A power grid fault analysis method based on high proportion of new energy access

Abstract

The application discloses a power grid fault analysis method based on high-proportion new energy access, which comprises the following analysis steps: S1, collecting various voltage and current data of a grid-connected gateway; S2, constructing an electromagnetic transient equivalent model of the grid-connected gateway; S3, performing precision evaluation and optimization on the electromagnetic transient equivalent model of the grid-connected gateway until the model precision reaches the standard; S4, based on the electromagnetic transient equivalent model of the grid-connected gateway, acquiring power grid operation data at the grid-connected gateway to construct a power grid historical operation data set; S5, constructing a power grid fault analysis model based on a convolutional neural network; S6, taking real-time operation data at the grid-connected gateway as input of the power grid fault analysis model, taking transient overvoltage at the grid-connected optical port as output of the model, and dividing the power grid risk level according to the transient overvoltage value, so that real-time analysis and monitoring of the power grid fault are realized; and the application can improve the accuracy and reliability of fault analysis.

Description

Technical Field

[0001] This invention belongs to the field of power grid fault analysis technology, specifically relating to a power grid fault analysis method based on a high proportion of new energy access. Background Technology

[0002] With the acceleration of global energy transition and the in-depth advancement of the "dual carbon" goal, the proportion of new energy sources, represented by wind and solar power, in the power system continues to increase rapidly. The scale of grid-connected operation of new energy units is expanding daily. The output of new energy units is highly random and volatile. The transient processes of power systems with a high proportion of new energy and a high proportion of power electronic equipment are becoming increasingly complex, leading to a corresponding increase in the amount of analysis and computation. The calculations are complex and slow. Key characteristics are contained in the massive amount of power system measurement data, making it difficult to extract them accurately. This makes it difficult for power electronic equipment to adjust its operating mode in real time according to the grid operating status under fault conditions, increasing the risk of system voltage instability and seriously threatening the safe and stable operation of power systems with a high proportion of new energy.

[0003] Therefore, in order to solve the above problems, it is necessary to develop a grid fault analysis method based on a high proportion of new energy access. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a grid fault analysis method based on a high proportion of new energy access, construct a high-precision simplified model, and improve the accuracy and reliability of fault analysis.

[0005] The objective of this invention is achieved as follows: a grid fault analysis method based on a high proportion of renewable energy access, comprising the following analysis steps:

[0006] S1. Collect various voltage and current data at the grid connection point by using a high-frequency sampling device installed at the grid connection point of new energy source;

[0007] S2. Construct an equivalent electromagnetic transient model of the parallel gateway interface. By adopting the time-domain equivalent method, the transient response characteristics at the parallel gateway interface are preserved, and the internal system is equivalent to a lumped parameter network of resistors R, inductors L, and capacitors C and a controlled source.

[0008] S3. Evaluate and optimize the accuracy of the constructed parallel gateway electromagnetic transient equivalent model until the model accuracy meets the standard.

[0009] S4. Based on the constructed equivalent electromagnetic transient model of the power grid connection interface, obtain the operating data at the power grid connection interface and construct a historical operating dataset of the power grid.

[0010] S5. Construct a power grid fault analysis model based on a convolutional neural network, and train the power grid fault analysis model using data from the historical operation dataset of the power grid as training data;

[0011] S6. The real-time operating data at the junction gate is used as the input of the power grid fault analysis model, and the transient overvoltage at the junction gate is used as the model output. The power grid risk level is divided according to the transient overvoltage value to realize real-time analysis and monitoring of power grid faults.

[0012] Furthermore, in step S1, the sampling rate of the high-frequency sampling device is ≥50Hz.

[0013] Furthermore, the topology of the equal value model in step S2 adopts an extended [structure / method]. Type equivalent circuit, including resistors in series. ,inductance and parallel capacitors Simultaneously, a controlled current source is introduced. Reflecting the dynamic characteristics of the internal system, the output of the controlled current source is related to the parallel gateway voltage and its derivative, expressed as: In the formula, , For proportional and differential coefficients, This is the impulse response function, used to reflect the memory characteristics of the system. For integration variables, This refers to the voltage at the parallel gateway interface.

[0014] Furthermore, in step S2, parameter identification of the equivalent model involves applying a perturbation signal to the model, recording the voltage and current responses at the cutoff points during the perturbation, and then using the least squares method for parameter identification. The objective function is expressed as: In the formula, For the measured current, Calculate the current for the equivalent model. The parameter vector to be identified , This represents the number of sampling points.

[0015] Furthermore, the evaluation metrics for evaluating the model in step S3 include peak error, harmonic distortion rate error, and energy error; wherein the peak error is less than 5%, the harmonic distortion rate error is less than 10%, and the energy error is less than 3%.

[0016] Furthermore, the optimization method for insufficient model accuracy in step S3 includes: ① If the error is large in the high-frequency band, increase the order of the equivalent model and introduce more RC branches; ② If the error exceeds the standard under nonlinear conditions, refine the piecewise linearization interval and increase the number of condition identification points.

[0017] Furthermore, step S5, which uses a convolutional neural network to construct a power grid fault analysis model, specifically includes an input layer, a convolutional layer, an activation function layer, a pooling layer, a fully connected layer, and an output layer.

[0018] Furthermore, in step S5, electrical quantity data at the parallel gateway interface at the moment the fault occurs and before and after the fault ends are used to construct the input feature, and the input feature X is represented as: In the formula, , , , For steady state, it is a parallel gateway interface. The short-circuit ratio at the point, These represent the instants before the fault begins, after the fault begins, before the fault ends, and after the fault ends, respectively. They represent the first At the entrance of the new energy connection The voltage amplitude, voltage phase angle, active power, and reactive power at each moment. This represents the total number of parallel gateway ports.

[0019] Furthermore, in step S6, the power grid risk level is divided into four levels based on the transient overvoltage value: In the formula, Indicates the wind direction level of the power grid. Transient overvoltage values ​​of Best This indicates that there is no risk. This indicates a low risk. The time indicates a moderate risk. This indicates a higher risk.

[0020] Due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0021] (1) By using the equivalent algorithm, the entire new energy unit is equivalent to one or more equivalent elements at the gate, and the gate electromagnetic transient equivalent model is constructed and connected to simplify the complexity when multiple new energy units are connected to the grid at the same time, reduce the amount of calculation, improve the analysis speed and efficiency, and at the same time ensure the accuracy of the model, so as to accurately reflect the overall impact of new energy on the grid.

[0022] (2) A power grid fault analysis model is constructed by using a convolutional neural network. The real-time operating data at the junction gate is used as the input of the power grid fault analysis model, and the transient overvoltage at the junction gate is used as the output of the model. Key information is effectively extracted from complex power grid data, and the power grid risk level is divided according to the transient overvoltage value, so as to realize the real-time analysis and monitoring of power grid faults. Attached Figure Description

[0023] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0024] The technical solution of the present invention will be further described in detail below through embodiments and in conjunction with the accompanying drawings.

[0025] like Figure 1 As shown, a grid fault analysis method based on a high proportion of renewable energy access includes the following analysis steps:

[0026] S1. Collect various voltage and current data from the grid connection outlet by using a high-frequency sampling device installed at the grid connection outlet of the new energy source; preferably, the sampling rate of the high-frequency sampling device in step S1 is ≥50Hz.

[0027] S2. Construct an equivalent electromagnetic transient model of the parallel gateway interface. By adopting the time-domain equivalent method, the transient response characteristics at the parallel gateway interface are preserved, and the internal system is equivalent to a network of lumped parameters of resistor R, inductor L, capacitor C and controlled source.

[0028] Preferably, the topology of the equal value model in step S2 adopts an extended [structure / structure]. Type equivalent circuit, including resistors in series. ,inductance and parallel capacitors Simultaneously, a controlled current source is introduced. Reflecting the dynamic characteristics of the internal system, the output of the controlled current source is related to the parallel gateway voltage and its derivative, expressed as: In the formula, , For proportional and differential coefficients, This is the impulse response function, used to reflect the memory characteristics of the system. For integration variables, This refers to the voltage at the parallel gateway interface.

[0029] Preferably, in step S2, parameter identification of the equivalent model is performed by applying a perturbation signal to the model, recording the voltage and current responses at the cutoff points during the perturbation, and then using the least squares method for parameter identification. The objective function is expressed as: In the formula, For the measured current, Calculate the current for the equivalent model. The parameter vector to be identified , This represents the number of sampling points.

[0030] S3. Evaluate and optimize the accuracy of the constructed parallel gateway electromagnetic transient equivalent model until the model accuracy meets the standard.

[0031] Preferably, the evaluation metrics for evaluating the model in step S3 include peak error, harmonic distortion rate error, and energy error; wherein the peak error is less than 5%, the harmonic distortion rate error is less than 10%, and the energy error is less than 3%.

[0032] Specifically, the peak error In the formula, , The peak voltage values ​​are those of the equivalent model and the detailed model, respectively; the harmonic distortion rate error is... In the formula, , The total harmonic distortion rates of the equivalent model and the detailed model are respectively; the energy error , Sampling time, , and , Equivalent model and detailed model, respectively. Voltage and current at any given moment.

[0033] Preferably, the optimization method for insufficient model accuracy in step S3 includes: ① if the error is large in the high-frequency band, increase the order of the equivalent model and introduce more RC branches; ② if the error exceeds the standard under nonlinear conditions, refine the piecewise linearization interval and increase the number of condition identification points.

[0034] S4. Based on the constructed electromagnetic transient equivalent model of the grid connection interface, obtain the operating data at the grid connection interface and construct the historical operating dataset of the power grid.

[0035] S5. Construct a power grid fault analysis model based on a convolutional neural network, and use data from the historical power grid operation dataset as training data to train the power grid fault analysis model.

[0036] Preferably, step S5, which uses a convolutional neural network to construct a power grid fault analysis model, specifically includes an input layer, a convolutional layer, an activation function layer, a pooling layer, a fully connected layer, and an output layer. The convolutional layer is used to perform convolution operations between the input data from the upper layer and the convolutional kernel, and uses the resulting output value as the input to the lower layer to achieve feature extraction. The output of the convolutional layer is represented as follows: In the formula, for The first order of the output matrix line, number Column elements, For The first convolution kernel line, number Column elements, For the input matrix The Middle line, number The elements of the column; the activation function layer uses the ReLU activation function: The pooling layer is used to perform feature selection and information filtering on the output of the convolutional layer, retaining key information. Its max pooling calculation formula is expressed as: The average pooling calculation formula is expressed as: In the formula, For pooling regions, The area of ​​the pooling region; the fully connected layer is used to convert the two-dimensional output data into one-dimensional data, thereby obtaining the mapping relationship between key features and output quantities, expressed as: In the formula, for Dimensional input variables, For connection weights, For bias, This is the output.

[0037] Preferably, in step S5, electrical quantity data at the parallel gateway interface at the moment the fault occurs and before and after the fault ends are used to construct the input feature, and the input feature X is represented as: In the formula, , , , For steady state, it is a parallel gateway interface. The short-circuit ratio at the point, These represent the instants before the fault begins, after the fault begins, before the fault ends, and after the fault ends, respectively. They represent the first At the entrance of the new energy connection The voltage amplitude, voltage phase angle, active power, and reactive power at each moment. This represents the total number of parallel gateway ports.

[0038] S6. The real-time operating data at the junction gate is used as the input of the power grid fault analysis model, and the transient overvoltage at the junction gate is used as the model output. The power grid risk level is divided according to the transient overvoltage value to realize real-time analysis and monitoring of power grid faults.

[0039] Preferably, in step S6, the power grid risk level is divided into four levels based on the transient overvoltage value: In the formula, Indicates the wind direction level of the power grid. Transient overvoltage values ​​of Best This indicates that there is no risk. This indicates a low risk. The time indicates a moderate risk. This indicates a higher risk.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of the claims of the present invention.

Claims

1. A grid fault analysis method based on a high proportion of renewable energy access, characterized in that: The analysis includes the following steps: S1. The voltage and current data at the grid connection point are collected by a high-frequency sampling device installed at the grid connection point of the new energy source. S2. Construct an equivalent electromagnetic transient model of the parallel gateway interface. By adopting the time-domain equivalent method, the transient response characteristics at the parallel gateway interface are preserved, and the internal system is equivalent to a network of lumped parameters of resistor R, inductor L, capacitor C and controlled source. S3. Evaluate and optimize the accuracy of the constructed parallel gateway electromagnetic transient equivalent model until the model accuracy meets the standard. S4. Based on the constructed equivalent electromagnetic transient model of the power grid connection interface, obtain the operating data at the power grid connection interface and construct a historical operating dataset of the power grid. S5. Construct a power grid fault analysis model based on a convolutional neural network, and train the power grid fault analysis model using data from the historical operation dataset of the power grid as training data; S6. The real-time operating data at the junction gate is used as the input of the power grid fault analysis model, and the transient overvoltage at the junction gate is used as the model output. The power grid risk level is divided according to the transient overvoltage value to realize real-time analysis and monitoring of power grid faults.

2. The grid fault analysis method based on a high proportion of new energy access according to claim 1, characterized in that: In step S1, the sampling rate of the high-frequency sampling device is ≥50Hz.

3. The grid fault analysis method based on a high proportion of new energy access according to claim 1, characterized in that: The topology of the equal value model in step S2 is extended. Type equivalent circuit, including resistors in series. ,inductance and parallel capacitors Simultaneously, a controlled current source is introduced. Reflecting the dynamic characteristics of the internal system, the output of the controlled current source is related to the parallel gateway voltage and its derivative, expressed as: In the formula, , For proportional and differential coefficients, This is the impulse response function, used to reflect the memory characteristics of the system. For integration variables, This refers to the voltage at the parallel gateway interface.

4. The grid fault analysis method based on a high proportion of new energy access according to claim 1, characterized in that: Step S2, in the parameter identification of the equivalent model, involves applying a perturbation signal to the model, recording the voltage and current responses at the threshold during the perturbation, and then using the least squares method for parameter identification. The objective function is expressed as: In the formula, For the measured current, Calculate the current for the equivalent model. The parameter vector to be identified , This represents the number of sampling points.

5. The grid fault analysis method based on a high proportion of new energy access according to claim 1, characterized in that: The evaluation metrics for evaluating the model in step S3 include peak error, harmonic distortion rate error, and energy error; wherein the peak error is less than 5%, the harmonic distortion rate error is less than 10%, and the energy error is less than 3%.

6. The grid fault analysis method based on high proportion of new energy access according to claim 1, characterized in that: The optimization methods for insufficient model accuracy in step S3 include: ① If the error is large in the high-frequency band, increase the order of the equivalent model and introduce more RC branches; ② If the error exceeds the standard under nonlinear conditions, refine the piecewise linearization interval and increase the number of condition identification points.

7. The grid fault analysis method based on a high proportion of new energy access according to claim 1, characterized in that: Step S5 uses a convolutional neural network to construct a power grid fault analysis model, specifically including an input layer, a convolutional layer, an activation function layer, a pooling layer, a fully connected layer, and an output layer.

8. The grid fault analysis method based on high proportion of new energy access according to claim 1, characterized in that: In step S5, electrical quantity data at the parallel gateway interface at the moment of fault occurrence and before and after the moment of fault termination are specifically used to construct the input feature, and the input feature X is represented as follows: In the formula, , , , For steady state, it is a parallel gateway interface. The short-circuit ratio at the point, These represent the instants before the fault begins, after the fault begins, before the fault ends, and after the fault ends, respectively. They represent the first At the entrance of the new energy connection The voltage amplitude, voltage phase angle, active power, and reactive power at each moment. This represents the total number of parallel gateway ports.

9. The grid fault analysis method based on high proportion of new energy access according to claim 1, characterized in that: In step S6, the power grid risk level is divided into four levels based on the transient overvoltage value: In the formula, Indicates the wind direction level of the power grid. Transient overvoltage values ​​of Best-Speed This indicates that there is no risk. This indicates a low risk. The time indicates a moderate risk. This indicates a higher risk.

Images