A signal interference separation and positioning method for a power distribution network with distributed power sources and sensitive loads

Abstract

This invention discloses a method for separating and locating signal interference in a distribution network with distributed power sources and sensitive loads. The method includes: acquiring grid signal data; decomposing the signal data using a wavelet transform algorithm to obtain preliminary separation results of fundamental, harmonic, and transient components; refining the harmonic and transient components using an empirical mode decomposition algorithm based on the preliminary separation results to determine the frequency and amplitude characteristics of each component; collecting real-time signal data at the preliminary location of the interference source; processing the real-time signal data using a sliding window method to obtain the time-domain boundary of the signal abrupt change point; fusing the amplitude characteristics of the fundamental component with the precise time label of the transient interference to generate an interference mitigation command sequence; adjusting response parameters using the command sequence to obtain optimized mitigation response data; verifying the stability of power quality from the optimized mitigation response data; obtaining verified stability indicators to determine that the power quality has returned to normal levels.

Description

Technical Field

[0001] This invention relates to the field of power grid technology, specifically to a method for separating and locating signal interference in a distribution network that includes distributed power sources and sensitive load access. Background Technology

[0002] Today, the large-scale integration of distributed power sources such as photovoltaic and wind power into modern power systems has placed unprecedented pressure on grid stability and increased the risk of power quality disturbances. The intermittency and volatility of these power sources directly amplify problems such as voltage instability and frequency deviations in the grid, especially in highly sensitive areas such as precision semiconductor manufacturing plants, hospital intensive care units, and the power supply buses of large data centers. Even minor power quality deviations can trigger equipment failures, production interruptions, or even safety accidents. For example, on semiconductor wafer fabrication lines, voltage fluctuations exceeding 1% can cause a surge in chip defect rates, resulting in millions in economic losses; in hospital operating rooms, transient harmonic interference can interfere with the accurate readings of electrocardiogram monitors, endangering patients' lives. These areas have extremely low tolerance for power quality issues, typically requiring voltage deviations to be controlled within 0.5% and harmonic distortion rates below 3%, making power quality monitoring and management a core challenge that the power system urgently needs to overcome.

[0003] While existing monitoring and analysis methods have diversified, they struggle to address the cumulative effects of multi-source interference in complex power grid environments. Traditional technologies may be adequate for single interferences, but their accuracy drops drastically and response is delayed when multiple issues such as fundamental distortion, harmonic pollution, voltage sags, and voltage rises occur simultaneously. Particularly in microgrids with high distributed power penetration, low-frequency oscillations and high-order harmonics in the signal couple, further amplifying the complexity of the interference and making it difficult to pinpoint the root cause. For example, in data center bus signals, voltage sags caused by sudden changes in photovoltaic output often overlap with odd-order harmonics generated by industrial loads. Without isolating these components, it becomes impossible to determine whether the issue stems from power supply fluctuations or load-side pollution, hindering fault location.

[0004] To address the above issues, Chinese Patent Publication No. CN215768805U discloses a smart grid signal acquisition device, including a mounting box and a display screen. The mounting box has a mounting cavity on one side, and a signal oscilloscope is installed inside the cavity. A connecting cable is connected to one side of the oscilloscope, and a acquisition probe is attached to the front end of the connecting cable. An anti-interference shell is installed on the outer wall of the acquisition probe, and a fixing rod is fixed to the outer wall of the anti-interference shell. A fixing bolt is attached to the front end of the fixing rod, and a limit plate is fixed to one side of the fixing bolt. This smart grid signal acquisition device, equipped with an acquisition probe, can acquire signals from the smart grid. The anti-interference shell on the acquisition probe provides anti-interference protection, preventing interference that could affect the accuracy of the acquisition. Furthermore, it eliminates the need for operators to hold the acquisition probe, facilitating subsequent operations and preventing poor contact due to unstable handling, further improving the accuracy of the acquisition.

[0005] For example, Chinese Patent CN113484666B discloses a method and system for analyzing characteristic frequency components of a power grid signal. The method includes acquiring an AC power grid signal x(n); filtering the AC signal x(n) based on a second-order sinusoidal window sequence ws2(n) and then performing a discrete Fourier transform to obtain a spectral sequence X(k); searching for local peak values ​​of the modulus sequence of X(k) within a specified width interval using a preset set of center frequencies as midpoints to obtain a local peak spectral line kh; and finally, using a complex frequency domain four-spectral-line equation for frequency correction to obtain the precise frequency value of the h-th frequency component of the analyzed power grid signal. This invention can reduce the spectral leakage influence of distant frequency components. The use of a complex frequency domain four-spectral-line equation for frequency correction fully considers the leakage interference of adjacent different frequency components, and by incorporating the imaginary part of the signal spectrum, it improves the accuracy of characteristic frequency component analysis.

[0006] Currently, existing power grid signal processing technologies still have shortcomings: while both patents have improved the reliability of power grid operation to some extent, they have not fundamentally solved the problem of "difficult separation of multiple interference couplings and difficulty in locating dynamic events" in power quality assurance in highly sensitive areas due to the complexity of signals from distributed power sources. Existing technologies still need improvement. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides a method for separating and locating signal interference in distribution networks with distributed power sources and sensitive load access, in order to solve the aforementioned problems.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a method for separating and locating signal interference in a distribution network with distributed power sources and sensitive load access, mainly comprising the following steps: S1. Acquire power grid signal data, decompose the signal data using wavelet transform algorithm, and obtain preliminary separation results of fundamental component, harmonic component and transient fluctuation component; S2. Based on the preliminary separation results, the harmonic components are refined using an empirical mode decomposition algorithm. And transient fluctuation components, determine the frequency and amplitude characteristics of each component; S3. Extract the interference superposition pattern from the frequency features and amplitude features. If the interference superposition pattern exceeds a preset threshold, it is determined to be multiple interference superposition, and the classification label of multiple interference superposition is obtained. S4. For the classification label, obtain bus signal samples, and directly extract the matching degree between the bus signal samples and the superposition of multiple interferences for the known interference frequency to determine the preliminary location of the interference source; S5. Based on the preliminary position, collect real-time signal data at the position, process the real-time signal data using a sliding window method, and obtain the time domain boundary of the signal change point; S6. Extract the start time and end time of the dynamic interference from the time domain boundary. If the interval between the start time and the end time is less than a preset threshold, it is determined to be transient interference, and the precise time label of the transient interference is obtained. S7. For the precise time tag, fuse the amplitude characteristics of the fundamental component to generate an interference mitigation command sequence, and adjust the response parameters through the command sequence to obtain optimized mitigation response data; S8. Verify the stability of power quality from the optimized governance response data, obtain the verified stability index, and determine that the power quality has returned to normal level.

[0009] Furthermore, the S1 wavelet transform decomposition is applied to the power grid signal. Perform discrete wavelet transform: In formula (1) For the first Scale, First Translation wavelet coefficients, These are discrete wavelet basis functions; Reconstructing the three components by scale: Fundamental component: ; Harmonic components: ; Transient fluctuation components: .

[0010] Furthermore, S1 extracts low-frequency approximation coefficients as the preliminary separation result of the fundamental component: the key to the accuracy of fundamental extraction lies in the fact that the lowest frequency band of the wavelet-divided frequency band has only the fundamental frequency, and the core lies in the determination of the boundary of the lowest frequency band: (1) Obtain the spectrum information of the power grid signal through fast Fourier transform, including the fundamental frequency of the power grid. and lowest harmonic frequency and highest harmonic frequency (2) Determine the lowest frequency band boundary based on the spectrum information. value, (3) Decompose the number of layers through iterative decomposition Calculate and determine the optimal sampling frequency ; Furthermore, S2 includes harmonic components. Empirical Mode Decomposition (EMD) is performed as follows: in For the first One eigenmode, for Number, For residual trend components; Frequency characteristics: ;in for The average interval between adjacent zero crossings, Amplitude characteristics: The same method was used for transient components. Do In order to extract its oscillation characteristics.

[0011] Furthermore, in S3, the frequency and amplitude characteristics of the harmonic components and transient fluctuation components are analyzed to determine the interference superposition mode, and the proportion of interference superposition energy is as follows: In formula (3), the numerator is the amplitude exceeding the threshold. Perform energy summation. The denominator is the total energy of the power grid signal. This is an indicator of the proportion of energy from multiple interferences. Multiple interference determination: To determine the threshold for multiple interferences, a threshold is set. :like It is determined to be multiple interference superposition; for those exceeding the threshold Classifying by frequency yields a set of characteristic frequencies. and corresponding amplitude If the number of categories Then mark it with the category label "multiple interference superposition".

[0012] Furthermore, the interference intensity of each bus in S4 is extracted: for the first bus... busbar signal The known interference frequencies of harmonic components and transient fluctuation components are extracted respectively. Amplitude at: Sampling interval in formula (4) , Sampling frequency, Number of sampling points; Busbar matching degree: In formula (5) The first one extracted for S2 Reference amplitude of each interfering component For the first busbar at frequency The amplitude at that point, The number of interference characteristic frequencies, For the first Interference correlation matching degree of each busbar The larger the value, the stronger the interference experienced by the busbar; Preliminary location of the interference source: Based on matching degree As weight, To monitor the total number of busbars, For the first The position coordinates of each busbar, for each busbar position A weighted average is calculated. Further, in step S5, a high-frequency sampling device is deployed at the initial location to acquire the real-time signal data; the length and step size of the sliding window are set, and the energy variance of the real-time signal data within the sliding window is calculated; when the derivative of the energy variance exceeds the abrupt change threshold, the corresponding window time interval is recorded as the time-domain boundary of the signal abrupt change point.

[0013] Furthermore, in S6, the leading edge extreme point of the time domain boundary is taken as the starting time; the trailing edge extreme point of the time domain boundary is taken as the ending time; the time difference between the ending time and the starting time is calculated; if the time difference is less than the preset threshold, the precise time label containing the starting time and the ending time is generated.

[0014] Furthermore, in S7, the amplitude characteristics of the fundamental component within the corresponding time period are extracted based on the precise time tag; the reactive power compensation and active power regulation are calculated based on the amplitude characteristics; the reactive power compensation and active power regulation are encoded into the interference mitigation instruction sequence; the interference mitigation instruction sequence is sent to the active power filter to adjust the output current parameters of the active power filter, and the adjusted grid signal is collected as the optimized mitigation response data.

[0015] Furthermore, the voltage deviation rate and total harmonic distortion rate of the optimized governance response data are calculated; the voltage deviation rate and the total harmonic distortion rate are used as the verified stability indicators; if the voltage deviation rate is less than a first standard value and the total harmonic distortion rate is less than a second standard value, then a confirmation message that the power quality has returned to normal is output.

[0016] Compared to existing technologies, the advantages of this invention are as follows: A method for signal interference separation and localization in distribution networks with distributed power sources and sensitive load access. This method initially separates signal components using wavelet transform, refines frequency and amplitude characteristics using empirical mode decomposition, accurately extracts interference superposition patterns, and locates the interference source and abrupt time-domain boundaries using Fourier transform and sliding window methods, ultimately generating precise time labels. Regarding interference characteristics, this invention integrates fundamental amplitude characteristics to generate a governance command sequence, optimizes response parameters, and ensures stable power quality. It achieves closed-loop optimization of the entire process from signal decomposition to interference localization and dynamic governance, significantly improving the reliability of power grid operation and the stability of power quality. Attached Figure Description

[0017] Figure 1 This is a flowchart of a method for separating and locating signal interference in a distribution network with distributed power sources and sensitive load access, according to the present invention. Figure 2 This diagram illustrates the influence of the vanishing moment order on the accuracy of wavelet extraction of fundamental wave feature information in a distribution network signal interference separation and localization method with distributed power sources and sensitive loads, as described in this invention. Figure 3 The impact of the vanishing moment order on the harmonic information extraction accuracy of a distribution network signal interference separation and location method with distributed power sources and sensitive load access in this invention; Figure 4 The influence of the vanishing moment order on the wavelet transform harmonic detection error rate and response time of a distribution network signal interference separation and location method with distributed power sources and sensitive load access in this invention; Figure 5 The present invention relates to a method for separating and locating signal interference in a distribution network with distributed power sources and sensitive loads, which examines the impact of different sampling frequencies on the accuracy of fundamental wavelet feature extraction using wavelet transform. Figure 6 The amplitude-frequency characteristics of wavelet transform in the method for separating and locating signal interference in a distribution network with distributed power sources and sensitive loads, as described in this invention; Figure 7 This invention provides a wavelet transform voltage sag start and end time location diagram for a distribution network signal interference separation and location method with distributed power sources and sensitive load access. Figure 8This is a wavelet transform voltage sag amplitude detection diagram for a distribution network signal interference separation and localization method with distributed power sources and sensitive load access according to the present invention. Figure 9 This diagram illustrates the influence of the vanishing moment order on amplitude analysis error and sag location error in a distribution network signal interference separation and location method with distributed power sources and sensitive load access, as described in this invention. Figure 10 This diagram illustrates the influence of the vanishing moment order on the accuracy of voltage sag time location and amplitude detection in a distribution network signal interference separation and location method with distributed power sources and sensitive load access, as described in this invention. Figure 11 This is a schematic diagram of wavelet transform multi-scale decomposition for a distribution network signal interference separation and localization method with distributed power sources and sensitive load access, according to the present invention. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0019] This invention provides a technical solution: a method for separating and locating signal interference in a distribution network with distributed power sources and sensitive loads, comprising: S1. Acquire power grid signal data, decompose the signal data using wavelet transform algorithm, and obtain preliminary separation results of fundamental component, harmonic component and transient fluctuation component; Wavelet transform decomposition for power grid signals Perform discrete wavelet transform: In formula (1) For the first Scale, First Translation wavelet coefficients, These are discrete wavelet basis functions; Reconstructing the three components by scale: Fundamental component: ; Harmonic components: ; Transient fluctuation components: ; S2. Based on the preliminary separation results, the harmonic components and transient fluctuation components are refined using the empirical mode decomposition algorithm to determine the frequency characteristics and amplitude characteristics of each component; Empirical mode decomposition for harmonic components Do : in For the first One eigenmode, for Number, For residual trend components; Frequency and amplitude characteristics: frequency: ; for The average interval between adjacent zero crossings, Amplitude: ; The same method was used for transient components. Do ; S3. Extract the interference superposition pattern from the frequency features and amplitude features. If the interference superposition pattern exceeds a preset threshold, it is determined to be multiple interference superposition, and the classification label of multiple interference superposition is obtained. Interference superposition mode was determined by analyzing the frequency and amplitude characteristics of the harmonic and transient fluctuation components, and the percentage of superposition energy was calculated. In formula (3), the numerator is the amplitude exceeding the threshold. Perform energy summation. The denominator is the total energy of the power grid signal. This is an indicator of the proportion of energy from multiple interferences. Multiple interference determination: To determine the threshold for multiple interferences, a threshold is set. :like It was determined to be a superposition of multiple interferences; For those exceeding the threshold Classifying by frequency yields a set of characteristic frequencies. and corresponding amplitude If the number of categories Then mark it with the category label "multiple interference superposition".

[0020] S4. For the classification label, obtain bus signal samples, and directly extract the matching degree between the bus signal samples and the superposition of multiple interferences for the known interference frequency to determine the preliminary location of the interference source; Interference intensity extraction for each bus: For the first bus... busbar signal The known interference frequencies of harmonic components and transient fluctuation components are extracted respectively. Amplitude at: Sampling interval in formula (4) , Sampling frequency, Number of sampling points; Busbar matching degree: In formula (5) The first one extracted for S2 Reference amplitude of each interfering component For the first busbar at frequency The amplitude at that point, The number of interference characteristic frequencies, For the first Interference correlation matching degree of each busbar The larger the value, the stronger the interference experienced by the busbar; Preliminary location of the interference source: Based on matching degree As weight, To monitor the total number of busbars, For the first The position coordinates of each busbar, for each busbar position Calculate a weighted average.

[0021] Location reliability: ; This represents the maximum matching degree of all buses. The mean of the matching degree. The standard deviation of the matching degree; For location reliability, the judgment rule is as follows: Clear positioning, take Corresponding busbar; Take the first 3 busbars; Expand the monitoring scope; S5. Based on the preliminary position, collect real-time signal data at the position, process the real-time signal data using a sliding window method, and obtain the time domain boundary of the signal change point; S6. Extract the start time and end time of the dynamic interference from the time domain boundary. If the interval between the start time and the end time is less than a preset threshold, it is determined to be transient interference, and the precise time label of the transient interference is obtained. S7. For the precise time tag, fuse the amplitude characteristics of the fundamental component to generate an interference mitigation command sequence, and adjust the response parameters through the command sequence to obtain optimized mitigation response data; S8. Verify the stability of power quality from the optimized governance response data, obtain the verified stability index, and determine that the power quality has returned to normal level.

[0022] If the vanishing moment order of a wavelet function is N, then its corresponding filter length is not less than 2N. The Daubechies wavelet, bioorthogonal wavelet basis series, Coifets wavelet basis, and Symlets wavelet basis series all have high vanishing moment orders N and correspondingly long filter lengths (>2N). From the perspective of filter properties, a higher vanishing moment order corresponds to a longer filter length, resulting in a more accurate description of the filter's amplitude-frequency characteristics. Correspondingly, the filter's transition band is narrower, and the passband is flatter, which is beneficial for the accurate extraction of fundamental frequency characteristics and harmonic components. Figure 2 .

[0023] For the Daubechies wavelet series function, the vanishing moment order ranges from 2 to 45. To study the influence of different vanishing moment orders, a 50Hz fundamental power grid signal containing third and seventh harmonics is used as an example. A power grid signal model is established using MATLAB, with a sampling frequency of 1000Hz and 1000 sampling points. For the aforementioned power grid signal, the extraction and analysis of fundamental and harmonic characteristic information of the power grid signal using Daubechies wavelets with vanishing moment orders of 11, 22, 33, and 44 are studied. The extraction results and accuracy errors are as follows: Figure 3 As shown; The analysis yielded the following conclusions: When the vanishing moment order is small, the extraction error of fundamental wavelet feature information by the db wavelet is relatively large; as the vanishing moment order N increases, the extraction accuracy of the fundamental wavelet feature information by the db wavelet becomes higher and the error becomes smaller, with the maximum error as low as 4.12% when N=44. This is because as the vanishing moment order increases, the transition band in the corresponding wavelet amplitude-frequency characteristics becomes steeper, and the degree of aliasing decreases. Spectral aliasing leads to unclear frequency division boundaries at each wavelet level, reducing the extraction accuracy of fundamental wavelet feature information. Furthermore, from a computational perspective, excessively high orders will increase the computational load and affect the immediacy of wavelet transform signal feature information extraction.

[0024] The reconstructed waveforms of the dB wavelets with different vanishing moment orders in each frequency band are as follows: Figure 4 As shown in the figure, according to Malat multiresolution analysis theory, with a sampling frequency of 1000Hz, the d1 band corresponds to 250~500Hz, containing the seventh harmonic at 350Hz; the d2 band corresponds to 125~250Hz, containing the third harmonic at 150Hz; and the a1 band corresponds to 0~62.5Hz, containing the fundamental frequency at 50Hz. The comparison in the figure shows that the extraction of the third harmonic, especially the fundamental frequency, has a significant impact. As the vanishing moment order increases, the waveform distortion gradually decreases, and the waveform extraction accuracy improves.

[0025] The filter length has a linear relationship with the wavelet order, such as... Figure 6The filter length gradually increases with the wavelet order (≥2N-1), meaning the response time becomes longer. This will have a certain impact on the immediacy of power quality management, resulting in delayed compensation and thus poor management effect. When the vanishing moment order N is greater than 30, the extraction error value changes relatively little. Considering the relationship between the vanishing moment order and the filter length, db30 is selected as the wavelet basis function for analysis.

[0026] Taking a 50Hz fundamental frequency power grid signal containing second, third, fifth, seventh, and ninth harmonics as an example, the db30 wavelet is used for harmonic information analysis. The extraction accuracy of the fundamental frequency feature information is represented by the error rate Rerror. The results are as follows Figure 6 When the sampling frequency is 950Hz, the fundamental frequency extraction error is relatively large, about 7%. As the sampling frequency increases, the fundamental frequency extraction error gradually decreases. When the sampling frequency is 1220Hz, the error rate reaches its minimum value, which corresponds to the lowest frequency band boundary value of 76.25Hz. As the sampling frequency increases, the fundamental frequency extraction error increases.

[0027] The amplitude-frequency characteristics of wavelet transform are as follows: Figure 7 As shown, when the sampling frequency is low, the center frequency is lower than the fundamental frequency; as the sampling frequency increases, the center frequency gradually approaches the fundamental frequency, at which point the filtering effect is optimal and the fundamental feature information extraction accuracy is best; as the sampling frequency increases further, the center frequency is higher than the fundamental frequency, and the filtering effect deteriorates.

[0028] Wavelet transform is used to detect abrupt changes such as voltage sags. After decomposing the signal into various scales, the abrupt change points correspond to the local extrema of the wavelet coefficients. By identifying the extrema in the wavelet components, the abrupt change points of the voltage signal can be identified, and the start and end times of the voltage abrupt change can be determined. Taking a voltage signal with an amplitude of 220V and a frequency of 50Hz as an example, a voltage sag detection analysis is performed to study the localization accuracy of the start and end points of voltage sag signals with different dB wavelets. The sampling frequency was 1600Hz, the time was 1s, a 10% voltage sag occurred between 0.2s and 0.5s, the wavelet transform decomposition level was 2, and the results are as follows. Figure 8 Different dB wavelets can accurately detect the start and end times of voltage sags. When a voltage sag occurs at 0.2 s, the wavelet coefficient amplitude reaches a maximum; when the voltage recovers at 0.5 s, the wavelet coefficient reaches a maximum again. In other cases, the wavelet coefficient is close to zero. The sag detection accuracy and wavelet coefficient extrema of different dB wavelet functions are shown in Table 1: Table 1. Accuracy of slope detection and extreme values ​​of wavelet coefficients for different dB wavelet functions.

[0029] Furthermore, the order of the vanishing moments affects the analytical accuracy of wavelet transform, further impacting the detection of sag amplitudes and the generation of command signals. As the order of the vanishing moments increases, the analytical accuracy of the wavelet transform gradually increases. The influence of the vanishing moment order on the analytical accuracy of wavelet transform is as follows: Figure 10 Under conditions free from harmonic and noise interference, the voltage amplitude detection accuracy is very high. As the vanishing moment order increases, the extracted voltage amplitude error shows a decreasing trend, but the reduction is not significant. The extraction error fluctuates to some extent before and after the voltage sag. Specific data on the wavelet transform sag amplitude detection accuracy are shown in Table 2. Table 2 Wavelet Transform Voltage Sag Detection Amplitude

[0030] Figure 11 This illustrates the impact of the vanishing moment order on the accuracy of voltage sag time localization and amplitude detection. The solid line graph shows the error in voltage amplitude extraction using wavelet transform. The dashed diagram shows the wavelet transform transient. Considering the impact on the compensation function of the series unit, the steady-state error during the voltage sag phase is mainly used as the voltage amplitude detection error. The average of the positioning errors at the time of voltage sag occurrence and the time of voltage sag termination is taken as the sag positioning error. As the order N of the vanishing moment increases, the amplitude analysis error gradually decreases. When N increases from 5 to 10, the amplitude analysis error... The decrease was significant, from 0.4% to below 0.1%. When N is greater than 10, the amplitude analysis error... The error remains at a low level, essentially unchanged. For voltage sag start and end time positioning, the positioning time error increases almost linearly with the increase of the vanishing moment order. Considering both amplitude analysis accuracy error and sag time positioning error, the db wavelet with a vanishing moment order between 10 and 15 is suitable for comprehensive sag analysis. In this embodiment, the db10 wavelet is used as the basis function for sag detection analysis.

[0031] Figure 11 The diagram illustrates wavelet transform multi-scale decomposition. According to multiresolution analysis theory, increasing the decomposition scale involves a layer-by-layer binary decomposition of the low-frequency band (ai, i=1,2,3… …), thus the scale increase has no effect on the highest frequency band d1. For singularity detection of sag signals, the modulus maxima of the highest frequency band d1 can fully reflect the time-domain characteristics of the sag.

[0032] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for signal interference separation and localization in a power distribution network with distributed generation and sensitive load access, characterized in that, include: S1. Acquire power grid signal data, decompose the signal data using wavelet transform algorithm, and obtain preliminary separation results of fundamental component, harmonic component and transient fluctuation component; S2. Based on the preliminary separation results, the harmonic components and transient fluctuation components are refined using the empirical mode decomposition algorithm to determine the frequency characteristics and amplitude characteristics of each component; S3. Extract the interference superposition pattern from the frequency features and amplitude features. If the interference superposition pattern exceeds a preset threshold, it is determined to be multiple interference superposition, and the classification label of multiple interference superposition is obtained. S4. For the classification label, obtain bus signal samples, and directly extract the matching degree between the bus signal samples and the superposition of multiple interferences for the known interference frequency to determine the preliminary location of the interference source; S5. Based on the preliminary position, collect real-time signal data at the position, process the real-time signal data using a sliding window method, and obtain the time domain boundary of the signal change point; S6. Extract the start time and end time of the dynamic interference from the time domain boundary. If the interval between the start time and the end time is less than a preset threshold, it is determined to be transient interference, and the precise time label of the transient interference is obtained. S7. For the precise time tag, fuse the amplitude characteristics of the fundamental component to generate an interference mitigation command sequence, and adjust the response parameters through the command sequence to obtain optimized mitigation response data; S8. Verify the stability of power quality from the optimized governance response data, obtain the verified stability index, and determine that the power quality has returned to normal level.

2. The method of claim 1, wherein, S1 includes: wavelet transform decomposition, on the power grid signal discrete wavelet transform is carried out: In formula (1) is the first scale, the first shift wavelet coefficient, is a discrete wavelet base function; Reconstructing the three components by scale: Fundamental wave component: ; Harmonic components: ; Transient fluctuation components: .

3. The method for separating and locating signal interference in a distribution network with distributed power sources and sensitive load access as described in claim 1, characterized in that, S2 include: For harmonic components Empirical Mode Decomposition (EMD) is performed as follows: in For the first One eigenmode, for Number, For residual trend components; Frequency characteristics: ;in for The average interval between adjacent zero crossings, Amplitude characteristics: ; The same method was used for transient components. Perform EMD to extract its oscillation characteristics.

4. The method for separating and locating signal interference in a distribution network with distributed power sources and sensitive load access according to claim 1, characterized in that, S3 include: Interference superposition mode was determined by analyzing the frequency and amplitude characteristics of the harmonic and transient fluctuation components, and the percentage of superposition energy was calculated. In formula (3), the numerator is the amplitude exceeding the threshold. Perform energy summation. The denominator is the total energy of the power grid signal. This is an indicator of the proportion of energy from multiple interferences. Multiple interference determination: To determine the threshold for multiple interferences, a threshold is set. :like It was determined to be a superposition of multiple interferences; For those exceeding the threshold Classifying by frequency yields a set of characteristic frequencies. and corresponding amplitude If the number of categories Then mark it with the category label "multiple interference superposition".

5. The method for separating and locating signal interference in a distribution network with distributed power sources and sensitive load access according to claim 1, characterized in that, S4 include: Interference intensity extraction for each bus: For the first bus... busbar signal The known interference frequencies of harmonic components and transient fluctuation components are extracted respectively. Amplitude at: Sampling interval in formula (4) , Sampling frequency, Number of sampling points; Busbar matching degree: In formula (5) The first one extracted for S2 Reference amplitude of each interfering component For the first busbar at frequency The amplitude at that point, The number of interference characteristic frequencies, For the first Interference correlation matching degree of each busbar The larger the value, the stronger the interference experienced by the busbar; Preliminary location of the interference source: Based on matching degree As weight, To monitor the total number of busbars, For the first The position coordinates of each busbar, for each busbar position Calculate a weighted average.

6. The method for separating and locating signal interference in a distribution network with distributed power sources and sensitive load access according to claim 1, characterized in that, S5 include: A high-frequency sampling device is deployed at the initial location to acquire the real-time signal data; Set the length and step size of the sliding window, and calculate the energy variance of the real-time signal data within the sliding window; When the derivative of the energy variance exceeds the mutation threshold, the corresponding window time interval is recorded as the time domain boundary of the signal mutation point.

7. The method for separating and locating signal interference in a distribution network with distributed power sources and sensitive load access according to claim 1, characterized in that, S6 include: The extreme point at the leading edge of the time domain boundary is taken as the starting time; The extreme point at the trailing edge of the time domain boundary is taken as the end time; Calculate the time difference between the end time and the start time; If the time difference is less than the preset threshold, then the precise time tag containing the start time and the end time is generated.

8. The method for separating and locating signal interference in a distribution network with distributed power sources and sensitive load access according to claim 1, characterized in that, S7 includes: The amplitude characteristics of the fundamental wave component within the corresponding time period are extracted based on the precise time label; Calculate the reactive power compensation and active power regulation based on the amplitude characteristics; The reactive power compensation amount and the active power adjustment amount are encoded into the interference control instruction sequence; The interference mitigation command sequence is sent to the active power filter, the output current parameters of the active power filter are adjusted, and the adjusted power grid signal is collected as the optimized mitigation response data.

9. The method for separating and locating signal interference in a distribution network with distributed power sources and sensitive load access according to claim 1, characterized in that, S8 includes: Calculate the voltage deviation rate and total harmonic distortion rate of the optimized governance response data; The voltage deviation rate and the total harmonic distortion rate are used as the stability indicators after verification. If the voltage deviation rate is less than the first standard value and the total harmonic distortion rate is less than the second standard value, then a confirmation message that the power quality has returned to normal level is output.

Images