Transient traveling wave adaptive denoising and wavefront identification methods, devices and storage media
By autonomously determining variational mode decomposition parameters and adaptive dynamic wavelet thresholds using the envelope entropy-kurtosis joint criterion in wind farms, and combining this with wavefront steep change region feature protection, refined denoising and wavefront identification of traveling wave signals are achieved. This solves the problem that fixed denoising parameters in wind farms cannot adapt to dynamic noise environments, and improves the accuracy and reliability of fault location.
Patent Information
- Application Number
- CN202611085489.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-21
- Publication Date
- 2026-11-17
AI Technical Summary
Existing technologies cannot adapt to dynamic noise environments in wind farms due to fixed denoising parameters, which makes the characteristics of traveling wave fronts easily damaged. The single wave front identification criterion has weak anti-interference ability, making it difficult to guarantee the accuracy and reliability of fault location.
The variational mode decomposition parameters are autonomously determined based on the envelope entropy-kurtosis joint criterion. Combined with adaptive dynamic wavelet threshold and wavefront steep change region feature protection, the traveling wave signal is refined for denoising and wavefront identification through multi-layer wavelet decomposition and time-domain modulus maxima screening.
It improves the noise reduction effect of transient traveling waves and the accuracy of wavefront identification, ensuring the accuracy and reliability of fault location in wind farm collection lines.
Smart Images

Figure CN123115592A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of power transmission technology, and in particular to a transient traveling wave adaptive denoising and wavefront identification method, device and storage medium. Background Technology
[0002] Wind farm collection lines often adopt a radial multi-branch structure with large line spans and complex terrain. The operation of power electronic equipment such as wind turbine converters and transformer substations generates a large amount of high-frequency electromagnetic interference, which makes the transient traveling wave signals generated by line faults easily submerged by noise. The noise reduction effect of the traveling wave signal and the accuracy of wavehead identification directly determine the accuracy and reliability of fault location.
[0003] Existing traveling wave denoising and wavefront identification technologies have many shortcomings. Traditional wavelet threshold denoising relies on manually preset wavelet bases, decomposition levels, and fixed thresholds, which cannot adapt to scenarios where the noise intensity changes in real time at wind farm sites. Fixed thresholds can easily lead to the loss of traveling wavefront details or noise residue. Conventional variational mode decomposition requires manual setting of the decomposition level and penalty parameters. Parameter settings rely on experience, and for non-stationary and strongly abrupt traveling wave signals in wind farms, mode aliasing is prone to occur, making it impossible to effectively separate effective traveling wave components from interference noise. Single-mode maximum wavefront identification methods rely solely on time-domain abrupt change characteristics to determine the wavefront. In scenarios with high-resistance grounding and weak fault signals, electromagnetic interference and branched reflected waves are easily misidentified as the initial fault wavefront. Furthermore, existing methods are not fully adapted to the specific characteristics of wind farm collector lines, making it difficult to guarantee the accuracy and reliability of fault location. Summary of the Invention
[0004] This application provides a transient traveling wave adaptive denoising and wavefront identification method, device, and storage medium. It can solve the problems in related technologies, such as the inability to adapt to the dynamic noise environment of wind farms due to fixed denoising parameters, the susceptibility of wavefront features to damage, and the weak anti-interference capability of single wavefront identification criteria.
[0005] According to a first aspect of this application, a transient traveling wave adaptive denoising and wavefront recognition method is provided, comprising:
[0006] S1, collects the noisy raw traveling wave signal of the wind farm's collector line; S2, using the envelope entropy-kurtosis joint criterion, autonomously determines the number of decomposition layers and penalty parameters in the variational mode decomposition method; S3, according to the number of decomposition layers and the penalty parameter, perform variational mode decomposition on the noisy original traveling wave signal to obtain multiple intrinsic mode components, calculate the energy proportion of each intrinsic mode component and its correlation coefficient with the noisy original traveling wave signal, remove intrinsic mode components with energy proportions lower than a first preset threshold and correlation coefficients lower than a second preset threshold, reconstruct the retained intrinsic mode components to obtain a preprocessed signal; S4, perform multi-layer wavelet decomposition on the preprocessed signal to obtain high-frequency detail components of each layer, and construct the corresponding adaptive dynamic threshold for each layer based on the real-time noise standard deviation, maximum signal amplitude and average amplitude of the high-frequency detail components of each layer. S5, determine the coefficient segment corresponding to the steep change region of the traveling wave front in each layer of high-frequency detail component by the slope extreme value detection method, set the feature protection coefficient for the coefficient segment, and correct the adaptive dynamic threshold according to the feature protection coefficient; S6, perform differentiated threshold processing on the high-frequency detail components of each layer: wavelet coefficients with amplitudes higher than the adaptive dynamic threshold are determined as effective traveling wave feature components and are shrunk; wavelet coefficients with amplitudes lower than the adaptive dynamic threshold are determined as noise components and are attenuated; for the coefficient segments corresponding to the steep change region of the traveling wave front, perform weak shrunk processing according to the corrected threshold. S7 performs inverse wavelet transform on the high-frequency detail components and low-frequency approximation components of each layer after processing by S6 to complete signal reconstruction and obtain a pure traveling wave signal. S8. Perform time-domain modulus maxima filtering on the pure traveling wave signal to obtain candidate wavefronts. Calculate the energy entropy value of the frequency band corresponding to each candidate wavefront. Determine the candidate wavefront with an energy entropy value lower than a preset energy entropy threshold as the actual fault initial wavefront and output the arrival time of the wavefront.
[0007] According to a second aspect of this application, an electronic device is provided, comprising: At least one processor; and memory that is communicatively connected to at least one processor; The memory stores instructions that can be executed by at least one processor, which are executed by at least one processor to enable the at least one processor to perform the transient traveling wave adaptive denoising and wavefront recognition method described in the first aspect above.
[0008] According to a third aspect of this application, a non-transient computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause a computer to execute the transient traveling wave adaptive denoising and wavefront recognition method described in the first aspect above.
[0009] According to a fourth aspect of this application, a computer program product is provided, including a computer program that, when executed by a processor, implements the transient traveling wave adaptive denoising and wavefront recognition method as described in the first aspect above.
[0010] This application addresses the problems in related technologies, such as the inability to adapt to the dynamic noise environment of wind farms due to fixed denoising parameters, the susceptibility of wavefront features to damage, and the weak anti-interference capability of single wavefront identification criteria. By adopting the envelope entropy-kurtosis joint criterion to autonomously determine the variational mode decomposition parameters and screen modes to complete the initial denoising, combined with the refined denoising strategy of adaptive dynamic wavelet threshold and wavefront steep change region feature protection, and the dual wavefront identification mechanism of time domain modulus maxima screening and frequency domain energy entropy verification, this application achieves the technical effect of improving the transient traveling wave denoising effect and wavefront identification accuracy, and ensuring the accuracy and reliability of fault location of wind farm collection lines.
[0011] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent from the following description. Attached Figure Description
[0012] To more clearly illustrate the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0013] Figure 1 This is a flowchart illustrating a transient traveling wave adaptive denoising and wavefront identification method provided in an embodiment of this application. Detailed Implementation
[0014] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of this application, including various details to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.
[0015] The following description, with reference to the accompanying drawings, describes a transient traveling wave adaptive denoising and wavefront identification method, device, and storage medium according to embodiments of this application.
[0016] Figure 1 This is a flowchart illustrating a transient traveling wave adaptive denoising and wavefront identification method provided in an embodiment of this application.
[0017] like Figure 1 As shown, the method includes the following steps: S1 collects the noisy raw traveling wave signal of the wind farm's collector lines.
[0018] In some embodiments, a traveling wave acquisition device is installed at the outlet of the 35kV collector line of the wind farm's booster station. This device is used to acquire voltage and current transient traveling wave signals generated under line fault conditions in real time. The sampling frequency of the traveling wave acquisition device is set to 10MHz, which can cover the main high-frequency band of transient traveling waves, fully capture the abrupt change characteristics and time-domain details of the traveling wave front, and avoid the loss of fault features due to insufficient sampling accuracy. The acquired raw signal contains the effective component of the fault traveling wave and electromagnetic interference noise in the field environment. This signal is denoted as the noisy raw traveling wave signal f(t) and serves as the input data source for subsequent signal processing.
[0019] The beneficial effect of this step is that by adapting the acquisition point configuration and high sampling frequency setting to the wind farm's collector line topology, it is possible to accurately acquire the original waveform data of the fault transient traveling wave, completely retain core fault information such as wavefront amplitude and arrival time, provide reliable data support for subsequent signal processing, and adapt to the operating characteristics of wind farm traveling wave propagation attenuation and dense interference.
[0020] S2 utilizes the envelope entropy-kurtosis joint criterion to autonomously determine the number of decomposition layers and penalty parameters in the variational mode decomposition method.
[0021] In some embodiments, envelope entropy is used to characterize the concentration of signal features; a smaller envelope entropy indicates more prominent effective signal features and a lower noise ratio. Kurtosis is used to characterize the severity of signal abrupt changes; a larger kurtosis indicates more significant abrupt impact features in the signal and higher identification of fault components such as traveling wave fronts. When determining variational mode decomposition parameters based on the envelope entropy-kurtosis joint criterion, the envelope entropy and kurtosis baseline values of the noisy original traveling wave signal are first calculated. Then, different decomposition levels and penalty parameter combinations are iterated, and the corresponding envelope entropy and kurtosis indices are calculated synchronously for the decomposition results of each set of parameters.
[0022] By comparing and filtering multiple sets of parameters, the parameter combination that minimizes the envelope entropy and maximizes the kurtosis of the decomposition result is finally selected as the number of decomposition layers and penalty parameters used in variational mode decomposition, thereby realizing the autonomous optimization and dynamic adaptation of decomposition parameters.
[0023] The advantage of this step is that it eliminates the need to rely on manual experience to set decomposition parameters. It can adaptively match the optimal parameters based on the actual state of the traveling wave signal at the wind farm site, effectively avoiding the mode aliasing problem caused by fixed parameters in conventional variational mode decomposition, and providing reliable parameter support for the subsequent effective separation of traveling wave components and interference noise.
[0024] S3. Based on the number of decomposition layers and the penalty parameter, perform variational mode decomposition on the noisy original traveling wave signal to obtain multiple intrinsic mode components. Calculate the energy proportion of each intrinsic mode component and its correlation coefficient with the noisy original traveling wave signal. Remove intrinsic mode components whose energy proportion is lower than a first preset threshold and whose correlation coefficient is lower than a second preset threshold. Reconstruct the retained intrinsic mode components to obtain a preprocessed signal.
[0025] In some embodiments, variational mode decomposition is performed on the noisy original traveling wave signal according to the self-determined number of decomposition layers and penalty parameters, decomposing the original signal into multiple intrinsic mode components. Each component corresponds to signal components in different frequency bands, covering high-frequency fault traveling wave components, low-frequency characteristic components and noise components, thereby achieving preliminary separation of signals in different frequency bands.
[0026] After decomposition, the energy proportion of each intrinsic mode component in the original signal and the correlation coefficient between each component and the noisy original traveling wave signal are calculated. The energy proportion reflects the signal energy scale carried by the component, and the correlation coefficient reflects the degree of correlation between the component and the original fault signal. Intrinsic mode components with an energy proportion lower than a first preset threshold and a correlation coefficient lower than a second preset threshold are judged as pure noise modes and directly eliminated; high-frequency fault modes and low-frequency characteristic modes containing effective traveling wave information are retained. The retained intrinsic mode components are then reconstructed to complete the initial denoising process, resulting in a preprocessed signal.
[0027] The beneficial effect of this step is that, through the dual screening criteria of energy ratio and correlation coefficient, pure noise components can be accurately separated, and the effective steep change characteristics of traveling waves can be preserved to the greatest extent while performing preliminary denoising. This avoids the loss of fault information caused by mode aliasing and improves the accuracy and scene adaptability of the preliminary denoising process.
[0028] S4. Perform multi-layer wavelet decomposition on the preprocessed signal to obtain high-frequency detail components of each layer. Based on the real-time noise standard deviation, maximum signal amplitude, and average amplitude of the high-frequency detail components of each layer, construct the corresponding adaptive dynamic threshold for each layer.
[0029] In some embodiments, multi-level wavelet decomposition is performed on the preprocessed signal, using the db4 wavelet as the wavelet basis function. The number of decomposition levels is adaptively determined by combining the frequency band distribution and noise intensity of the preprocessed signal, ranging from 3 to 5 levels. After decomposition, high-frequency detail components and low-frequency approximation components corresponding to different frequency bands are obtained. The high-frequency detail components mainly carry the traveling wave abrupt change characteristics and residual electromagnetic interference noise, achieving fine-grained decomposition of signals in different frequency bands.
[0030] For each layer of high-frequency detail components, the noise standard deviation of the layer is calculated using the noise standard deviation estimation method based on adjacent coefficients, and the maximum and average amplitude of the wavelet detail coefficients of the layer are statistically analyzed simultaneously.
[0031] The beneficial effect of this step is that by constructing a dynamic threshold that matches the signal characteristics in layers, it can adapt to the differences in noise intensity in different frequency bands, avoid the problem of noise residue or weakening of effective features caused by fixed thresholds, provide an accurate basis for subsequent fine noise reduction, and adapt to the complex noise environment of wind farms that are not stable.
[0032] S5. The coefficient segment corresponding to the steep change region of the traveling wave front in each layer of high-frequency detail component is determined by the slope extreme value detection method. A feature protection coefficient is set for the coefficient segment, and the adaptive dynamic threshold is corrected according to the feature protection coefficient.
[0033] In some embodiments, the slope extreme value detection method is used to locate the abrupt change region of the traveling wave front for each layer of high-frequency detail components. The slope of the high-frequency detail coefficients is calculated point by point. If there is a sharp change in the signal amplitude at the corresponding position of the traveling wave front, it will exhibit significant slope extreme value characteristics. By identifying continuous slope extreme value intervals, the high-frequency coefficient segment corresponding to the abrupt change region of the traveling wave front can be determined, and the distribution range of the wave front fault characteristics can be accurately located.
[0034] For the wavefront abrupt change coefficient segment obtained from localization, a feature protection coefficient is set, with a value ranging from 0.3 to 0.6. Based on this feature protection coefficient, the corresponding adaptive dynamic threshold for this region is corrected using the following formula:
[0035] In the formula, The corrected threshold. For characteristic protection coefficient, The original adaptive dynamic threshold is used. The corrected threshold is lower than the original dynamic threshold, so a weak contraction processing strategy is adopted to reduce the denoising contraction intensity in the wavehead region.
[0036] The beneficial effect of this step is that by accurately locating the wavefront abrupt change region and specifically reducing the threshold contraction intensity, it is possible to prioritize ensuring that the key fault characteristics of the traveling wavefront, such as amplitude, steepness, and arrival time, are not weakened during the refined denoising process, maintaining the sharpness and integrity of the wavefront abrupt change, and preserving complete and reliable feature basis for the subsequent wavefront identification process.
[0037] S6, perform differentiated threshold processing on the high-frequency detail components of each layer: wavelet coefficients with amplitudes higher than the adaptive dynamic threshold are determined as effective traveling wave feature components and are shrunk; wavelet coefficients with amplitudes lower than the adaptive dynamic threshold are determined as noise components and are attenuated; for the coefficient segments corresponding to the steep change region of the traveling wave front, perform weak shrunk processing according to the corrected threshold.
[0038] In some embodiments, a differentiated thresholding mechanism is used for each layer of high-frequency detail components, and different processing strategies are executed based on the matching relationship between the coefficient amplitude and the corresponding threshold.
[0039] For wavelet coefficients with amplitudes below the adaptive dynamic threshold, they are identified as noise components. A significantly attenuated effect is applied using an improved threshold function to thoroughly filter out residual broadband electromagnetic interference, Gaussian white noise, and noise redundancy not completely eliminated in the initial denoising stage, achieving deep signal band purification. For wavelet coefficients with amplitudes above the adaptive dynamic threshold, they are identified as valid traveling wave characteristic components. A moderate shrinkage correction is applied using the improved threshold function to preserve the amplitude and phase characteristics of the coefficients, ensuring that the original form of the traveling wave signal is not distorted and maintaining the authenticity of the fault characteristics.
[0040] For the high-frequency coefficient segment corresponding to the steep change region of the traveling wave front, weak shrinkage processing is performed according to the corrected threshold. Coefficients with amplitude close to the corrected threshold are not shrunk, so as to maintain the sharpness and integrity of the steep change of the wave front to the greatest extent and avoid the loss of key fault features such as wave front amplitude, steepness, and arrival time during the denoising process.
[0041] The beneficial effect of this step is that, through differentiated threshold processing based on hierarchical classification, it takes into account both the overall noise filtering effect and the protection of key wavefront features. While deeply removing residual noise, it avoids damage to effective traveling wave features, improves the signal-to-noise ratio and feature fidelity of the denoised signal, and adapts to the traveling wave signal processing needs in strong interference scenarios in wind farms.
[0042] S7 performs inverse wavelet transform on the high-frequency detail components and low-frequency approximate components of each layer after processing by S6 to complete signal reconstruction and obtain a pure traveling wave signal.
[0043] In some embodiments, the high-frequency detail components of each layer after differential thresholding are phase-synchronized with the low-frequency approximate components obtained by wavelet decomposition to ensure accurate temporal phase alignment of components at different scales and eliminate phase shift caused by layered processing. Inverse discrete wavelet transform is then performed on all calibrated components to restore the multi-scale frequency domain components to a complete time-domain traveling wave signal, completing signal reconstruction and obtaining a clean traveling wave signal. The reconstruction process fully preserves the amplitude and phase characteristics of each component, avoiding waveform distortion after reconstruction and ensuring that key fault characteristics such as the steepness of the traveling wavefront and arrival time are not affected by the reconstruction process.
[0044] The beneficial effect of this step is that, through phase-synchronized inverse wavelet transform reconstruction, the components after layered fine denoising can be completely restored into a high-fidelity time-domain traveling wave signal. While deeply purifying noise, the original waveform shape of the traveling wave is maintained, providing a reliable signal basis for subsequent accurate wavefront identification.
[0045] S8. Perform time-domain modulus maxima filtering on the pure traveling wave signal to obtain candidate wavefronts. Calculate the energy entropy value of the frequency band corresponding to each candidate wavefront. Determine the candidate wavefront with an energy entropy value lower than a preset energy entropy threshold as the actual fault initial wavefront and output the arrival time of the wavefront.
[0046] In some embodiments, continuous wavelet transform is performed on the pure traveling wave signal to extract the modulus maxima of wavelet coefficients at each scale. These modulus maxima correspond to the locations of abrupt changes in the time domain of the signal. The initial wavefront of a fault traveling wave exhibits significant abrupt amplitude changes, forming a clear sequence of modulus maxima in the wavelet transform results. By selecting modulus maxima with prominent time-domain abrupt changes, a candidate wavefront point set is obtained, completing the initial localization of the wavefront in the time domain.
[0047] Calculate the energy entropy value of each candidate wavefront point in the corresponding frequency band. The energy of the initial wavefront of the fault traveling wave is concentrated in a specific frequency band, with a high energy distribution concentration, and the corresponding energy entropy value is low and stable. The energy of false wavefronts such as branched reflected waves and power electronic interference is dispersed across multiple frequency bands, with energy entropy values that are high and fluctuate significantly. Candidate wavefront points with energy entropy values lower than a preset energy entropy threshold are identified as the true initial wavefronts of the fault, and all kinds of false interference wavefronts are eliminated. Finally, the arrival time of the true initial wavefront of the fault is output.
[0048] The beneficial effect of this step is that by integrating time-domain abrupt change characteristics and frequency-domain energy entropy as dual criteria for wavefront identification, it can effectively avoid the problems of wavefront misjudgment and missed judgment caused by multi-branch reflections and strong electromagnetic interference in wind farms, improve the wavefront identification accuracy in weak fault signal scenarios, and provide a precise time reference for subsequent line fault location.
[0049] Compared with related technologies, in this embodiment, the noisy original traveling wave signal of the wind farm's collector line is collected; the number of decomposition layers and penalty parameters in the variational mode decomposition method are autonomously determined using the envelope entropy-kurtosis joint criterion; variational mode decomposition is performed on the noisy original traveling wave signal according to the number of decomposition layers and the penalty parameters to obtain multiple intrinsic mode components; the energy proportion of each intrinsic mode component and its correlation coefficient with the noisy original traveling wave signal are calculated; intrinsic mode components with an energy proportion lower than a first preset threshold and a correlation coefficient lower than a second preset threshold are removed; the retained intrinsic mode components are reconstructed to obtain a preprocessed signal; multi-level wavelet decomposition is performed on the preprocessed signal to obtain high-frequency detail components of each layer; adaptive dynamic thresholds corresponding to each layer are constructed based on the real-time noise standard deviation, maximum signal amplitude, and average amplitude of each layer's high-frequency detail components; and the traveling wave signal in each layer's high-frequency detail components is determined by the slope extreme value detection method. For the coefficient segment corresponding to the abrupt change region of the traveling wave front, a feature protection coefficient is set for this coefficient segment, and the adaptive dynamic threshold is corrected according to the feature protection coefficient. Differential threshold processing is performed on the high-frequency detail components of each layer: wavelet coefficients with amplitudes higher than the adaptive dynamic threshold are determined as valid traveling wave feature components and are shrunk; wavelet coefficients with amplitudes lower than the adaptive dynamic threshold are determined as noise components and are attenuated; the coefficient segment corresponding to the abrupt change region of the traveling wave front is weakly shrunk according to the corrected threshold; inverse wavelet transform is performed on the processed high-frequency detail components and low-frequency approximate components of each layer to complete signal reconstruction and obtain a pure traveling wave signal; the pure traveling wave signal is screened for time-domain modulus maxima to obtain candidate wavefront points, the energy entropy value of the frequency band corresponding to each candidate wavefront point is calculated, and the candidate wavefront points with energy entropy values lower than the preset energy entropy threshold are determined as the actual fault initial wavefronts, and the arrival time of the wavefront is output. It can solve the problems in related technologies, such as the inability to adapt to the dynamic noise environment of wind farms due to fixed denoising parameters, the easy damage of wavefront features, and the weak anti-interference of single wavefront identification criteria. It can improve the denoising effect of transient traveling waves and the accuracy of wavefront identification, and ensure the accuracy and reliability of fault location of wind farm collection lines.
[0050] As a specific implementation of this application, based on the basic scheme, the step of S2, which involves autonomously determining the number of decomposition layers and penalty parameters in the variational mode decomposition method using the envelope entropy-kurtosis joint criterion, includes: Calculate the envelope entropy and kurtosis value of the noisy original traveling wave signal; By iterating through different combinations of decomposition levels and penalty parameters, the parameter combination that minimizes the envelope entropy and maximizes the kurtosis is selected as the autonomously determined decomposition level and penalty parameters.
[0051] Specifically, when autonomously determining the number of decomposition levels and penalty parameters for variational mode decomposition, the envelope entropy and kurtosis values of the noisy original traveling wave signal are first calculated. Envelope entropy characterizes the degree of clustering of signal features; a smaller value indicates more prominent effective signal features and a lower proportion of noise components. Kurtosis characterizes the degree of abrupt signal impact; a larger value indicates more significant abrupt fault features such as the traveling wave front. These two calculated indicators reflect the inherent characteristics of the original signal and serve as a criterion for subsequent parameter optimization.
[0052] The decomposition level and penalty parameter range are defined. Different parameter combinations are iteratively substituted into the variational mode decomposition process. For each parameter set, the envelope entropy and kurtosis are calculated simultaneously. After calculating all parameter combinations, the index results for all groups are compared. The parameter combination that minimizes the envelope entropy and maximizes the kurtosis of the decomposition result is selected as the final decomposition level and penalty parameter, thus completing the autonomous optimization configuration of the decomposition parameters.
[0053] The beneficial effect of this embodiment is that, through the traversal iterative parameter optimization mechanism, the optimal decomposition parameters can be automatically matched based on the characteristics of the signal itself, without relying on manual experience for debugging, effectively avoiding the mode mixing problem caused by improper parameter settings, and improving the adaptability of variational mode decomposition to non-stationary traveling wave signals of wind farms.
[0054] As a specific implementation of this application, based on the basic scheme, the calculation formulas for the real-time noise standard deviation, the maximum signal amplitude, and the average amplitude in S4 are further defined as follows:
[0055] Let be the standard deviation of the real-time noise, and median() be the median function. For the i-th wavelet detail coefficient of the current layer, This refers to the (i-1)th wavelet detail coefficient of the current layer; The maximum value of the signal amplitude; The average amplitude is N, and N is the total number of wavelet detail coefficients in the current layer.
[0056] Specifically, when calculating the real-time noise standard deviation, maximum signal amplitude, and average amplitude of each high-frequency detail component, the corresponding formulas are used for quantization. The real-time noise standard deviation is obtained through an estimation algorithm based on adjacent coefficients, and the calculation formula is as follows:
[0057] Let be the standard deviation of the real-time noise, and median() be the median function. For the i-th wavelet detail coefficient of the current layer, This represents the (i-1)th wavelet detail coefficient of the current layer. This algorithm uses the difference between adjacent coefficients to filter out the interference of effective traveling wave components, and can accurately extract the statistical characteristics of residual noise within a single layer, adapting to the working conditions of real-time changes in wind farm noise intensity.
[0058] The maximum signal amplitude is taken as the maximum absolute value of all wavelet detail coefficients in the current layer, calculated using the following formula:
[0059] The average amplitude is the arithmetic mean of the absolute values of all wavelet detail coefficients in the current layer, and is calculated using the following formula:
[0060] In the formula, This represents the maximum amplitude of the signal. The average amplitude is given by , and N is the total number of wavelet detail coefficients in the current layer. These two amplitude parameters quantify the intensity level of the effective traveling wave signal within the corresponding frequency band, providing a quantitative basis for the adaptive adjustment of the dynamic threshold.
[0061] The beneficial effect of this embodiment is that, through a clear quantitative calculation method, the noise intensity and signal amplitude characteristics of each frequency band can be accurately obtained, ensuring the accuracy and rationality of adaptive dynamic threshold calculation, and providing a reliable judgment benchmark for subsequent layered differentiated denoising.
[0062] As a specific implementation of this application, based on the basic scheme, the calculation formula for the adaptive dynamic threshold in S4 is further defined as follows:
[0063] The adaptive dynamic threshold, The adjustment coefficient ranges from 1.0 to 1.5. The real-time noise standard deviation is... The average amplitude is given.
[0064] Specifically, when constructing the adaptive dynamic thresholds corresponding to the high-frequency detail components of each layer, a quantization calculation model is used to solve for the threshold values. The calculation formula is as follows:
[0065] In the formula, For adaptive dynamic threshold, This is an adjustment coefficient, with a value ranging from 1.0 to 1.5. This represents the real-time noise standard deviation of the high-frequency detail components of the corresponding layer. This represents the average amplitude of the signal for the corresponding high-frequency detail components.
[0066] This computational model enables dynamic adaptive adjustment of the threshold. The dynamic threshold increases synchronously with the increase of the noise standard deviation, preventing strong interference from being misjudged as valid traveling wave characteristics under conditions of increased on-site noise intensity. Simultaneously, the threshold adjusts accordingly with the increase of the average signal amplitude, adapting to scenarios with fluctuating traveling wave signal amplitudes, balancing the thoroughness of noise filtering with the integrity of valid signal characteristics. The adjustment coefficient can be adapted to the actual electromagnetic environment of the wind farm's collector lines, flexibly controlling the overall sensitivity of the threshold and matching the differences in operating conditions of different wind farms.
[0067] The beneficial effect of this embodiment is that by combining the dynamic threshold calculation logic of noise intensity and signal amplitude, it replaces the traditional fixed threshold processing method. It can adaptively adjust the denoising judgment criteria according to the signal conditions, effectively balance the denoising depth and wavefront feature fidelity, and improve the adaptability of the wavelet threshold denoising process to the variable noise environment of wind farms.
[0068] As a specific implementation of this application, based on the basic scheme, S7 further includes a quality verification step after performing inverse wavelet transform on the high-frequency detail components and low-frequency approximate components of each layer to complete signal reconstruction: Calculate the signal-to-noise ratio and waveform distortion of the reconstructed signal. If the signal-to-noise ratio is not less than the signal-to-noise ratio threshold and the waveform distortion is not greater than the waveform distortion threshold, then the reconstructed signal is determined to be a qualified pure traveling wave signal. If the qualified standard is not met, adjust the adjustment coefficient and / or the characteristic protection coefficient, and repeat S4 to S7 until a qualified pure traveling wave signal is obtained.
[0069] Specifically, after completing the inverse wavelet transform and signal reconstruction, the quality of the reconstructed traveling wave signal is verified by calculating the signal-to-noise ratio (SNR) and waveform distortion. The SNR threshold is set to 35 dB, and the waveform distortion threshold is set to 5%. If the SNR of the reconstructed signal is not less than 35 dB and the waveform distortion is not greater than 5%, the signal is determined to be a qualified pure traveling wave signal and can proceed to the subsequent wavefront identification process.
[0070] If the reconstructed signal fails to meet the acceptable standard, it indicates that the current denoising parameters do not match the actual signal conditions sufficiently. Therefore, the adjustment coefficient of the dynamic threshold and the feature protection coefficient of the wavefront abrupt change region are adjusted accordingly. Multi-level wavelet decomposition, adaptive dynamic threshold construction, wavefront region threshold correction, and differentiated threshold processing are then re-executed to complete signal reconstruction and quality verification once more. The denoising parameters are gradually optimized through iterative calibration until a clean traveling wave signal meeting the quality requirements is obtained.
[0071] The beneficial effect of this embodiment is that by constructing a closed-loop verification and iterative optimization mechanism for denoising quality, it can ensure that the output traveling wave signal simultaneously meets the requirements of high signal-to-noise ratio and low waveform distortion, avoid the problem of incomplete denoising or waveform distortion caused by single parameter configuration, effectively improve the reliability and stability of signal processing results, and provide a stable data source for subsequent accurate wavefront identification.
[0072] As a specific implementation of this application, based on the basic scheme, S8 further includes, after determining the initial wavefront of the actual fault and outputting its arrival time, the following: The arrival time of the initial wavefront of the actual fault is corrected by linear interpolation to eliminate the time deviation caused by sampling discreteness, and the corrected wavefront arrival time is output.
[0073] Specifically, after determining the initial wavefront of the actual fault and obtaining its initial arrival time, linear interpolation is used to correct the wavefront arrival time. Due to the limitations of the discrete sampling mechanism, the initially identified wavefront time can only correspond to the time nodes of discrete sampling points, resulting in a sub-sampling-level time deviation from the continuous arrival time of the wavefront, which will affect the accuracy of subsequent fault location. By extracting the amplitude data of adjacent sampling points before and after the wavefront abrupt change region, linear interpolation fitting is performed based on the amplitude change trend to reconstruct the continuous change process of the traveling wave wavefront abrupt change, and the accurate arrival time of the actual wavefront is calculated, eliminating the time deviation caused by sampling discreteness, and finally outputting the corrected high-precision wavefront arrival time.
[0074] The beneficial effect of this embodiment is that it overcomes the accuracy limitations of discrete sampling by using linear interpolation correction, further improves the calibration accuracy of wavefront arrival time, reduces measurement errors in the time dimension, provides a more accurate time reference for subsequent line fault location, and ensures the reliability and accuracy of wind farm collector line fault location results.
[0075] Based on the above, Figure 1 The present embodiment describes a transient traveling wave adaptive denoising and wavefront identification method. Correspondingly, this embodiment also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method. Figure 1 This paper presents an adaptive denoising and wavefront identification method for transient traveling waves.
[0076] Based on the above, Figure 1 The present embodiment describes a transient traveling wave adaptive denoising and wavefront identification method. Correspondingly, this embodiment also provides a computer program product on which a computer program is stored. When executed by a processor, this computer program implements the above-described method. Figure 1 This paper presents an adaptive denoising and wavefront identification method for transient traveling waves.
[0077] Based on this understanding, the technical solution of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as CD-ROM, USB flash drive, mobile hard drive, etc.) and includes several instructions to cause a computer device (such as personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of this application.
[0078] Based on the above, Figure 1 The transient traveling wave adaptive denoising and wavefront recognition method shown herein, in order to achieve the above objectives, also provides an electronic device, such as a personal computer or server, which includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to implement the above-described method. Figure 1 This paper presents an adaptive denoising and wavefront identification method for transient traveling waves.
[0079] In some embodiments, the aforementioned physical device may further include a user interface, a network interface, a camera, radio frequency (RF) circuitry, sensors, audio circuitry, a Wi-Fi module, etc. The user interface may include a display screen, an input unit such as a keyboard, etc., and optionally, a USB interface, a card reader interface, etc. In some embodiments, the network interface may include a standard wired interface, a wireless interface (such as a Wi-Fi interface), etc.
[0080] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the aforementioned physical device, supporting the operation of information processing programs and other software and / or programs. The network communication module is used to enable communication between the various components within the storage medium, as well as communication with other hardware and software in the information processing physical device.
[0081] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, 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 the element.
[0082] The above are merely specific embodiments of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to these embodiments, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.
Claims
1. A transient traveling wave adaptive denoising and wavefront identification method, characterized in that, include: S1, collects the noisy raw traveling wave signal of the wind farm's collector line; S2, using the envelope entropy-kurtosis joint criterion, autonomously determines the number of decomposition layers and penalty parameters in the variational mode decomposition method; S3, according to the number of decomposition layers and the penalty parameter, perform variational mode decomposition on the noisy original traveling wave signal to obtain multiple intrinsic mode components, calculate the energy proportion of each intrinsic mode component and its correlation coefficient with the noisy original traveling wave signal, remove intrinsic mode components with energy proportions lower than a first preset threshold and correlation coefficients lower than a second preset threshold, reconstruct the retained intrinsic mode components to obtain a preprocessed signal; S4, perform multi-layer wavelet decomposition on the preprocessed signal to obtain high-frequency detail components of each layer, and construct the corresponding adaptive dynamic threshold for each layer based on the real-time noise standard deviation, maximum signal amplitude and average amplitude of the high-frequency detail components of each layer. S5, determine the coefficient segment corresponding to the steep change region of the traveling wave front in each layer of high-frequency detail component by the slope extreme value detection method, set the feature protection coefficient for the coefficient segment, and correct the adaptive dynamic threshold according to the feature protection coefficient; S6, perform differentiated threshold processing on the high-frequency detail components of each layer: wavelet coefficients with amplitudes higher than the adaptive dynamic threshold are determined as effective traveling wave feature components and are shrunk; wavelet coefficients with amplitudes lower than the adaptive dynamic threshold are determined as noise components and are attenuated; for the coefficient segments corresponding to the steep change region of the traveling wave front, perform weak shrunk processing according to the corrected threshold. S7 performs inverse wavelet transform on the high-frequency detail components and low-frequency approximation components of each layer after processing by S6 to complete signal reconstruction and obtain a pure traveling wave signal. S8. Perform time-domain modulus maxima filtering on the pure traveling wave signal to obtain candidate wavefronts. Calculate the energy entropy value of the frequency band corresponding to each candidate wavefront. Determine the candidate wavefront with an energy entropy value lower than a preset energy entropy threshold as the actual fault initial wavefront and output the arrival time of the wavefront.
2. The transient traveling wave adaptive denoising and wavefront identification method according to claim 1, characterized in that, In step S2, the step of autonomously determining the number of decomposition layers and penalty parameters in the variational mode decomposition method based on the envelope entropy-kurtosis joint criterion includes: Calculate the envelope entropy and kurtosis value of the noisy original traveling wave signal; By iterating through different combinations of decomposition levels and penalty parameters, the parameter combination that minimizes the envelope entropy and maximizes the kurtosis is selected as the autonomously determined decomposition level and penalty parameters.
3. The transient traveling wave adaptive denoising and wavefront identification method according to claim 1, characterized in that, In step S4, the formulas for calculating the real-time noise standard deviation, the maximum signal amplitude, and the average amplitude are as follows: Let be the standard deviation of the real-time noise, and median() be the median function. For the i-th wavelet detail coefficient of the current layer, This refers to the (i-1)th wavelet detail coefficient of the current layer; The maximum value of the signal amplitude; The average amplitude is N, and N is the total number of wavelet detail coefficients in the current layer.
4. The transient traveling wave adaptive denoising and wavefront identification method according to claim 3, characterized in that, In step S4, the formula for calculating the adaptive dynamic threshold is: The adaptive dynamic threshold, The adjustment coefficient ranges from 1.0 to 1.
5. The real-time noise standard deviation is... The average amplitude is given.
5. The transient traveling wave adaptive denoising and wavefront identification method according to claim 1, characterized in that, In step S7, after performing inverse wavelet transform on the high-frequency detail components and low-frequency approximate components of each layer to complete signal reconstruction, a quality verification step is also included: Calculate the signal-to-noise ratio and waveform distortion of the reconstructed signal. If the signal-to-noise ratio is not less than the signal-to-noise ratio threshold and the waveform distortion is not greater than the waveform distortion threshold, then the reconstructed signal is determined to be a qualified pure traveling wave signal. If the qualified standard is not met, adjust the adjustment coefficient and / or the characteristic protection coefficient, and repeat S4 to S7 until a qualified pure traveling wave signal is obtained.
6. The transient traveling wave adaptive denoising and wavefront identification method according to claim 1, characterized in that, In step S8, after determining the initial wavefront of the actual fault and outputting its arrival time, the method further includes: The arrival time of the initial wavefront of the actual fault is corrected by linear interpolation to eliminate the time deviation caused by sampling discreteness, and the corrected wavefront arrival time is output.
7. An electronic device, characterized in that, include: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform the transient traveling wave adaptive denoising and wavefront recognition method according to any one of claims 1-6.
8. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to execute the transient traveling wave adaptive denoising and wavefront identification method according to any one of claims 1-6.
9. A computer program product, characterized in that, The method includes a computer program that, when executed by a processor, implements the transient traveling wave adaptive denoising and wavefront identification method according to any one of claims 1-6.
