Fault detection method for flexible direct-current power transmission system

Abstract

The invention provides a fault detection method for a flexible DC power transmission system. The method comprises the following steps: acquiring an original current signal; setting a search range for the parameters of the variational mode decomposition so as to obtain original set parameters; decomposing the original current signal into modal functions with different center frequencies; selecting a value corresponding to the minimum relative entropy as an optimal parameter; carrying out final variational mode decomposition on the original current signal by adopting the optimal parameter to obtain orderly arranged final mode functions, selecting a second final mode function, and carrying out Hilbert-Huang transform on the second final mode function to obtain a Hilbert spectrum of the second final mode function; and performing synchronous compression transformation processing on the Hilbert spectrum to obtain the instantaneous energy density of the Hilbert spectrum, and comparing the amplitude of the instantaneous energy density with a steady-state threshold to realize fault detection. According to the invention, accurate and rapid fault detection of the flexible DC power transmission system can be realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of circuit system detection, and in particular to a flexible direct current power transmission system fault detection method. BACKGROUND

[0002] The flexible direct current power transmission system has the advantages of long transmission distance, large energy transmission, low line loss and flexible control, and is of great significance for large-scale renewable energy consumption and rational allocation of resources, so it has been widely concerned by researchers in recent years.

[0003] However, the flexible direct current power transmission system is prone to failure during operation, which will generate a large fault transient current. The fault transient current will reach a peak value within a few milliseconds, causing a huge impact on the system equipment in a short time, which seriously limits its development.

[0004] Due to the long distance of the flexible direct current power transmission line, the complex working terrain environment and the high short-circuit fault rate, accurate and rapid fault detection is of great significance to improve the reliability of the direct current system. SUMMARY

[0005] In view of this, the present application provides a flexible direct current power transmission system fault detection method to realize accurate and rapid fault detection of the flexible direct current power transmission system.

[0006] A flexible direct current power transmission system fault detection method comprises: Step S1, real-time sampling of the direct current line by a current sensor to obtain an original current signal, and inputting the collected original current signal to a window; Step S2, setting a search range for the parameters of variational mode decomposition to obtain original setting parameters, the parameters including mode number And penalty factor ; Step S3, applying variational mode decomposition to the original current signal using the original setting parameters to decompose the original current signal into Mode functions with different center frequencies; Step S4, adding the Mode functions with different center frequencies to obtain a reconstructed signal, calculating the relative entropy of the original current signal and the reconstructed signal based on kernel density estimation, searching every pair of Value in the search range, and selecting the Value corresponding to the minimum relative entropy as the optimal parameter; Step S5, performing final variational mode decomposition on the original current signal using the optimal parameter to obtain final mode functions arranged in order, selecting the second final mode function, performing Hilbert-Huang transform on the second final mode function, and obtaining the Hilbert spectrum of the second final mode function; In step S6, synchronous compression transform is performed on the Hilbert spectrum of the second final modal function to remove the fuzzy energy caused by noise, harmonics, or modulation sidebands, thereby obtaining the instantaneous energy density of the Hilbert spectrum. The amplitude of the instantaneous energy density is compared with the steady-state threshold to achieve fault detection.

[0007] The method for detecting faults in a flexible direct current transmission system provided by the present invention has the following beneficial effects: (1) In a flexible DC transmission system, the DC fault detection time should be limited to 2ms in order to reliably protect the entire system and converter components. However, the training process of the artificial intelligence algorithm is time-consuming and lengthy, and model reasoning may also involve complex calculations, which is an unbearable burden for the fast-paced fault detection requirements of the flexible DC system that emphasizes millisecond-level response. The present invention only requires fast variational mode decomposition, with low computational complexity, and can complete the modal data in real time or quasi-real time. The setting meets the speed requirement of fault detection.

[0008] (2) The present invention can automatically adapt the most suitable decomposition parameters for different lines, different operating states, and different fault types, and has greater versatility. It ensures that the difference between the reconstructed signal and the original signal is minimized, ensuring that the decomposed modal function can truly and completely retain all the characteristics of the original signal, avoiding information loss or distortion. In addition, the second final modal function obtained by applying variational modal decomposition to the original current signal is used for fault detection. Since the second final modal function oscillates at the frequency of the important feature, it can reduce the noise interference of the fault detection and improve the accuracy of fault detection.

[0009] (3) The present invention introduces synchronous compression transformation to recompress the diffused energy and concentrate it onto the real instantaneous frequency ridge, making the clear, short-lived impact energy band generated by the fault stand out, significantly improving the ability to identify weak fault characteristics in complex noise environments. In addition, the present invention uses the amplitude of the instantaneous energy density as the judgment basis for fault detection, emphasizing the fault characteristics within the selected frequency range, and can effectively suppress the influence of the steady-state component. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic diagram of a flow chart of a method for detecting faults in a flexible direct current transmission system according to an embodiment of the present invention; Figure 2 is an exemplary waveform of the original current signal; Figure 3a response diagram of the instantaneous energy density of an exemplary faulty line; Figure 4 a response diagram of the instantaneous energy density of an exemplary non-faulty line; Figure 5 a response diagram of the instantaneous energy density of the second final modal function after adding Gaussian white noise with a signal-to-noise ratio of 25 dB; Figure 6 a response diagram of the instantaneous energy density of the second final modal function after adding Gaussian white noise with a signal-to-noise ratio of 55 dB. DETAILED DESCRIPTION

[0011] Embodiments of the present application are described in detail below with reference to examples illustrated in the accompanying drawings, in which the same or similar components or components having the same or similar functions are denoted by the same or similar reference numerals throughout. The embodiments described below by reference to the accompanying drawings are exemplary and are intended to explain the embodiments of the present application, and cannot be understood as limiting the present application.

[0012] Referring to Figure 1 , the embodiments of the present application provide a flexible DC power transmission system fault detection method, comprising steps S1 to S6: Step S1, real-time sampling of the DC line by a current sensor to obtain an original current signal, and inputting the collected original current signal to a window.

[0013] Specifically, an exemplary waveform of the original current signal is as shown in Figure 2 In this embodiment, step S1 satisfies the following formula:

[0014] wherein, represents a one-dimensional discrete time domain signal sequence with a length of , represents the 1st sampling point of the sampled current discrete time domain signal, represents the 2nd sampling point of the sampled current discrete time domain signal, represents the 1st sampling point of the sampled current discrete time domain signal.

[0015] Step S2, setting a search range for the parameters of the variational modal decomposition, thereby obtaining original setting parameters, the parameters including the modal number and the penalty factor .

[0016] For example, according to prior knowledge, the search range of the modal number is 2 to 10, and the search range of the penalty factor is 1000 to 5000.​​

[0017] Step S3, applying the variational mode decomposition to the original current signal with the original setting parameters, so as to decompose the original current signal into modal functions with different central frequencies.

[0018] Wherein, the step S3 specifically comprises: Step S3.1, obtaining a one-dimensional discrete time domain signal sequence with a length of based on the original current signal, performing the variational mode decomposition to the one-dimensional discrete time domain signal sequence with a length of and introducing a modal bandwidth weight factor, so as to obtain a constrained variational problem; In the step S3.1, the expression of the constrained variational problem is :

[0019] Wherein, represents taking the minimum value, is the i-th modal function to be decomposed out, is the central frequency of the i-th modal function, represents the modal bandwidth weight factor, represents the partial derivative of time, represents the Dirac function, represents the imaginary unit, represents the convolution operation, represents the square of L2 norm. Step S3.2, obtaining an extended Lagrange expression by introducing a Lagrange penalty operator, and converting the constrained variational problem into an unconstrained variational problem based on the extended Lagrange expression; In the step S3.2, the extended Lagrange expression is:

[0020] Wherein, is the extended Lagrange expression, is the Lagrange penalty operator,

[0021] is the original current signal, represents the inner product. Step S3.3, performing a loop iteration calculation on the unconstrained variational problem to obtain an optimal solution, updating the modal function based on the optimal solution, and outputting modal functions with different central frequencies.

[0022]

[0023] ​​​In step S3.3, when updating the modal function, a relaxation factor is introduced to predict the trend of the change of the Lagrange multiplier by extrapolation method, so as to accelerate the update of the Lagrange multiplier, and the expression is as follows:

[0024] wherein, is the updated Lagrange multiplier, is the updated Lagrange multiplier, is the relaxation factor, denotes the noise tolerance parameter, is the Fourier transform of the original current signal, denotes the Fourier transform of the updated modal function.

[0025] Step S4, the modal functions of different center frequencies are added to obtain a reconstructed signal, the relative entropy of the original current signal and the reconstructed signal is calculated based on the kernel density estimation, each pair of values in the search range is traversed, and the value corresponding to the minimum relative entropy is selected as the optimal parameter.

[0026] wherein, the reconstructed signal satisfies the following formula:

[0027] wherein, , , , are the first modal function, the second modal function, the th modal function, and the th modal function, respectively.

[0028] The formula for calculating the relative entropy is as follows:

[0029] wherein, is the relative entropy of the original current signal and the reconstructed signal, denotes the kernel density estimation of the original current signal, denotes the kernel density estimation of the reconstructed signal, denotes the difference of viewed by the distribution of , denotes the difference of viewed by the distribution of .

[0030] Step S5, the optimal parameter is used to perform the final variational modal decomposition on the original current signal, and the​​ The second final modal function is selected from the ordered arranged final modal functions, and a Hilbert-Huang transform is performed on the second final modal function to obtain a Hilbert spectrum of the second final modal function.

[0031] In this embodiment, the second final modal function is selected because the second final modal function oscillates at the frequency of the important feature, can reduce noise interference on fault detection, and thus improves the fault detection accuracy.

[0032] In step S6, a synchronous compression transform is performed on the Hilbert spectrum of the second final modal function to remove blurred energy caused by noise or harmonics or modulation sidebands, so as to obtain an instantaneous energy density of the Hilbert spectrum, and the amplitude of the instantaneous energy density is compared with a steady-state threshold, so as to realize fault detection.

[0033] In step S6, a synchronous compression transform is performed on the Hilbert spectrum of the second final modal function to remove blurred energy caused by noise or harmonics or modulation sidebands, so as to obtain an instantaneous energy density of the Hilbert spectrum, and the amplitude of the instantaneous energy density is compared with a steady-state threshold, so as to realize fault detection. In step S6.1, a synchronous compression transform is performed on the Hilbert spectrum of the second final modal function, a Dirac function is used as a selector to compress the Hilbert spectrum into a time-frequency energy ridge line to obtain a first reconstructed spectrum. In step S6.1, the following formula is satisfied:

[0034]

[0035] wherein, is the first reconstructed spectrum, is a normalization constant, is the Hilbert spectrum of the second final modal function, is a preset center frequency range, is a differential of the compressed instantaneous angular frequency is an instantaneous amplitude, is an instantaneous angular frequency, is a differential of time

[0036] In step S6.2, blurred energy caused by noise or harmonics or modulation sidebands is removed from the first reconstructed spectrum to obtain a second reconstructed spectrum. In step S6.2, the following formula is satisfied:

[0037] wherein, is the second reconstructed spectrum, is an instantaneous frequency function, represents an integral bandwidth.

[0038] ​​Step S6.3, according to the secondary reconstructed spectrum, the instantaneous energy density is calculated, the amplitude of the instantaneous energy density is compared with the steady-state threshold, if the amplitude of the instantaneous energy density exceeds the steady-state threshold, it is determined that the fault working condition occurs.

[0039] Specifically, the instantaneous energy density response results of the example fault line and the instantaneous energy density response results of the non-fault line are shown in Figure 3 and Figure 4 As can be seen from Figure 3 and Figure 4 , the instantaneous energy density can effectively detect whether the fault working condition occurs.

[0040] In addition, since the background noise will affect the spectrum distribution, the existing fault detection method based on frequency domain is easy to be disturbed by noise. The present application adds Gaussian white noise with different signal-to-noise ratios to the direct current line current signal when the load changes, and evaluates the robustness of the proposed method to noise. After adding Gaussian white noise with a signal-to-noise ratio of 25dB, the instantaneous energy density of the second final modal function is shown in Figure 5 , after adding Gaussian white noise with a signal-to-noise ratio of 55dB, the instantaneous energy density of the second final modal function is shown in Figure 6 , from Figure 5 and Figure 6 , it can be seen that the instantaneous energy density response of the method proposed in the present application remains below the steady-state threshold, is not disturbed by noise, and has good robustness.

[0041] In summary, the flexible DC power transmission system fault detection method according to the above embodiment has the following beneficial effects: (1) In the flexible DC power transmission system, the DC fault detection time should be limited within 2ms in order to reliably protect the entire system and converter components, while the training process of artificial intelligence algorithm is time-consuming and lengthy, and model reasoning may involve complex calculations, which is a heavy burden for the speed requirement of flexible DC system fault detection which emphasizes millisecond-level response. The present application only needs to perform fast variational modal decomposition, the calculation complexity is low, and the number of modes can be set in real time or quasi-real time, meeting the speed requirement of fault detection.

[0042] (2) The application can automatically adapt the most suitable decomposition parameters for different lines, different operating states and different fault types, is more versatile, ensures that the reconstructed signal and the original signal have the smallest difference as the criterion, ensures that the decomposed modal function can truly and completely retain all the characteristics of the original signal, avoids information loss or distortion. In addition, the second final modal function obtained by applying the original current signal to the variational mode decomposition is used for fault detection, since the second final modal function oscillates at the frequency of important features, thus can reduce the noise interference received by the fault detection, and improve the fault detection accuracy.

[0043] (3) The application introduces a synchronous compression transformation, re-compresses and collects the diffused energy on the real instantaneous frequency ridge line, makes the clear and short impact energy band generated by the fault highlighted, greatly improves the ability to identify weak fault characteristics in a complex noise environment. And the application takes the amplitude of the instantaneous energy density as the basis for fault detection, emphasizes the fault characteristics in the selected frequency range, and can effectively suppress the influence of the steady-state component.

[0044] The above-mentioned embodiments only express several embodiments of the application, and the description is more specific and detailed, but it cannot be understood as a limitation on the scope of the patent. It should be noted that for ordinary skilled persons in the art, without departing from the concept of the application, a number of modifications and improvements can be made, which are all within the protection scope of the application. Therefore, the protection scope of the patent of the application should be subject to the appended claims.

Claims

1. A method for detecting faults in a flexible direct current transmission system, characterized in that: include: Step S1, sampling the DC line in real time through a current sensor to obtain an original current signal, and inputting the collected original current signal into a window; Step S2, setting a search range for the parameters of variational modal decomposition, thereby obtaining the original setting parameters, which include the modal number and penalty factor ; Step S3, applying variational mode decomposition to the original current signal using the original set parameters, thereby decomposing the original current signal into Mode functions with different center frequencies; Step S4: The reconstructed signal is obtained by adding the modal functions of different center frequencies, and the relative entropy of the original current signal and the reconstructed signal is calculated based on the kernel density estimation. Each pair of value, select the value corresponding to the minimum relative entropy value as the optimal parameter; Step S5, use the optimal parameters to perform final variational modal decomposition on the original current signal to obtain final mode functions arranged in order, selecting the second final mode function, performing Hilbert-Huang transform on the second final mode function, and obtaining the Hilbert spectrum of the second final mode function; In step S6, synchronous compression transform is performed on the Hilbert spectrum of the second final modal function to remove the fuzzy energy caused by noise, harmonics, or modulation sidebands, thereby obtaining the instantaneous energy density of the Hilbert spectrum. The amplitude of the instantaneous energy density is compared with the steady-state threshold to achieve fault detection.

2. The method for detecting faults in a flexible DC transmission system according to claim 1, wherein: Step S3 specifically includes: Step S3.1, based on the original current signal, obtain the length The one-dimensional discrete time domain signal sequence of length is The one-dimensional discrete time domain signal sequence is subjected to variational mode decomposition and the modal bandwidth weight factor is introduced to obtain the constrained variational problem. Step S3.2, obtaining an extended Lagrangian expression by introducing a Lagrangian penalty operator, and converting the constrained variational problem into an unconstrained variational problem based on the extended Lagrangian expression; Step S3.3, perform cyclic iterative calculation on the unconstrained variational problem to obtain the optimal solution, and update the modal function based on the optimal solution to output Mode functions with different center frequencies.

3. The method for detecting faults in a flexible DC transmission system according to claim 2, wherein: In step S3.1, the expression of the constrained variational problem is for: in, Indicates taking the minimum value, is the first A modal function, It is The center frequency of the mode function, represents the modal bandwidth weighting factor, Indicates time The partial derivative of represents the Dirac function, represents the imaginary unit, represents the convolution operation, Represents the square of the L2 norm.

4. The method for detecting faults in a flexible DC transmission system according to claim 3, wherein: In step S3.2, the extended Lagrangian expression is: in, is the extended Lagrangian expression, is the Lagrangian penalty operator, is the original current signal, represents the inner product.

5. The method for detecting faults in a flexible DC transmission system according to claim 4, wherein: In step S3.3, when updating the modal function, a relaxation factor is introduced to predict the change trend of the Lagrangian penalty operator through extrapolation, thereby accelerating the update of the Lagrangian penalty operator. The expression is: in, is the updated Lagrangian penalty operator, is the Lagrangian penalty operator before update, is the relaxation factor, represents the noise margin parameter, is the Fourier transform of the original current signal, Indicates the updated Fourier transform of the modal function.

6. The method for detecting faults in a flexible DC transmission system according to claim 5, wherein: In step S4, the calculation formula of relative entropy is: in, is the relative entropy of the original current signal and the reconstructed signal, represents the kernel density estimate of the original current signal, represents the kernel density estimate of the reconstructed signal, Indicates The distribution of The difference, Indicates The distribution of difference.

7. The method for detecting faults in a flexible HVDC system according to claim 6, wherein: Step S6 specifically includes: Step S6.1, performing synchronous compression transform on the Hilbert spectrum of the second final mode function, using the Dirac function as a selector to compress the Hilbert spectrum into a time-frequency energy ridge to obtain a reconstructed spectrum; Step S6.2, removing the blurred energy caused by noise, harmonics, or modulation sidebands in the primary reconstructed spectrum to obtain a secondary reconstructed spectrum; Step S6.3, calculating the instantaneous energy density based on the secondary reconstructed spectrum, comparing the amplitude of the instantaneous energy density with the steady-state threshold, and determining a fault condition if the amplitude of the instantaneous energy density exceeds the steady-state threshold.

8. The method for detecting faults in a flexible HVDC system according to claim 7, wherein: Step S6.1 satisfies the following equation: in, To reconstruct the spectrum once, is the normalization constant, is the Hilbert spectrum of the second final mode function, To preset the center frequency range, is the instantaneous angular frequency after compression The differential of is the instantaneous amplitude, is the instantaneous angular frequency, Indicates time The differential of .

9. The method for detecting faults in a flexible HVDC system according to claim 8, wherein: Step S6.2 satisfies the following equation: in, For the secondary reconstruction spectrum, is the instantaneous frequency function, represents the integration bandwidth.

Images