Distribution cable branch box state monitoring method and device

Abstract

The invention belongs to the technical field of power system monitoring, and particularly relates to a distribution cable branch box state monitoring method and device, and the method comprises the steps: collecting load current data and electromagnetic pulse signals at a cable branch box, and carrying out the variational mode decomposition of the electromagnetic pulse signals, so as to obtain a pure partial discharge signal sequence; the pure partial discharge signal sequence is processed into a partial discharge intensity sequence, a data pair is constructed in a sliding time window, correlation parameters are obtained, and an upper tail coupling drift coefficient is calculated according to the correlation parameters; calculating signal energy in the current sliding time window, and constructing an insulation degradation comprehensive risk index; and when the insulation degradation comprehensive risk index is greater than or equal to a red threshold value, judging that the insulation degradation is serious and triggering an alarm. According to the invention, signals can be adaptively reconstructed through the spectrum entropy, the coupling features are extracted by using the connection function, and the risk index is constructed to realize accurate early warning of the insulation state in the interference environment.

Description

Technical Field

[0001] This invention relates to the field of power system monitoring technology. More specifically, this invention relates to a method and apparatus for monitoring the status of distribution cable branch boxes. Background Technology

[0002] As a critical node in power distribution networks, distribution cable branch boxes primarily undertake the important tasks of cable branching and power distribution. However, because distribution cable branch boxes are usually installed outdoors, they are affected by harsh conditions such as diurnal temperature variations, humidity, and rain and snow erosion. Core components inside the distribution cable branch boxes, such as stress cones and insulators, are highly susceptible to insulation aging, surface creep, and even poor contact, posing safety hazards. If these early defects are not detected and addressed in a timely manner, they can easily lead to safety accidents such as short circuits and fires, affecting the reliability of power supply.

[0003] Existing methods for monitoring distribution cable branch boxes mainly include regular manual inspections and online monitoring. In regular manual inspections, infrared thermal imagers are typically used to monitor the usage status of distribution cable branch boxes, but this method is inefficient and has blind spots. In online monitoring, partial discharge is generally identified by monitoring ultra-high frequency signals or transient voltage to ground. However, the outdoor environment is filled with various random electromagnetic interferences such as communication base station signals, radio broadcasts, and vehicle ignition. These noises often overlap with early partial discharge signals in the spectrum. Existing technologies usually use filters with fixed thresholds for processing, which can easily filter out the real weak partial discharge signals or misjudge strong interference signals as faults, resulting in a high false alarm rate for online monitoring.

[0004] Furthermore, existing monitoring systems often view partial discharge amplitude as a single indicator in isolation. However, the intensity of partial discharge caused by insulation defects is usually strongly coupled with the load current, meaning that the greater the load current, the stronger the partial discharge intensity. On the other hand, external interference signals are unrelated to changes in the load current. Existing technologies ignore this nonlinear coupling characteristic between the load current and the intensity of partial discharge, which makes it impossible for monitoring systems to fundamentally distinguish between real faults and false interference, and makes it difficult to provide accurate early warning information in the early stages of a fault. Summary of the Invention

[0005] To address the technical problems of existing monitoring methods being prone to false alarms due to environmental interference and having low judgment accuracy, this invention proposes a method and device for monitoring the status of distribution cable branch boxes. This method can adaptively reconstruct signals through spectral entropy and extract coupling features using connection functions to construct a risk index, thereby achieving accurate early warning of insulation status under interference environments.

[0006] In a first aspect, the present invention provides a method for monitoring the status of a power distribution cable branch box, comprising: collecting load current data and electromagnetic pulse signals at the cable branch box, obtaining the spectral entropy value of the current signal and calculating the optimal decomposition mode number, performing variational mode decomposition on the electromagnetic pulse signal, and reconstructing a clean partial discharge signal sequence based on the kurtosis values ​​of the decomposed mode components;

[0007] The pure partial discharge signal sequence is processed into a partial discharge intensity sequence with the same sampling rate as the load current data. Data pairs are constructed within a sliding time window. The load current data and the partial discharge intensity sequence are subjected to probability integral transformation and mapped to uniformly distributed data. The uniformly distributed data is fitted to obtain the correlation parameters. The upper tail coupling drift coefficient is calculated based on the correlation parameters. The root mean square value of the clean partial discharge signal sequence within the current sliding time window is calculated as the signal energy, and a comprehensive insulation degradation risk index is constructed. The comprehensive insulation degradation risk index is positively correlated with the signal energy and the upper tail coupling drift coefficient. When the comprehensive risk index of insulation degradation is less than the preset yellow threshold, the equipment is determined to be in good working order. When the comprehensive risk index of insulation degradation is greater than or equal to the yellow threshold and less than the preset red threshold, it is determined to be an early latent fault and an early warning is issued. When the comprehensive risk index of insulation degradation is greater than or equal to the red threshold, it is determined to be severe insulation degradation and an alarm is triggered.

[0008] By adopting the above technical solution, load current data and electromagnetic pulse signals are first collected, and the optimal decomposition mode number is adaptively determined based on spectral entropy. This overcomes the shortcomings of traditional filters, which filter out weak partial discharge signals or misjudge strong interference signals due to fixed thresholds, thus ensuring the purity of the extracted partial discharge signal from the source. Furthermore, by constructing a comprehensive insulation degradation risk index that includes the upper tail coupling drift coefficient, the problem of existing monitoring systems viewing partial discharge amplitude in isolation and struggling to identify random electromagnetic interference is effectively solved. This method utilizes the strong coupling characteristics between partial discharge intensity and load current under extreme conditions to accurately distinguish between real insulation defects that vary with the load and random noise unaffected by the load. This significantly improves the anti-interference capability and early warning accuracy of the monitoring system in strong interference environments, fundamentally reducing the false alarm rate of online monitoring caused by misjudgment.

[0009] Preferably, the optimal number of decomposition modes satisfies the following relationship: ; In the formula, The nearest integer is taken as the optimal number of decomposed modes for the current signal. This is the preset minimum number of decomposition levels; This is the preset maximum number of decomposition levels; This represents the measured spectral entropy value of the electromagnetic pulse signal at the current acquisition moment. The statistical mean of the spectral entropy of pure background noise collected historically; The maximum statistical value of the spectral entropy of typical broadband fault signals collected historically; and when season ,when season .

[0010] By adopting the above technical solution, a nonlinear mapping relationship between the optimal number of decomposed modes and the signal spectral entropy was established. The mean historical background noise entropy and the maximum entropy of typical fault signals were introduced as boundary constraints, enabling the monitoring system to adaptively adjust the signal decomposition depth according to the complexity of the electromagnetic environment. In environments with strong interference, the system achieves fine noise removal by increasing the number of decomposition layers; when the signal is relatively clean, the number of decomposition layers is reduced to improve processing efficiency. This significantly optimizes the dynamic allocation of computing resources and enhances the monitoring device's adaptability and operational efficiency in complex and variable outdoor electromagnetic environments.

[0011] Preferably, reconstructing a clean partial discharge signal sequence based on the kurtosis values ​​of the decomposed modal components includes: calculating the kurtosis value of each modal component obtained by variational mode decomposition; determining whether the kurtosis value is greater than a set kurtosis threshold; and accumulating the modal components with kurtosis values ​​greater than the kurtosis threshold to obtain the reconstructed clean partial discharge signal sequence.

[0012] By employing the aforementioned technical solution, and using the kurtosis values ​​of each modal component after variational mode decomposition as a basis, a clean partial discharge signal sequence is screened and reconstructed by comparing it with a preset threshold. This approach accurately captures partial discharge pulse signals with significant impact characteristics while effectively suppressing stable background noise, thereby enhancing the salience of partial discharge characteristics during signal preprocessing. The reconstruction strategy based on the statistical property of kurtosis avoids the problem of potentially filtering out weak but valuable partial discharge signals in traditional filtering methods, improving the completeness and reliability of fault signal extraction.

[0013] Preferably, the partial discharge intensity sequence is processed to be consistent with the sampling rate of the load current data. Data pairs are constructed within a sliding time window, and probability integral transformation is performed on the load current data and the partial discharge intensity sequence respectively, including: calculating the marginal distribution of the load current data and the partial discharge intensity sequence respectively using an empirical distribution function; and using the calculated marginal distribution values ​​as the transformed uniform distribution data, wherein the load current data corresponds to a first uniform distribution variable, and the partial discharge intensity sequence corresponds to a second uniform distribution variable.

[0014] By employing the above technical solution, the load current data and partial discharge intensity sequence are subjected to probability integral transformation using an empirical distribution function, mapping them to data following a uniform distribution. This effectively eliminates the differences in physical meaning and dimensions of the original signals and overcomes the interference of amplitude inconsistencies on correlation analysis. Thus, it provides standardized, dimensionless input variables for accurately extracting the nonlinear coupling characteristics between the load and partial discharge using a connection function. This method improves the robustness and comparability of the feature extraction process, laying the foundation for reliable identification of insulation status.

[0015] Preferably, fitting the uniformly distributed data to obtain the correlation parameters includes: constructing a density function expression for the Gumbel connection function; substituting the uniformly distributed data into the density function expression to establish a log-likelihood function; and using the maximum likelihood estimation method to solve for the maximum value of the log-likelihood function to obtain the corresponding parameter estimate as the correlation parameter.

[0016] By adopting the above technical solution, we can deeply analyze the correlation and evolution law of load current and partial discharge intensity in the tail region of the distribution, and obtain the key parameters of the connection function based on maximum likelihood estimation. This allows us to more accurately characterize the intrinsic correlation of high-intensity partial discharge induced under high load conditions. From the algorithm level, we can effectively distinguish between real fault signals that fluctuate with the load and random noise that is not affected by the load, thereby enhancing the objectivity and reliability of the state judgment.

[0017] Preferably, the upper tail coupling drift coefficient includes: ; In the formula, This represents the upper tail coupling drift coefficient within the current time window; These are the correlation parameters obtained by fitting using the Gumbel connection function.

[0018] By employing the above technical solution, the correlation and evolution law of load current and partial discharge intensity in the tail region of the distribution can be deeply revealed, and the key parameters of the connection function can be obtained based on maximum likelihood estimation, thereby more accurately characterizing the intrinsic correlation mechanism of high-intensity partial discharge induced under high load conditions. This mechanism effectively distinguishes between real fault signals that fluctuate with the load and random interference that is not affected by the load at the algorithmic level, improving the objectivity and reliability of state judgment.

[0019] Preferably, the comprehensive risk index of insulation degradation satisfies the following relationship: ; In the formula, This is a comprehensive risk index for insulation degradation. This represents the average energy value of the reconstructed clean partial discharge signal sequence within the current time window; The baseline background noise energy when the equipment is in a healthy state; These are the preset coupling weight coefficients; This represents the upper tail coupling drift coefficient.

[0020] Preferably, the load current data is acquired through a Rogowski coil, and the electromagnetic pulse signal is acquired through an ultra-high frequency antenna mounted on the inner wall of the branch box.

[0021] Preferably, the determination logic of the comprehensive risk index of insulation degradation includes: before calculating the comprehensive risk index of insulation degradation, the root mean square algorithm is used when calculating the energy of the pure partial discharge signal sequence; after determining that the insulation degradation is serious and triggering an alarm, a trip command or an emergency maintenance command is generated.

[0022] Secondly, the present invention provides a power distribution cable branch box status monitoring device, comprising: a data acquisition module for acquiring load current data and electromagnetic pulse signals of the power distribution cable branch box; a signal reconstruction module for calculating the spectral entropy value of the electromagnetic pulse signal, adaptively calculating the optimal decomposition mode number based on the spectral entropy value, and performing variational mode decomposition and reconstruction on the electromagnetic pulse signal based on the optimal decomposition mode number to obtain a clean partial discharge signal sequence; a feature analysis module for performing probability integral transformation and connection function fitting on the load current data and the partial discharge intensity sequence obtained from the clean partial discharge signal sequence within a sliding time window to calculate the upper tail coupling drift coefficient; and a decision warning module for constructing a comprehensive insulation degradation risk index based on the energy of the clean partial discharge signal sequence and the upper tail coupling drift coefficient, and outputting a status judgment result by comparing it with a preset threshold.

[0023] By integrating the acquisition, reconstruction, analysis, and early warning functions of the monitoring method through a modular structure, the entire process from signal perception to state decision-making is automated. It can not only execute all the key steps of the aforementioned methods to ensure the integrity and consistency of the monitoring process, but also enable the algorithm to run efficiently and stably in actual equipment through hardware and software co-design. Thus, at the engineering application level, it realizes real-time, accurate, and reliable online monitoring of the insulation status of power distribution cable branch boxes.

[0024] The beneficial effects of this invention are as follows: This invention dynamically assesses the degree of signal disorder by calculating the spectral entropy of electromagnetic pulse signals in real time, and adaptively adjusts the number of modes decomposition in the variational mode decomposition algorithm accordingly. This increases the number of decomposition layers to finely remove noise when the signal is complex, and reduces the number of decomposition layers to improve efficiency when the signal is clean. This mechanism ensures the purity of partial discharge signal extraction from the source, laying the foundation for subsequent analysis.

[0025] Furthermore, this invention breaks through the limitation of traditional monitoring that only focuses on signal amplitude. It uses the Gumbel connection function to characterize the nonlinear correlation between load current and partial discharge signal in the extreme region and generates an upper tail coupling drift coefficient that is immune to interference signals. The upper tail coupling drift coefficient can effectively distinguish between real insulation defects that change with load and random noise that does not fluctuate with load, thus achieving high-reliability early warning even in strong noise background and significantly improving the anti-interference capability of the monitoring system. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating a method for monitoring the status of a power distribution cable branch box according to the present invention. Figure 2 This is a diagram illustrating the nonlinear coupling distribution characteristics of load current and partial discharge intensity in this invention. Figure 3 This is a schematic diagram of the fault state decision plane based on energy and coupling two-dimensional features in this invention. Detailed Implementation

[0027] 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 some embodiments of the present invention, but not all embodiments.

[0028] This invention discloses a method for monitoring the status of a power distribution cable branch box, referring to... Figure 1 This includes steps S1-S4: S1. Collect load current data and electromagnetic pulse signal at the cable branch box, obtain the spectral entropy value of the current signal and calculate the optimal decomposition mode number, perform variational mode decomposition on the electromagnetic pulse signal, and reconstruct a clean partial discharge signal sequence based on the kurtosis values ​​of the decomposed mode components.

[0029] In an optional embodiment, load current data is first acquired in real time using a Rogowski coil installed at the cable connector of the branch box, and recorded as a sequence. Simultaneously, an ultra-high frequency antenna mounted on the inner wall of the enclosure is used to collect electromagnetic pulse signals, which are recorded as the original high-frequency signal sequence. It should be noted that the sampling frequency for acquiring electromagnetic pulse signals should be set to no less than 20MHz in order to acquire high-frequency partial discharge characteristics.

[0030] Traditional variational mode decomposition algorithms require manual specification of the number of mode decompositions. If improperly formulated, it can lead to under- or over-decomposition of the signal. Therefore, this invention dynamically calculates the optimal value based on the spectral entropy, which represents the degree of disorder in the currently acquired signal. value.

[0031] Specifically, firstly, the original high-frequency signal sequence Perform an FFT transform and normalize to obtain the spectral entropy value of the current signal. Then the optimal decomposition mode number is calculated. Calculate the optimal decomposition mode number The formula is: ; in, The number of optimal decomposition modes for the current signal is calculated, and it is taken as the closest integer in the actual calculation. This is the preset minimum number of decomposition layers, used to handle simple background noise; This is the preset maximum number of decomposition layers, used to process complex mixed signals; Signal at the current acquisition time The measured spectral entropy value; The statistical mean of the spectral entropy of pure background noise collected historically; This represents the statistical maximum value of the spectral entropy of typical broadband fault signals collected historically.

[0032] It is important to note that when calculating the optimal decomposition mode number... Previously, boundary constraints were required; specifically, if Then let ;like Then let This is to avoid errors in formula calculation.

[0033] To more clearly illustrate the optimal decomposition mode number The function and calculation process will be illustrated with examples below: First, let's assume the system is pre-configured. It is 3; The value is 8; the entropy value of pure background noise obtained from historical statistics. =0.5; typical fault signal entropy value The value is 2.5; if the signal acquired at the current moment is chaotic, the calculated value is... If it is 1.5; then: .

[0034] This means that the system determines the current signal is complex, automatically decomposes it into 7 modal components, and finally uses the calculated... right Variational mode decomposition is performed to obtain multiple mode components. The kurtosis value of each mode component is calculated, and components with kurtosis values ​​greater than a set threshold are retained. These components are then accumulated and reconstructed into a clean partial discharge signal sequence. .

[0035] Thus, by using adaptive variational mode decomposition and information entropy differences, physical separation of noise and features is achieved at the very front end of signal processing, solving the problem that traditional filters cannot handle broadband random noise.

[0036] S2. Process the pure partial discharge signal sequence into a partial discharge intensity sequence with the same sampling rate as the load current data. Construct data pairs within the sliding time window, perform probability integral transformation on the load current data and the partial discharge intensity sequence respectively, and map them to uniformly distributed data. Fit the uniformly distributed data to obtain the correlation parameters, and calculate the upper tail coupling drift coefficient based on the correlation parameters.

[0037] In an optional embodiment, the pure partial discharge signal sequence is first... By performing integration or envelope calculation, the relationship with the load current can be obtained. A partial discharge intensity sequence with a consistent sampling rate is used, and a sliding time window containing N consecutive sampling points is set. Data pairs are constructed within each window. .

[0038] In one specific implementation, to eliminate the influence of dimensions and adapt to the input requirements of the connection function, probability integral transformations are performed on the current data and partial discharge data within the window, mapping them to... The uniform distribution of intervals is denoted as and Since true insulation degradation manifests as high partial discharge under high load, meaning the two variables are strongly correlated in the upper tail of their distributions, the Gumbel connection function, which is sensitive to the upper tail, is chosen for modeling. The correlation parameters of the connection function are estimated from the current window of data using the maximum likelihood estimation method. Finally, the tail coupling drift coefficient is calculated. : ; in, This represents the upper tail coupling drift coefficient within the current time window, and its value range is... ; The correlation parameters obtained by fitting the Gumbel connection function are, in this invention, The range is .

[0039] To more clearly illustrate the upper tail coupling drift coefficient The function and calculation process will be illustrated with examples below: First, assume that the correlation parameters obtained by fitting the Gumbel connection function are... If it is 2, then This indicates a moderately strong positive correlation between the current load and partial discharge; If the correlation parameters obtained by fitting using the Gumbel connection function A value of 1 indicates that the two variables are completely independent. This also indicates that there is no correlation between the current load and the partial discharge.

[0040] Thus, through dynamic connection function analysis, this invention not only considers the signal strength but also the logical consistency between the signal and the load. Any external interference that is out of sync with load changes, such as mobile phone signals or lightning, will not cause [the problem]. The increase is suppressed through subsequent multiplication mechanisms, thus solving the false alarm problem.

[0041] S3. Calculate the root mean square value of the pure partial discharge signal sequence within the current sliding time window as the signal energy, and construct the comprehensive risk index of insulation degradation. The comprehensive risk index of insulation degradation is positively correlated with the signal energy and the upper tail coupling drift coefficient.

[0042] In an alternative embodiment, due to the single dependence on signal energy It is susceptible to false alarms due to strong interference, and relies solely on the upper tail coupling drift coefficient. The error is large when the signal is extremely weak, so this invention constructs a comprehensive risk index for insulation degradation. This is used to comprehensively evaluate the signal energy and the upper tail coupling drift coefficient. A comprehensive risk index for insulation degradation is constructed. The process is as follows: First calculate the current window content. The root mean square value is used as the signal energy. The comprehensive risk index of insulation degradation was then calculated using the following method. : ; in, The final output is the comprehensive risk index of insulation degradation, which is a dimensionless value. Reconstruct the signal within the current monitoring window The average energy value; The baseline background noise energy is the energy of the equipment when it is in a healthy state. It is a fixed constant greater than 0 to prevent abnormalities in the denominator and other calculations. These are preset coupling weight coefficients used to amplify the contribution of coupling features to the risk index; This represents the upper tail coupling drift coefficient.

[0043] To more clearly illustrate the role and calculation process of the comprehensive risk index for insulation degradation, the following example will demonstrate this: First, assume =1; It is 5; In one specific embodiment, the measurement at this time is It is 100, but A value of 0.05 indicates a large signal energy but a low correlation between the load current data and the partial discharge intensity sequence. .

[0044] In another specific embodiment, the measurement at this time is It is 10, but A value of 0.8 indicates that the signal energy is relatively low, but the correlation between the load current data and the partial discharge intensity sequence is relatively high. .

[0045] As can be seen from the two specific examples above, although the signal energy in the second example is only one-tenth of that in the first example, the comprehensive risk index of insulation degradation in the second example is more than twice that in the first example because there is a clear coupling characteristic between the load current data and the partial discharge intensity sequence. This indicates that the system has successfully identified and processed the fault and suppressed the interference.

[0046] Thus, only when the energy increases significantly and there is a clear load coupling between the load current data and the partial discharge intensity sequence will the product of the load current data and the partial discharge intensity sequence lead to the comprehensive risk index of insulation degradation. The temperature rises sharply, triggering an alarm.

[0047] S4. When the comprehensive risk index of insulation degradation is less than the preset yellow threshold, the equipment is judged to be in good operating condition. When the comprehensive risk index of insulation degradation is greater than or equal to the yellow threshold and less than the preset red threshold, it is judged to be an early latent fault and an early warning is issued. When the comprehensive risk index of insulation degradation is greater than or equal to the red threshold, it is judged to be severe insulation degradation and an alarm is triggered.

[0048] In an optional embodiment, the present invention classifies the comprehensive risk index of insulation degradation by pre-set yellow and red thresholds, thereby enabling the system to perform different levels of processing for different comprehensive risk indices of insulation degradation.

[0049] Specifically, when the comprehensive risk index of insulation degradation is less than the yellow threshold, the system is judged to be in healthy operation; When the comprehensive risk index of insulation degradation is not less than the yellow threshold and is less than the red threshold, the system determines it to be an early latent fault and issues an early warning signal. When the comprehensive risk index of insulation deterioration is not less than the red threshold, the system determines it as severe insulation deterioration and triggers a trip or emergency maintenance command.

[0050] Thus, in the early stages of insulation degradation, although the partial discharge energy... The growth may not be significant, but its sensitivity to load will be the first to drift. This solution can keenly detect this micro-change and provide an early warning before the equipment experiences thermal breakdown.

[0051] Reference Figure 2 The distribution of sample points representing normal operation is flat and disordered, showing no correlation, while the distribution of sample points representing insulation degradation shows an exponential increase with increasing current. The curve in the figure accurately fits this nonlinear growth trend, intuitively revealing the core mechanism of this invention for distinguishing faults using coupling characteristics.

[0052] Reference Figure 3 The data clusters representing the health status are clustered in the safe zone in the lower left corner of the coordinate system; the data clusters representing early latent faults mainly rise along the vertical axis and approach the warning trigger boundary line but do not cross the boundary, which confirms the ability of this invention to achieve early warning through the coupling coefficient before the energy bursts; the data clusters representing severe breakdown risk have both high energy and high coupling coefficient and are located in the upper right region of the boundary line.

[0053] This invention also discloses a power distribution cable branch box status monitoring device, comprising: a data acquisition module for acquiring load current data and electromagnetic pulse signals of the power distribution cable branch box; a signal reconstruction module for calculating the spectral entropy value of the electromagnetic pulse signal, adaptively calculating the optimal decomposition mode number based on the spectral entropy value, and performing variational mode decomposition and reconstruction on the electromagnetic pulse signal based on the optimal decomposition mode number to obtain a clean partial discharge signal sequence; a feature analysis module for performing probability integral transformation and connection function fitting on the load current data and the partial discharge intensity sequence obtained from the clean partial discharge signal sequence within a sliding time window to calculate the upper tail coupling drift coefficient; and a decision warning module for constructing a comprehensive insulation degradation risk index based on the energy of the clean partial discharge signal sequence and the upper tail coupling drift coefficient, and outputting a status judgment result by comparing it with a preset threshold.

[0054] In the description of this specification, "multiple" or "several" means at least two, such as two, three or more, unless otherwise expressly and specifically defined.

Claims

1. A method for monitoring the status of a power distribution cable branch box, characterized in that, include: The load current data and electromagnetic pulse signal at the cable branch box are collected to obtain the spectral entropy value of the current signal and calculate the optimal decomposition mode number. Variational mode decomposition is performed on the electromagnetic pulse signal, and a clean partial discharge signal sequence is reconstructed based on the kurtosis values ​​of the decomposed mode components. The pure partial discharge signal sequence is processed into a partial discharge intensity sequence with the same sampling rate as the load current data. Data pairs are constructed within a sliding time window. The load current data and the partial discharge intensity sequence are subjected to probability integral transformation and mapped to uniformly distributed data. The uniformly distributed data is fitted to obtain the correlation parameters. The upper tail coupling drift coefficient is calculated based on the correlation parameters. The root mean square value of the clean partial discharge signal sequence within the current sliding time window is calculated as the signal energy, and a comprehensive insulation degradation risk index is constructed. The comprehensive insulation degradation risk index is positively correlated with the signal energy and the upper tail coupling drift coefficient. When the comprehensive risk index of insulation degradation is less than the preset yellow threshold, the equipment is determined to be in good working order. When the comprehensive risk index of insulation degradation is greater than or equal to the yellow threshold and less than the preset red threshold, it is determined to be an early latent fault and an early warning is issued. When the comprehensive risk index of insulation degradation is greater than or equal to the red threshold, it is determined to be severe insulation degradation and an alarm is triggered.

2. The method for monitoring the status of a power distribution cable branch box according to claim 1, characterized in that, The optimal number of decomposition modes satisfies the following relationship: ； In the formula, The nearest integer is taken as the optimal number of decomposed modes for the current signal. This is the preset minimum number of decomposition levels; This is the preset maximum number of decomposition levels; This represents the measured spectral entropy value of the electromagnetic pulse signal at the current acquisition moment. The statistical mean of the spectral entropy of pure background noise collected historically; The maximum statistical value of the spectral entropy of typical broadband fault signals collected historically; and when season ,when season .

3. The method for monitoring the status of a power distribution cable branch box according to claim 1, characterized in that, The clean partial discharge signal sequence reconstructed from the kurtosis values ​​of the decomposed modal components includes: Calculate the kurtosis value of each mode component obtained from variational mode decomposition; Determine whether the kurtosis value is greater than the set kurtosis threshold; The modal components with kurtosis values ​​greater than the kurtosis threshold are accumulated to obtain the reconstructed pure partial discharge signal sequence.

4. The method for monitoring the status of a power distribution cable branch box according to claim 1, characterized in that, The partial discharge intensity sequence is processed to match the sampling rate of the load current data. Data pairs are constructed within a sliding time window, and probability integral transformations are performed on the load current data and the partial discharge intensity sequence, respectively, including: The edge distributions of the load current data and the partial discharge intensity sequence are calculated using empirical distribution functions, respectively. The calculated edge distribution values ​​are used as the transformed uniform distribution data, wherein the load current data corresponds to the first uniform distribution variable and the partial discharge intensity sequence corresponds to the second uniform distribution variable.

5. The method for monitoring the status of a power distribution cable branch box according to claim 1, characterized in that, The fitting of uniformly distributed data to obtain correlation parameters includes: Construct the density function expression for the Gumbel connection function; Substitute the uniformly distributed data into the density function expression to establish the log-likelihood function; The maximum value of the log-likelihood function is obtained by using the maximum likelihood estimation method, and the corresponding parameter estimate is used as the correlation parameter.

6. The method for monitoring the status of a power distribution cable branch box according to claim 5, characterized in that, The upper tail coupling drift coefficient satisfies the following relationship: ； In the formula, This represents the upper tail coupling drift coefficient within the current time window; These are the correlation parameters obtained by fitting using the Gumbel connection function.

7. The method for monitoring the status of a power distribution cable branch box according to claim 6, characterized in that, The comprehensive risk index of insulation degradation satisfies the following relationship: ； In the formula, This is a comprehensive risk index for insulation degradation. This is the root mean square value of the reconstructed clean partial discharge signal sequence within the current time window; The baseline background noise energy when the equipment is in a healthy state; These are the preset coupling weight coefficients; This represents the upper tail coupling drift coefficient.

8. The method for monitoring the status of a power distribution cable branch box according to claim 1, characterized in that, The load current data is acquired through a Rogowski coil, and the electromagnetic pulse signal is acquired through an ultra-high frequency antenna mounted on the inner wall of the branch box.

9. The method for monitoring the status of a power distribution cable branch box according to claim 1, characterized in that, The determination logic for the comprehensive risk index of insulation degradation includes: Before calculating the comprehensive risk index of insulation degradation, the root mean square algorithm is used to calculate the energy of the pure partial discharge signal sequence. After determining that the insulation is severely deteriorated and triggering an alarm, a trip command or emergency maintenance command is generated.

10. A power distribution cable branch box status monitoring device, the monitoring device being configured to perform a power distribution cable branch box status monitoring method as described in any one of claims 1 to 9, characterized in that, The monitoring device includes: The data acquisition module is used to acquire load current data and electromagnetic pulse signals from the power distribution cable branch box; The signal reconstruction module is used to calculate the spectral entropy value of the electromagnetic pulse signal, adaptively calculate the optimal decomposition mode number based on the spectral entropy value, and perform variational mode decomposition and reconstruction on the electromagnetic pulse signal based on the optimal decomposition mode number to obtain a clean partial discharge signal sequence. The feature analysis module is used to perform probability integral transformation and connection function fitting on the load current data and the partial discharge intensity sequence obtained from the pure partial discharge signal sequence within a sliding time window, so as to calculate the upper tail coupling drift coefficient. The decision-making and early warning module is used to construct a comprehensive risk index for insulation degradation based on the energy of the pure partial discharge signal sequence and the upper tail coupling drift coefficient, and output the state judgment result by comparing it with a preset threshold.

Images