A frequency stability quantification discrimination method and system adapted to different scenarios

Abstract

The application discloses a frequency stability quantitative discrimination method and system suitable for different scenes and belongs to the technical field of power systems. The method comprises the following steps: obtaining model parameters of a target frequency response physical model of a target power system; identifying the model parameters based on a discrimination model to obtain a frequency response curve of the target frequency response physical model for the target power system; and quantifying the frequency stability of the target power system by discriminating the frequency response curve. The application can identify the model parameters of the frequency response physical model to quickly discriminate the discrimination frequency response curve of the power system, thereby quantifying the frequency stability of the power system.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and more specifically, to a frequency stability quantification and discrimination method and system adapted to different scenarios. Background Technology

[0002] In recent years, new energy sources such as wind power and photovoltaic power have been integrated into the grid on a large scale through power electronic converters, gradually reducing the installed capacity of traditional synchronous power sources. Compared with traditional synchronous systems, the large-scale integration of new energy sources in high-proportion power electronic power systems has replaced some synchronous units. The decoupling characteristics of new energy power electronics and their maximum power point tracking mode have gradually reduced the system inertia level and weakened the frequency regulation capability. In addition, the commissioning of ultra-high-voltage large-capacity inter-regional DC transmission has disrupted the inter-regional inertia support and power response under disturbances, severely deteriorating the system frequency stability under large disturbances.

[0003] Traditional methods for analyzing the frequency-dependent electromechanical transient characteristics of power systems after disturbances are based on physical models. These primarily include full-time-domain simulation methods that solve high-order nonlinear differential-algebraic equations encompassing the entire network model, and equivalent methods using single-machine load models, represented by average system frequency models and system frequency response models. Machine learning methods, as a representative approach, can operate entirely independently of physical models, utilizing historical data to mine the correlations between system inputs and outputs. These methods offer significant advantages in data analysis and processing speed. Theoretically, with sufficient and accurate samples, machine learning methods can accurately fit the response characteristics of various nonlinear components. However, current technology has limited accuracy in identifying the model parameters of the frequency response physical model. Therefore, the identified model parameters cannot be used to determine the discriminant frequency response curve of the power system, and consequently, the frequency stability of the power system cannot be quantified using model parameters. Summary of the Invention

[0004] To address the above problems, this invention proposes a frequency stability quantization discrimination method adapted to different scenarios, comprising:

[0005] Obtain the model parameters of the physical model of the target frequency response of the target power system;

[0006] The model parameters are identified based on the discriminant model to obtain the frequency response curve of the target frequency response physical model for the target power system;

[0007] The frequency stability of the target power system is quantified by analyzing the frequency response curve.

[0008] Optionally, the method also includes: building a discriminative model, including:

[0009] Based on the frequency response physical model, the power system under different scenarios is simulated and run to obtain simulation data, and the input and output feature quantities of the frequency response physical model are extracted from the simulation data.

[0010] The input and output features are normalized to obtain sample data, which is then divided into training and testing sample data according to a preset ratio. The preset model is trained based on the training and testing sample data to obtain a discriminant model.

[0011] Optionally, different scenarios include at least one of the following: a three-phase short-circuit fault in the power system and a power step disturbance fault in the power system.

[0012] Optionally, the power system load power fluctuates randomly under different scenarios.

[0013] Optionally, the input feature quantities for the frequency response physical model are extracted from the simulation data, including: extracting the frequency information of the power system at the moment of disturbance and within a preset time period after the moment of disturbance from the simulation data as the input feature quantities;

[0014] The frequency information includes: natural oscillation angular frequency and damping ratio.

[0015] Optionally, the output features of the frequency response physical model can be extracted from the simulation data, including:

[0016] The mechanism parameters are determined based on the response data in the simulation data;

[0017] The mechanistic parameters of the physical model of the frequency response to be identified in the simulation data are extracted as the output feature quantity.

[0018] Optional, the default ratio is 70%:30%.

[0019] Optionally, the preset model is trained based on training sample data and test sample data, including:

[0020] Determine the number of neural network layers and the number of neurons in each layer of the preset model. Based on the number of neural network layers and the number of neurons in each layer, input training sample data as needed for training until the training times or error indicators meet the requirements, and obtain the initial discrimination model.

[0021] The test sample data is input into the initial discrimination model to verify the accuracy of the initial discrimination model. The initial discrimination model that passes the verification is the discrimination model.

[0022] The error index includes at least one of the following: absolute error, average relative error, average absolute error, and root mean square error.

[0023] Optionally, the frequency stability of the target power system can be quantified by judging the frequency response curve, including:

[0024] Determine the maximum frequency deviation value of the frequency response curve, and compare the maximum frequency deviation value with the set frequency deviation constraint value;

[0025] If the frequency deviation constraint value is greater than the maximum frequency deviation value, the target power system is in a stable state; otherwise, the target power system is in an unstable state.

[0026] Furthermore, this invention proposes a frequency stability quantization discrimination system adapted to different scenarios, comprising:

[0027] The data acquisition unit is used to acquire the model parameters of the physical model of the target frequency response of the target power system;

[0028] The identification unit identifies the model parameters based on the discrimination model to obtain the frequency response curve of the target frequency response physical model for the target power system;

[0029] A quantization unit is used to quantify the frequency stability of the target power system by judging the frequency response curve.

[0030] Optionally, the data acquisition unit is further configured to build a discrimination model, wherein building the discrimination model includes:

[0031] Based on the frequency response physical model, the power system under different scenarios is simulated and run to obtain simulation data, and the input and output feature quantities of the frequency response physical model are extracted from the simulation data.

[0032] The input and output features are normalized to obtain sample data, which is then divided into training and testing sample data according to a preset ratio. The preset model is trained based on the training and testing sample data to obtain a discriminant model.

[0033] Optionally, different scenarios include at least one of the following: a three-phase short-circuit fault in the power system and a power step disturbance fault in the power system.

[0034] Optionally, the power system load power fluctuates randomly under different scenarios.

[0035] Optionally, the input feature quantities for the frequency response physical model are extracted from the simulation data, including: extracting the frequency information of the power system at the moment of disturbance and within a preset time period after the moment of disturbance from the simulation data as the input feature quantities;

[0036] The frequency information includes: natural oscillation angular frequency and damping ratio.

[0037] Optionally, the output features of the frequency response physical model can be extracted from the simulation data, including:

[0038] The mechanism parameters are determined based on the response data in the simulation data;

[0039] The mechanistic parameters of the physical model of the frequency response to be identified in the simulation data are extracted as the output feature quantity.

[0040] Optional, the default ratio is 70%:30%.

[0041] Optionally, the preset model is trained based on training sample data and test sample data, including:

[0042] Determine the number of neural network layers and the number of neurons in each layer of the preset model. Based on the number of neural network layers and the number of neurons in each layer, input training sample data as needed for training until the training times or error indicators meet the requirements, and obtain the initial discrimination model.

[0043] The test sample data is input into the initial discrimination model to verify the accuracy of the initial discrimination model. The initial discrimination model that passes the verification is the discrimination model.

[0044] The error index includes at least one of the following: absolute error, average relative error, average absolute error, and root mean square error.

[0045] Optionally, the quantization unit quantifies the frequency stability of the target power system by judging the frequency response curve, including:

[0046] Determine the maximum frequency deviation value of the frequency response curve, and compare the maximum frequency deviation value with the set frequency deviation constraint value;

[0047] If the frequency deviation constraint value is greater than the maximum frequency deviation value, the target power system is in a stable state; otherwise, the target power system is in an unstable state.

[0048] Furthermore, this invention proposes a method for building a discriminative model, comprising:

[0049] Based on the frequency response physical model, the power system under different scenarios is simulated and run to obtain simulation data, and the input and output feature quantities of the frequency response physical model are extracted from the simulation data.

[0050] The input and output features are normalized to obtain sample data, which is then divided into training and testing sample data according to a preset ratio. The preset model is trained based on the training and testing sample data to obtain a discriminant model.

[0051] In another aspect, the present invention also provides a computing device, comprising: one or more processors;

[0052] A processor is used to execute one or more programs;

[0053] When the one or more programs are executed by the one or more processors, the method described above is implemented.

[0054] In another aspect, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed, implements the method described above.

[0055] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0056] This invention provides a frequency stability quantification and discrimination method adaptable to different scenarios, comprising: acquiring model parameters of a target frequency response physical model of a target power system; identifying the model parameters based on a discrimination model to obtain the frequency response curve of the target frequency response physical model for the target power system; and quantifying the frequency stability of the target power system by discriminating the frequency response curve. This invention can identify the model parameters of the frequency response physical model to quickly determine the discrimination frequency response curve of the power system, thereby enabling the quantification of the frequency stability of the power system. Attached Figure Description

[0057] Figure 1 This is a flowchart of an embodiment of the method of the present invention;

[0058] Figure 2 This is a schematic diagram of an embodiment of the method of the present invention;

[0059] Figure 3 This is a diagram showing the model parameter identification results of an embodiment of the method of the present invention;

[0060] Figure 4 This is a structural diagram of an embodiment of the system of the present invention. Detailed Implementation

[0061] Exemplary embodiments of the invention will now be described with reference to the accompanying drawings. However, the invention may be embodied in many different forms and is not limited to the embodiments described herein. These embodiments are provided to fully and completely disclose the invention and to fully convey its scope to those skilled in the art. The terminology used in the exemplary embodiments illustrated in the drawings is not intended to limit the invention. In the drawings, the same units / elements are referred to by the same reference numerals.

[0062] Unless otherwise stated, the terms used herein (including technical terms) have their common meaning as understood by one of ordinary skill in the art. Furthermore, it is understood that terms defined in commonly used dictionaries should be understood to have a meaning consistent with the context of their relevant field, and not to be interpreted as having an idealized or overly formal meaning.

[0063] Example 1:

[0064] This invention proposes a frequency stability quantization discrimination method adapted to different scenarios, such as... Figure 1 As shown, it includes:

[0065] Step 1: Obtain the model parameters of the physical model of the target frequency response of the target power system;

[0066] Step 2: Identify the model parameters based on the discriminant model to obtain the frequency response curve of the target frequency response physical model for the target power system;

[0067] Step 3: Quantify the frequency stability of the target power system by judging the frequency response curve.

[0068] The method further includes: building a discriminant model, including:

[0069] Based on the frequency response physical model, the power system under different scenarios is simulated and run to obtain simulation data, and the input and output feature quantities of the frequency response physical model are extracted from the simulation data.

[0070] The input and output features are normalized to obtain sample data, which is then divided into training and testing sample data according to a preset ratio. The preset model is trained based on the training and testing sample data to obtain a discriminant model.

[0071] Among them, different scenarios include at least one of the following: a three-phase short-circuit fault in the power system and a power step disturbance fault in the power system.

[0072] Among them, the power load of the power system fluctuates randomly under different scenarios.

[0073] The process of extracting input features from the simulation data for the frequency response physical model includes: extracting the frequency information of the power system at the moment of disturbance and within a preset time period after the moment of disturbance from the simulation data as the input features.

[0074] The frequency information includes: natural oscillation angular frequency and damping ratio.

[0075] Specifically, the output features of the frequency response physical model extracted from the simulation data include:

[0076] The mechanism parameters are determined based on the response data in the simulation data;

[0077] The mechanistic parameters of the physical model of the frequency response to be identified in the simulation data are extracted as the output feature quantity.

[0078] The preset ratio is 70%:30%.

[0079] The process of training a pre-defined model based on training sample data and test sample data includes:

[0080] Determine the number of neural network layers and the number of neurons in each layer of the preset model. Based on the number of neural network layers and the number of neurons in each layer, input training sample data as needed for training until the training times or error indicators meet the requirements, and obtain the initial discrimination model.

[0081] The test sample data is input into the initial discrimination model to verify the accuracy of the initial discrimination model. The initial discrimination model that passes the verification is the discrimination model.

[0082] The error index includes at least one of the following: absolute error, average relative error, average absolute error, and root mean square error.

[0083] The process of quantifying the frequency stability of the target power system by analyzing the frequency response curve includes:

[0084] Determine the maximum frequency deviation value of the frequency response curve, and compare the maximum frequency deviation value with the set frequency deviation constraint value;

[0085] If the frequency deviation constraint value is greater than the maximum frequency deviation value, the target power system is in a stable state; otherwise, the target power system is in an unstable state.

[0086] The invention will be further described below with reference to specific implementation processes:

[0087] Implementation process, such as Figure 2 As shown, it includes:

[0088] Step 1: Acquire simulation data, considering different types of disturbances in the power system and setting random fluctuations in system load levels to obtain system frequency response data. Extract the required input and output features from the acquired simulation data. The input features are the frequency information at the instant of the disturbance and for a subsequent period, and the outputs are the model parameters to be identified.

[0089] Step 2, Model Training: First, normalize the input and output quantities. Then, divide the normalized sample data into training samples and test samples according to a certain ratio. Next, determine the parameters of the intelligent model, such as the number of neural network layers and the number of neurons in each layer. Finally, input the training samples into the configured model for training until the required number of training iterations or error indicators are met.

[0090] Step 3: Perform frequency stability determination. Input the parameters into the trained model to quickly calculate the frequency response curve. Compare the maximum frequency deviation with the stability threshold to determine the system's frequency stability.

[0091] In step 1, simulation data is acquired, considering different types of disturbances to the power system, and setting random fluctuations in the system load level to obtain system frequency response data. The required input and output features are extracted from the acquired simulation data. The input features are the frequency information at the instant of the disturbance and for a subsequent period, and the outputs are the model parameters to be identified, mainly including:

[0092] (1) Obtain sample data, set the system to experience fault modes such as three-phase short circuit and power step disturbance, and simultaneously allow the load power to fluctuate randomly, as shown in the following formula:

[0093]

[0094] In the formula, r is the load level coefficient, and P i Q represents the active power of the node. i This represents the reactive power of the node.

[0095] (2) Extract the required input features and outputs from the acquired simulation data. Select the frequency information at the moment of the disturbance and for a period of time thereafter as the model input, and select the mechanism parameters to be identified as the model output.

[0096] The formulas for calculating the natural oscillation angular frequency and damping ratio of the second-order model of the system frequency response are as follows:

[0097]

[0098]

[0099] In the formula, R, D, H, T R F HThese are the unit governor speed control gain, load frequency regulation coefficient, inertia time constant, reheater time constant, and high-pressure cylinder power ratio.

[0100] The system frequency time-domain expression is:

[0101]

[0102] In the formula:

[0103] Based on the response data, the accurate values ​​of the parameters to be enhanced are obtained through numerical methods and used as the model output.

[0104] In step 2, model training begins by normalizing the input and output data. Then, the normalized sample data is divided into training and testing samples according to a certain ratio. Next, parameters such as the number of neural network layers and the number of neurons in each layer are determined. Finally, the training samples are input into the configured model for training until the required number of training iterations or error metrics are met. This mainly includes:

[0105] (1) First, normalize the input and output values ​​to the range [0,1]:

[0106]

[0107] In the formula, x i For input or output samples, x imin x is the minimum value of the sample. imax This represents the maximum value of the sample.

[0108] (2) Divide the normalized sample data into training samples and test samples according to a certain ratio, such as dividing the training samples and test samples in the network according to a ratio of 70% and 30%.

[0109] (3) Determine the parameters of the intelligent model, such as the number of neural network layers and the number of neurons in each layer. For example, if an LSTM network is selected, its parameters are an LSTM network with 4 hidden layers, with 256 / 128 / 64 / 32 hidden layer neurons, ReLU activation function, Adam optimization algorithm, and 500 iterations.

[0110] (4) Input the training samples into the configured model for training until the required number of training iterations or error metrics are met:

[0111] By selecting evaluation indicators and comparing them with actual values ​​to calculate errors, the predictive performance of the model is evaluated. In the formula, a... i and a p These are the accurate value and the predicted value, respectively.

[0112] Absolute error (AE):

[0113] AE = a i -a p (6)

[0114] Mean relative error (MRE):

[0115]

[0116] Mean absolute error (MAE):

[0117]

[0118] Root mean square error (RMSE):

[0119]

[0120] In step 3, frequency stability is determined by inputting the parameters into the trained model, which allows for rapid calculation of the frequency response curve, such as... Figure 3 As shown, comparing the maximum frequency deviation with the stability threshold to determine the system's frequency stability mainly includes:

[0121] (1) Input the input parameters into the trained model to obtain accurate mechanistic model parameters.

[0122] (2) Differentiate equation (4) and set its derivative to 0 to obtain the time when the maximum frequency deviation occurs:

[0123]

[0124] (3) Substituting equation (10) into the time-domain expression (4), we obtain the maximum frequency deviation:

[0125]

[0126] (4) Set the frequency deviation constraint value to Δf sh Compare Δf m and Δf sh If Δf sh >Δf m If the system is stable, then it is stable; otherwise, it is unstable.

[0127] Example 2:

[0128] This invention also proposes a frequency stability quantization discrimination system 200 adapted to different scenarios, such as... Figure 4 As shown, it includes:

[0129] The data acquisition unit 201 is used to acquire the model parameters of the physical model of the target frequency response of the target power system;

[0130] The identification unit 202 identifies the model parameters based on the discrimination model to obtain the frequency response curve of the target frequency response physical model for the target power system;

[0131] The quantization unit 203 is used to quantify the frequency stability of the target power system by judging the frequency response curve.

[0132] The data acquisition unit 201 is also used to build a discrimination model, wherein building the discrimination model includes:

[0133] The discriminant model is constructed, including:

[0134] Based on the frequency response physical model, the power system under different scenarios is simulated and run to obtain simulation data, and the input and output feature quantities of the frequency response physical model are extracted from the simulation data.

[0135] The input and output features are normalized to obtain sample data, which is then divided into training and testing sample data according to a preset ratio. The preset model is trained based on the training and testing sample data to obtain a discriminant model.

[0136] Among them, different scenarios include at least one of the following: a three-phase short-circuit fault in the power system and a power step disturbance fault in the power system.

[0137] Among them, the power load of the power system fluctuates randomly under different scenarios.

[0138] The process of extracting input features from the simulation data for the frequency response physical model includes: extracting the frequency information of the power system at the moment of disturbance and within a preset time period after the moment of disturbance from the simulation data as the input features.

[0139] The frequency information includes: natural oscillation angular frequency and damping ratio.

[0140] Specifically, the output features of the frequency response physical model extracted from the simulation data include:

[0141] The mechanism parameters are determined based on the response data in the simulation data;

[0142] The mechanistic parameters of the physical model of the frequency response to be identified in the simulation data are extracted as the output feature quantity.

[0143] The preset ratio is 70%:30%.

[0144] The process of training a pre-defined model based on training sample data and test sample data includes:

[0145] Determine the number of neural network layers and the number of neurons in each layer of the preset model. Based on the number of neural network layers and the number of neurons in each layer, input training sample data as needed for training until the training times or error indicators meet the requirements, and obtain the initial discrimination model.

[0146] The test sample data is input into the initial discrimination model to verify the accuracy of the initial discrimination model. The initial discrimination model that passes the verification is the discrimination model.

[0147] The error index includes at least one of the following: absolute error, average relative error, average absolute error, and root mean square error.

[0148] The quantization unit 203 quantifies the frequency stability of the target power system by judging the frequency response curve, including:

[0149] Determine the maximum frequency deviation value of the frequency response curve, and compare the maximum frequency deviation value with the set frequency deviation constraint value;

[0150] If the frequency deviation constraint value is greater than the maximum frequency deviation value, the target power system is in a stable state; otherwise, the target power system is in an unstable state.

[0151] Example 3:

[0152] This invention proposes a method for building a discriminative model, comprising:

[0153] Step 1: Simulate the power system under different scenarios based on the frequency response physical model, obtain simulation data, and extract the input and output features of the frequency response physical model from the simulation data.

[0154] Step 2: Normalize the input and output features to obtain sample data, and divide the sample data into training sample data and test sample data according to a preset ratio. Train the preset model based on the training sample data and test sample data to obtain the discriminant model.

[0155] This invention can identify the model parameters of the physical model of frequency response to achieve fast and accurate frequency calculation, thereby determining whether the system frequency is unstable.

[0156] Example 4:

[0157] Based on the same inventive concept, this invention also provides a computer device, which includes a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to implement corresponding method flows or corresponding functions, thereby implementing the steps of the methods in the above embodiments.

[0158] Example 5:

[0159] Based on the same inventive concept, this invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the steps of the method in the above embodiments.

[0160] 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.

[0161] 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.

[0162] 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.

[0163] 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.

[0164] 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.

[0165] 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.

Claims

1. A frequency stability quantization discrimination method adaptable to different scenarios, characterized in that, The method includes: Obtain the model parameters of the physical model of the target frequency response of the target power system; The model parameters are identified based on the discriminant model to obtain the frequency response curve of the target frequency response physical model for the target power system; The frequency stability of the target power system is quantified by analyzing the frequency response curve. The construction of the discrimination model includes: Based on the frequency response physical model, the power system under different scenarios is simulated and run to obtain simulation data, and the input and output feature quantities of the frequency response physical model are extracted from the simulation data. The input and output features are normalized to obtain sample data, and the sample data is divided into training sample data and test sample data according to a preset ratio. The preset model is trained based on the training sample data and test sample data to obtain a discriminant model. The training of the preset model based on training sample data and test sample data includes: Determine the number of neural network layers and the number of neurons in each layer of the preset model. Based on the number of neural network layers and the number of neurons in each layer, input training sample data as needed for training until the training times or error indicators meet the requirements, and obtain the initial discrimination model. The test sample data is input into the initial discrimination model to verify the accuracy of the initial discrimination model. The initial discrimination model that passes the verification is the discrimination model. The error index includes at least one of the following: absolute error, mean relative error, mean absolute error, and root mean square error. The process of quantifying the frequency stability of the target power system by analyzing the frequency response curve includes: Determine the maximum frequency deviation value of the frequency response curve, and compare the maximum frequency deviation value with the set frequency deviation constraint value; If the frequency deviation constraint value is greater than the maximum frequency deviation value, the target power system is in a stable state; otherwise, the target power system is in an unstable state. The formula for calculating the occurrence time of the maximum frequency deviation is as follows: The formula for calculating the maximum frequency deviation is as follows: in, The time when the frequency deviation value occurs. The reheater time constant is... This represents the maximum frequency deviation. For the speed control gain of the unit's speed governor, This is the load frequency regulation coefficient. It is the natural oscillation angular frequency. The damping ratio; in: 。 2. The method according to claim 1, characterized in that, The different scenarios include at least one of the following: a three-phase short-circuit fault in the power system and a power step disturbance fault in the power system.

3. The method according to claim 1, characterized in that, The power system load fluctuates randomly under different scenarios.

4. The method according to claim 1, characterized in that, The extraction of input features for the frequency response physical model from the simulation data includes: extracting the frequency information of the power system at the moment of disturbance and within a preset time period after the moment of disturbance from the simulation data as the input features; The frequency information includes: natural oscillation angular frequency and damping ratio.

5. The method according to claim 1, characterized in that, The extracted output features of the frequency response physical model from the simulation data include: The mechanism parameters are determined based on the response data in the simulation data; The mechanistic parameters of the physical model of the frequency response to be identified in the simulation data are extracted as the output feature quantity.

6. The method according to claim 1, characterized in that, The preset ratio is 70%:30%.

7. A frequency stability quantization discrimination system adaptable to different scenarios, characterized in that, The system includes: The data acquisition unit is used to acquire the model parameters of the physical model of the target frequency response of the target power system; The identification unit identifies the model parameters based on the discrimination model to obtain the frequency response curve of the target frequency response physical model for the target power system; A quantization unit is used to quantify the frequency stability of the target power system by judging the frequency response curve; The data acquisition unit is also used to build a discrimination model, which includes: Based on the frequency response physical model, the power system under different scenarios is simulated and run to obtain simulation data, and the input and output feature quantities of the frequency response physical model are extracted from the simulation data. The input and output features are normalized to obtain sample data, and the sample data is divided into training sample data and test sample data according to a preset ratio. The preset model is trained based on the training sample data and test sample data to obtain a discriminant model. The training of the preset model based on training sample data and test sample data includes: Determine the number of neural network layers and the number of neurons in each layer of the preset model. Based on the number of neural network layers and the number of neurons in each layer, input training sample data as needed for training until the training times or error indicators meet the requirements, and obtain the initial discrimination model. The test sample data is input into the initial discrimination model to verify the accuracy of the initial discrimination model. The initial discrimination model that passes the verification is the discrimination model. Among them, the error index includes at least one of the following: absolute error, average relative error, average absolute error, and root mean square error; The quantization unit quantifies the frequency stability of the target power system by judging the frequency response curve, including: Determine the maximum frequency deviation value of the frequency response curve, and compare the maximum frequency deviation value with the set frequency deviation constraint value; If the frequency deviation constraint value is greater than the maximum frequency deviation value, the target power system is in a stable state; otherwise, the target power system is in an unstable state. The formula for calculating the occurrence time of the maximum frequency deviation is as follows: The formula for calculating the maximum frequency deviation is as follows: in, The time when the frequency deviation value occurs. The reheater time constant is... This represents the maximum frequency deviation. For the speed control gain of the unit's speed governor, This is the load frequency regulation coefficient. It is the natural oscillation angular frequency. The damping ratio; in: 。 8. The system according to claim 7, characterized in that, The different scenarios include at least one of the following: a three-phase short-circuit fault in the power system and a power step disturbance fault in the power system.

9. The system according to claim 7, characterized in that, The power system load fluctuates randomly under different scenarios.

10. The system according to claim 7, characterized in that, The extraction of input features for the frequency response physical model from the simulation data includes: extracting the frequency information of the power system at the moment of disturbance and within a preset time period after the moment of disturbance from the simulation data as the input features; The frequency information includes: natural oscillation angular frequency and damping ratio.

11. The system according to claim 7, characterized in that, Extract the output features of the frequency response physical model from the simulation data, including: The mechanism parameters are determined based on the response data in the simulation data; The mechanistic parameters of the physical model of the frequency response to be identified in the simulation data are extracted as the output feature quantity.

12. The system according to claim 7, characterized in that, The preset ratio is 70%:30%.

13. A computer device, characterized in that, include: One or more processors; A processor is used to execute one or more programs; When the one or more programs are executed by the one or more processors, the method described in any one of claims 1-6 is implemented.

14. A computer-readable storage medium, characterized in that, It contains a computer program, which, when executed, implements the method as described in any one of claims 1-6.

Images