A method and device for identifying broadband multi-mode signals in active power distribution networks.
By filtering out noise in the broadband polymorphic signals of active distribution networks using local weighted regression and local mean decomposition methods, high-precision decomposition is achieved, solving the problem of inaccurate signal decomposition in existing technologies and improving the accuracy and reliability of power grid condition monitoring and fault identification.
Patent Information
- Application Number
- CN202211481376.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-11-24
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2042-11-24
AI Technical Summary
Existing methods for identifying wideband multi-mode signals in active distribution networks suffer from problems such as spectral leakage, low frequency resolution, computational complexity, and difficulty in accurate decomposition under high-noise environments, which affect the accuracy and reliability of power grid condition monitoring and fault identification.
By employing local weighted regression and local mean decomposition methods, combined with Gaussian kernel function and polynomial fitting, and filtering out polymorphic noise components, high-precision decomposition and identification of wideband signals are achieved, including local waveform regression prediction, weight correction, and signal feature identification.
It improves the precision and accuracy of signal decomposition, enabling comprehensive and accurate extraction of various frequency signal components from active distribution networks, thereby enhancing the reliability of power grid condition monitoring and fault identification.
Smart Images

Figure CN115828062B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of active power distribution network signal decomposition and identification technology, and particularly relates to a method and device for identifying broadband polymorphic signals in active power distribution networks. Background Technology
[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art known to those skilled in the art.
[0003] With the large-scale integration and application of controllable loads such as distributed power sources and energy storage systems, as well as intelligent terminal devices, a large number of new power electronic devices, such as converters, inverters, and unified power flow controllers, have been introduced into the power grid. These power electronic devices generate a large number of high-order harmonics, interharmonics, and random noise, causing the active distribution network signals to exhibit obvious broadband and noise polymorphic characteristics. Condition monitoring and fault identification in active distribution networks are based on analyzing the amplitude and phasor characteristics of voltage / current signal components. If broadband polymorphic signals cannot be accurately decomposed and identified, the accuracy of power grid condition monitoring and fault identification will be affected, thereby impacting the reliability and safety of power grid operation. Therefore, how to decompose and identify broadband polymorphic signals in active distribution networks has become a key concern for the power sector.
[0004] The inventors discovered that among existing methods for identifying wideband multi-mode signals in active distribution networks, the Discrete Fourier Transform (DFT) is the most commonly used method for identifying wideband deterministic signals. However, it suffers from severe spectral leakage, low frequency resolution, and complex computation. The Rotationally Invariant Signal Parameter Estimation (ESPRIT) method can achieve high-resolution decomposition, but in high-noise conditions, the boundaries between the signal subspace and the noise subspace are difficult to distinguish, easily leading to order determination errors. Improper setting of the noise covariance matrix in the Kalman filter method can affect the accuracy of deterministic component decomposition and increase the error of random noise component decomposition. The signal model order and initial values of each component frequency in the Taylor Fourier Transform (TFT) method need to be determined in advance. Other signal decomposition methods include wavelet decomposition and Infinite Impulse Response (IIR) filter decomposition, but they usually only analyze narrowband fundamental frequency components, making it difficult to comprehensively and accurately extract the signal components of each frequency in the active distribution network. Summary of the Invention
[0005] To overcome the shortcomings of the prior art, the present invention provides a method and device for identifying wideband polymorphic signals in active power distribution networks, which can filter out polymorphic noise components in active power distribution network signals and achieve high-precision decomposition and identification of wideband signals.
[0006] To achieve the above objectives, the present invention mainly includes the following aspects: In a first aspect, embodiments of the present invention provide a method for identifying broadband multi-mode signals in an active distribution network, comprising: Acquire wideband multi-state signals of active power distribution networks and determine multiple sampling points of the wideband multi-state signals; The local weighted regression method is used to determine the residual of the local waveform regression prediction centered on each sampling point. Based on the residual, it is determined whether each sampling point meets the polymorphic noise screening criterion. If it does, the weights are corrected using the Gaussian kernel function. The filtered value of the filtered position point in the local waveform is determined using the corrected weights. Based on the filtered value, the polymorphic noise components of the wideband polymorphic signal are filtered out to obtain the deterministic component. The deterministic components are processed by local mean decomposition to obtain multiple PF components; The signal characteristics of multiple power distribution components are identified to determine the operating status of the active power distribution network.
[0007] In one possible implementation, in the local weighted regression method, the objective function of the local waveform is set as: ; In the formula, x i For sampling point location parameters, y i for x i Point waveform measurement value, For weights, Let m be the parameters of the j-th order fitting polynomial, m be the order of the fitting polynomial, and n be the number of sampling points in the local waveform. in, x i Point weight The initial value is calculated as follows: ; In the formula, x s The location point being filtered. x i For sampling point location parameters, For x s The average distance between the filter and other points within the filter window.
[0008] In one possible implementation, the multimorphic noise screening criterion is:
[0009] In the formula, Let |e| be the average value, and |·| represent the absolute value; If the sampling point satisfies the polymorphic noise screening criterion, the weights are corrected using the Gaussian kernel function, and the filtered value of the filtered position point is determined using the corrected weights.
[0010] In one possible implementation, the weights are corrected using the following formula: .
[0011] In one possible implementation, the method further includes: if a sampling point does not meet the polymorphic noise screening criteria, then parabolic interpolation is used to process the sampling point.
[0012] In one possible implementation, the step of using local mean decomposition to perform signal decomposition processing on the deterministic components to obtain multiple PF components includes: The first envelope signal is determined based on the envelope function of each local waveform; The first envelope signal is separated from the deterministic component to obtain the residual component; If the residual component does not meet the preset iteration conditions, the residual component is determined as a deterministic component. Based on the envelope function of each local waveform, the second envelope signal and its corresponding residual component are determined until the residual component meets the preset iteration conditions.
[0013] Determining the first envelope signal based on the envelope function of each local waveform includes: Identify the local extreme points of each local waveform; Based on the local mean values of each adjacent extreme point, a local mean function is obtained through Newton's mean difference interpolation; and based on the local magnitude values of each adjacent extreme point, a first envelope function is obtained through Hermite interpolation. The local mean function is separated from the deterministic components and demodulated using the first envelope function to obtain the demodulation function; Calculate the second envelope function of the demodulation function, and determine whether the demodulation function is a pure frequency modulation function based on the second envelope function; if the demodulation function is not a pure frequency modulation function, multiply the first envelope function and the second envelope function to obtain the first envelope signal.
[0014] In one possible implementation, before determining the local extrema of each local waveform, the method further includes: extending the endpoints of the deterministic components using a polynomial fitting-based extension method.
[0015] In one possible implementation, the signal characteristics include, but are not limited to, one or more of frequency, amplitude, and phase.
[0016] Secondly, embodiments of the present invention also provide an identification device for broadband multi-mode signals in active power distribution networks, comprising: The acquisition module is used to acquire wideband multi-state signals of active power distribution networks and determine multiple sampling points of the wideband multi-state signals; The filtering module is used to determine the residual of the local waveform regression prediction centered on each sampling point using the local weighted regression method. Based on the residual, it is determined whether each sampling point meets the polymorphic noise screening criterion. If it does, the weights are corrected using the Gaussian kernel function. The filtered value of the filtered position point in the local waveform is determined using the corrected weights. Based on the filtered value, the polymorphic noise components of the wideband polymorphic signal are filtered out to obtain the deterministic component. The decomposition module is used to perform signal decomposition processing on the deterministic components using the local mean decomposition method to obtain multiple PF components; The identification module is used to identify the signal characteristics of multiple PF components and determine the operating status of the active distribution network.
[0017] The above one or more technical solutions have the following beneficial effects: This invention provides a method for identifying wideband multi-mode signals in active power distribution networks. It uses a local weighted regression method to determine the residual of local waveform regression prediction centered on each sampling point of the wideband multi-mode signal, and introduces a multi-mode noise screening criterion to judge random noise. The weights of the multi-mode noise points are corrected to improve the filtering performance. The extension method using polynomial fitting improves the end-point effect problem, and by improving the calculation method of the local mean function and envelope function and introducing an iterative formula, the problem of over-decomposition can be effectively avoided, thus improving the signal decomposition accuracy. Therefore, the identification method provided by this invention can effectively filter out polymorphic noise in the wideband polymorphic signal of the active distribution network, and extract the signal components of each frequency of the active distribution network more comprehensively and accurately, thereby achieving high-precision decomposition and identification of wideband signals. Attached Figure Description
[0018] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the 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.
[0019] Figure 1 This is a flowchart illustrating the method for identifying broadband multi-mode signals in active power distribution networks provided in Embodiment 1 of the present invention. Figure 2 This is a schematic diagram of the structure of the active power distribution network broadband multi-mode signal identification device provided in Embodiment 2 of the present invention. Detailed Implementation
[0020] The present invention will be further described below with reference to the accompanying drawings and embodiments.
[0021] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0022] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.
[0023] Example 1 like Figure 1 As shown, this embodiment of the invention provides a method for identifying broadband multi-mode signals in an active distribution network, specifically including the following steps: S101: Acquire the wideband multi-state signal of the active power distribution network and determine multiple sampling points of the wideband multi-state signal.
[0024] S102: The local weighted regression method is used to determine the residual of the local waveform regression prediction centered on each sampling point. Based on the residual, it is determined whether each sampling point meets the polymorphic noise screening criterion. If it does, the weight is corrected using the Gaussian kernel function. The filtered value of the filtered position point in the local waveform is determined using the corrected weight. The polymorphic noise component of the wideband polymorphic signal is filtered out according to the filtered value to obtain the deterministic component.
[0025] Among them, the local weighted regression method is used to filter out polymorphic noise components in the sampled signal. In its specific implementation, it can be divided into the following steps: (1) Assume a local waveform Set the objective function as:
[0026] In the formula, x i For sampling point location parameters, y i for x i Point waveform measurement value, For weights, For the parameters of the j-th order fitted polynomial, m To fit the order of the polynomial, n This represents the number of sampling points in the local waveform.
[0027] (2)x i Point weight The initial value is calculated as follows: , In the formula, x s The location point being filtered. x i For sampling point location parameters, for x s The average distance to other points within the filter window.
[0028] (3) According to the weighted regression model Calculate And calculate the residuals .in, X For the position parameter matrix, Y This is a waveform measurement matrix. W It is a weighted diagonal matrix.
[0029] (4) Let Given the average value of |e|, determine whether the polymorphic noise screening criterion is satisfied as follows: .
[0030] (5) If satisfied, then the point is determined to be polymorphic noise, and the Gaussian kernel function is used to adjust the weights. Make corrections and repeat steps (3) to (4) above to obtain the following results. x s Point filter value y s Otherwise, parabolic interpolation is used to process the point.
[0031] Using Gaussian kernel function to apply weights The formula for correction is:
[0032] In the formula, Let |e| be the average value, and n be the number of sampling points in the local waveform. x i x is the position parameter. s The location point being filtered. For x s The average distance to other points within the filter window.
[0033] (6) Move to the next sampling point and repeat steps (2) to (5) to obtain the deterministic component of the local waveform after filtering out the polymorphic noise component. .
[0034] S103: The deterministic component is processed by the local mean decomposition method to obtain multiple PF components.
[0035] The step of using local mean decomposition to decompose the deterministic component into multiple PF components includes: determining a first envelope signal based on the envelope function of each local waveform; separating the first envelope signal from the deterministic component to obtain a residual component; if the residual component does not meet a preset iteration condition, then determining the residual component as a deterministic component, and determining a second envelope signal and its corresponding residual component based on the envelope function of each local waveform, until the residual component meets the preset iteration condition.
[0036] The step of determining the first envelope signal based on the envelope function of each local waveform includes: determining the local extreme points of each local waveform; obtaining a local mean function by Newton's mean-difference interpolation based on the local mean values of each adjacent extreme point; obtaining a first envelope function by Hermite interpolation based on the local amplitude values of each adjacent extreme point; separating the local mean function from the deterministic components and demodulating it with the first envelope function to obtain a demodulation function; calculating a second envelope function of the demodulation function; determining whether the demodulation function is a pure frequency modulation function based on the second envelope function; and multiplying the first envelope function by the second envelope function to obtain the first envelope signal if the demodulation function is not a pure frequency modulation function.
[0037] This embodiment improves the calculation method of the local mean function and envelope function by introducing an iterative formula, which can effectively avoid the problem of over-decomposition and improve the accuracy of signal decomposition.
[0038] Optionally, before determining the local extrema of each local waveform, the method further includes: extending the endpoints of the deterministic components using a polynomial fitting-based extension method, which can improve the endpoint effect problem.
[0039] In specific implementation, (1) the extension method based on polynomial fitting is used to... The left and right endpoints are extended.
[0040] (2) Determine all local extreme points n i Calculate the local mean of adjacent extreme points. m i , all local means m i The local mean function is obtained through Newton's mean difference interpolation. .
[0041] Calculate the local mean of adjacent extreme points m i The formula is:
[0042] In the formula, It is the i-th local extremum point n i Time, It is the first i +1 local extremum point n i+1 The time.
[0043] (3) Calculate the local amplitude of adjacent extreme points a i : All local amplitudes a i The envelope function is obtained through cubic Hermite interpolation. .
[0044] (4) The local mean function From deterministic components Separate from the middle, and obtain:
[0045] Will Using envelope function Demodulation, to obtain :
[0046] (5) Calculation envelope function ,if ,but If it is a pure frequency modulation function, the iteration ends; if Then repeat steps 2-3 until a pure frequency modulation function is obtained. .
[0047] (6) Multiply all the envelope functions from steps (2) to (4) to obtain the first envelope signal: Deterministic components The first PF component is: .
[0048] (7) Separate the first PF component from the original signal to obtain a new signal. : ,judge Does it satisfy the iterative formula? If not, The original signal is treated as a repeat of steps (2)-(6); if the condition is met, the iteration terminates, and the deterministic component is determined. Decomposed into k One PF component and one residual component :
[0049] The iterative formula is defined as:
[0050] S104: Identify the signal characteristics of multiple PF components to determine the operating status of the active distribution network.
[0051] The signal characteristics include, but are not limited to, one or more of frequency, amplitude, and phase. By identifying the signal characteristics of the multiple PF components decomposed from the signal characteristics, the operating status of the active distribution network can be monitored, and the location of the fault can be determined when a fault occurs in the active distribution network.
[0052] Example 2 like Figure 2 As shown, this embodiment of the invention also provides an identification device for broadband multi-mode signals in active power distribution networks. The identification device 200 includes: The acquisition module 210 is used to acquire the wideband multi-state signal of the active power distribution network and determine multiple sampling points of the wideband multi-state signal; The filtering module 220 is used to determine the residual of the local waveform regression prediction centered on each sampling point using the local weighted regression method, and to determine whether each sampling point meets the polymorphic noise screening criterion based on the residual. If it does, the weight is corrected using the Gaussian kernel function, and the filtered value of the filtered position point in the local waveform is determined using the corrected weight. The polymorphic noise component of the wideband polymorphic signal is filtered out based on the filtered value to obtain the deterministic component. The decomposition module 230 is used to perform signal decomposition processing on the deterministic components using the local mean decomposition method to obtain multiple PF components; The identification module 240 is used to identify the signal characteristics of multiple PF components and determine the operating status of the active distribution network.
[0053] The active distribution network broadband multi-mode signal identification device provided in this embodiment is used to implement the aforementioned active distribution network broadband multi-mode signal identification method. Therefore, the specific implementation method of the active distribution network broadband multi-mode signal identification device can be found in the embodiment section of the active distribution network broadband multi-mode signal identification method above, and will not be repeated here.
[0054] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A method for identifying a broadband multi-modal signal of an active power distribution network, characterized in that, include: Acquire wideband multi-state signals of active power distribution networks and determine multiple sampling points of the wideband multi-state signals; The local weighted regression method is used to determine the residual of the local waveform regression prediction centered on each sampling point. Based on the residual, it is determined whether each sampling point meets the polymorphic noise screening criterion. If it does, the weights are corrected using the Gaussian kernel function. The filtered value of the filtered position point in the local waveform is determined using the corrected weights. Based on the filtered value, the polymorphic noise components of the wideband polymorphic signal are filtered out to obtain the deterministic component. In the local weighted regression method, the objective function of the local waveform is set as follows: ; wherein x i is a sampling point position parameter, y i is x i a point waveform measurement value, is a weight value, is a j-th fitting polynomial parameter, m is a fitting polynomial order, and n is a number of sampling points in a local waveform. wherein, x i point weight The initial value of the point weight is calculated as follows: ; where x s is the filtered position point, x i is the sample point position parameter, is x s the average distance to other points within the filter window. The multi-mode noise screening criterion is as follows: wherein is the average of |e|, e being the residual, | · | denoting the absolute value; If the sampling point satisfies the polymorphic noise screening criterion, the weights are corrected using the Gaussian kernel function, and the filtered value of the filtered position point is determined using the corrected weights. The weights are adjusted using the following formula: ; The deterministic components are processed by local mean decomposition to obtain multiple PF components. The signal characteristics of multiple power distribution components are identified to determine the operating status of the active power distribution network.
2. The method of claim 1, wherein the active power distribution network wideband multimodal signal is a wideband multimodal signal of a power distribution network. Also includes: If a sampling point does not meet the polymorphic noise screening criteria, then parabolic interpolation is used to process the sampling point.
3. The method of claim 1, wherein the active power distribution network wideband multimodal signal is a wideband multimodal signal of a power distribution network. The deterministic components are decomposed using the local mean decomposition method to obtain multiple PF components, including: The first envelope signal is determined based on the envelope function of each local waveform; The first envelope signal is separated from the deterministic component to obtain the residual component; If the residual component does not meet the preset iteration conditions, the residual component is determined as a deterministic component. Based on the envelope function of each local waveform, the second envelope signal and its corresponding residual component are determined until the residual component meets the preset iteration conditions.
4. The method of claim 3, wherein the active power distribution network wideband multimodal signal is a power distribution network wideband multimodal signal. Determining the first envelope signal based on the envelope function of each local waveform includes: Identify the local extreme points of each local waveform; Based on the local mean values of each adjacent extreme point, a local mean function is obtained through Newton's mean difference interpolation; and based on the local magnitude values of each adjacent extreme point, a first envelope function is obtained through Hermite interpolation. The local mean function is separated from the deterministic components and demodulated using the first envelope function to obtain the demodulation function; Calculate the second envelope function of the demodulation function, and determine whether the demodulation function is a pure frequency modulation function based on the second envelope function; if the demodulation function is not a pure frequency modulation function, multiply the first envelope function and the second envelope function to obtain the first envelope signal.
5. The method of claim 4, wherein the active power distribution network wideband multimodal signal is identified by: Before determining the local extrema of each local waveform, the process also includes: extending the endpoints of the deterministic components using a polynomial fitting-based extension method.
6. The method for identifying wideband multi-mode signals in an active distribution network as described in claim 1, characterized in that, The signal characteristics include, but are not limited to, one or more of frequency, amplitude, and phase.
7. A device for identifying wideband multi-mode signals in an active power distribution network, characterized in that, include: The acquisition module is used to acquire wideband multi-state signals of active power distribution networks and determine multiple sampling points of the wideband multi-state signals; The filtering module is used to determine the residual of the local waveform regression prediction centered on each sampling point using the local weighted regression method. Based on the residual, it is determined whether each sampling point meets the polymorphic noise screening criterion. If it does, the weights are corrected using the Gaussian kernel function. The filtered value of the filtered position point in the local waveform is determined using the corrected weights. Based on the filtered value, the polymorphic noise components of the wideband polymorphic signal are filtered out to obtain the deterministic component. In the local weighted regression method, the objective function of the local waveform is set as follows: ; wherein x i is a sampling point position parameter, y i is x i point waveform measurement value, is a weight, is a j-th fitting polynomial parameter, m is a fitting polynomial order, and n is a number of sampling points in a local waveform. wherein, x i point weight The initial value of the point weight is calculated as follows: ; where x s is the filtered position point, x i is the sample point position parameter, is x s the average distance between the filtered position point and other points within the filter window. The multi-mode noise screening criterion is as follows: wherein is the average of |e|, e being the residual, | · | denoting the absolute value; If the sampling point satisfies the polymorphic noise screening criterion, the weights are corrected using the Gaussian kernel function, and the filtered value of the filtered position point is determined using the corrected weights. The weights are adjusted using the following formula: ; The decomposition module is used to perform signal decomposition processing on the deterministic components using the local mean decomposition method to obtain multiple PF components; The identification module is used to identify the signal characteristics of multiple PF components and determine the operating status of the active distribution network.
Citation Information
Patent Citations
Water turbine vibration fault diagnosis method
CN104089778A
Deformation monitoring GNSS multipath effect correction method based on local mean decomposition filtering
CN112099067A
