A high-precision ranging method and system of time-refined direct current counter-wave

Abstract

This invention discloses a high-precision ranging method and system for time-refined DC back-traveling waves, belonging to the field of power system fault location technology. The method includes: acquiring signals and performing mode transformation to extract first-mode voltage and current signals; establishing a time-domain expression for the back-traveling wave; performing wavelet denoising on the signals to extract the first-mode DC voltage signal; reconstructing the time scale in the time-refined model; performing variational mode decomposition and Hilbert transform to obtain instantaneous amplitude and instantaneous energy sequences, constructing an energy kurtosis index, detecting sudden increases in energy kurtosis, and identifying the arrival time of the first DC back-traveling wavefront generated by the fault; calculating the wavefront arrival time difference based on the arrival times of the first DC back-traveling wavefront measured at both ends of the DC line, and determining the location of the fault point by combining the DC line length and traveling wave propagation characteristics. This invention has low dependence on traveling wave propagation parameters, provides reliable ranging results, and is applicable to deep-sea offshore wind power flexible DC systems.

Description

Technical Field

[0001] This invention belongs to the field of power system fault location technology, specifically relating to a high-precision fault location method and system for time-refined DC back-traveling waves. Background Technology

[0002] Currently, with the rapid development of offshore wind power, grid connection methods based on flexible DC transmission (Voltage Source Converter-High Voltage Direct Current, VSC-HVDC) are widely used due to their suitability for long-distance, high-capacity power transmission. However, offshore wind power VSC-HVDC grid connection systems face challenges such as long line lengths and complex operating environments. If a fault occurs in the DC line, the inability to quickly and accurately locate the fault could significantly impact system safety and power supply reliability. Therefore, higher requirements are placed on the accuracy and real-time performance of DC line fault location.

[0003] Among existing methods for fault location in DC lines, traveling wave (TW) ranging has been widely studied due to its independence from line parameters and high positioning speed. This type of method typically detects the traveling wave signal generated by the fault and calculates the fault distance using the arrival time of the traveling wave front at the measurement end. Especially in dual-end ranging scenarios, only the first arriving TW front needs to be identified to complete the ranging calculation, showing promising engineering application prospects. However, TW signals are characterized by strong transients, wide bandwidth, and abrupt changes. Their wavefront duration is extremely short, and the accuracy of TW ranging largely depends on the sampling frequency of the measurement system. In practical engineering, limitations in measurement device cost, communication bandwidth, and real-time requirements often prevent the sampling frequency from reaching ideal levels (typically 10-20 kHz). This results in the wavefront arrival time being located only at a limited number of discrete sampling points, introducing significant time quantization errors, which are further amplified into significant ranging deviations in long-distance DC lines. Meanwhile, in the actual operating environment of deep-sea wind power flexible DC grid-connected systems, measurement signals are susceptible to electromagnetic interference, converter station control actions, and random noise. Traveling wavefront characteristics are easily distorted or even submerged in strong noise, significantly reducing the reliability of wavefront identification. The combined effect of insufficient sampling frequency and noise interference becomes a key factor restricting the accuracy and stability of existing traveling wave ranging methods.

[0004] As an improvement, some studies have attempted to preprocess traveling wave signals using interpolation, filtering, or signal decomposition. However, traditional interpolation methods struggle to characterize the non-stationary abrupt changes in traveling wave signals, while conventional filtering methods may weaken the sharp features of the traveling wave front. This makes it difficult to simultaneously achieve both time resolution and wave front recognition accuracy under low sampling rates and strong noise conditions. In summary, current DC traveling wave ranging methods are highly dependent on high sampling frequencies, and wave front features are easily obscured under low sampling rates and strong noise environments. Therefore, the accuracy and reliability of ranging require further optimization. Summary of the Invention

[0005] This invention provides a high-precision ranging method and system for time-refined DC traveling wave, aiming to solve the problems in current DC traveling wave ranging methods, such as high dependence on high sampling frequency, easy submersion of wavefront features in low sampling rate and strong noise environment, and the need for further optimization of ranging accuracy and reliability.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: This invention provides a high-precision ranging method for time-refined DC back-traveling waves, comprising the following steps: S1. Collect DC voltage and DC current signals at the positive and negative terminals at both ends of the DC line after the fault occurs; perform mode transformation to extract the first-mode voltage and first-mode current signals that can characterize the traveling wave information generated by the DC line fault. S2. Based on DC traveling wave theory and the equivalent model of the system after the fault, combined with the equivalent parameters of the DC side of the converter and the parameters of the DC line, the time-domain expressions of the first-mode voltage and the first-mode current reverse traveling wave after the fault are established; wavelet denoising processing is performed on the first-mode DC voltage signal, and noise components are removed by multi-scale decomposition and threshold suppression, and the denoised first-mode DC voltage signal reflecting the sudden change characteristics of the DC reverse traveling wave is extracted. S3. Input the denoised low-sampling-rate inverse traveling wave signal into the time refinement model for time-scale reconstruction to generate an equivalent high-time-resolution DC inverse traveling wave voltage signal. S4. Perform variational mode decomposition on the equivalent high time resolution DC back traveling wave voltage signal to obtain several intrinsic mode components; perform Hilbert transform on each intrinsic mode component to obtain the instantaneous amplitude and instantaneous energy sequence, construct an energy kurtosis index, and identify the arrival time of the first DC back traveling wave front generated by the fault by detecting the sudden increase characteristics of energy kurtosis. S5. Calculate the arrival time difference of the wavefront based on the arrival time of the first DC reverse traveling wavefront measured at both ends of the DC line, and determine the location of the fault point on the DC line by combining the length of the DC line and the propagation characteristics of the traveling wave.

[0007] In some implementations, in S1, discrete DC positive voltage time series, DC negative voltage time series, DC positive current time series, and DC negative current time series are obtained after acquisition. The mode transformation is performed by linearly combining the voltage and current signals of the positive and negative terminals of the DC line to obtain zero-mode components and one-mode components. The time-domain expressions of the one-mode voltage signal and the one-mode current signal are as follows: ; in, It is a modal voltage time-domain signal. It is a modal current time-domain signal. , The voltage signals are measured at the positive and negative terminals respectively. , The positive and negative terminals are used to measure the current signal, respectively.

[0008] In some implementations, in S2, the time-domain expressions for the first-mode voltage and first-mode current back-traveling waves after a fault are established. First, the DC side of the converter is equivalent to a lumped parameter model, and the DC line is equivalent to a distributed parameter transmission line model. The propagation distance is introduced in the form of an exponential term, resulting in the following expressions for the voltage and current of the first-mode DC back-traveling wave in the Laplace domain: ; in, The frequency domain expression of the mode DC back-traveling current at the MN measurement point of the DC line after Laplace transform; The frequency domain representation of the mode reverse traveling wave voltage at the MN measurement point of the DC line after Laplace transform; This refers to the sudden change in the equivalent voltage of the first-order module caused by the fault. This is a model characteristic impedance for a DC line; This is the equivalent inductance on the DC side of the converter station; The attenuation coefficient is related to the line's distributed parameters; The line time equivalent constant; The distance from the fault point to the measurement point; The propagation speed of a traveling wave; For the Laplace operator; Performing the inverse Laplace transform and combining it with the propagation delay characteristics of the traveling wave represented by the exponential term, we obtain the following expressions for the current and voltage of a first-mode DC anti-traveling wave in the time domain: ; in, This represents the time-domain expression of the mode DC reverse traveling wave current at the measurement point MN of the DC line. This represents the time-domain expression of the DC reverse traveling wave voltage at the MN measurement point of the DC line.

[0009] In some implementations, in S2, wavelet denoising processing uses the biorthogonal wavelet Bior4 to perform multi-scale discrete wavelet decomposition on the mode DC inverse traveling wave voltage and current signals, and applies threshold suppression to the high-frequency detail coefficients to remove noise components, resulting in the denoised signal. and .

[0010] In some implementations, in S3, the time refinement model is constructed based on CNN and BiLSTM; the denoised DC inverse traveling wave... and Using discrete time series as input, the abrupt change features of the anti-traveling wave signal within the local time window are extracted by CNN, and the bidirectional temporal correlation between adjacent sampling points is established by BiLSTM, realizing the mapping and reconstruction from the original sampling time scale to a higher time resolution scale, and generating an equivalent high time resolution DC anti-traveling wave voltage signal.

[0011] In some implementations, in S3, the denoised DC voltage signal and DC current signal are represented as discrete sequences: ; in, No. The discrete sequence of the amplitude of the DC reverse traveling wave voltage at each sampling time, after mode transformation and wavelet denoising; No. The amplitude of the DC reverse traveling wave current at each sampling time; Number of sample points; For the time refinement model in the first The multi-channel input vector at each time step, with voltage and current as independent input channels; The actual sampling interval of the inverse traveling wave signal is as follows: ; in, The actual sampling frequency of the DC traveling wave signal ranges from 10 to 20 kHz. The time interval between adjacent sampling points; The formula for local temporal feature extraction using CNN is as follows: ; In the formula: For the first l Layer convolutional network at time k The output feature vector; This is the input layer of the CNN, corresponding to a voltage-current multi-channel inverse traveling wave signal; No. The first layer of convolutional network The input channel in the first Weight parameters at each convolutional kernel position; No. The kernel length of the convolutional layer is used to characterize the abrupt change features of the anti-traveling wave within a local time window; It is a non-linear activation function. For the first Bias terms in a convolutional network.

[0012] In some implementations, in S3, the time correlation is modeled using BiLSTM using the following formula: ; in, For the positive LSTM at time The hidden state is used to characterize the dynamic characteristics of the forward evolution of the anti-traveling wave signal over time; For the inverse LSTM at time The hidden state is used to characterize the dynamic characteristics of the reverse traveling wave signal's reverse correlation over time; The fused output vector of BiLSTM takes into account the forward and backward correlation of the inverse traveling wave signal in the time dimension. Time refinement and variable reconstruction at low sampling frequencies for DC traveling wave signals based on CNN and BiLSTM are shown in the following formula: ; ; in, The time refinement factor represents the factor by which the equivalent sampling rate of the signal after time refinement is increased relative to the original sampling rate. For the time-refined first n The DC reverse traveling wave voltage amplitude at each equivalent high time resolution sampling point; For the time-refined first n The DC reverse traveling wave voltage amplitude at each equivalent high time resolution sampling point; This is a time refinement mapping operator composed of CNN and BiLSTM, used to realize the time-scale reconstruction of low sampling rate inverse traveling wave signals into equivalent high time resolution signals; This represents the set of all trainable parameters in the time-refined model, including convolutional kernel weights, LSTM gating parameters, and fully connected layer parameters.

[0013] In some implementations, the formula for variational mode decomposition in S4 is as follows: ; in, For the first Each variational mode component represents the intrinsic components of the inverse traveling wave signal in different frequency bands; For the first The center frequencies of the modal components; K is the number of modes in the VMD decomposition; The Dirac impulse function in VMD decomposition; The time-refined DC inverse traveling wave signal is composed of multiple intrinsic mode components of different frequency bands superimposed, among which the high-frequency modes contain obvious wavefront abrupt change characteristics.

[0014] In some implementations, the formula for performing the Hilbert transform on each intrinsic mode component in S5 is as follows: ; in, No. Hilbert transform results for each modal component; For Hilbert transform operators; For the first The instantaneous amplitude of each modal component; No. The instantaneous energy of each modal component; The formula for the energy kurtosis index is: ; in, For the first Energy kurtosis index of each modal component; The mean of the instantaneous energy sequence; The variance of the instantaneous energy sequence; For mathematical expectation operators.

[0015] This invention also provides a high-precision ranging system for time-refined DC backtraveling waves to realize the above-mentioned high-precision ranging method for time-refined DC backtraveling waves. The system includes a signal acquisition and extraction module, a time-domain modeling and signal processing module, a time-scale reconstruction module, a variational mode decomposition and wavefront identification module, and a dual-end fault ranging module, wherein: The signal acquisition and extraction module is used to: acquire DC voltage and DC current signals after a fault occurs at the positive and negative terminals at both ends of the DC line; perform mode transformation to extract a mode voltage signal and a mode current signal that can characterize the traveling wave information generated by the DC line fault. The time-domain modeling and signal processing module is used to: establish time-domain expressions for the reverse traveling waves of the first-mode voltage and first-mode current after a fault, based on DC traveling wave theory and the equivalent model of the system after a fault, combined with the equivalent parameters of the DC side of the converter and the parameters of the DC line; perform wavelet denoising on the first-mode DC voltage signal, remove noise components through multi-scale decomposition and threshold suppression, and extract the denoised first-mode DC voltage signal that reflects the abrupt change characteristics of the DC reverse traveling wave; The time-scale reconstruction module is used to: input the denoised low-sampling-rate inverse traveling wave signal into the time refinement model for time-scale reconstruction, and generate an equivalent high-time-resolution DC inverse traveling wave voltage signal; The variational mode decomposition and wavefront identification module is used to: perform variational mode decomposition on the equivalent high time resolution DC back traveling wave voltage signal to obtain several intrinsic mode components; perform Hilbert transform on each intrinsic mode component to obtain the instantaneous amplitude and instantaneous energy sequence, construct an energy kurtosis index, and identify the arrival time of the first DC back traveling wave wavefront generated by the fault by detecting the sudden increase characteristics of energy kurtosis; The dual-end fault location module is used to: calculate the arrival time difference of the wavefront based on the arrival time of the first DC reverse traveling wavefront measured at both ends of the DC line, and determine the location of the fault point on the DC line by combining the length of the DC line and the propagation characteristics of the traveling wave.

[0016] Compared with the prior art, the high-precision ranging method and system of time-refined DC back-traveling wave of the present invention has the following advantages: This invention discloses a high-precision ranging method for time-refined DC reverse traveling wave (DRW). It uses the DC DSW voltage and current signals as multi-channel joint inputs and constructs a time-refining model based on a convolutional neural network and a bidirectional long short-term memory (LSTM) network. This model performs synchronous time-scale reconstruction of the DSW signal under low actual sampling frequency conditions. By utilizing the complementary information of voltage and current signals, it improves the temporal positioning accuracy and stability of the DSW's abrupt change features. Furthermore, this invention adaptively extracts the local abrupt change features of the DSW signal using a convolutional neural network and utilizes a bidirectional LSM network to characterize the forward and backward correlations of the DSW signal in the time dimension. This achieves the mapping of low-sampling-rate DSW signals to equivalent high-time-resolution signals, compensating for insufficient time resolution without increasing the hardware sampling frequency. This invention performs variational mode decomposition on the time-refined DC back-traveling wave signal, adaptively decomposing the signal into multiple intrinsic mode components of different frequency bands. This effectively separates the high-frequency abrupt change components and low-frequency background components in the back-traveling wave, providing an analytical basis for subsequent wavefront feature extraction. Based on variational mode decomposition, Hilbert transform is introduced to construct an energy kurtosis index. By detecting the abrupt change characteristics of energy kurtosis, the arrival time of the first DC back-traveling wavefront is accurately identified, effectively reducing the impact of noise interference, multiple reflections, and amplitude attenuation on the wavefront determination results. Fault location is achieved based on the arrival time difference of the first DC back-traveling wavefront measured at both ends of the DC line. Fault location can be completed using only the information of the first back-traveling wavefront, with low dependence on traveling wave propagation parameters and stable and reliable ranging results. It is suitable for long-distance, high-noise operating scenarios such as deep-sea wind power flexible DC grid-connected systems. Attached Figure Description

[0017] The accompanying drawings are provided to further understand the invention and constitute a part of this invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0018] Figure 1 This is a flowchart illustrating a high-precision ranging method for time-refined DC reverse traveling wave according to the present invention.

[0019] Figure 2 This is a schematic diagram of the fault data measurement location of a offshore wind power flexible DC grid-connected system in an embodiment of a time-refined DC back-traveling wave high-precision ranging method of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0021] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0022] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0023] It should be noted that the apparatus and methods disclosed in the embodiments herein can also be implemented in other ways. The apparatus embodiments described above are merely illustrative; for example, the flowcharts and block diagrams in the accompanying drawings show the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments herein. In this regard, each block in a flowchart or block diagram may represent a module, program, or part of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system to perform the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0024] In addition, the functional modules in the various embodiments of this article can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0025] How to provide a high-precision ranging method for DC inverse traveling wave head that can refine the time scale of traveling wave signals, suppress noise interference, and reliably identify them without increasing the physical sampling frequency.

[0026] Based on this, the present invention provides a high-precision ranging method for time-refined DC back-traveling waves, comprising the following steps: S1. Collect DC voltage and DC current signals at the positive and negative terminals at both ends of the DC line after the fault occurs; perform mode transformation to extract the first-mode voltage and first-mode current signals that can characterize the traveling wave information generated by the DC line fault. S2. Based on DC traveling wave theory and the equivalent model of the system after the fault, combined with the equivalent parameters of the DC side of the converter and the parameters of the DC line, the time-domain expressions of the first-mode voltage and the first-mode current reverse traveling wave after the fault are established; wavelet denoising processing is performed on the first-mode DC voltage signal, and noise components are removed by multi-scale decomposition and threshold suppression, and the denoised first-mode DC voltage signal reflecting the sudden change characteristics of the DC reverse traveling wave is extracted. S3. Input the denoised low-sampling-rate inverse traveling wave signal into the time refinement model for time-scale reconstruction to generate an equivalent high-time-resolution DC inverse traveling wave voltage signal. S4. Perform variational mode decomposition on the equivalent high time resolution DC back traveling wave voltage signal to obtain several intrinsic mode components; perform Hilbert transform on each intrinsic mode component to obtain the instantaneous amplitude and instantaneous energy sequence, construct an energy kurtosis index, and identify the arrival time of the first DC back traveling wave front generated by the fault by detecting the sudden increase characteristics of energy kurtosis. S5. Calculate the arrival time difference of the wavefront based on the arrival time of the first DC reverse traveling wavefront measured at both ends of the DC line, and determine the location of the fault point on the DC line by combining the length of the DC line and the propagation characteristics of the traveling wave.

[0027] This invention integrates DC traveling wave theory and low-sampling-rate signal processing methods, combined with a data-driven time-scale reconstruction model. Under the condition of acquiring only DC voltage and current signals at actual sampling frequencies of 10–20 kHz, it achieves the reconstruction and wavefront feature enhancement of an equivalent high-time-resolution DC inverse traveling wave signal. Furthermore, by combining variational mode decomposition and energy kurtosis criteria, a DC traveling wave ranging method suitable for high-noise environments is constructed. By jointly refining the DC inverse traveling wave voltage and current signals over time, the problem of missing traveling wave abrupt change information under low sampling frequency conditions is compensated for. Based on this, stable identification of the arrival time of the first inverse traveling wave wavefront is achieved. Finally, a dual-end inverse traveling wave ranging method is used to achieve accurate fault location. This invention does not require improving sampling hardware performance or communication bandwidth, and can adapt to the operating characteristics of strong converter interference and significant signal attenuation in deep-sea wind power flexible DC systems, exhibiting certain engineering applicability and robustness.

[0028] like Figure 1 and Figure 2 As shown, in some embodiments, a high-precision ranging method for time-refined DC back-traveling waves is specifically performed according to the following steps: In this embodiment, voltage and current measuring devices are installed at the protection installation positions at both ends of the DC line.

[0029] Step 1: Measure the DC voltage and DC current signals at the positive and negative terminals. , , , Synchronous data acquisition is performed. The actual sampling frequency of the sampling device is set to 10–20 kHz to meet the requirements of cost, communication bandwidth, and real-time performance for on-site engineering measurement devices.

[0030] ; in, and These are DC voltage measurement signals and DC current measurement signals, respectively. , The voltage signals are measured at the positive and negative terminals respectively; , The positive and negative terminals are used to measure the current signal, respectively.

[0031] When a DC line fault occurs, the sampling device acquires the discrete time series of positive and negative voltage and current during the fault transient period, and transmits the acquired data to the ranging calculation unit for subsequent signal processing and analysis.

[0032] Step 2: In the ranging calculation unit, the positive and negative voltage and current signals acquired in Step 1 are processed by mode transformation. By linearly combining the positive and negative signals, the zero-mode component and the first-mode component are obtained respectively.

[0033] Among them, the first-mode voltage signal With a modal current signal It can effectively reflect the propagation characteristics of traveling waves generated by DC line faults and weaken the influence of common-mode interference on traveling wave feature extraction. Therefore, a single-mode component is selected as the input signal for subsequent DC reverse traveling wave analysis.

[0034] Step 3: Based on DC traveling wave theory and post-fault system equivalent modeling method, the DC side of the flexible DC converter station is equivalent to a lumped parameter model, and the DC line is equivalent to a distributed parameter transmission line model.

[0035] Combined with the DC-side equivalent inductance of a Modular Multilevel Converter (MMC) Line characteristic impedance and speed of spread A time-domain expression for a mode DC reverse traveling wave is established. The voltage and current expressions for the mode DC reverse traveling wave generated by the fault in the Laplace domain can be expressed as: ; In the formula: The frequency domain expression of the mode DC back-traveling current at the MN measurement point of the DC line after Laplace transform; The frequency domain representation of the mode reverse traveling wave voltage at the MN measurement point of the DC line after Laplace transform; This refers to the sudden change in the equivalent voltage of the first-order module caused by the fault. This is a model characteristic impedance for a DC line; The equivalent inductance on the DC side of the converter station; The attenuation coefficient is related to the line's distributed parameters; The line time equivalent constant; The distance from the fault point to the measurement point; The propagation speed of a traveling wave; For the Laplace operator; The above equation can be expressed in its time domain form through the inverse Laplace transform. The first-mode DC inverse traveling wave current can be expressed as: ; in: ; The corresponding expression for the DC reverse traveling wave voltage is: ; in: ; In the formula: This represents the time-domain expression of the mode DC reverse traveling wave current at the measurement point MN of the DC line. This represents the time-domain expression of the DC reverse traveling wave voltage at the MN measurement point of the DC line.

[0036] Its time-domain expression is obtained by inverse Laplace transform, which is used to describe the propagation delay and attenuation characteristics of DC anti-traveling waves, providing a theoretical basis for subsequent wavefront identification.

[0037] Step 4: Before refining the time of the reverse traveling wave signal, use biorthogonal wavelets to perform multi-scale discrete wavelet decomposition on the DC reverse traveling wave voltage and current signals.

[0038] By applying threshold suppression to the high-frequency detail coefficients, non-fault-related components such as measurement noise and high-frequency disturbances of the converter are removed. The denoised reverse traveling wave voltage and current signals are reconstructed while maintaining the abrupt change characteristics of the reverse traveling wave wavehead, providing a high signal-to-noise ratio input for subsequent time-scale reconstruction.

[0039] Step 5: Using the denoised single-mode DC inverse traveling wave voltage and current signals obtained in Step 4 as multi-channel joint inputs, construct a time-refined model composed of a Convolutional Neural Network (CNN) and a Bidirectional Long Short-Term Memory (BiLSTM) network. The wavelet-denoised single-mode voltage and current inverse traveling wave time-domain signals are shown. and Represented as the following discrete voltage and current sequences: ; In the formula: No. k The discrete sequence of the amplitude of the DC reverse traveling wave voltage at each sampling time, after mode transformation and wavelet denoising; No. k The amplitude of the DC reverse traveling wave current at each sampling time; Number of sample points; For the time refinement model in the first k The multi-channel input vector at each time step is used, where voltage and current are used as independent input channels.

[0040] Its corresponding actual sampling interval for: ; In the formula: The actual sampling frequency of the DC traveling wave signal ranges from 10 to 20 kHz. This represents the time interval between adjacent sampling points.

[0041] The following is a detailed explanation of how CNNs are used for local temporal feature extraction: ; In the formula: For the first l Layer convolutional network at time k The output feature vector. This is the input layer of the convolutional neural network, corresponding to a voltage-current multi-channel inverse traveling wave signal. No. l The first layer of convolutional network c The input channel in the first m The weight parameters at each convolutional kernel position, where c=1 represents the voltage channel and c=2 represents the current channel; No. l The kernel length of the convolutional layer is used to characterize the abrupt change features of the anti-traveling wave within a local time window; It is a non-linear activation function. For the first l Bias terms in a convolutional network.

[0042] The temporal correlation is modeled using BiLSTM, as follows: ; in: For the positive LSTM at time k The hidden state is used to characterize the dynamic characteristics of the forward evolution of the anti-traveling wave signal over time. For the inverse LSTM at time k The hidden state is used to characterize the dynamic characteristics of the reverse traveling wave signal as it is correlated in reverse over time. It is the fused output vector of the bidirectional LSTM, which takes into account the forward and backward correlation of the reverse traveling wave signal in the time dimension.

[0043] Based on this, time refinement and variable reconstruction of DC traveling wave signals at low sampling frequencies are achieved using CNN-BiLSTM, as detailed below: ; ; In the formula, The time refinement factor represents the factor by which the equivalent sampling rate of the signal after time refinement is increased relative to the original sampling rate. For the time-refined first n The DC reverse traveling wave voltage amplitude at each equivalent high time resolution sampling point; For the time-refined first n The DC reverse traveling wave voltage amplitude at each equivalent high time resolution sampling point; This is a time refinement mapping operator composed of CNN and BiLSTM, used to realize the time-scale reconstruction of low sampling rate inverse traveling wave signals into equivalent high time resolution signals; This represents the set of all trainable parameters in the time-refined model, including convolutional kernel weights, LSTM gating parameters, and fully connected layer parameters.

[0044] Among them, CNN is used to extract the abrupt change features of the inverse traveling wave signal within the local time window, and BiLSTM is used to model the forward and backward correlations of the inverse traveling wave signal in the time dimension. Through this time refinement model, the time scale mapping and reconstruction of the low sampling rate inverse traveling wave signal into an equivalent high time resolution signal is realized, thereby making up for the problem of insufficient time resolution under the condition of limited physical sampling frequency.

[0045] Step 6: Perform variational mode decomposition on the equivalent high-time-resolution DC inverse traveling wave signal obtained in Step 5, adaptively decomposing the inverse traveling wave signal into several eigenmode components with different center frequencies and bandwidth characteristics. Specifically: ; In the formula: No. k Each variational mode function (IMF) represents the intrinsic components of the inverse traveling wave signal in different frequency bands. No. k The center frequencies of the modal components; K is the number of modes in the VMD (Variational Mode Decomposition) decomposition; The Dirac impulse function in VMD decomposition; The time-refined DC inverse traveling wave signal is composed of multiple intrinsic mode components of different frequency bands superimposed, among which the high-frequency modes contain obvious wavefront abrupt change characteristics.

[0046] Through the above decomposition process, the high-frequency abrupt change component and the low-frequency background component in the inverse traveling wave signal are effectively separated, making the high-frequency mode containing wavefront abrupt change information more prominent and reducing the interference of background fluctuations on wavefront identification.

[0047] Step 7: Perform Hilbert transform on each intrinsic mode component obtained in Step 6 to obtain the corresponding instantaneous amplitude and instantaneous energy sequence.

[0048] ; In the formula: The Hilbert transform result of the k-th modal component; This is the Hilbert transform operator. For the first k The instantaneous amplitude of each modal component; No. k The instantaneous energy of each modal component.

[0049] Based on this, an energy kurtosis index for each modal component is constructed. By detecting the abrupt increase in energy kurtosis over time, the arrival time of the first DC back-traveling wavefront caused by the fault is determined, thereby achieving accurate identification of the back-traveling wavefront. The energy kurtosis index is defined as follows: ; In the formula: No. k Energy kurtosis index of each modal component; The mean of the instantaneous energy sequence; The variance of the instantaneous energy sequence; For mathematical expectation operators.

[0050] The criterion for the arrival of the first DC reverse traveling wave at the protection installation point is defined as follows: ; In the formula: This is the moment when the first DC reverse traveling wavefront arrives at the measurement end; This is the threshold for the energy kurtosis abrupt change criterion.

[0051] Step 8: Based on the arrival times of the first DC reverse traveling wave fronts identified at both ends of the DC line, calculate the arrival time difference of the reverse traveling wave fronts, and combine this with the length of the DC line and the propagation speed of the first-mode traveling wave to calculate the location of the fault point on the DC line.

[0052] This invention can complete fault location using only the information of the first traveling wave front, which reduces the dependence on complex traveling wave reflection processes and multi-parameter modeling, improves the stability and reliability of the location results, and has certain engineering application value.

[0053] This invention also provides a high-precision ranging system for time-refined DC backtraveling waves to realize the above-mentioned high-precision ranging method for time-refined DC backtraveling waves. The system includes a signal acquisition and extraction module, a time-domain modeling and signal processing module, a time-scale reconstruction module, a variational mode decomposition and wavefront identification module, and a dual-end fault ranging module, wherein: The signal acquisition and extraction module is used to: acquire DC voltage and DC current signals after a fault occurs at the positive and negative terminals at both ends of the DC line; perform mode transformation to extract a mode voltage signal and a mode current signal that can characterize the traveling wave information generated by the DC line fault. The time-domain modeling and signal processing module is used to: establish time-domain expressions for the reverse traveling waves of the first-mode voltage and first-mode current after a fault, based on DC traveling wave theory and the equivalent model of the system after a fault, combined with the equivalent parameters of the DC side of the converter and the parameters of the DC line; perform wavelet denoising on the first-mode DC voltage signal, remove noise components through multi-scale decomposition and threshold suppression, and extract the denoised first-mode DC voltage signal that reflects the abrupt change characteristics of the DC reverse traveling wave; The time-scale reconstruction module is used to: input the denoised low-sampling-rate inverse traveling wave signal into the time refinement model for time-scale reconstruction, and generate an equivalent high-time-resolution DC inverse traveling wave voltage signal; The variational mode decomposition and wavefront identification module is used to: perform variational mode decomposition on the equivalent high time resolution DC back traveling wave voltage signal to obtain several intrinsic mode components; perform Hilbert transform on each intrinsic mode component to obtain the instantaneous amplitude and instantaneous energy sequence, construct an energy kurtosis index, and identify the arrival time of the first DC back traveling wave wavefront generated by the fault by detecting the sudden increase characteristics of energy kurtosis; The dual-end fault location module is used to: calculate the arrival time difference of the wavefront based on the arrival time of the first DC reverse traveling wavefront measured at both ends of the DC line, and determine the location of the fault point on the DC line by combining the length of the DC line and the propagation characteristics of the traveling wave.

[0054] In summary, this invention provides a high-precision ranging method for time-refined DC reverse traveling wave (RTW) signals. It acquires DC voltage and current signals after a fault at actual sampling frequencies of 10–20 kHz at both ends of the DC line, and extracts a first-mode component reflecting the traveling wave information through mode transformation. Based on DC traveling wave theory and an equivalent model of the system after the fault, a time-domain expression for the first-mode DC RTD is established. Biorthogonal wavelets are used to perform multi-scale denoising on the RTD voltage signal, thereby suppressing measurement noise and high-frequency disturbances in the converter. The time-refining model of this invention reconstructs the time scale of the denoised low-sampling-rate RTD signal, generating an equivalent high-time-resolution DC RTD signal. By constructing an energy kurtosis index through variational mode decomposition and Hilbert transform, the arrival time of the first RTD wavefront is accurately identified, and precise fault ranging is achieved based on the arrival time difference of the two wavefronts. This method can improve the accuracy and reliability of DC RTD ranging to a certain extent without increasing the hardware sampling frequency.

[0055] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Anyone skilled in the art can readily implement the present invention according to the description and above. Any modifications, alterations, or equivalent variations made using the technical content disclosed above are equivalent embodiments of the present invention. Furthermore, any modifications, alterations, or variations made to the above embodiments based on the essential technology of the present invention are still within the protection scope of the present invention.

Claims

1. A high-precision ranging method for time-refined DC back-traveling waves, characterized in that, Includes the following steps: S1. Collect DC voltage and DC current signals at the positive and negative terminals at both ends of the DC line after the fault occurs; perform mode transformation to extract the first-mode voltage and first-mode current signals that can characterize the traveling wave information generated by the DC line fault. S2. Based on DC traveling wave theory and the equivalent model of the system after the fault, combined with the equivalent parameters of the DC side of the converter and the parameters of the DC line, the time-domain expressions of the first-mode voltage and the first-mode current reverse traveling wave after the fault are established; wavelet denoising processing is performed on the first-mode DC voltage signal, and noise components are removed by multi-scale decomposition and threshold suppression, and the denoised first-mode DC voltage signal reflecting the sudden change characteristics of the DC reverse traveling wave is extracted. S3. Input the denoised low-sampling-rate inverse traveling wave signal into the time refinement model for time-scale reconstruction to generate an equivalent high-time-resolution DC inverse traveling wave voltage signal. S4. Perform variational mode decomposition on the equivalent high time resolution DC reverse traveling wave voltage signal to obtain several intrinsic mode components. Hilbert transform is performed on each intrinsic mode component to obtain the instantaneous amplitude and instantaneous energy sequence. An energy kurtosis index is constructed, and the arrival time of the first DC back traveling wave front generated by the fault is identified by detecting the sudden increase characteristics of energy kurtosis. S5. Calculate the arrival time difference of the wavefront based on the arrival time of the first DC reverse traveling wavefront measured at both ends of the DC line, and determine the location of the fault point on the DC line by combining the length of the DC line and the propagation characteristics of the traveling wave.

2. The high-precision ranging method for time-refined DC reverse traveling wave according to claim 1, characterized in that, In step S1, discrete DC positive voltage time series, DC negative voltage time series, DC positive current time series, and DC negative current time series are obtained after acquisition. Mode transformation is performed by linearly combining the voltage and current signals of the positive and negative terminals of the DC line to obtain zero-mode components and one-mode components. The time-domain expressions of the one-mode voltage signal and the one-mode current signal are as follows: ； in, It is a modal voltage time-domain signal. It is a modal current time-domain signal. , The voltage signals are measured at the positive and negative terminals respectively. , The positive and negative terminals are used to measure the current signal, respectively.

3. The high-precision ranging method for time-refined DC reverse traveling wave according to claim 1, characterized in that, In S2, time-domain expressions for the first-mode voltage and first-mode current reverse traveling waves after a fault are established. First, the DC side of the converter is equivalent to a lumped parameter model, and the DC line is equivalent to a distributed parameter transmission line model. The propagation distance is introduced in exponential form, resulting in the following expressions for the voltage and current of the first-mode DC reverse traveling wave in the Laplace domain: ； in, The frequency domain expression of the mode DC back-traveling current at the MN measurement point of the DC line after Laplace transform; The frequency domain representation of the mode reverse traveling wave voltage at the MN measurement point of the DC line after Laplace transform; This refers to the sudden change in the equivalent voltage of the first-order module caused by the fault. This is a model characteristic impedance for a DC line; This is the equivalent inductance on the DC side of the converter station; The attenuation coefficient is related to the line's distributed parameters; The line time equivalent constant; The distance from the fault point to the measurement point; The propagation speed of a traveling wave; For the Laplace operator; Performing the inverse Laplace transform and combining it with the propagation delay characteristics of the traveling wave represented by the exponential term, we obtain the following expressions for the current and voltage of a first-mode DC anti-traveling wave in the time domain: ； in, This represents the time-domain expression of the mode DC reverse traveling wave current at the measurement point MN of the DC line. This represents the time-domain expression of the DC reverse traveling wave voltage at the MN measurement point of the DC line.

4. The high-precision ranging method for time-refined DC reverse traveling wave according to claim 1, characterized in that, In step S2, wavelet denoising is performed using the Bior4 biorthogonal wavelet to decompose the single-mode DC inverse traveling wave voltage and current signals into multi-scale discrete wavelet components. Threshold suppression is applied to the high-frequency detail coefficients to remove noise components, resulting in the denoised signal. and .

5. The high-precision ranging method for time-refined DC reverse traveling wave according to claim 1, characterized in that, In S3, the time refinement model is constructed based on CNN and BiLSTM; the denoised DC inverse traveling wave and Using discrete time series as input, the abrupt change features of the anti-traveling wave signal within the local time window are extracted by CNN, and the bidirectional temporal correlation between adjacent sampling points is established by BiLSTM, realizing the mapping and reconstruction from the original sampling time scale to a higher time resolution scale, and generating an equivalent high time resolution DC anti-traveling wave voltage signal.

6. The high-precision ranging method for time-refined DC reverse traveling wave according to claim 1, characterized in that, In step S3, the denoised DC voltage signal and DC current signal are represented as discrete sequences: ； in, No. The discrete sequence of the amplitude of the DC reverse traveling wave voltage at each sampling time, after mode transformation and wavelet denoising; No. The amplitude of the DC reverse traveling wave current at each sampling time; Number of sample points; For the time refinement model in the first The multi-channel input vector at each time step, with voltage and current as independent input channels; The actual sampling interval of the inverse traveling wave signal is as follows: ； in, The actual sampling frequency of the DC traveling wave signal ranges from 10 to 20 kHz. The time interval between adjacent sampling points; The formula for local temporal feature extraction using CNN is as follows: ； In the formula: For the first l Layer convolutional network at time k The output feature vector; This is the input layer of the CNN, corresponding to a voltage-current multi-channel inverse traveling wave signal; No. The first layer of convolutional network The input channel in the first Weight parameters at each convolutional kernel position; No. The kernel length of the convolutional layer is used to characterize the abrupt change features of the anti-traveling wave within a local time window; It is a non-linear activation function. For the first Bias terms in a convolutional network.

7. The high-precision ranging method for time-refined DC reverse traveling wave according to claim 1, characterized in that, In S3, the time correlation is modeled using BiLSTM using the following formula: ； in, For the positive LSTM at time The hidden state is used to characterize the dynamic characteristics of the forward evolution of the anti-traveling wave signal over time; For the inverse LSTM at time The hidden state is used to characterize the dynamic characteristics of the reverse traveling wave signal's reverse correlation over time; The fused output vector of BiLSTM takes into account the forward and backward correlation of the inverse traveling wave signal in the time dimension. Time refinement and variable reconstruction at low sampling frequencies for DC traveling wave signals based on CNN and BiLSTM are shown in the following formula: ； ； in, The time refinement factor represents the factor by which the equivalent sampling rate of the signal after time refinement is increased relative to the original sampling rate. For the time-refined first n The DC reverse traveling wave voltage amplitude at each equivalent high time resolution sampling point; For the time-refined first n The DC reverse traveling wave voltage amplitude at each equivalent high time resolution sampling point; This is a time refinement mapping operator composed of CNN and BiLSTM, used to realize the time-scale reconstruction of low sampling rate inverse traveling wave signals into equivalent high time resolution signals; This represents the set of all trainable parameters in the time-refined model, including convolutional kernel weights, LSTM gating parameters, and fully connected layer parameters.

8. The high-precision ranging method for time-refined DC reverse traveling wave according to claim 1, characterized in that, In S4, the formula for variational mode decomposition is as follows: ； in, For the first Each variational mode component represents an intrinsic component of the inverse traveling wave signal in different frequency bands; For the first The center frequencies of the modal components; K is the number of modes in the VMD decomposition; The Dirac impulse function in VMD decomposition; The time-refined DC inverse traveling wave signal is composed of multiple intrinsic mode components of different frequency bands superimposed, among which the high-frequency modes contain obvious wavefront abrupt change characteristics.

9. The high-precision ranging method for time-refined DC reverse traveling wave according to claim 1, characterized in that, In S5, the formula for performing the Hilbert transform on each intrinsic mode component is as follows: ； in, No. Hilbert transform results for each modal component; For Hilbert transform operators; For the first The instantaneous amplitude of each modal component; No. The instantaneous energy of each modal component; The formula for the energy kurtosis index is: ； in, For the first Energy kurtosis index of each modal component; The mean of the instantaneous energy sequence; The variance of the instantaneous energy sequence; For mathematical expectation operators.

10. A high-precision ranging system for time-refined DC back-traveling waves, used to implement the high-precision ranging method for time-refined DC back-traveling waves as described in any one of claims 1-9, characterized in that, It includes a signal acquisition and extraction module, a time-domain modeling and signal processing module, a time-scale reconstruction module, a variational mode decomposition and wavefront identification module, and a dual-end fault location module, wherein: The signal acquisition and extraction module is used to: acquire DC voltage and DC current signals after a fault occurs at the positive and negative terminals at both ends of the DC line; perform mode transformation to extract a mode voltage signal and a mode current signal that can characterize the traveling wave information generated by the DC line fault. The time-domain modeling and signal processing module is used to: establish time-domain expressions for the reverse traveling waves of the first-mode voltage and first-mode current after a fault, based on DC traveling wave theory and the equivalent model of the system after a fault, combined with the equivalent parameters of the DC side of the converter and the parameters of the DC line; perform wavelet denoising on the first-mode DC voltage signal, remove noise components through multi-scale decomposition and threshold suppression, and extract the denoised first-mode DC voltage signal that reflects the abrupt change characteristics of the DC reverse traveling wave; The time-scale reconstruction module is used to: input the denoised low-sampling-rate inverse traveling wave signal into the time refinement model for time-scale reconstruction, and generate an equivalent high-time-resolution DC inverse traveling wave voltage signal; The variational mode decomposition and wavefront identification module is used to: perform variational mode decomposition on the equivalent high time resolution DC back traveling wave voltage signal to obtain several intrinsic mode components; perform Hilbert transform on each intrinsic mode component to obtain the instantaneous amplitude and instantaneous energy sequence, construct an energy kurtosis index, and identify the arrival time of the first DC back traveling wave wavefront generated by the fault by detecting the sudden increase characteristics of energy kurtosis; The dual-end fault location module is used to: calculate the arrival time difference of the wavefront based on the arrival time of the first DC reverse traveling wavefront measured at both ends of the DC line, and determine the location of the fault point on the DC line by combining the length of the DC line and the propagation characteristics of the traveling wave.

Images