A method and system for evaluating wind farm disconnection during power system restoration

Abstract

The application belongs to the technical field of power system recovery period evaluation, and particularly relates to a power system recovery period wind farm off-grid evaluation method and system, which comprises the following steps: obtaining a power system recovery scenario set; obtaining electromechanical transient response data based on the obtained recovery scenario data set and an electromechanical transient simulation analysis platform; constructing a wind farm off-grid evaluation model according to the obtained electromechanical transient response data; obtaining a wind farm grid connection point voltage in combination with the constructed wind farm off-grid evaluation model; and judging the off-grid state of the wind farm according to the obtained wind farm grid connection point voltage, thereby completing the off-grid evaluation of the wind farm during the power system recovery period.

Description

Technical Field

[0001] This invention belongs to the field of power system recovery period assessment technology, specifically relating to a method and system for assessing wind farm disconnection during power system recovery. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] With a high proportion of wind power being integrated into the power system, the risk of large-scale power outages caused by extreme weather is gradually increasing. Cold waves, strong winds, and snowstorms can easily cause physical damage to transmission lines, such as short circuits, line breaks, and tower collapses. Post-disaster damage to power grid components typically exhibits spatial correlation and randomness, and may be accompanied by the failure of monitoring, communication, and protection control systems. Limited on-site inspection and fault location capabilities result in the true fault state of transmission lines remaining unknown, posing a more severe challenge to the power system recovery process.

[0004] Power system restoration refers to the process of restoring the power system to normal operation after a major power outage through black start, grid reconfiguration, and gradual load restoration. Unlike traditional restoration processes primarily driven by synchronous turbines, the large-scale integration of wind power reduces system rotational inertia and short-circuit capacity, weakening voltage support. Wind turbines connected to the grid via converters have fault ride-through capabilities constrained by control strategies, making them more sensitive to voltage dips at the grid connection point. When the fault status is uncertain and a three-phase short-circuit fault actually exists on the transmission line, and trial commissioning fails, a large-scale voltage dip will occur before the fault is cleared. If the magnitude and duration of the voltage dip at the wind farm's grid connection point exceed the low voltage ride-through (LVRT) tolerance of the wind turbines, large-scale wind turbine disconnection will occur, causing a significant instantaneous loss of active power in the system, and restored loads will face the risk of secondary power outages. Therefore, before making any line commissioning decision, it is crucial to quickly assess whether the operation will trigger the potential risk of wind farm disconnection, thus providing a safety basis for dispatchers' restoration operation selection.

[0005] While widely used electromechanical transient simulation analysis platforms for power systems can accurately characterize transient responses, their large scale and complex component models in actual power systems result in lengthy simulation times for each transient operation, making it difficult to meet the real-time requirements of online search frameworks. Furthermore, analytical calculation methods based on short-circuit voltage formulas can only roughly estimate the voltage minimum point and cannot obtain a complete voltage drop time curve, making it difficult to support comprehensive grid disconnection judgments at multiple critical moments in wind farm low-voltage ride-through standards, such as the fault occurrence time, fault clearing time, and voltage recovery endpoint. Therefore, there is an urgent need for a rapid assessment method for wind farm grid disconnection risk during the power system recovery process. This method should be able to quickly predict the voltage response at the wind farm's grid connection point and determine its grid disconnection status under conditions of dynamic changes in the recovery topology and uncertain fault conditions, providing online transient safety support for dispatchers' recovery operation selection. Summary of the Invention

[0006] To address the aforementioned issues, this invention proposes a method and system for assessing wind farm disconnection during power system restoration. By combining a power system electromechanical transient simulation analysis platform and a rapid assessment model for wind farm disconnection, the voltage curve of the wind farm's grid connection point is obtained, thereby determining the wind farm's disconnection status. This enables online and rapid assessment of wind farm disconnection risks, significantly improving the efficiency of transient safety judgment during power system restoration operations.

[0007] According to some embodiments, the first aspect of the present invention provides a method for assessing wind farm disconnection during power system restoration, employing the following technical solution: A method for assessing wind farm disconnection during power system restoration includes: Acquire a set of power system recovery scenarios; Based on the acquired recovery scenario dataset and electromechanical transient simulation analysis platform, electromechanical transient response data is obtained; Based on the obtained electromechanical transient response data, a wind farm grid disconnection assessment model is constructed; Based on the constructed wind farm disconnection assessment model, the grid connection point voltage of the wind farm is obtained; Based on the obtained grid connection voltage of the wind farm, determine the grid disconnection status of the wind farm and complete the grid disconnection assessment of the wind farm during the power system restoration period.

[0008] As a further technical limitation, the constructed wind farm detachment assessment model is a wind farm detachment assessment model based on a gradient-enhanced physical information neural network; the gradient-enhanced physical information neural network adopts a fully connected deep neural network structure.

[0009] As a further technical limitation, the loss function of the wind farm disconnection assessment model includes data fitting loss, generator swing equation residual loss, network power balance residual loss, wind farm grid connection point short-circuit voltage equation residual term, and physical residual gradient enhancement loss.

[0010] As a further technical constraint, the obtained grid connection point voltage of the wind farm is compared with the fault low voltage ride-through curve of the wind turbine at several key moments. If the voltage at the time of fault occurrence, the time of lowest voltage, and the time of fault clearing are not lower than the first preset value, the voltage at 1 second after fault clearing is not lower than the second preset value, and the voltage at 3 and 4 seconds after fault clearing is not lower than the third preset value, then the wind farm is judged to have successfully crossed the grid; otherwise, the wind farm is judged to be disconnected from the grid.

[0011] As a further technical limitation, the recovery topologies in the acquired recovery scenario dataset are imported in batches into the electromechanical transient simulation analysis platform. Under each recovery topology, a three-phase short-circuit fault is applied to different fault locations on the line. The electromechanical transient response curves of the recovered generators, recovered nodes, and recovered wind farm grid connection points collected under the fault disturbance are used as electromechanical transient response data.

[0012] As a further technical limitation, in the process of obtaining the power system recovery scenario set, based on the topology information and parameters of the power system model, power system recovery scenarios are generated from single-machine island operation state to multi-machine interconnected operation state; among them, each recovery scenario corresponds to a different combination of generator, line and load commissioning, and each recovery scenario constitutes a scenario set for each time period of power system recovery, that is, the power system recovery scenario set is obtained.

[0013] According to some embodiments, a second aspect of the present invention provides a wind farm disconnection assessment system during power system recovery, employing the following technical solution: A system for assessing wind farm disconnection during power system restoration includes: The acquisition module is configured to acquire a set of power system recovery scenarios; based on the acquired recovery scenario dataset and the electromechanical transient simulation analysis platform, electromechanical transient response data is obtained. The module is configured to build a wind farm disconnection assessment model based on the obtained electromechanical transient response data; The calculation module is configured to combine the constructed wind farm disconnection assessment model to obtain the grid connection point voltage of the wind farm; The assessment module is configured to determine the off-grid status of the wind farm based on the obtained grid connection voltage of the wind farm, and to complete the off-grid assessment of the wind farm during the power system recovery period.

[0014] According to some embodiments, a third aspect of the present invention provides a computer-readable storage medium, employing the following technical solution: A computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in a method for assessing wind farm disconnection during power system recovery as described in the first aspect of the present invention.

[0015] According to some embodiments, the fourth aspect of the present invention provides an electronic device, which adopts the following technical solution: An electronic device includes a memory, a processor, and a program stored in the memory and running on the processor, wherein the processor executes the program to implement the steps in a method for assessing wind farm disconnection during power system recovery as described in the first aspect of the present invention.

[0016] According to some embodiments, the fifth aspect of the present invention provides a computer program product, which adopts the following technical solution: A computer program product includes software code, wherein the program in the software code performs the steps of a wind farm disconnection assessment method during power system recovery as described in the first aspect of the present invention.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention embeds the differential-algebraic equations describing electromechanical transient behavior and their gradient information into the neural network loss function, enabling the gradient-enhanced physical information neural network gPINN to still provide voltage prediction results that conform to the physical laws of electromechanical transients even when the topology of the early network structure is incomplete. This overcomes the problem of poor generalization ability of pure data-driven methods when the training samples are sparse in the early stage of recovery. This invention introduces a residual term of the short-circuit voltage equation at the wind farm grid connection point into the gPINN loss function. This residual term constrains the neural network output to satisfy the physical relationship that the wind farm grid connection point voltage is determined by the network impedance matrix during a fault, which significantly improves the accuracy of gPINN in predicting the voltage drop trajectory during a fault.

[0018] The gPINN trained by this invention uses neural network forward inference to complete the judgment of the wind farm disconnection status under the commissioning of a single line in one go. Compared with the time required for a single complete numerical solution of the power system electromechanical transient simulation analysis platform PSASP, it can significantly reduce the time required to meet the timeliness requirements of dispatchers to make decisions on a line-by-line basis, and provide an effective means for rapid safety assessment during the recovery of new power systems. Attached Figure Description

[0019] The accompanying drawings, which form part of this embodiment, are used to provide a further understanding of this embodiment. The illustrative embodiments and their descriptions are used to explain this embodiment and do not constitute an improper limitation of this embodiment.

[0020] Figure 1 This is a flowchart of a wind farm disconnection assessment method during power system restoration according to Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the gradient-enhanced physical information neural network in Embodiment 1 of the present invention; Figure 3 Add to the loss function in Embodiment 1 of the present invention Before and after the item, t A comparative diagram of the predicted scatter points at time 1=1.01s; Figure 4 Add to the loss function in Embodiment 1 of the present invention Before and after the item, t c =1.1s Prediction scatter plot comparison diagram; Figure 5 Add to the loss function in Embodiment 1 of the present invention Before and after the item t A schematic diagram comparing the predicted scatter points at time 2=2.0s; Figure 6 Add to the loss function in Embodiment 1 of the present invention Before and after the item, t A comparative diagram of the predicted scatter points at time 3=4.0s; Figure 7 Add to the loss function in Embodiment 1 of the present invention Before and after the item, t A comparative diagram of the predicted scatter points at time 4=5.0s; Figure 8 Add to the loss function in Embodiment 1 of the present invention A scatter plot comparing the predicted scatter points before and after the fault, showing the lowest voltage during the fault period; Figure 9 This is a structural block diagram of a wind farm disconnection assessment system during power system recovery, as shown in Embodiment 2 of the present invention. Detailed Implementation

[0021] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0022] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0023] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0024] In this invention, terms such as "upper," "lower," "left," "right," "front," "back," "vertical," "horizontal," "side," and "bottom" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used only to facilitate the description of the structural relationships of the various components or elements of this invention and do not specifically refer to any component or element in this invention. They should not be construed as limiting the invention.

[0025] In this invention, terms such as "fixed connection," "connected," and "linked" should be interpreted broadly, indicating a fixed connection, an integral connection, or a detachable connection; a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can determine the specific meaning of these terms in this invention based on the specific circumstances, and they should not be construed as limitations on the invention.

[0026] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0027] Example 1 Embodiment 1 of this invention introduces a method for assessing wind farm disconnection during power system restoration.

[0028] like Figure 1 The method shown is for assessing wind farm disconnection during power system restoration, including: Acquire a set of power system recovery scenarios; Based on the acquired recovery scenario dataset and electromechanical transient simulation analysis platform, electromechanical transient response data is obtained; Based on the obtained electromechanical transient response data, a wind farm grid disconnection assessment model is constructed; Based on the constructed wind farm disconnection assessment model, the grid connection point voltage of the wind farm is obtained; Based on the obtained grid connection voltage of the wind farm, determine the grid disconnection status of the wind farm and complete the grid disconnection assessment of the wind farm during the power system restoration period.

[0029] The following is a detailed description of the evaluation method in this embodiment: This embodiment generates a large number of power system recovery scenarios, ranging from single-machine islanded operation to multi-machine interconnected operation, by inputting the topology information and parameters of the power system model under study. Each generated scenario corresponds to a different combination of generators, lines and loads, forming a set of scenarios for each period of system recovery, which serves as the network topology input for subsequent simulation data acquisition.

[0030] The topology information and parameters of the power system model input in this embodiment include: line connection relationships and electrical parameters, generator connection locations and unit parameters, load node locations and load levels, etc. Based on the above input, and under the condition of satisfying the system recovery feasibility constraints, a recovery scenario is generated that gradually transitions from the initial islanded operation state to a multi-machine interconnected operation state.

[0031] Each recovery scenario is jointly determined by the set of currently operational lines, the set of grid-connected generators, and the set of restored loads, and is used to characterize the network topology and operating status at a certain moment during the recovery process. 0-1 variables are used to encode the recovery scenario, and the topology coding vector is defined as follows: ;in, This is the line commissioning state vector. This is the state vector of a conventional generator in operation. This represents the recovered state vector of the wind farm. This is the load recovery state vector; , , and These represent the number of lines, conventional generators, wind farms, and load nodes in the system under study, respectively. A value of 1 for a vector element indicates that the corresponding component is in operation, connected to the grid, or has been restored, while a value of 0 indicates that the corresponding component is not yet in operation, connected to the grid, or has not been restored.

[0032] Under the premise of satisfying network connectivity, operational security, and recovery feasibility, different combinations of line commissioning, generator grid connection, and load restoration are traversed or extracted to form a large number of recovery scenarios covering different recovery stages. The resulting scenario set can cover various network structures that may occur during initial islanded operation, island expansion, multi-regional interconnection, and multi-machine interconnection operation, serving as the network topology input for subsequent PSASP electromechanical transient simulation data acquisition. For components that have not yet been commissioned or restored, their corresponding positions in the topology encoding vector are set to 0, and the relevant physical parameters are set to 0 in subsequent gPINN inputs.

[0033] This embodiment imports the generated recovery topology into the PSASP electromechanical transient simulation platform. Under each recovery topology, a three-phase short-circuit fault is applied at different fault locations along the line. Dynamic response curves of the recovered generator, including its power angle, speed deviation, transient potential amplitude, and mechanical power, as well as the voltage amplitude and voltage phase angle time-series curves of the recovered nodes, are collected under the fault scenario. These are used as label samples for subsequent gPINN training. All samples are then divided into training, validation, and test sets at 70%, 15%, and 15% respectively. Specifically: The generated large amount of restored topology information is imported in batches into the PSASP electromechanical transient simulation platform, and three-phase short-circuit fault disturbances are applied to different fault locations on the traversed lines for each type of restored topology. Fault feature vector φ Used to describe the location and time of the fault, including the line number where the fault occurred and the relative location of the fault. α ∈ (0,1), fault occurrence time and fault clearing time.

[0034] For each simulation sample, the total time-domain simulation duration is set to 5 seconds, and the fault occurrence time is set to 1 second. The following data are collected during the simulation: (1) The generator has been restored The dynamic response curve, including the power angle Speed ​​deviation transient potential amplitude Mechanical power and electromagnetic power It is used to characterize the electromechanical transient response of a generator under fault disturbance.

[0035] (2) Recovered Nodes The timing response curves, including node voltage amplitudes. Voltage phase angle Node active power injection and node reactive power injection It is used to characterize the voltage and power changes of system nodes under fault disturbances.

[0036] (3) Wind farms that have been restored Voltage timing curve of the grid connection point bus The labeled data used for training gPINN is used to characterize the dynamic voltage response of the wind farm grid connection point during fault occurrence, fault duration, fault clearing, and voltage recovery.

[0037] (4) Time sampling point information, including labeled data sampling points, physical residual sampling points, and gradient constraint sampling points. Among them, labeled data sampling points are used to calculate the data fitting loss, and physical residual sampling points and gradient constraint sampling points are used for residual evaluation in the subsequent gPINN physical constraint training process.

[0038] All simulation samples were randomly divided into training, validation, and test sets at a ratio of 70%, 15%, and 15%, respectively. The training set was used for gPINN parameter learning, the validation set was used for loss monitoring and hyperparameter selection during training, and the test set was used to finally evaluate the generalization ability of gPINN in unrecovered topologies and unrecovered fault locations.

[0039] As one or more implementation methods, this embodiment uses a weighted combination of the data fitting loss, generator swing equation residual loss, network power balance residual loss, wind farm grid connection point short-circuit voltage residual loss, and the gradient enhancement loss of the above physical residuals as the total loss function of gPINN. This ensures that the neural network output strictly satisfies the differential-algebraic physical laws of the electromechanical transients of the power system while fitting the simulation data. The gPINN network structure takes time variables, topology, system parameters, and fault characteristics as inputs and the wind farm grid connection point voltage time series curve as output. The loss function of gPINN consists of the data fitting loss, generator swing equation residual loss, network power balance residual loss, wind farm grid connection point short-circuit voltage equation residual term, and physical residual gradient enhancement loss.

[0040] The gradient-enhanced physical information neural network structure in this embodiment is as follows: Figure 2 As shown, a fully connected deep neural network structure is adopted. The input dimension is composed of time variables, topology encoding vectors, system parameter vectors and fault feature vectors. The output is the voltage prediction curve of the wind farm grid connection point.

[0041] In the process of designing the gPINN model structure, the time variable t Including the time of failure t 1. Fault clearance time t c Here t ∈[0, T ], T The simulation duration is 5 seconds in this example.

[0042] In this embodiment, the topology vector s uses 0-1 variables to describe the network recovery state, that is... ;in This is the line commissioning state vector. This is the generator commissioning state vector. This is the load recovery state vector.

[0043] The system parameter vector in this embodiment Includes key physical parameters and operating point information of operational components, such as the inertial constant of operational generators. Damping coefficient Transient reactance Resistance of the lines already in operation With reactance For components that have not been restored, their corresponding parameters are uniformly set to 0.

[0044] The fault feature vector in this embodiment Used to describe the location of a short circuit, including the faulty line number and the relative location of the fault. In this experiment, the simulation dataset is set to include a three-phase ground fault.

[0045] The model input belongs to a high-dimensional parameter space, which can be represented as: The output of gPINN is the voltage prediction curve of the bus at the grid connection point of the wind farm.

[0046] The loss function of gPINN will describe the differential-algebraic equations that characterize the electromechanical transient behavior of the power system and their gradients embedded in the neural network training; specifically: (1) Data fitting loss By fitting the time-domain simulation results onto the labeled data points, we obtain... ; (2) Residual loss of generator dynamic equation For generators already in operation, the residual of the swing equation should be zero, i.e. ; ; ; (3) Network power balance residual loss The power balance residuals at each node should be zero, i.e. ; ; ; (4) Residual loss of short-circuit voltage equation That is, the short-circuit voltage residual at the grid connection point of the wind farm should be zero.

[0047] Each synchronous generator is represented using a classical second-order model, which is equivalent to transient reactance. The subsequent constant transient potential Therefore, its equivalent injection current is obtained as ;in, For generator The equivalent injection current, The transient electromotive force amplitude of the generator. The generator power angle, For generator Axis transient reactance.

[0048] Let the node admittance matrix corresponding to the current restored topology be... The node impedance matrix is Based on the superposition of the equivalent injected currents of the grid-connected generators, the wind farm before the fault can be obtained. The bus voltage at the grid connection point is When the system is at the node A three-phase short-circuit fault occurred at [location], and the fault impedance was [value]. When the fault point boundary conditions are met, they can be expressed as follows: Its contribution to the grid connection voltage of the wind farm is Based on the superposition relationship between the pre-fault components and the fault components, the wind farm during the fault period can be obtained. The bus voltage at the grid connection point is ; The residual loss of the short-circuit voltage equation is defined as follows: ; (5) Gradient Boosting Loss: Based on the physical residual, its gradient with respect to the time variable t is further constrained, so that the network output not only satisfies the physical equation at the sampling point, but also maintains physical consistency in nearby time steps, i.e. ;in, N d This represents the number of labeled data points. N r This represents the number of residual sampling points within the domain. N s The number of sampling points constrained by the gradient; t d For the first d Each data sampling time; t r For the first r Each residual sampling time; t s For the first s Each gradient-constrained sampling time; For wind farm w exist t d Predicted voltage at grid connection point at sampling time. For wind farm w exist t d The actual voltage value at the grid connection point of the wind farm at the sampling time; n g This represents the number of generators that have been restored. n i This represents the number of nodes that have been recovered. n w This refers to the number of wind turbine units that have been restored. δ g For generator g The angle of attack; ω g For generator g The rotor angular velocity; M g For generator g The inertial time constant; P mg For generator g Mechanical power; P eg For generator g Electromagnetic power; D g For generator gDamping coefficient; P i For nodes i Active power injection; V i , θ i For nodes i Voltage amplitude and phase angle; Z wj For wind farm nodes w With nodes j Transfer impedance between; For the first j generator q Amplitude of shaft transient potential; For the first j generator d The parameters, including shaft transient reactance, together constitute the loss function of gPINN. By minimizing these loss terms, the neural network predictions can satisfy the physical laws of electromechanical transients in the power system.

[0049] Combining the above factors, the total loss function of gPINN is: ;in, , , , and These are the weight coefficients for each loss term, and their initial values ​​and dynamic adjustment strategies during training are obtained from the preceding and subsequent steps.

[0050] This embodiment employs a two-stage training strategy using the Adam optimizer and the L-BFGS optimizer, combined with an adaptive weight dynamic adjustment mechanism for each loss term to train the network parameters. The first stage uses the Adam optimizer to quickly reduce the total loss function, allowing the network to converge to a better initial solution. The second stage switches to the L-BFGS optimizer for fine-tuning the parameters. Specifically: gPINN is trained using a two-stage training strategy that combines the Adam optimizer and the L-BFGS optimizer. In the first phase, Adam was used. Adaptive Moment Estimation The optimizer trains the network parameters. Adam combines adaptive estimation of first-order and second-order momentum, which can quickly reduce the total loss function value in a high-dimensional parameter space with multiple loss terms coupled, enabling the network to converge to a better initial solution. In this embodiment, the initial learning rate of Adam is 0.001, which is adjusted according to the exponential decay law; the batch size is 64; training continues until the total loss decreases gradually, then proceeds to the second stage.

[0051] In this embodiment, the second stage switch is L-BFGS. Limited-memory Broyden-Fletcher-Goldfarb-Shanno algorithm The quasi-Newton optimizer fine-tunes the network parameters. L-BFGS utilizes recent gradient information to construct a low-rank approximation of the Hessian matrix, exhibiting strong local fine-grained convergence capability in the physical residual loss term, further reducing the values ​​of each physical residual loss term, and improving the degree to which the gPINN output satisfies the electromechanical transient physical laws.

[0052] To address the issue of significant differences in the order of magnitude between data fitting loss and various physical residual loss terms, this embodiment employs an adaptive weight dynamic adjustment mechanism to balance the order of magnitude of each loss term: , , Furthermore, during training, the weights are dynamically updated based on the relative magnitude of each loss term, so that each loss term remains at a similar order of magnitude, thereby improving training stability and convergence efficiency.

[0053] This embodiment employs a multi-criteria termination strategy during model training. Network training is stopped when any of the following conditions are met: 1) The total loss function value is lower than the preset threshold. This embodiment takes ; 2) Validation set loss in continuous No significant decrease was observed within one epoch; 3) The training rounds reach the maximum number of iterations. After training, the model parameters that minimize the loss on the validation set are selected as the final model parameters.

[0054] Loss function added For example, scatter plots of LVRT key moment predictions before and after the item. Figure 3 , Figure 4 , Figure 5 , Figure 6 , Figure 7 and Figure 8 As shown, at the time of the fault occurrence t 1 = 1.01s, fault clearing time t c and the lowest voltage point t min At this point, introduce L faultThe terms reduced the RMSE from 0.0132, 0.0378, and 0.0123 to 0.0110, 0.0282, and 0.0108, respectively, representing decreases of approximately 16.7%, 25.4%, and 12.2%. The scatter points became more densely distributed on both sides of the ideal y=x line; while at the crossing judgment point... t 2 = 2s (relative to 1s fault), recovery endpoint t 3 = 4s (relative fault 3s) t 4 = 5s, the RMSE of both schemes are in the range of 0.003~0.005, and the difference is negligible. The results show that... L fault This study specifically improves the prediction accuracy of the fault depth drop zone without compromising steady-state performance, providing a more reliable prediction basis for subsequent LVRT crossing / disconnection determination.

[0055] This embodiment calls the wind farm disconnection rapid assessment model when a fault occurs during the recovery period, thereby outputting the wind farm grid connection point voltage time series curve; according to the wind farm low voltage ride-through capability requirements (GB / T 19963-2011), the predicted key moment voltage value is selected to determine the wind farm disconnection status; specifically: The trained gPINN is used as a fast evaluator for the off-grid status of wind farms during the restoration of a new power system. Before restoring any line with an unknown fault state, the current restored topology vector, system parameter vector, fault feature vector, and the fault occurrence and disconnection times are taken as inputs. gPINN is then used to quickly predict the voltage drop curves at each wind farm's grid connection point. Based on the low-voltage ride-through capability requirements for wind farms (GB / T 19963-2011), the following key voltage values ​​predicted by gPINN are selected: fault occurrence time... t 1 = 1.01s, requiring ≥0.2pu; lowest voltage point t min The lowest voltage drop during a fault must be ≥0.2 pu; the fault clearing time... t c The requirement is ≥0.2 pu. Crossing judgment point. t 2 = 2s (relative fault 1s), requiring ≥0.55 pu; recovery endpoint t 3 = 4s (relative fault 3s) t 4 = 5s, requiring recovery to ≥0.9pu; if the above conditions are met simultaneously, the wind farm w During the period The wind farm is not disconnected from the grid, otherwise the wind farm w During the period t The network is disconnected.

[0056] This embodiment embeds the differential-algebraic equations describing electromechanical transient behavior and their gradient information into the neural network loss function. This enables the gradient-enhanced physical information neural network gPINN to provide voltage prediction results that conform to the physical laws of electromechanical transients, even under incomplete network topology in the early stages of recovery. This overcomes the problem of poor generalization ability of pure data-driven methods under sparse training sample topology in the early stages of recovery. The residual term of the short-circuit voltage equation at the wind farm grid connection point is introduced into the gPINN loss function. This residual term constrains the neural network output to satisfy the physical relationship that the voltage at the wind farm grid connection point is determined by the network impedance matrix during the fault period, which significantly improves the accuracy of gPINN in predicting the voltage drop trajectory during the fault period. The trained gPINN completes the judgment of the wind farm disconnection status under the single line commissioning in one neural network forward inference. Compared with the time required for a single complete numerical solution of the power system electromechanical transient simulation analysis platform PSASP, it greatly reduces the time required to meet the second-level response time requirements of dispatchers when making decisions line by line, and provides an effective means for rapid safety assessment during the recovery of new power systems.

[0057] Example 2 Embodiment 2 of the present invention introduces a wind farm disconnection assessment system during power system restoration.

[0058] like Figure 9 The system shown is a wind farm disconnection assessment system during power system restoration, comprising: The acquisition module is configured to acquire a set of power system recovery scenarios; based on the acquired recovery scenario dataset and the electromechanical transient simulation analysis platform, electromechanical transient response data is obtained. The module is configured to build a wind farm disconnection assessment model based on the obtained electromechanical transient response data; The calculation module is configured to combine the constructed wind farm disconnection assessment model to obtain the grid connection point voltage of the wind farm; The assessment module is configured to determine the off-grid status of the wind farm based on the obtained grid connection voltage of the wind farm, and to complete the off-grid assessment of the wind farm during the power system recovery period.

[0059] The detailed steps are the same as those provided in Example 1 for assessing wind farm disconnection during power system restoration, and will not be repeated here.

[0060] Example 3 Embodiment 3 of the present invention provides a computer-readable storage medium.

[0061] A computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in a method for assessing wind farm disconnection during power system recovery as described in Embodiment 1 of the present invention.

[0062] The detailed steps are the same as those provided in Example 1 for assessing wind farm disconnection during power system restoration, and will not be repeated here.

[0063] Example 4 Embodiment 4 of the present invention provides an electronic device.

[0064] An electronic device includes a memory, a processor, and a program stored in the memory and running on the processor. When the processor executes the program, it implements the steps in a method for assessing wind farm disconnection during power system recovery as described in Embodiment 1 of the present invention.

[0065] The detailed steps are the same as those provided in Example 1 for assessing wind farm disconnection during power system restoration, and will not be repeated here.

[0066] Example 5 Embodiment 5 of the present invention provides a computer program product.

[0067] A computer program product includes software code, wherein the program in the software code performs the steps of a wind farm disconnection assessment method during power system recovery as described in Embodiment 1 of the present invention.

[0068] The detailed steps are the same as those provided in Example 1 for assessing wind farm disconnection during power system restoration, and will not be repeated here.

[0069] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0070] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0071] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0072] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0073] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0074] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

[0075] The above description is merely a preferred embodiment of this practice and is not intended to limit the scope of this practice. Various modifications and variations can be made to this practice by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this practice should be included within the protection scope of this practice.

Claims

1. A method for assessing wind farm disconnection during power system restoration, characterized in that, include: Acquire a set of power system recovery scenarios; Based on the acquired recovery scenario dataset and electromechanical transient simulation analysis platform, electromechanical transient response data is obtained; Based on the obtained electromechanical transient response data, a wind farm grid disconnection assessment model is constructed; Based on the constructed wind farm disconnection assessment model, the grid connection point voltage of the wind farm is obtained; Based on the obtained grid connection voltage of the wind farm, determine the grid disconnection status of the wind farm and complete the grid disconnection assessment of the wind farm during the power system restoration period.

2. The method for assessing wind farm disconnection during power system restoration as described in claim 1, characterized in that, The constructed wind farm degrid assessment model is a wind farm degrid assessment model based on gradient-enhanced physical information neural network; the gradient-enhanced physical information neural network adopts a fully connected deep neural network structure.

3. The method for assessing wind farm disconnection during power system restoration as described in claim 1, characterized in that, The loss function of the wind farm disconnection assessment model includes data fitting loss, generator swing equation residual loss, network power balance residual loss, wind farm grid connection point short-circuit voltage equation residual term, and physical residual gradient enhancement loss.

4. The method for assessing wind farm disconnection during power system restoration as described in claim 1, characterized in that, Several key moments of the obtained wind farm grid connection point voltage are compared with the wind turbine fault low voltage ride-through curve. If the voltage at the fault occurrence time, the time of the lowest voltage point, and the time of fault clearing are not lower than the first preset value, the voltage at 1 second after fault clearing is not lower than the second preset value, and the voltage at 3 and 4 seconds after fault clearing is not lower than the third preset value, then the wind farm is judged to have successfully crossed the grid; otherwise, the wind farm is judged to be disconnected from the grid.

5. The method for assessing wind farm disconnection during power system restoration as described in claim 1, characterized in that, The recovery topologies in the acquired recovery scenario dataset are imported in batches into the electromechanical transient simulation analysis platform. Under each recovery topology, a three-phase short-circuit fault is applied to different fault locations on the line. The electromechanical transient response curves of the recovered generators, recovered nodes, and recovered wind farm grid connection points under the collected fault disturbances are used as electromechanical transient response data.

6. The method for assessing wind farm disconnection during power system restoration as described in claim 1, characterized in that, In the process of acquiring the power system recovery scenario set, based on the topology information and parameters of the power system model, power system recovery scenarios are generated from single-machine island operation to multi-machine interconnected operation. Each recovery scenario corresponds to a different combination of generators, lines and loads. Each recovery scenario constitutes a scenario set for each time period of power system recovery, thus obtaining the power system recovery scenario set.

7. A wind farm disconnection assessment system during power system restoration, characterized in that, include: The acquisition module is configured to acquire a set of power system recovery scenarios; Based on the acquired recovery scenario dataset and electromechanical transient simulation analysis platform, electromechanical transient response data is obtained; The module is configured to build a wind farm disconnection assessment model based on the obtained electromechanical transient response data; The calculation module is configured to combine the constructed wind farm disconnection assessment model to obtain the grid connection point voltage of the wind farm; The assessment module is configured to determine the off-grid status of the wind farm based on the obtained grid connection voltage of the wind farm, and to complete the off-grid assessment of the wind farm during the power system recovery period.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the steps of a method for assessing wind farm disconnection during power system recovery as described in any one of claims 1-6.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the program, it implements the steps of a wind farm disconnection assessment method during power system recovery as described in any one of claims 1-6.

10. A computer program product, comprising software code, characterized in that, The program in the software code performs the steps of a wind farm disconnection assessment method during power system recovery as described in any one of claims 1-6.

Images