring that the samples conform to the physical characteristics of the power system, laying the foundation for building a reliable sample set. Then, by integrating a predetermined number of samples (N), a sample set X = {X1, X2, ..., X...} is obtained. N This transforms scattered sampled data into structured training data, providing standardized input for training the initial neural network model and ensuring that the model can learn mapping relationships based on comprehensive power system information.

[0099] Further, the iterative update of the initial neural network model based on the sample set, the initial neural network model, and the preset weight matrix until a preset convergence condition is met to obtain the neural network model includes:

[0100] Based on the sample set, the preset weight matrix, and the initial neural network model, obtain the first Lyapunov function value;

[0101] Based on the first Lyapunov function value and the preset weight matrix, a risk function is constructed;

[0102] The initial neural network model is iteratively updated based on the risk function until the preset convergence condition is met, thus obtaining the neural network model.

[0103] In this embodiment, an initial neural network model for analyzing the transient stability of a power system is provided. This model uses a sample set, a preset weight matrix (initialized using a random normal distribution, such as a normal distribution matrix with a mean of 0 and a variance of 0.1), and an initial neural network model (with its bias term initialized to 0) to ensure the randomness and diversity of the initial neural network output. The input dimension, output dimension, hidden layer dimension, learning rate, and maximum number of iterations are preset. Then, the sample set X is input into the neural network. After activation operations in the hidden layers, the first Lyapunov function value under the current weights can be calculated using the forward propagation of the neural network. Since determining transient stability through the Lyapunov function requires function differentiation, the activation function must be differentiable. Therefore, the hypothesis class of the Lyapunov function is designed as a multilayer feedforward network using the tanh activation function. The expressions for the tanh function and its derivative are as follows:

[0104]

[0105] tanh(x) = 1 - tanh 2 x

[0106] This allows the physical parameters of the power system in the sample set to be transformed into a quantitative expression of a Lyapunov function, providing a computational basis for subsequent risk function construction and model optimization, and ensuring that model learning revolves around Lyapunov stability determination. Next, using the first Lyapunov function value and a preset weight matrix, the risk function R is constructed:

[0107]

[0108] in, (The gradient is calculated via automatic differentiation), where λ is the regularization coefficient (used to prevent overfitting, and can be taken as 10). -5 ~10 -2 W is a preset weight matrix, which transforms the Lyapunov stability condition into an optimizable mathematical index, while preventing overfitting and providing a clear optimization objective for the iterative update of the initial neural network model. Finally, the initial neural network model is iteratively updated based on the risk function, using stochastic gradient descent to update the weight matrix. This allows the model to continuously optimize and accurately learn the functional relationship that conforms to the Lyapunov stability condition until the preset convergence condition is met, ensuring that the output neural network model can reliably output the Lyapunov function value used for transient stability determination. The preset convergence condition can be analyzed from two dimensions: the convergence of the loss function and the gradient norm. Iteration stops when the iterative change of the risk function R is less than a preset accuracy threshold. Let the risk function for the k-th iteration be R. o The convergence condition is:

[0109] |R o -R o-1 |<ε

[0110] Where ε is the preset precision threshold, which can be empirically set to 10. -5 ~10 -3 .

[0111] The iteration stops when the gradient norm of the neural network parameters is less than a preset gradient threshold, as shown in the formula:

[0112]

[0113] in, It is the gradient of the risk function with respect to the neural network parameters. The preset gradient threshold can be empirically set to 10. -4 ~10 -2 This indicates that the parameter update magnitude is sufficiently small and the model tends to stabilize. By setting preset accuracy thresholds and preset gradient thresholds, it can be ensured that the output neural network model can reliably output the Lyapunov function value used for transient stability determination.

[0114] This embodiment provides a system for determining transient stability after a power system fault. Please refer to [link to relevant documentation]. Figure 2 It includes a parameter acquisition module, an energy calculation module, an energy determination module, a neural network construction module, an iterative optimization module, and a transient determination module, specifically:

[0115] The parameter acquisition module is used to acquire initial power system operating status parameters;

[0116] The energy calculation module is used to determine several PEBS exit points and critical energy function values ​​based on the initial power system operating state parameters.

[0117] The energy determination module is used to determine the dominant unstable equilibrium point and calculate the transient critical energy threshold based on several PEBS exit points.

[0118] The neural network construction module is used to obtain the difference based on the critical energy function value and the transient critical energy threshold. If the difference is less than or equal to the preset difference threshold, then a sample set is obtained and an initial neural network model is established.

[0119] The iterative optimization module is used to iteratively update the initial neural network model based on the sample set, the initial neural network model and the preset weight matrix until the preset convergence condition is met, thereby obtaining the neural network model.

[0120] The transient determination module is used to obtain the Lyapunov function value corresponding to the initial power system operating state parameters and the neural network model. If the Lyapunov function value is less than the preset stability threshold, the power system is determined to be stable. If the Lyapunov function value is greater than or equal to the preset stability threshold, the power system is determined to be unstable, thereby realizing the determination of transient stability after a power system fault.

[0121] This embodiment provides a system for determining transient stability after a power system fault. In practical applications, it only requires a parameter acquisition module to initially determine the PEBS exit point and critical energy function value using initial power system operating state parameters, laying the groundwork for subsequent calculations of more accurate unstable equilibrium points. Next, an energy determination module, using the PEBS exit point, can more accurately pinpoint the dominant unstable equilibrium point of the power system, obtaining a transient critical energy threshold that better reflects the actual fault scenario and improving the reliability of the critical energy index. Then, a neural network construction module calculates the difference between the critical energy function value and the transient critical energy threshold. If the difference is less than or equal to a preset difference threshold, a sample set is acquired and an initial neural network model is built, ensuring data validity before acquiring the samples and initial model framework required for subsequent neural network model training, providing a data and model foundation. Subsequently, an iterative optimization module continuously optimizes the neural network model parameters using the sample set, combining a preset weight matrix to ensure the rationality of the training direction, so that the final neural network model reaches the preset convergence condition, meeting the model requirements for transient stability determination. Finally, a transient determination module uses the initial power system operating state parameters and the neural network model to obtain the corresponding Lyapunov function value, and then compares this value with the preset stability threshold c. * The magnitude of the value is used to determine whether the power system is transiently stable after a fault, thus achieving accurate determination of the transient stability state after a power system fault. The preset stability threshold c is used. * The specific process for obtaining the threshold is as follows: Candidate points randomly generated within the preset feasible region are input into the neural network model, and the corresponding Lyapunov function is calculated. Through numerical optimization, possible state combinations are traversed, and the maximum value is selected as the preset stability threshold c. * After a power system fault, the initial power system operating state parameters are input into a neural network model to calculate the corresponding Lyapunov function value V. If V <c * If V≥c, then the power system is in a stable region and the power system is transiently stable; * If the power system exceeds its stability region, it will become transiently unstable. The above steps avoid the problems of strong parameter dependence and low prediction accuracy, achieving accurate analysis and judgment of power system steady-state conditions. The Lyapunov function construction scheme based on neural networks in this embodiment is relatively conservative. However, on the one hand, conservatism is unavoidable in the fields of direct methods and classical transient stability analysis; on the other hand, transient synchronous stability judgment needs to rely on relatively conservative results—this not only provides a safety margin for fault identification and elimination, but also avoids the situation where the actual power system instability is not detected, leading to the failure of protection actions, thereby preventing the further expansion of the fault's impact.

[0122] Furthermore, the neural network building module is also used for:

[0123] Based on the critical energy function value and the transient critical energy threshold, the difference is obtained. If the difference is greater than the preset difference threshold, the dominant unstable equilibrium point is determined again based on the PEBS exit point, and the transient critical energy threshold is calculated.

[0124] In this embodiment, if the difference is greater than the preset difference threshold, the dominant unstable equilibrium point is re-determined based on the PEBS exit point and the transient critical energy threshold is recalculated. This avoids distortion of subsequent analysis results due to single calculation deviations. By recalculating and correcting the deviations, the reliability of the transient critical energy threshold is ensured, providing accurate data for subsequent steps.

[0125] Furthermore, the energy calculation module is used to determine several PEBS exit points and critical energy function values ​​based on initial power system operating state parameters; including:

[0126] Based on the initial power system operating state parameters, the generator motion equations are established;

[0127] The transient energy function is obtained based on the generator's equation of motion.

[0128] Based on the transient energy function and the preset fault trajectory, several PEBS exit points and critical energy function values ​​are determined.

[0129] In this embodiment, the generator motion equations are established using the initial power system operating state parameters. For a multi-machine system, the following assumptions are made: the n generators adopt a classical second-order model, the load adopts a constant impedance model, the network transfer conductance is ignored, and the network nodes are shrunk to nodes within the generators. Then, the generator motion equations can be expressed as follows:

[0130]

[0131] In the formula, Let ω represent the rotor angle of the i-th generator. i Let P represent the rotor angular velocity of the i-th generator. mi M represents the mechanical power of the i-th generator. i Represents the unit inertia constant E of the i-th generator. i B represents the amplitude of the internal electromotive force of the i-th generator. ij The transfer susceptance from node i to node j is sin(δ). i -δ j The sine value of the rotor angle difference is represented by δ. All parameters mentioned above are obtained from the initial power system operating state parameters. The nth generator is taken as the reference generator. (n-1) relative rotor angles and (n-1) relative angular velocities are defined as follows: δ m =δ i -δ n ,ω m=ω i -ω n If i = 1, 2, ..., n-1, then the system equations after the fault can be simplified as follows:

[0132]

[0133] The equilibrium point of the power system is the solution where ω = 0 and f(δ) = 0. In the formula, δ and ω are both (n-1)-dimensional vectors; f is an (n-1)-dimensional function vector with the following components:

[0134]

[0135] In the formula, P mn M represents the mechanical power of the nth generator. n The inertial constant of the nth generator, E n B represents the amplitude of the internal electromotive force of the nth generator. nj The transfer susceptance from node n to node j is sin(δ). in -δ jn ) represents the sine value of the rotor angular difference, sinδ nj This represents the sine value of the rotor angle difference between the nth reference machine and the jth generator. By constructing a dynamic mathematical model that conforms to the physical mechanism of the power system, a fundamental physical basis is provided for subsequent derivation of the transient energy function and fault trajectory analysis, ensuring that subsequent calculations are consistent with the actual dynamic characteristics of the power system. Next, the transient energy function is obtained through the generator's equations of motion:

[0136]

[0137] In the formula, V(δ,ω) represents the transient energy function, V KE (ω) represents the transient kinetic energy, V PE (δ) represents transient potential energy, used to describe the "potential energy" stored in a power system due to rotor angular deviation and electromagnetic power interaction. It is a core physical quantity for judging transient stability and consists of two parts: mechanical power potential energy and electromagnetic power potential energy. Electromagnetic power potential energy ω represents the steady-state rotor angle of the i-th generator; i M represents the rotor angular velocity of the i-th generator, which is the difference between the actual angular velocity and the synchronous angular velocity ω0 (a preset value), reflecting the change in the generator rotor speed; i E represents the moment of inertia coefficient of the i-th generator, used to characterize the rotor's ability to store kinetic energy; i E j Let represent the internal electromotive force amplitudes of the i-th and j-th generators, respectively; This represents the steady-state rotor angular difference. Based on the transient energy function expression, the generator's equations of motion can be rewritten as follows:

[0138]

[0139] By transforming the dynamic behavior of the power system into a quantitative indicator of energy change, a mathematical tool is provided for determining the PEBS exit point and calculating the critical energy. Subsequently, the PEBS exit point is located using a transient energy function and a pre-defined fault trajectory. Specifically, the PEBS (Potential Energy Boundary Surface) is formed by the trajectory connecting the points of maximum potential energy along any direction from the equilibrium point. Taking the time derivative of the transient energy function and expanding the derivation, the following result is obtained: Inside the PEBS, V... PE >0; outside of PEBS, V PE <0; and above PEBS, V PE =0. At this point, the directional derivative is equal to zero, that is:

[0140]

[0141] m represents the participating transient energy (V) PE The number of generators analyzed. Therefore, when V PE When (δ) = 0, this point is the PEBS exit point. Substituting several PEBS exit points into the transient energy function yields the critical energy function value, obtaining several sets of results to improve the data reliability and coverage of subsequent calculations of dominant unstable equilibrium points. Finally, based on maxV... PE (δ)=V cr Determine the critical energy function value.

[0142] Furthermore, the energy determination module is used to determine the dominant unstable equilibrium point and calculate the transient critical energy threshold based on several PEBS exit points; including:

[0143] Integrate for any PEBS exit point, obtain the corresponding integral value, and determine the PEBS exit point corresponding to the integral value that meets the preset target conditions as the initial value;

[0144] Based on the initial values, solve for the dominant unstable equilibrium point;

[0145] The transient critical energy threshold is calculated based on the dominant unstable equilibrium point.

[0146] In this embodiment, the dominant unstable equilibrium point is solved using the BCU method. The BCU method is developed based on the PEBS method and is used to calculate unstable equilibrium points more accurately. This is achieved by obtaining the integral value corresponding to any PEBS exit point: Then, select the integral values ​​that meet the preset target conditions (the first time the minimum value is reached; the specific value of this minimum value needs to be obtained through professional analysis and calculation based on the actual situation of the power system) and determine them as the initial values. This provides an initial iteration point for accurate subsequent solving of the dominant unstable equilibrium point, avoiding iteration divergence or result deviation due to improper initial value selection, and improving solution efficiency and accuracy. Next, the equation f(δ) = 0 is solved using the initial value and Newton's iteration method to determine the dominant unstable equilibrium point δ. UEP Compared to directly using the PEBS exit point as the unstable equilibrium point, this method can further correct errors, making the obtained dominant unstable equilibrium point more consistent with the actual dynamic characteristics of the power system after a fault, thus providing reliable parameters for subsequent transient critical energy calculations. Subsequently, δ... UEP Substitute V into the preset MSVR prediction model PE (δ) Calculate the transient critical energy threshold: V cr =V PE (δ UEP )=V(δ UEP By calculating the transient critical energy threshold, the energy boundary of transient stability of the power system is quantified. Furthermore, compared with the PEBS method, the BCU method can find a more accurate unstable equilibrium point, providing a key energy benchmark for subsequent comparison with the critical energy function value, neural network training, and final stability determination.

[0147] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for determining transient stability after a power system fault, characterized in that, include: Obtain initial power system operating status parameters; Based on the initial power system operating state parameters, several PEBS exit points and critical energy function values ​​are determined. Based on several PEBS exit points, the dominant unstable equilibrium point is determined and the transient critical energy threshold is calculated; Based on the critical energy function value and the transient critical energy threshold, the difference is obtained. If the difference is less than or equal to the preset difference threshold, the sample set is obtained and the initial neural network model is established. The initial neural network model is iteratively updated based on the sample set, the initial neural network model, and the preset weight matrix until the preset convergence condition is met, thereby obtaining the neural network model. Based on the initial power system operating state parameters and the neural network model, the Lyapunov function value corresponding to the initial power system operating state parameters is obtained. If the Lyapunov function value is less than the preset stability threshold, the power system is determined to be stable. If the Lyapunov function value is greater than or equal to the preset stability threshold, the power system is determined to be unstable, thus realizing the determination of transient stability after a power system fault.

2. The method for determining transient stability after a power system fault according to claim 1, characterized in that, Also includes: Based on the critical energy function value and the transient critical energy threshold, the difference is obtained. If the difference is greater than the preset difference threshold, the dominant unstable equilibrium point is determined again based on the PEBS exit point, and the transient critical energy threshold is calculated.

3. The method for determining transient stability after a power system fault according to claim 1, characterized in that, The determination of several PEBS exit points and critical energy function values ​​based on initial power system operating state parameters includes: Based on the initial power system operating state parameters, the generator motion equations are established; The transient energy function is obtained based on the generator's equation of motion. Based on the transient energy function and the preset fault trajectory, several PEBS exit points and critical energy function values ​​are determined.

4. The method for determining transient stability after a power system fault according to claim 1, characterized in that, The process of determining the dominant unstable equilibrium point and calculating the transient critical energy threshold based on several PEBS exit points includes: Integrate for any PEBS exit point, obtain the corresponding integral value, and determine the PEBS exit point corresponding to the integral value that meets the preset target conditions as the initial value; Based on the initial values, solve for the dominant unstable equilibrium point; The transient critical energy threshold is calculated based on the dominant unstable equilibrium point.

5. The method for determining transient stability after a power system fault according to claim 1, characterized in that, The step involves obtaining a difference based on the critical energy function value and the transient critical energy threshold. If the difference is less than or equal to a preset difference threshold, a sample set is acquired and an initial neural network model is established. This includes: Based on the critical energy function value and the transient critical energy threshold, the difference is obtained. If the difference is less than or equal to the preset difference threshold, sampling is performed in the preset feasible region to obtain a preset number of samples. Obtain a sample set based on a preset number of samples; Establish an initial neural network model.

6. The method for determining transient stability after a power system fault according to claim 1, characterized in that, The step of iteratively updating the initial neural network model based on a sample set, an initial neural network model, and a preset weight matrix until a preset convergence condition is met, thereby obtaining the neural network model, includes: Based on the sample set and the initial neural network model, obtain the first Lyapunov function value; Based on the first Lyapunov function value and the preset weight matrix, a risk function is constructed; The initial neural network model is iteratively updated based on the risk function until the preset convergence condition is met, thus obtaining the neural network model.

7. A system for determining transient stability after a power system fault, characterized in that, It includes a parameter acquisition module, an energy calculation module, an energy determination module, a neural network construction module, an iterative optimization module, and a transient determination module, specifically: The parameter acquisition module is used to acquire initial power system operating status parameters; The energy calculation module is used to determine several PEBS exit points and critical energy function values ​​based on the initial power system operating state parameters. The energy determination module is used to determine the dominant unstable equilibrium point and calculate the transient critical energy threshold based on several PEBS exit points. The neural network construction module is used to obtain the difference based on the critical energy function value and the transient critical energy threshold. If the difference is less than or equal to the preset difference threshold, then a sample set is obtained and an initial neural network model is established. The iterative optimization module is used to iteratively update the initial neural network model based on the sample set, the initial neural network model and the preset weight matrix until the preset convergence condition is met, thereby obtaining the neural network model. The transient determination module is used to obtain the Lyapunov function value corresponding to the initial power system operating state parameters and the neural network model. If the Lyapunov function value is less than the preset stability threshold, the power system is determined to be stable. If the Lyapunov function value is greater than or equal to the preset stability threshold, the power system is determined to be unstable, thereby realizing the determination of transient stability after a power system fault.

8. The system for determining transient stability after a power system fault according to claim 7, characterized in that, The neural network building module is also used for: Based on the critical energy function value and the transient critical energy threshold, the difference is obtained. If the difference is greater than the preset difference threshold, the dominant unstable equilibrium point is determined again based on the PEBS exit point, and the transient critical energy threshold is calculated.

9. The system for determining transient stability after a power system fault according to claim 7, characterized in that, The energy calculation module is used to determine several PEBS exit points and critical energy function values ​​based on initial power system operating state parameters; including: Based on the initial power system operating state parameters, the generator motion equations are established; The transient energy function is obtained based on the generator's equation of motion. Based on the transient energy function and the preset fault trajectory, several PEBS exit points and critical energy function values ​​are determined.

10. The system for determining transient stability after a power system fault according to claim 7, characterized in that, The energy determination module is used to determine the dominant unstable equilibrium point and calculate the transient critical energy threshold based on several PEBS exit points; including: Integrate for any PEBS exit point, obtain the corresponding integral value, and determine the PEBS exit point corresponding to the integral value that meets the preset target conditions as the initial value; Based on the initial values, solve for the dominant unstable equilibrium point; The transient critical energy threshold is calculated based on the dominant unstable equilibrium point.

Images