Method and device for determining frequency safety of power system and storage medium

Abstract

The invention discloses a method and a device for determining frequency safety of a power system and a storage medium. The method comprises the following steps: acquiring a multi-machine frequency response model of a plurality of generator sets of a to-be-detected power system; performing frequency simulation on the corresponding generator set by using each frequency response model to obtain a frequency time domain curve, and performing parameter fitting on each frequency time domain curve to obtain a frequency time domain formula; performing Laplace transformation on the frequency time domain formula to obtain a single-machine frequency response model; establishing a nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model by using a generalization regression neural network, determining single-machine frequency response parameters of the single-machine frequency response model based on the nonlinear relationship, and determining an aggregation power system frequency response model based on the single-machine frequency response parameters; and simulating the frequency response process of the to-be-detected power system by using the aggregated power system frequency response model, and determining the frequency safety of the to-be-detected power system based on a simulation result.

Description

Technical Field

[0001] The present invention relates to the field of power system frequency, and in particular to a method, device and storage medium for determining power system frequency safety. Background Art

[0002] Frequency is an important operating indicator of the power system. With the construction and development of new power systems, frequency security issues have become prominent. Therefore, it is urgent to propose a method that can quickly and accurately estimate the frequency dynamic response of the power system and determine whether the frequency of the power system is safe, so as to respond to power system frequency issues in a timely manner.

[0003] At present, the frequency safety of each unit is usually predicted through the frequency response model corresponding to each unit. However, this frequency safety prediction method has a low efficiency due to the large number of units in the power system. At the same time, the frequency safety needs to be predicted as many times as there are generators. When the number of frequency safety predictions increases, the number of frequency safety prediction errors will increase accordingly. Summary of the invention

[0004] The present invention provides a method, device and storage medium for determining the frequency security of an electric power system, which are mainly capable of improving the prediction efficiency and prediction accuracy of the frequency security of an electric power system.

[0005] According to a first aspect of the present invention, there is provided a method for determining frequency security of a power system, comprising:

[0006] Obtaining a multi-machine frequency response model corresponding to multiple generator sets of the power system to be detected;

[0007] Using each of the frequency response models to perform frequency simulation on the corresponding generator sets in the power system to be detected, obtaining the frequency-time domain curves corresponding to each of the generator sets, and performing parameter fitting on each of the frequency-time domain curves to obtain the frequency-time domain formula of the power system to be detected;

[0008] Performing Laplace transformation on the frequency-time domain formula to obtain a single-machine frequency response model of the power system to be detected;

[0009] A nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model is established by using a preset generalized regression neural network, and based on the nonlinear relationship, a single-machine frequency response parameter of the single-machine frequency response model is determined, and based on the single-machine frequency response parameter, an aggregated power system frequency response model of the power system to be detected is determined;

[0010] The aggregated power system frequency response model is used to simulate the frequency response process of the power system to be detected in a predetermined disturbance scenario to obtain a simulation result, and based on the simulation result, the frequency security of the power system to be detected is determined.

[0011] Optionally, the multi-machine frequency response model includes a thermal power frequency response model, a nuclear power frequency response model, a hydropower frequency response model, a wind, solar and storage frequency response model, and an ultra-high voltage frequency response model; the frequency time domain curve includes a thermal power frequency time domain curve corresponding to the thermal power frequency response model, a nuclear power frequency time domain curve corresponding to the nuclear power frequency response model, a hydropower frequency time domain curve corresponding to the hydropower frequency response model, a wind and solar frequency time domain curve corresponding to the wind, solar and storage frequency response model, and a high-voltage frequency time domain curve corresponding to the ultra-high voltage frequency response model.

[0012] Optionally, performing parameter fitting on each of the frequency-time domain curves to obtain a frequency-time domain formula of the power system to be detected includes:

[0013] Respectively determine the parameters to be fitted corresponding to the thermal power frequency time domain curve, the nuclear power frequency time domain curve, the hydropower frequency time domain curve, the wind and solar frequency time domain curve, and the high-voltage frequency time domain curve, as well as the parameter value ranges of the parameters to be fitted;

[0014] Based on the parameters to be fitted and their corresponding parameter value ranges, the nonlinear least squares method is used to perform parameter fitting on the thermal power frequency time domain curve, the nuclear power frequency time domain curve, the hydropower frequency time domain curve, the wind and solar frequency time domain curve, and the high-voltage frequency time domain curve to obtain the frequency time domain formula of the power system to be detected.

[0015] Optionally, performing Laplace transform on the frequency-time domain formula to obtain a single-machine frequency response model of the power system to be detected includes:

[0016] Determine the Laplace transform formula;

[0017] Substitute the frequency-time domain formula into the Laplace transform formula to obtain the substituted formula, perform time integration from zero to positive infinity on the substituted formula, and determine the single-machine frequency response model of the power system to be detected based on the integration result.

[0018] Optionally, the preset generalized regression neural network includes an input layer, a radial base layer and an output layer;

[0019] The method of establishing a nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model by using a preset generalized regression neural network, and determining a single-machine frequency response parameter of the single-machine frequency response model based on the nonlinear relationship, includes:

[0020] Determine a multi-machine frequency response parameter vector corresponding to the multi-machine frequency response model, a unit matrix composed of model parameters of the single-machine frequency response model, a multidimensional matrix composed of center points of the input layer radial basis functions, and a comprehensive parameter matrix composed of model parameters of the single-machine frequency response model and model parameters of the multi-machine frequency response model;

[0021] Based on the multi-dimensional matrix, the Euclidean distance between the multi-machine frequency response parameter vector and the center point of each radial basis function is calculated to obtain the multi-machine frequency spacing, and the unit matrix is ​​feature processed by using the input layer to obtain a single-machine feature vector, and the multi-machine frequency spacing is feature processed by using the input layer to obtain a multi-machine feature vector;

[0022] Performing Hadamard product processing on the single-machine feature vector and the multi-machine feature vector, and inputting the Hadamard product processing result into the radial base layer, and performing activation processing on the Hadamard product processing result through the activation function in the radial base layer to obtain an activated feature vector;

[0023] Performing a normalized dot product operation on the activation feature vector and the comprehensive parameter matrix to obtain a dot product feature vector;

[0024] The point product feature vector is input to the output layer, and linear weighting is performed at the output layer to obtain the single-machine frequency response parameters of the single-machine frequency response model.

[0025] Optionally, simulating the frequency response process of the power system to be detected in a predetermined disturbance scenario by using the aggregated power system frequency response model to obtain a simulation result includes:

[0026] Setting a predetermined disturbance scenario based on the actual operating environment, actual operating characteristics, and actual interference conditions of the power system to be detected;

[0027] Setting model parameters of the aggregated power system frequency response model based on the inertia constant of the generator in the power system to be detected, the gain of the speed regulator, and the frequency regulation effect of the load;

[0028] The aggregated power system frequency response model after setting parameters is controlled by preset simulation software to simulate the frequency response process of the power system to be detected in the predetermined disturbance scenario to obtain a simulation result.

[0029] Optionally, determining the frequency security of the power system to be detected based on the simulation result includes:

[0030] Based on the simulation result, determining a disturbance frequency value of the power system to be detected after being disturbed by the predetermined disturbance scenario;

[0031] If the disturbance frequency value is greater than a preset frequency threshold, it is determined that the frequency of the power system to be detected is unsafe; otherwise, it is determined that the frequency of the power system to be detected is safe.

[0032] According to a second aspect of the present invention, there is provided a device for determining frequency security of a power system, comprising:

[0033] An acquisition unit, used for acquiring a multi-machine frequency response model corresponding to a plurality of generator sets of the power system to be detected;

[0034] A simulation unit, used for performing frequency simulation on the corresponding generator sets in the power system to be detected by using each of the frequency response models, obtaining the frequency-time domain curves corresponding to each of the generator sets, and performing parameter fitting on each of the frequency-time domain curves to obtain the frequency-time domain formula of the power system to be detected;

[0035] A transformation unit, used for performing Laplace transformation on the frequency-time domain formula to obtain a single-machine frequency response model of the power system to be detected;

[0036] A determination unit, configured to establish a nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model by using a preset generalized regression neural network, and determine a single-machine frequency response parameter of the single-machine frequency response model based on the nonlinear relationship, and determine an aggregated power system frequency response model of the power system to be detected based on the single-machine frequency response parameter;

[0037] A simulation unit is used to simulate the frequency response process of the power system to be detected in a predetermined disturbance scenario using the aggregated power system frequency response model to obtain a simulation result, and determine the frequency security of the power system to be detected based on the simulation result.

[0038] Optionally, the multi-machine frequency response model includes a thermal power frequency response model corresponding to the thermal power generating set, a nuclear power frequency response model corresponding to the nuclear power generating set, a hydropower frequency response model corresponding to the hydropower generating set, a wind-solar-storage frequency response model corresponding to the wind-solar generating set, and an ultra-high voltage frequency response model corresponding to the ultra-high voltage generating set; the frequency time domain curve includes a thermal power frequency time domain curve corresponding to the thermal power frequency response model, a nuclear power frequency time domain curve corresponding to the nuclear power frequency response model, a hydropower frequency time domain curve corresponding to the hydropower frequency response model, a wind-solar frequency time domain curve corresponding to the wind-solar-storage frequency response model, and a high-voltage frequency time domain curve corresponding to the ultra-high voltage frequency response model.

[0039] Optionally, the simulation unit includes a first determination module and a fitting module;

[0040] The first determination module is used to respectively determine the parameters to be fitted corresponding to the thermal power frequency time domain curve, the nuclear power frequency time domain curve, the hydropower frequency time domain curve, the wind and solar frequency time domain curve, and the high-voltage frequency time domain curve, as well as the parameter value range of the parameters to be fitted;

[0041] The fitting module is used to perform parameter fitting on the thermal power frequency time domain curve, the nuclear power frequency time domain curve, the hydropower frequency time domain curve, the wind and solar frequency time domain curve, and the high-voltage frequency time domain curve based on the parameters to be fitted and their corresponding parameter value ranges, using the nonlinear least squares method to obtain the frequency time domain formula of the power system to be detected.

[0042] Optionally, the transformation unit includes a second determination module and an integration module;

[0043] The second determination module is used to determine the Laplace transform formula;

[0044] The integration module is used to substitute the frequency-time domain formula into the Laplace transform formula to obtain the substituted formula, perform time integration from zero to positive infinity on the substituted formula, and determine the single-machine frequency response model of the power system to be detected based on the integration result.

[0045] Optionally, the preset generalized regression neural network includes an input layer, a radial base layer and an output layer; the determination unit includes a third determination module, a calculation module, a processing module, a dot multiplication operation module, and a weighting module;

[0046] The third determination module is used to determine the multi-machine frequency response parameter vector corresponding to the multi-machine frequency response model, the unit matrix composed of the model parameters of the single-machine frequency response model, the multidimensional matrix composed of the center points of the input layer radial basis function, and the comprehensive parameter matrix composed of the model parameters of the single-machine frequency response model and the model parameters of the multi-machine frequency response model;

[0047] The calculation module is used to calculate the Euclidean distance between the multi-machine frequency response parameter vector and the center point of each radial basis function based on the multi-dimensional matrix to obtain the multi-machine frequency spacing, and use the input layer to perform feature processing on the unit matrix to obtain a single-machine feature vector, and use the input layer to perform feature processing on the multi-machine frequency spacing to obtain a multi-machine feature vector;

[0048] The processing module is used to perform Hadamard product processing on the single-machine feature vector and the multi-machine feature vector, and input the Hadamard product processing result to the radial base layer, and activate the Hadamard product processing result through the activation function in the radial base layer to obtain an activated feature vector;

[0049] The dot product operation module is used to perform a normalized dot product operation on the activation feature vector and the comprehensive parameter matrix to obtain a dot product feature vector;

[0050] The weighting module is used to input the point product feature vector into the output layer, perform linear weighting at the output layer, and obtain the single-machine frequency response parameters of the single-machine frequency response model.

[0051] Optionally, the simulation unit includes a setting module and a simulation module;

[0052] The setting module is used to set a predetermined disturbance scenario based on the actual operating environment, actual operating characteristics, and actual interference conditions faced by the power system to be detected;

[0053] The setting module is used to set the model parameters of the aggregated power system frequency response model based on the inertia constant of the generator, the gain of the speed regulator, and the frequency regulation effect of the load in the power system to be detected;

[0054] The simulation module is used to use preset simulation software to control the aggregated power system frequency response model after setting parameters to simulate the frequency response process of the power system to be detected in the predetermined disturbance scenario to obtain a simulation result.

[0055] Optionally, the simulation unit further includes a determination module;

[0056] The determination module is used to determine the disturbance frequency value of the power system to be detected after being disturbed by the predetermined disturbance scenario based on the simulation result;

[0057] The determination module is used to determine that the frequency of the power system to be detected is unsafe if the disturbance frequency value is greater than a preset frequency threshold, and otherwise, determine that the frequency of the power system to be detected is safe.

[0058] According to a third aspect of the present invention, there is provided a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above method for determining frequency security of a power system.

[0059] According to a fourth aspect of the present invention, there is provided a computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method for determining frequency security of a power system when executing the program.

[0060] According to a method, device and storage medium for determining frequency security of a power system provided by the present invention, compared with the current method of predicting the frequency security of each unit separately through the frequency response model corresponding to each unit, the present invention performs frequency simulation on the multi-machine frequency response model of each generator set to obtain a frequency time domain curve, and performs parameter fitting on the frequency time domain curve to obtain a frequency time domain formula, then performs Laplace transform on the frequency time domain formula to obtain a single-machine frequency response model of the power system, and establishes a nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model, determines the single-machine frequency response parameters of the single-machine frequency response model based on the nonlinear relationship, and then determines the aggregated power system frequency response model of the power system based on the single-machine frequency response parameters, and finally uses the single aggregated power system frequency response model to predict the frequency security of the entire power system, that is, only one aggregated power system frequency response model is used to predict the frequency security of each generator set in the entire power system, thereby avoiding the time and resources wasted by using multiple models to predict frequency security, and also avoiding the risk of increased error due to an increase in the number of predictions, so that the present invention can improve the prediction efficiency and prediction accuracy of the frequency security of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0062] Figure 1 A flow chart of a method for determining frequency security of a power system provided by an embodiment of the present invention is shown;

[0063] Figure 2 A schematic diagram of a single-machine frequency response model provided by an embodiment of the present invention is shown;

[0064] Figure 3 A flow chart of another method for determining power system frequency security provided by an embodiment of the present invention is shown;

[0065] Figure 4 A schematic diagram showing the structure of a device for determining frequency security of a power system provided by an embodiment of the present invention is shown;

[0066] Figure 5 A schematic diagram of the physical structure of a computer device provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0067] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0068] At present, the frequency safety of each unit is predicted separately through the frequency response model corresponding to each unit. However, due to the large number of units in the power system, the efficiency of frequency safety prediction is low. When the number of frequency safety predictions increases, the number of frequency safety prediction errors will increase accordingly.

[0069] In order to solve the above problems, an embodiment of the present invention provides a method for determining power system frequency security, such as Figure 1 As shown, the method includes:

[0070] 101. Obtain a multi-machine frequency response model corresponding to multiple generator sets of a power system to be detected.

[0071] Among them, the power system to be detected contains multiple generator sets, each generator set corresponds to a frequency response model, and the frequency response models corresponding to each generator set are collectively referred to as multi-machine frequency response models. The frequency response model of each generator set has been constructed in advance. Specifically, a differential equation modeling method can be adopted, and modern modeling technologies such as neural networks can also be used to pre-construct the frequency response models corresponding to each generator set.

[0072] 102. Use each frequency response model to perform frequency simulation on the corresponding generator set in the power system to be tested, obtain the frequency-time domain curve corresponding to each generator set, and perform parameter fitting on each frequency-time domain curve to obtain the frequency-time domain formula of the power system to be tested.

[0073] Among them, the frequency time domain curve refers to the frequency characteristic curve of each generator set, that is, the relationship curve between the generator set frequency and the active load when the load is constantly changing.

[0074] For the embodiment of the present invention, simulation parameters are set, and the simulation parameters include parameters such as the initial state of each generator set, load changes, fault type and its duration, and output variables that need to be monitored, such as system frequency, etc., and then the frequency response model and simulation parameters of each generator set are respectively input into the simulation tool, and the simulation program is run to realize frequency simulation of the corresponding generator set in the power system to be detected. During the simulation process, the frequency response characteristics of the power system to be detected are monitored in real time, so as to obtain the frequency time domain curve of each generator set, and then the nonlinear least squares method is used to perform parameter fitting on each frequency time domain curve, so as to fit each frequency time domain curve into a frequency time domain formula. Specifically, the frequency time domain fitting formula is first selected as follows:

[0075]

[0076] Among them, ΔP dis a step disturbance, C0, C1, C2, T1, T2, ω2, θ2 are all fitting parameters, t is time, and according to the above frequency-time domain fitting formula, the nonlinear least squares method is used to fit each frequency-time domain curve. The expression is as follows:

[0077]

[0078] Where Δf represents the frequency Lapse transformation, ΔP d is a step disturbance, C0, C1, C2, T1, T2, ω2, θ2 are all fitting parameters in the frequency-time domain curves, t is the time, and the frequency-time domain formula can be obtained by solving the above nonlinear equations.

[0079] 103. Perform Laplace transform on the frequency-time domain formula to obtain the single-machine frequency response model of the power system to be tested.

[0080] For the embodiment of the present invention, Laplace transform is performed on the frequency-time domain formula, as shown in the following formula:

[0081]

[0082] Among them, Δf(s) represents the complex frequency domain form, ΔP d is a step disturbance, s represents a complex variable, and C0, C1, C2, T1, T2, ω2, and θ2 are fitting parameters in the time domain curves of each frequency.

[0083] The order of the denominator of the transfer function of the real system is greater than the order of the numerator. The above formula can be transformed into the following formula:

[0084]

[0085] Wherein, Δf represents the Laplace transformation of frequency, G(s) represents the second-order transfer function, and B0, B1, A0, A1, A2, and A3 are coefficients.

[0086] Then G m (s) is the second-order transfer function, and its standard form is as follows:

[0087]

[0088] Among them, b0, b1, a0, and a1 are all coefficients.

[0089] The transfer function of the single-machine frequency response model is shown in the following formula:

[0090]

[0091] Among them, T J Inertia, D Σ Indicates damping.

[0092] The following formulas can be used to determine a0, a1, T J , D Σ , b0.

[0093] a0=B0

[0094] a1=B1

[0095]

[0096] b0=A2-T J -D Σ a1

[0097] b1 can be determined by the following formula.

[0098]

[0099] Among them, C0 represents the fitting parameter, Δf +∞ Represents the steady-state value of frequency, ΔP d+∞ Represents the steady-state value of the disturbance. Therefore, through the transfer function of the above single-machine frequency response model, the single-machine frequency response model can be determined, such as Figure 2 A single-machine frequency response model is shown, where ΔP d is a step disturbance, ΔP a represents the net input power, Σ represents summation, s represents a complex variable, Δf represents the Laplacian transformation of frequency, ΔP m Indicates the frequency modulation voltage response power.

[0100] 104. A nonlinear relationship between a single-machine frequency response model and a multi-machine frequency response model is established using a preset generalized regression neural network, and based on the nonlinear relationship, the single-machine frequency response parameters of the single-machine frequency response model are determined, and based on the single-machine frequency response parameters, the aggregated power system frequency response model of the power system to be tested is determined.

[0101] Among them, the single-machine frequency response parameters refer to the dynamic response characteristics of the power system when the frequency changes, including inertia time constant, speed regulator parameters, load frequency control parameters, etc.

[0102] For the embodiment of the present invention, after determining the single-machine frequency response model, a nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model is established using a preset generalized regression neural network, and based on the nonlinear relationship, the single-machine frequency response parameters of the single-machine frequency response model are determined, and the single-machine frequency response model is updated using the single-machine frequency response parameters to obtain an aggregated power system frequency response model of the power system to be detected. Thus, by establishing the aggregated power system frequency response model through a generalized regression neural network, the construction efficiency and construction accuracy of the aggregated power system frequency response model can be improved.

[0103] 105. Use the aggregated power system frequency response model to simulate the frequency response process of the power system to be tested in a predetermined disturbance scenario, obtain simulation results, and determine the frequency security of the power system to be tested based on the simulation results.

[0104] The predetermined disturbance scenarios include sudden load changes, generator failures, line failures and other scenarios of the power system to be detected. For each scenario, it is necessary to determine parameters such as the type, size, occurrence time and duration of the disturbance.

[0105] For the embodiment of the present invention, firstly, the simulation parameters are set, and the simulation parameters include disturbance data. The disturbance data of the predetermined disturbance scenario is input into the frequency response model of the aggregated power system. The disturbance data includes the type, size, occurrence time and duration of the disturbance. And set the simulation conditions: according to actual needs, set the initial conditions, time step, simulation duration and other parameters of the simulation. Run the simulation and analyze the results: Run the simulation: Use the established frequency response model of the aggregated power system and set the simulation parameters and simulation conditions, run the simulation program, and the simulation program will calculate the frequency response process of the power system to be tested under the predetermined disturbance scenario. Analyze the results: After the simulation is completed, analyze the simulation results, such as observing the frequency change curve over time, calculating the frequency deviation, and analyzing the frequency security of the power system to be tested. Therefore, by comprehensively analyzing the multi-machine frequency response models corresponding to multiple generator sets, a single aggregated power system frequency response model capable of performing frequency analysis on the entire power system is constructed. Only one aggregated power system frequency response model can be used to predict the frequency safety of each generator set in the entire power system, thereby avoiding the waste of time and resources caused by using multiple models to predict frequency safety, and also avoiding the risk of increased error due to an increase in the number of predictions. Therefore, the present invention can improve the prediction efficiency and accuracy of the frequency safety of the power system.

[0106] According to a method for determining the frequency security of a power system provided by the present invention, compared with the current method of predicting the frequency security of each unit separately through the frequency response model corresponding to each unit, the present invention performs frequency simulation on the multi-machine frequency response model of each generator set to obtain a frequency time domain curve, and performs parameter fitting on the frequency time domain curve to obtain a frequency time domain formula, then performs Laplace transform on the frequency time domain formula to obtain a single-machine frequency response model of the power system, and establishes a nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model, determines the single-machine frequency response parameters of the single-machine frequency response model based on the nonlinear relationship, and then determines the aggregated power system frequency response model of the power system based on the single-machine frequency response parameters, and finally uses the single aggregated power system frequency response model to predict the frequency security of the entire power system, that is, only one aggregated power system frequency response model is used to predict the frequency security of each generator set in the entire power system, thereby avoiding the time and resources wasted by using multiple models to predict the frequency security, and also avoiding the risk of increased error due to an increase in the number of predictions, so that the present invention can improve the prediction efficiency and prediction accuracy of the frequency security of the power system.

[0107] Further, in order to better illustrate the above process of determining the frequency security of the power system, as a refinement and extension of the above embodiment, the embodiment of the present invention provides another method for determining the frequency security of the power system, such as Figure 3 As shown, the method includes:

[0108] 201. Obtain a multi-machine frequency response model corresponding to multiple generator sets of a power system to be detected.

[0109] Among them, multiple generator sets include: thermal power generator sets, nuclear power generator sets, hydropower generator sets, wind and solar generator sets, ultra-high voltage generator sets, etc.: the multi-machine frequency response model includes: the thermal power frequency response model corresponding to the thermal power generator set, the nuclear power frequency response model corresponding to the nuclear power generator set, the hydropower frequency response model corresponding to the hydropower generator set, the wind and solar storage frequency response model corresponding to the wind and solar generator set, and the ultra-high voltage frequency response model corresponding to the ultra-high voltage generator set; the multi-machine frequency response model is specifically a mathematical model of each generator set. It should be noted that each parameter in the multi-machine frequency response model is a fixed parameter. In order to improve the calculation accuracy, the power gain coefficient can also be obtained, and the power gain coefficient is determined by the unit capacity ratio to the total capacity.

[0110] 202. Use each frequency response model to perform frequency simulation on the corresponding generator set in the power system to be tested, obtain the frequency-time domain curve corresponding to each generator set, and perform parameter fitting on each frequency-time domain curve to obtain the frequency-time domain formula of the power system to be tested.

[0111] Among them, the frequency time domain curves include: the thermal power frequency time domain curve corresponding to the thermal power frequency response model, the nuclear power frequency time domain curve corresponding to the nuclear power frequency response model, the hydropower frequency time domain curve corresponding to the hydropower frequency response model, the wind and solar frequency time domain curve corresponding to the wind and solar storage frequency response model, and the high-voltage frequency time domain curve corresponding to the UHV frequency response model. The frequency time domain formula refers to the relationship between the frequency of the power system to be tested and time.

[0112] For the embodiment of the present invention, the relevant parameters and data of each generator set are collected, including the type of generator, rated power, inertia time constant, speed regulator parameters, etc. Based on the collected data, the mathematical model of the generator set, i.e., the frequency response model, is input into the simulation tool. The frequency response model can accurately reflect the dynamic response characteristics of the generator set when the frequency changes. Then, the disturbance is defined: that is, according to the actual situation of the power system to be tested and the test requirements, a suitable disturbance scenario is defined, including load mutation, generator fault removal, line fault and other scenarios. At the same time, the simulation parameters are set, including simulation time, time step, output variables, etc., to ensure that the simulation parameters can accurately reflect the operating status of the actual power system. Finally, the pre-established frequency response models of each generator set are run in the simulation tool, and the defined disturbance scenario is applied to observe the frequency response process of the generator set under the disturbance, so as to obtain the frequency time domain curve corresponding to each generator set.

[0113] Furthermore, after determining the frequency-time domain curves corresponding to each generator set, it is also necessary to perform parameter fitting on each frequency-time domain curve. Based on this, step 202 specifically includes: respectively determining the parameters to be fitted corresponding to the thermal power frequency-time domain curve, the nuclear power frequency-time domain curve, the hydropower frequency-time domain curve, the wind-solar frequency-time domain curve, and the high-voltage frequency-time domain curve, as well as the parameter value range of the parameters to be fitted; based on the parameters to be fitted and their corresponding parameter value ranges, the thermal power frequency-time domain curve, the nuclear power frequency-time domain curve, the hydropower frequency-time domain curve, the wind-solar frequency-time domain curve, and the high-voltage frequency-time domain curve are parameter fitted using the nonlinear least squares method to obtain the frequency-time domain formula of the power system to be detected.

[0114] Specifically, first of all, it is necessary to clarify the form of the objective function of the fitting. In the analysis of the power system, the frequency-time domain curve may show complex nonlinear characteristics, so it is very important to select an appropriate nonlinear function form for fitting. Common nonlinear function forms include exponential function, power function, trigonometric function, etc. The specific selection should be determined according to the actual situation and the characteristics of the data. In the fitting process, the frequency-time domain data of each generator set is first collected, including the time point and the corresponding frequency value. Then define the error function: The core of the nonlinear least squares method is to minimize the error function, that is, the residual sum of squares. For the parameter fitting of the frequency-time domain curve, the error function can be defined as the sum of the squares of the difference between the actual frequency value and the fitting function value. Select initial parameters: Select a set of initial parameter values ​​for the fitting function based on actual needs, and use these initial parameter values ​​as the starting point for iterative optimization. Construct the objective function: Take the error function as the objective function and minimize the objective function value by adjusting the parameters of the fitting function. Then select the solution algorithm of the nonlinear least squares method to continuously adjust the parameters of the fitting function through iterative optimization until the objective function value reaches the minimum or meets the predetermined convergence conditions, so as to obtain the frequency-time domain formula of the power system to be tested. The embodiment of the present invention performs parameter fitting through the nonlinear least square method, which can process the nonlinear relationship between various frequency-time domain curves, capture the nonlinear characteristics in the data, and thus provide a more accurate fitting result.

[0115] 203. Determine the Laplace transform formula.

[0116] 204. Substitute the frequency-time domain formula into the Laplace transform formula to obtain the substituted formula, and integrate the substituted formula from zero to positive infinity, and determine the single-machine frequency response model of the power system to be tested based on the integration result.

[0117] For the embodiment of the present invention, first select the Laplace transform formula, such as F(s) = ∫[0,∞]f(t)e∧(-st)dt, where f(t) is a time domain function, F(s) is a complex frequency domain function, s is a complex variable, t is a time variable, and e^(-st) is an exponential function. First, substitute the time domain function f(t) in the frequency time domain formula into the Laplace transform formula and perform an integral operation. Through the integral operation, the representation F(s) of the frequency in the complex frequency domain is obtained. This function describes the distribution characteristics of the frequency on the complex plane and is the basis of the single-machine frequency response model. Then, the obtained complex frequency domain function F(s) is analyzed to understand its poles, zeros and other characteristics on the complex plane. These characteristics are closely related to the frequency response characteristics of the power system. Based on the analysis results of the complex frequency domain function, a single-machine frequency response model of the power system to be tested is established. The single-machine frequency response model usually includes key elements such as the transfer function and frequency response curve of the system. Finally, the accuracy and reliability of the model can be verified by simulation experiments or actual data. If there is a large difference between the model and the actual data, the model needs to be adjusted and optimized.

[0118] 205. A nonlinear relationship between a single-machine frequency response model and a multi-machine frequency response model is established using a preset generalized regression neural network, and based on the nonlinear relationship, a single-machine frequency response parameter of the single-machine frequency response model is determined, and based on the single-machine frequency response parameter, an aggregated power system frequency response model of the power system to be detected is determined.

[0119] Among them, the preset generalized regression neural network includes an input layer, a radial base layer and an output layer, and the output layer is a special linear network layer; the nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model is a nonlinear mapping between the power gain coefficient and the fitting parameters of the multi-machine frequency response model.

[0120] For the embodiment of the present invention, in order to construct an aggregated power system frequency response model of the power system to be detected, it is first necessary to determine the single-machine frequency response parameters of the single-machine frequency response model. Based on this, step 205 specifically includes: determining the multi-machine frequency response parameter vector corresponding to the multi-machine frequency response model, the unit matrix composed of the model parameters of the single-machine frequency response model, the multi-dimensional matrix composed of the center points of the input layer radial basis function, and the comprehensive parameter matrix composed of the model parameters of the single-machine frequency response model and the model parameters of the multi-machine frequency response model; based on the multi-dimensional matrix, calculating the Euclidean distance between the multi-machine frequency response parameter vector and the center points of each radial basis function to obtain the multi-machine frequency spacing, The input layer is used to perform feature processing on the unit matrix to obtain a single-machine feature vector, and the input layer is used to perform feature processing on the multi-machine frequency spacing to obtain a multi-machine feature vector; the single-machine feature vector and the multi-machine feature vector are processed by Hadamard product, and the Hadamard product processing result is input into the radial base layer, and the Hadamard product processing result is activated by the activation function in the radial base layer to obtain an activated feature vector; the activated feature vector is normalized point multiplication operation is performed with the comprehensive parameter matrix to obtain a point product feature vector; the point product feature vector is input into the output layer, and linear weighting is performed at the output layer to obtain the single-machine frequency response parameters of the single-machine frequency response model.

[0121] The multi-machine frequency response parameters refer to the response characteristics of each generator set to different frequency signals, which can be obtained by simulating the frequency response models corresponding to each generator set.

[0122] Specifically, first determine the multi-machine frequency response parameter vector of the frequency response model corresponding to each generator set, and at the same time determine the unit matrix composed of the model parameters of the single-machine frequency response model, the multi-dimensional matrix composed of the center points of the radial basis function of the input layer, such as the q×r-dimensional matrix, calculate the Euclidean distance between the multi-machine frequency response parameter vector and the center points of each radial basis function, obtain the multi-machine frequency spacing, and input the multi-machine frequency spacing to the input layer, output the multi-machine feature vector through the input layer, use the input layer to perform feature processing on the unit matrix, obtain the single-machine feature vector, perform Hadamard product processing on the single-machine feature vector and the multi-machine feature vector, and input the Hadamard product processing result to the radial base layer for activation processing to obtain the activated feature vector, then perform normalized point multiplication operation on the activated feature vector and the comprehensive parameter matrix composed of the model parameters of the single-machine frequency response model and the model parameters of the multi-machine frequency response model to obtain the point product feature vector, and finally input the point product feature vector to the output layer for linear weighting to obtain the single-machine frequency response parameter, and its weight parameter is all 1.

[0123] Furthermore, the frequency response parameters of individual machines are aggregated based on the topological structure of the power system to be detected to obtain an aggregated power system frequency response model.

[0124] 206. Use the aggregated power system frequency response model to simulate the frequency response process of the power system to be tested in a predetermined disturbance scenario to obtain simulation results, and determine the frequency security of the power system to be tested based on the simulation results.

[0125] For the embodiment of the present invention, after determining the frequency response model of the aggregated power system, in order to determine the frequency safety of the power system to be detected, it is first necessary to simulate the frequency response process of the power system to be detected. Based on this, step 206 specifically includes: setting a predetermined disturbance scenario based on the actual operating environment, actual operating characteristics, and actual interference conditions faced by the power system to be detected; setting model parameters of the frequency response model of the aggregated power system based on the inertia constant of the generator in the power system to be detected, the gain of the speed regulator, and the frequency regulation effect of the load; using preset simulation software to control the aggregated power system frequency response model after setting parameters to simulate the frequency response process of the power system to be detected in the predetermined disturbance scenario to obtain simulation results.

[0126] Among them, the actual operating environment includes temperature, humidity, geographical location, etc.; the actual operating characteristics refer to the load characteristics, power generation characteristics, network structure characteristics, etc. of the power system to be tested; the actual interference situation refers to the load disturbance, generator failure, line failure, external interference, etc. of the power system to be tested.

[0127] Specifically, based on the analysis of the actual operating environment, actual operating characteristics, and actual interference faced by the power system to be tested, a predetermined disturbance scenario is set. The set disturbance scenario should be able to reflect various situations that the power system may encounter in actual operation, such as setting a scenario where the load suddenly increases or decreases in a short period of time. Determine the parameters of the aggregated power system frequency response model, including the inertia constant of the generator, the speed regulator parameters, the load characteristics, etc. These parameters are set according to the actual situation of the power system, and configure the aggregated power system frequency response model, such as setting the inertia constant, speed regulator parameters, load characteristics, etc. of each generator in the power system to be tested, and configure the predetermined disturbance scenario. Run the configured aggregated power system frequency response model in the simulation tool to simulate the power system frequency response process under the predetermined disturbance scenario. During the simulation process, monitor the frequency response characteristics of the power system in real time, including key indicators such as frequency deviation and frequency change rate, so as to obtain simulation results.

[0128] Furthermore, the frequency response process of the power system to be detected is simulated using the aggregated power system frequency response model. After the simulation results are obtained, the frequency security of the power system to be detected needs to be determined according to the simulation results. Based on this, the method includes: based on the simulation results, determining the disturbance frequency value of the power system to be detected after the disturbance in the predetermined disturbance scenario; if the disturbance frequency value is greater than a preset frequency threshold, determining that the frequency of the power system to be detected is unsafe, otherwise, determining that the frequency of the power system to be detected is safe.

[0129] Among them, the preset frequency threshold is set according to actual needs. Specifically, according to the frequency response process simulation results, the disturbance frequency value of the power system to be detected after the disturbance in the predetermined disturbance scenario can be obtained. If the disturbance frequency value is greater than the preset frequency threshold, it is determined that the power system to be detected is unstable, that is, the frequency is unsafe. If the disturbance frequency value is less than or equal to the preset frequency threshold, it is determined that the power system to be detected is stable, that is, the frequency is safe.

[0130] According to another method for determining the frequency security of a power system provided by the present invention, compared with the current method of predicting the frequency security of each unit separately through the frequency response model corresponding to each unit, the present invention performs frequency simulation on the multi-machine frequency response model of each generator set to obtain a frequency time domain curve, and performs parameter fitting on the frequency time domain curve to obtain a frequency time domain formula, then performs Laplace transform on the frequency time domain formula to obtain a single-machine frequency response model of the power system, and establishes a nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model, determines the single-machine frequency response parameters of the single-machine frequency response model based on the nonlinear relationship, and then determines the aggregated power system frequency response model of the power system based on the single-machine frequency response parameters, and finally uses the single aggregated power system frequency response model to predict the frequency security of the entire power system, that is, only one aggregated power system frequency response model is used to predict the frequency security of each generator set in the entire power system, thereby avoiding the time and resources wasted by using multiple models for frequency security prediction, and also avoiding the risk of increased error due to an increase in the number of predictions, so that the present invention can improve the prediction efficiency and prediction accuracy of the frequency security of the power system.

[0131] Further, as Figure 1 In a specific implementation, an embodiment of the present invention provides a device for determining the frequency security of a power system, such as Figure 4 As shown, the device includes: an acquisition unit 31, a simulation unit 32, a transformation unit 33, a determination unit 34, and a simulation unit 35.

[0132] The acquisition unit 31 may be used to acquire a multi-machine frequency response model corresponding to multiple generator sets of the power system to be detected.

[0133] The simulation unit 32 can be used to use each of the frequency response models to perform frequency simulation on the corresponding generator sets in the power system to be detected, obtain the frequency-time domain curves corresponding to each of the generator sets, and perform parameter fitting on each of the frequency-time domain curves to obtain the frequency-time domain formula of the power system to be detected.

[0134] The transformation unit 33 may be used to perform Laplace transformation on the frequency-time domain formula to obtain a single-machine frequency response model of the power system to be detected.

[0135] The determination unit 34 can be used to establish a nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model using a preset generalized regression neural network, and based on the nonlinear relationship, determine the single-machine frequency response parameters of the single-machine frequency response model, and based on the single-machine frequency response parameters, determine the aggregated power system frequency response model of the power system to be detected.

[0136] The simulation unit 35 can be used to simulate the frequency response process of the power system to be detected in a predetermined disturbance scenario using the aggregated power system frequency response model to obtain simulation results, and determine the frequency security of the power system to be detected based on the simulation results.

[0137] In a specific application scenario, the multi-machine frequency response model includes a thermal power frequency response model, a nuclear power frequency response model, a hydropower frequency response model, a wind, solar and storage frequency response model, and an ultra-high voltage frequency response model; the frequency time domain curve includes a thermal power frequency time domain curve corresponding to the thermal power frequency response model, a nuclear power frequency time domain curve corresponding to the nuclear power frequency response model, a hydropower frequency time domain curve corresponding to the hydropower frequency response model, a wind and solar frequency time domain curve corresponding to the wind, solar and storage frequency response model, and a high-voltage frequency time domain curve corresponding to the ultra-high voltage frequency response model.

[0138] In a specific application scenario, in order to determine the frequency-time domain formula, the simulation unit 32 includes a first determination module 321 and a fitting module 322 .

[0139] The first determination module 321 can be used to respectively determine the parameters to be fitted corresponding to the thermal power frequency time domain curve, the nuclear power frequency time domain curve, the hydropower frequency time domain curve, the wind and solar frequency time domain curve, and the high-voltage frequency time domain curve, as well as the parameter value range of the parameters to be fitted.

[0140] The fitting module 322 can be used to perform parameter fitting on the thermal power frequency time domain curve, the nuclear power frequency time domain curve, the hydropower frequency time domain curve, the wind and solar frequency time domain curve, and the high-voltage frequency time domain curve based on the parameters to be fitted and their corresponding parameter value ranges, using the nonlinear least squares method to obtain the frequency time domain formula of the power system to be detected.

[0141] In a specific application scenario, in order to determine the single-machine frequency response model, the transformation unit 33 includes a second determination module 331 and an integration module 332 .

[0142] The second determination module 331 may be used to determine a Laplace transform formula.

[0143] The integration module 332 can be used to substitute the frequency-time domain formula into the Laplace transform formula to obtain the substituted formula, and perform time integration from zero to positive infinity on the substituted formula, and determine the single-machine frequency response model of the power system to be detected based on the integration result.

[0144] In a specific application scenario, the preset generalized regression neural network includes an input layer, a radial base layer and an output layer. In order to determine the single-machine frequency response parameters of the single-machine frequency response model, the determination unit 34 includes a third determination module 341, a calculation module 342, a processing module 343, a point multiplication operation module 344, and a weighting module 345.

[0145] The third determination module 341 can be used to determine the multi-machine frequency response parameter vector corresponding to the multi-machine frequency response model, the unit matrix composed of the model parameters of the single-machine frequency response model, the multi-dimensional matrix composed of the center points of the input layer radial basis function, and the comprehensive parameter matrix composed of the model parameters of the single-machine frequency response model and the model parameters of the multi-machine frequency response model.

[0146] The calculation module 342 can be used to calculate the Euclidean distance between the multi-machine frequency response parameter vector and the center point of each radial basis function based on the multi-dimensional matrix to obtain the multi-machine frequency spacing, and use the input layer to perform feature processing on the unit matrix to obtain a single-machine feature vector, and use the input layer to perform feature processing on the multi-machine frequency spacing to obtain a multi-machine feature vector.

[0147] The processing module 343 can be used to perform Hadamard product processing on the single-machine feature vector and the multi-machine feature vector, and input the Hadamard product processing result into the radial base layer, and activate the Hadamard product processing result through the activation function in the radial base layer to obtain an activated feature vector.

[0148] The dot product operation module 344 may be used to perform a normalized dot product operation on the activated feature vector and the comprehensive parameter matrix to obtain a dot product feature vector.

[0149] The weighting module 345 may be used to input the point product feature vector into the output layer, perform linear weighting at the output layer, and obtain the single-machine frequency response parameters of the single-machine frequency response model.

[0150] In a specific application scenario, in order to simulate the frequency response of the power system to be detected, the simulation unit 35 includes a setting module 351 and a simulation module 352 .

[0151] The setting module 351 may be used to set a predetermined disturbance scenario based on the actual operating environment, actual operating characteristics, and actual interference conditions faced by the power system to be detected.

[0152] The setting module 351 may also be used to set model parameters of the aggregated power system frequency response model based on the inertia constant of the generator in the power system to be detected, the gain of the speed regulator, and the frequency regulation effect of the load.

[0153] The simulation module 352 can be used to use preset simulation software to control the aggregated power system frequency response model with set parameters to simulate the frequency response process of the power system to be detected in the predetermined disturbance scenario to obtain simulation results.

[0154] In a specific application scenario, in order to determine the frequency safety of the power system to be detected, the simulation unit 35 further includes a determination module 353 .

[0155] The determination module 353 can be specifically used to determine the disturbance frequency value of the power system to be detected after being disturbed by the predetermined disturbance scenario based on the simulation results; if the disturbance frequency value is greater than a preset frequency threshold, it is determined that the frequency of the power system to be detected is unsafe; otherwise, it is determined that the frequency of the power system to be detected is safe.

[0156] It should be noted that for other corresponding descriptions of the functional modules involved in the device for determining the frequency safety of a power system provided in an embodiment of the present invention, reference can be made to Figure 1 The corresponding description of the method shown will not be repeated here.

[0157] Based on the above Figure 1The method shown, accordingly, an embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the following steps are implemented: obtaining a multi-machine frequency response model corresponding to multiple generator sets of the power system to be detected; using each of the frequency response models to perform frequency simulation on the corresponding generator sets in the power system to be detected, to obtain the frequency time domain curves corresponding to each of the generator sets, and performing parameter fitting on each of the frequency time domain curves to obtain the frequency time domain formula of the power system to be detected; performing Laplace transform on the frequency time domain formula to obtain a single-machine frequency response model of the power system to be detected; using a preset generalized regression neural network to establish a nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model, and based on the nonlinear relationship, determining the single-machine frequency response parameters of the single-machine frequency response model, and based on the single-machine frequency response parameters, determining the aggregate power system frequency response model of the power system to be detected; using the aggregate power system frequency response model to simulate the frequency response process of the power system to be detected in a predetermined disturbance scenario, to obtain a simulation result, and based on the simulation result, determining the frequency security of the power system to be detected.

[0158] Based on the above Figure 1 The method shown and Figure 4 The embodiment of the device shown in the figure, the embodiment of the present invention also provides a physical structure diagram of a computer device, such as Figure 5 As shown, the computer device includes: a processor 41, a memory 42, and a computer program stored in the memory 42 and executable on the processor, wherein the memory 42 and the processor 41 are both arranged on a bus 43, and when the processor 41 executes the program, the following steps are implemented: obtaining a multi-machine frequency response model corresponding to a plurality of generator sets of the power system to be detected; using each of the frequency response models to perform frequency simulation on the corresponding generator sets in the power system to be detected, obtaining a frequency time domain curve corresponding to each of the generator sets, and performing parameter fitting on each of the frequency time domain curves to obtain a frequency time domain formula of the power system to be detected; fitting the frequency time domain formula Perform Laplace transform to obtain a single-machine frequency response model of the power system to be detected; use a preset generalized regression neural network to establish a nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model, and based on the nonlinear relationship, determine the single-machine frequency response parameters of the single-machine frequency response model, and based on the single-machine frequency response parameters, determine the aggregated power system frequency response model of the power system to be detected; use the aggregated power system frequency response model to simulate the frequency response process of the power system to be detected in a predetermined disturbance scenario to obtain a simulation result, and based on the simulation result, determine the frequency security of the power system to be detected.

[0159] Through the technical solution of the present invention, the present invention performs frequency simulation on the multi-machine frequency response model of each generator set to obtain a frequency time domain curve, and performs parameter fitting on the frequency time domain curve to obtain a frequency time domain formula, and then performs Laplace transform on the frequency time domain formula to obtain a single-machine frequency response model of the power system, and establishes a nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model, and determines the single-machine frequency response parameters of the single-machine frequency response model based on the nonlinear relationship, and then determines the aggregated power system frequency response model of the power system based on the single-machine frequency response parameters, and finally uses the single aggregated power system frequency response model to predict the frequency security of the entire power system, that is, only one aggregated power system frequency response model is used to predict the frequency security of each generator set in the entire power system, thereby avoiding the time and resources wasted by using multiple models for frequency security prediction, and also avoiding the risk of increased error due to an increase in the number of predictions, so that the present invention can improve the prediction efficiency and prediction accuracy of the frequency security of the power system.

[0160] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0161] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for determining power system frequency security, characterized in that: include: Obtaining a multi-machine frequency response model corresponding to multiple generator sets of the power system to be detected; Using each of the frequency response models to perform frequency simulation on the corresponding generator sets in the power system to be detected, obtaining the frequency-time domain curves corresponding to each of the generator sets, and performing parameter fitting on each of the frequency-time domain curves to obtain the frequency-time domain formula of the power system to be detected; Performing Laplace transformation on the frequency-time domain formula to obtain a single-machine frequency response model of the power system to be detected; A nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model is established by using a preset generalized regression neural network, and based on the nonlinear relationship, a single-machine frequency response parameter of the single-machine frequency response model is determined, and based on the single-machine frequency response parameter, an aggregated power system frequency response model of the power system to be detected is determined; The aggregated power system frequency response model is used to simulate the frequency response process of the power system to be detected in a predetermined disturbance scenario to obtain a simulation result, and based on the simulation result, the frequency security of the power system to be detected is determined.

2. The method according to claim 1, characterized in that The multi-machine frequency response model includes a thermal power frequency response model corresponding to the thermal power generating set, a nuclear power frequency response model corresponding to the nuclear power generating set, a hydropower frequency response model corresponding to the hydropower generating set, a wind-solar-storage frequency response model corresponding to the wind-solar generating set, and an ultra-high voltage frequency response model corresponding to the ultra-high voltage generating set; the frequency time domain curve includes a thermal power frequency time domain curve corresponding to the thermal power frequency response model, a nuclear power frequency time domain curve corresponding to the nuclear power frequency response model, a hydropower frequency time domain curve corresponding to the hydropower frequency response model, a wind-solar frequency time domain curve corresponding to the wind-solar-storage frequency response model, and a high-voltage frequency time domain curve corresponding to the ultra-high voltage frequency response model.

3. The method according to claim 2, characterized in that The performing parameter fitting on each of the frequency-time domain curves to obtain the frequency-time domain formula of the power system to be detected includes: Respectively determine the parameters to be fitted corresponding to the thermal power frequency time domain curve, the nuclear power frequency time domain curve, the hydropower frequency time domain curve, the wind and solar frequency time domain curve, and the high-voltage frequency time domain curve, as well as the parameter value ranges of the parameters to be fitted; Based on the parameters to be fitted and their corresponding parameter value ranges, the nonlinear least squares method is used to perform parameter fitting on the thermal power frequency time domain curve, the nuclear power frequency time domain curve, the hydropower frequency time domain curve, the wind and solar frequency time domain curve, and the high-voltage frequency time domain curve to obtain the frequency time domain formula of the power system to be detected.

4. The method according to claim 1, characterized in that: The Laplace transform is performed on the frequency time domain formula to obtain the single-machine frequency response model of the power system to be detected, including: Determine the Laplace transform formula; Substitute the frequency-time domain formula into the Laplace transform formula to obtain the substituted formula, perform time integration from zero to positive infinity on the substituted formula, and determine the single-machine frequency response model of the power system to be detected based on the integration result.

5. The method according to claim 1, characterized in that The preset generalized regression neural network includes an input layer, a radial base layer and an output layer; The method of establishing a nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model by using a preset generalized regression neural network, and determining a single-machine frequency response parameter of the single-machine frequency response model based on the nonlinear relationship, includes: Determine a multi-machine frequency response parameter vector corresponding to the multi-machine frequency response model, a unit matrix composed of model parameters of the single-machine frequency response model, a multidimensional matrix composed of center points of the input layer radial basis functions, and a comprehensive parameter matrix composed of model parameters of the single-machine frequency response model and model parameters of the multi-machine frequency response model; Based on the multi-dimensional matrix, the Euclidean distance between the multi-machine frequency response parameter vector and the center point of each radial basis function is calculated to obtain the multi-machine frequency spacing, and the unit matrix is ​​feature processed by using the input layer to obtain a single-machine feature vector, and the multi-machine frequency spacing is feature processed by using the input layer to obtain a multi-machine feature vector; Performing Hadamard product processing on the single-machine feature vector and the multi-machine feature vector, and inputting the Hadamard product processing result into the radial base layer, and performing activation processing on the Hadamard product processing result through the activation function in the radial base layer to obtain an activated feature vector; Performing a normalized dot product operation on the activation feature vector and the comprehensive parameter matrix to obtain a dot product feature vector; The point product feature vector is input to the output layer, and linear weighting is performed at the output layer to obtain the single-machine frequency response parameters of the single-machine frequency response model.

6. The method according to claim 1, characterized in that The method of simulating the frequency response process of the power system to be detected in a predetermined disturbance scenario by using the aggregated power system frequency response model to obtain a simulation result includes: Setting a predetermined disturbance scenario based on the actual operating environment, actual operating characteristics, and actual interference conditions of the power system to be detected; Setting model parameters of the aggregated power system frequency response model based on the inertia constant of the generator in the power system to be detected, the gain of the speed regulator, and the frequency regulation effect of the load; The aggregated power system frequency response model after setting parameters is controlled by preset simulation software to simulate the frequency response process of the power system to be detected in the predetermined disturbance scenario to obtain a simulation result.

7. The method according to claim 1, characterized in that Determining the frequency security of the power system to be detected based on the simulation result includes: Based on the simulation result, determining a disturbance frequency value of the power system to be detected after being disturbed by the predetermined disturbance scenario; If the disturbance frequency value is greater than a preset frequency threshold, it is determined that the frequency of the power system to be detected is unsafe; otherwise, it is determined that the frequency of the power system to be detected is safe.

8. A device for determining frequency security of a power system, characterized in that: include: An acquisition unit, used for acquiring a multi-machine frequency response model corresponding to a plurality of generator sets of the power system to be detected; A simulation unit, used for performing frequency simulation on the corresponding generator sets in the power system to be detected by using each of the frequency response models, obtaining the frequency-time domain curves corresponding to each of the generator sets, and performing parameter fitting on each of the frequency-time domain curves to obtain the frequency-time domain formula of the power system to be detected; A transformation unit, used for performing Laplace transformation on the frequency-time domain formula to obtain a single-machine frequency response model of the power system to be detected; A determination unit, configured to establish a nonlinear relationship between the single-machine frequency response model and the multi-machine frequency response model by using a preset generalized regression neural network, and determine a single-machine frequency response parameter of the single-machine frequency response model based on the nonlinear relationship, and determine an aggregated power system frequency response model of the power system to be detected based on the single-machine frequency response parameter; A simulation unit is used to simulate the frequency response process of the power system to be detected in a predetermined disturbance scenario using the aggregated power system frequency response model to obtain a simulation result, and determine the frequency security of the power system to be detected based on the simulation result.

9. The determination device according to claim 8, characterized in that: The multi-machine frequency response model includes a thermal power frequency response model corresponding to the thermal power generating set, a nuclear power frequency response model corresponding to the nuclear power generating set, a hydropower frequency response model corresponding to the hydropower generating set, a wind-solar-storage frequency response model corresponding to the wind-solar generating set, and an ultra-high voltage frequency response model corresponding to the ultra-high voltage generating set; the frequency time domain curve includes a thermal power frequency time domain curve corresponding to the thermal power frequency response model, a nuclear power frequency time domain curve corresponding to the nuclear power frequency response model, a hydropower frequency time domain curve corresponding to the hydropower frequency response model, a wind-solar frequency time domain curve corresponding to the wind-solar-storage frequency response model, and a high-voltage frequency time domain curve corresponding to the ultra-high voltage frequency response model.

10. The determination device according to claim 9, characterized in that: The simulation unit includes a first determination module and a fitting module; The first determination module is used to respectively determine the parameters to be fitted corresponding to the thermal power frequency time domain curve, the nuclear power frequency time domain curve, the hydropower frequency time domain curve, the wind and solar frequency time domain curve, and the high-voltage frequency time domain curve, as well as the parameter value range of the parameters to be fitted; The fitting module is used to perform parameter fitting on the thermal power frequency time domain curve, the nuclear power frequency time domain curve, the hydropower frequency time domain curve, the wind and solar frequency time domain curve, and the high-voltage frequency time domain curve based on the parameters to be fitted and their corresponding parameter value ranges, using the nonlinear least squares method to obtain the frequency time domain formula of the power system to be detected.

11. The determination device according to claim 8, characterized in that: The transformation unit includes a second determination module and an integration module; The second determination module is used to determine the Laplace transform formula; The integration module is used to substitute the frequency-time domain formula into the Laplace transform formula to obtain the substituted formula, perform time integration from zero to positive infinity on the substituted formula, and determine the single-machine frequency response model of the power system to be detected based on the integration result.

12. The determination device according to claim 8, characterized in that: The preset generalized regression neural network includes an input layer, a radial base layer and an output layer; the determination unit includes a third determination module, a calculation module, a processing module, a point multiplication operation module, and a weighting module; The third determination module is used to determine the multi-machine frequency response parameter vector corresponding to the multi-machine frequency response model, the unit matrix composed of the model parameters of the single-machine frequency response model, the multidimensional matrix composed of the center points of the input layer radial basis function, and the comprehensive parameter matrix composed of the model parameters of the single-machine frequency response model and the model parameters of the multi-machine frequency response model; The calculation module is used to calculate the Euclidean distance between the multi-machine frequency response parameter vector and the center point of each radial basis function based on the multi-dimensional matrix to obtain the multi-machine frequency spacing, and use the input layer to perform feature processing on the unit matrix to obtain a single-machine feature vector, and use the input layer to perform feature processing on the multi-machine frequency spacing to obtain a multi-machine feature vector; The processing module is used to perform Hadamard product processing on the single-machine feature vector and the multi-machine feature vector, and input the Hadamard product processing result to the radial base layer, and activate the Hadamard product processing result through the activation function in the radial base layer to obtain an activated feature vector; The dot product operation module is used to perform a normalized dot product operation on the activation feature vector and the comprehensive parameter matrix to obtain a dot product feature vector; The weighting module is used to input the point product feature vector into the output layer, perform linear weighting at the output layer, and obtain the single-machine frequency response parameters of the single-machine frequency response model.

13. The determination device according to claim 8, characterized in that: The simulation unit includes a setting module and a simulation module; The setting module is used to set a predetermined disturbance scenario based on the actual operating environment, actual operating characteristics, and actual interference conditions faced by the power system to be detected; The setting module is used to set the model parameters of the aggregated power system frequency response model based on the inertia constant of the generator, the gain of the speed regulator, and the frequency regulation effect of the load in the power system to be detected; The simulation module is used to use preset simulation software to control the aggregated power system frequency response model with set parameters to simulate the frequency response process of the power system to be detected in the predetermined disturbance scenario to obtain a simulation result.

14. The determination device according to claim 8, characterized in that: The simulation unit further includes a determination module; The determination module is used to determine the disturbance frequency value of the power system to be detected after being disturbed by the predetermined disturbance scenario based on the simulation result; The determination module is used to determine that the frequency of the power system to be detected is unsafe if the disturbance frequency value is greater than a preset frequency threshold, and otherwise, determine that the frequency of the power system to be detected is safe.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

16. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images