Cable terminal potential breakdown fault early warning method based on space charge measurement

Abstract

The invention relates to the technical field of cable-based breakdown early warning, and discloses a cable terminal potential breakdown fault early warning method based on space charge measurement, and the method comprises the following steps: building a space charge test platform based on an electro-acoustic pulse method; designing a cable terminal aging test and judging a breakdown threshold value; s2, testing a to-be-detected cable terminal by adopting the space charge test platform built in S1, and calculating a degradation risk index; if yes, it is judged that the to-be-detected cable terminal enters the aging later stage, and a potential breakdown fault early warning signal is sent to an operation and maintenance department. Through a technical chain of multi-physics field coupling quantization, adaptive clustering staging and probability type dynamic threshold, manual subjective errors are eliminated, evaluation standards are ensured, and the accuracy of breakdown early warning is ensured.

Description

Technical Field

[0001] This invention relates to the field of cable base breakdown early warning technology, and more specifically, to a method for early warning of potential breakdown faults in cable terminals based on space charge measurement. Background Technology

[0002] In high-voltage power systems, cable terminals are the core connection nodes between cables and power equipment such as transformers and switches. They serve three major functions: insulation, electric field homogenization, and power transmission. Their operational reliability directly determines the safety and stability of the power system. Cable terminals are subjected to complex stress environments over long periods: on the one hand, the Joule heat generated by the load current and the electric field formed by the grid voltage constitute electrothermal composite stress, accelerating the molecular chain breakage and performance degradation of the insulation layer; on the other hand, environmental stresses such as humidity, pollution, and temperature fluctuations further exacerbate the evolution of insulation defects.

[0003] The stress cone region is a weak point in cable terminations. While the electric field distribution can be controlled through structural design, the interface between the semiconducting and insulating layers in this region is prone to microscopic defects due to electrothermal aging, leading to the accumulation of space charge. Long-term accumulation of space charge not only causes local electric field distortion but also exacerbates insulation molecule degradation through charge injection into the de-trapping process, ultimately inducing breakdown faults and causing large-scale power outages.

[0004] Existing technologies suffer from the following shortcomings: Multiphysics coupling effects are ignored, leading to inaccurate quantification of degradation risk. Traditional space charge testing only calculates the total charge amount, failing to consider the radial electric field gradient, axial electric field attenuation characteristics, and circumferential structural differences within the stress cone region. The degradation hazard of the same charge amount differs significantly between high and low field strength regions, resulting in the total charge characterization failing to accurately reflect the actual risk. Furthermore, existing technologies largely rely on manual experience, such as aging time and charge density thresholds, lacking data-driven adaptive algorithms. This leads to inconsistent judgments from different personnel, hindering the formation of a unified evaluation standard. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method for early warning of potential breakdown faults in cable terminals based on space charge measurement. This method overcomes the shortcomings of existing technologies by employing a technical chain of multi-physics field coupling quantization, adaptive clustering in stages, and probabilistic dynamic thresholds.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows:

[0007] A method for early warning of potential breakdown faults in cable terminals based on space charge measurement includes the following steps:

[0008] S1. A space charge testing platform was built based on the electroacoustic pulse method;

[0009] S2. Design cable termination aging tests and determine breakdown thresholds. ;

[0010] S3. Obtain the space charge data of the cable terminal under test: Using the space charge testing platform built according to S1, test the cable terminal under test and calculate the degradation risk index. ;

[0011] S4. Fault Warning Judgment: [The following is a separate, unrelated sentence:] and In comparison, if If the cable terminal under test is determined to be in the late stage of aging, a potential breakdown fault warning signal is sent to the operation and maintenance department.

[0012] As a preferred embodiment of the present invention, the space charge testing platform described in S1 includes a high-voltage AC power supply, a high-voltage pulse power supply, an AC matching circuit, a portable space charge sensor, an amplifier, and a data acquisition and processing system. The AC matching circuit synchronizes the charge testing time with the AC phase, applies a high-voltage pulse to the cable terminal to excite the space charge vibration, the sensor captures the sound wave and converts it into an electrical signal, and then amplifies it and processes it through the amplifier and the data acquisition and processing system to obtain the three-dimensional distribution data of space charge at the installation position of the stress cone of the cable terminal.

[0013] As a preferred embodiment of the present invention, the portable space charge sensor in S1 includes a ring buckle and a detection element. The detection element consists of three parts: a measuring electrode, a piezoelectric film, and a sensor. The measuring electrode is on the top layer, the sensor is on the bottom layer, and the piezoelectric film is sandwiched between the two. The area of ​​the piezoelectric film is the same as the area of ​​the inner aluminum plate of the sensor and it only contacts the inner aluminum plate. The measuring electrode and the sensor are fixed together by screws. The contact surface of the measuring electrode has an arc-shaped structure. The sensor support structure is composed of four nested materials: the inner layer is polymethyl methacrylate, the second layer from the inside out is aluminum plate, the third layer is polytetrafluoroethylene, and the outermost layer is aluminum plate.

[0014] As a preferred embodiment of the present invention, step S2 includes the following steps:

[0015] S2.1 Sample preparation: Select several complete cable terminals of the same specification, containing stress cone structure and with cross-linked polyethylene as insulation layer as test samples. At the same time, select new cable terminals of the same specification that have not been aged as blank control samples. Clean the surface of all samples. Prepare an accelerated aging device with stable temperature control and AC voltage output.

[0016] S2.2 Initial test of the control sample: Using the space charge test platform in step S1, test several positions along the axial direction of the blank control sample, record the space charge distribution waveform, and confirm that there is no space charge accumulation in the unaged state.

[0017] S2.3 Accelerated aging test: Place the test sample in the accelerated aging device, set the preset aging temperature and AC voltage parameters for accelerated aging, stop the aging at least three preset time nodes, and take out a number of samples each time to cool to room temperature.

[0018] S2.4, Space charge testing and multi-dimensional quantification of aging samples;

[0019] S2.5. Aging stage division and dynamic threshold determination based on the modified DBSCAN algorithm.

[0020] As a preferred embodiment of the present invention, the number of test specimens described in S2.1 Number of roots, blank control samples For root cleaning, wipe the surface with isopropyl alcohol and then let it air dry.

[0021] As a preferred embodiment of the present invention, the preset aging temperature in S2.3 is... AC voltage is The preset time interval for nodes is sky.

[0022] As a preferred embodiment of the present invention, S2.4 further includes:

[0023] S2.4.1. Using the testing platform of step S1, obtain the three-dimensional distribution data of space charge in the stress cone region of the test specimen, define a three-dimensional coordinate system, and introduce an electric field coupling model:

[0024] radial : (Radius of the inner semiconducting layer) (Outer semiconducting layer radius), electric field strength ,in The effective value of the applied voltage;

[0025] Axial The origin is the inner edge of the stress cone. The left side (outer side) is The right side (inner side) is axial coverage area of ​​stress cone electric field strength Pre-obtained through finite element simulation to meet the requirements ,in This is the attenuation coefficient, which is related to the stress cone structure;

[0026] Zhou Xiang : Define the circumferential weight coefficient ,when When in the area covered by aluminum foil The rest of the area ;

[0027] Space charge density matrix: ,in , , These represent the number of discrete grids in the radial, axial, and circumferential directions, respectively. Corresponding grid ; charge density;

[0028] S2.4.2 Introduce the space charge degradation risk index algorithm, the formula is as follows:

[0029]

[0030] in:

[0031] This is the risk index for degradation.

[0032] For grid The resultant electric field strength at the location;

[0033] The aging sensitivity coefficient, ,in , These are the fitting coefficients. , , For aging time, The radius of the electric field peak is... ;

[0034] For the mesh volume element, ;

[0035] S2.4.3 Calculate the values ​​of all samples at the same node. The values ​​are then subjected to a Box-Cox transformation to eliminate data skewness, resulting in a standardized risk index. ,in , These represent the mean and standard deviation of the samples in the same batch.

[0036] As a preferred embodiment of the present invention, S2.5 includes the following specific steps:

[0037] S2.5.1 Constructing the Feature Matrix ,in The rate of change of the risk index. For cumulative degradation;

[0038] S2.5.2. Clustering using the improved DBSCAN algorithm, defining a density threshold. The core point judgment condition: in The neighborhood contains There are 10 samples, and the similarity between samples is calculated using cosine distance:

[0039]

[0040] Clustering yields sample sets for three aging stages: early aging, middle aging, and late aging.

[0041] S2.5.3 For the late-stage aging set, the breakdown probability function is calculated using a survival analysis model:

[0042]

[0043] in Let be the risk rate function. As the benchmark risk rate, For regression coefficients; take the breakdown probability. corresponding value as critical value The actual breakdown threshold is obtained by combining the Box-Cox inverse transform: .

[0044] As a preferred embodiment of the present invention, the finite element simulation described in S2.4.1 uses COMSOL Multiphysics software, and the electric field calculation error is... Radial grid number axial grid number Circumferential grid number ;

[0045] The fitting coefficients described in S2.4.2 The data was obtained through least squares fitting, specifically: for the sample data in the initial aging stage, using... The objective is to minimize the sum of squared residuals between measured and theoretical values. The formula for the sum of squared residuals is: , The initial sample size; the formula for the Box-Cox transformation described in S2.4.3 is:

[0046]

[0047] in The transformation parameters are determined using the maximum likelihood estimation method. The maximum likelihood function is , This is the mean of the transformed samples.

[0048] As a preferred embodiment of the present invention, the risk index change rate described in S2.5.1 Calculated using the central difference method:

[0049]

[0050] in The time interval between adjacent aging duration nodes. , They are respectively Standardized risk index of adjacent nodes;

[0051] The clustering process of the improved DBSCAN algorithm described in S2.5.2 includes:

[0052] Randomly select unlabeled sample points Calculate its cosine distance to all other sample points. Statistical satisfaction The number of samples; if the number of samples Then mark Using this as the core, an initial cluster is established. ;

[0053] Traversing the initial clusters All core points, including each core point Unlabeled samples in the neighborhood are added This step is repeated recursively until no new samples can be added.

[0054] All isolated samples that are not core points are marked as noise points and removed.

[0055] The survival analysis model described in S2.5.3 uses the Kaplan-Meier estimation method to calculate the baseline hazard rate. The estimation formula is: , For the first Number of breakdown samples at each time point For the first The number of surviving samples at each time point; the regression coefficients are obtained using Cox proportional hazards regression. The regression model is The significance level was set at 1. The effectiveness of the model was verified by the likelihood ratio test.

[0056] The beneficial technical effects of this invention are:

[0057] Multi-dimensional coupling quantification significantly improves the accuracy of degradation risk characterization. By introducing an electric field coupling model, circumferential weight coefficient, and aging sensitivity coefficient, a degradation risk index formula is constructed, which comprehensively reflects the coupling effect of charge density, electric field strength, structural differences, and aging time. Compared with the traditional total integral method, the degradation risk characterization error is reduced, and high-risk areas of stress cones can be accurately located, providing a reliable quantitative basis for subsequent phased and threshold setting.

[0058] Adaptive clustering in stages eliminates subjective human error and is based on feature matrices. The improved DBSCAN algorithm can automatically divide the aging stage without pre-setting the number of categories, solving the problem of division bias caused by human experience and forming a unified aging stage evaluation standard.

[0059] The statistical significance of the threshold was optimized through a survival analysis model: the Kaplan-Meier method was used to estimate the baseline hazard rate, and the regression coefficients were solved using Cox proportional hazards regression. Ultimately, the breakdown probability is taken. The corresponding critical value is obtained by inverse Box-Cox transformation. Compared with the traditional fixed threshold, this dynamic threshold can reduce the false alarm rate and the missed alarm rate, which can avoid the waste of resources caused by excessive maintenance and prevent safety accidents caused by missed faults.

[0060] The portable space charge sensor features a lightweight design, making it easy to transport to cable operation sites. The space charge testing platform is easy to set up and meets the measurement needs of complex scenarios such as underground cable wells and outdoor substations. Based on the accurate on-site measurement of the space charge testing platform, this invention provides a reliable basis for predicting insulation aging and troubleshooting potential faults, effectively reducing the risk of power outages caused by cable terminal breakdown and ensuring the safe and stable operation of the power system. Attached Figure Description

[0061] Figure 1 This is a flowchart of the method of the present invention;

[0062] Figure 2 This is a schematic diagram of the measurement system of the present invention;

[0063] Figure 3 This is a schematic diagram of the structure of the portable space charge sensor in this invention. Detailed Implementation

[0064] In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0065] Combination Figure 1 - Figure 3 The present invention provides the following embodiments:

[0066] A method for early warning of potential breakdown faults in cable terminals based on space charge measurement includes the following steps:

[0067] S1. A space charge testing platform was built based on the electroacoustic pulse method;

[0068] S2. Design cable termination aging tests and determine breakdown thresholds. ;

[0069] S3. Obtain space charge data of the cable terminal under test: Using the space charge testing platform built in S1, test the cable terminal under test and calculate the degradation risk index. ;

[0070] S4. Fault Warning Judgment: [The following is a separate, unrelated sentence:] and In comparison, if If the cable terminal under test is determined to be in the late stage of aging, a potential breakdown fault warning signal is sent to the operation and maintenance department.

[0071] Furthermore, the space charge testing platform described in S1 includes a high-voltage AC power supply, a high-voltage pulse power supply, an AC matching circuit, a portable space charge sensor, an amplifier, and a data acquisition and processing system. The AC matching circuit synchronizes the charge testing time and AC phase, applying a high-voltage pulse to the cable terminal to excite the space charge vibration. The sensor captures the sound wave and converts it into an electrical signal, which is then amplified by the amplifier and processed by the data acquisition and processing system to obtain the three-dimensional distribution data of the space charge at the stress cone installation location of the cable terminal. This constructs a complete hardware path, synchronizing the testing timing and AC phase to ensure efficient excitation of charge vibration by the high-voltage pulse, while avoiding attenuation of the weak signal transmission after acoustic-to-electric conversion.

[0072] Furthermore, the portable space charge sensor described in S1 includes a ring-shaped clip and a detection element. The detection element comprises three parts: a measuring electrode, a piezoelectric film, and a sensor. The measuring electrode is on the top layer, the sensor is on the bottom layer, and the piezoelectric film is sandwiched between them. The area of ​​the piezoelectric film is the same as the area of ​​the inner aluminum plate of the sensor and it only contacts the inner aluminum plate. The measuring electrode and the sensor are fixed together by screws. The contact surface of the measuring electrode has an arc-shaped structure. The sensor support structure is composed of four nested materials: the inner layer is polymethyl methacrylate, the second layer from the inside out is aluminum plate, the third layer is polytetrafluoroethylene, and the outermost layer is aluminum plate. The arc-shaped contact surface adapts to the terminal shape and reduces acoustic signal loss; the multi-layer support isolates electromagnetic interference and ensures signal purity; the overall lightweight design combined with the ring-shaped clip enables convenient on-site deployment, facilitating on-site setup and measurement.

[0073] Furthermore, S2 includes the following steps:

[0074] S2.1 Sample preparation: Select several complete cable terminals of the same specification, containing stress cone structure and with cross-linked polyethylene as insulation layer as test samples. At the same time, select new cable terminals of the same specification that have not been aged as blank control samples. Clean the surface of all samples. Prepare an accelerated aging device with stable temperature control and AC voltage output.

[0075] S2.2 Initial test of the control sample: According to the space charge test platform built in S1, several positions along the axial direction of the blank control sample are selected for testing, and the space charge distribution waveform is recorded to confirm that there is no space charge accumulation in the unaged state.

[0076] S2.3 Accelerated aging test: Place the test sample in the accelerated aging device, set the preset aging temperature and AC voltage parameters for accelerated aging, stop the aging at least three preset time nodes, and take out a number of samples each time to cool to room temperature.

[0077] S2.4, Space charge testing and multi-dimensional quantification of aging samples;

[0078] S2.5. Aging stage division and dynamic threshold determination based on the modified DBSCAN algorithm.

[0079] The design of adding blank control to the test sample can eliminate charge accumulation caused by non-aging factors, such as impurities at the interface of the semiconductive layer and microbubbles in the insulation during the manufacturing process. This clarifies that the generation of charge in subsequent tests is dominated only by electrothermal aging. The initial test results of the blank control sample provide a benchmark for the aging test, avoiding misjudging manufacturing defects as aging defects.

[0080] The temperature and pressure control functions of the accelerated aging device can simulate the electrothermal composite stress in the actual operation of cable terminals, and the accelerated aging parameters can be precisely adjusted to ensure that the aging mechanism is consistent with the actual operation.

[0081] The multi-time-term, multi-stage testing method can capture the dynamic evolution of space charge during the aging process, avoiding the problem that a single test cannot reflect the characteristics of the aging stage; the sample is cooled to room temperature for testing, which can eliminate the influence of temperature on the sensitivity of the piezoelectric film and ensure the comparability of test data at different aging stages.

[0082] Furthermore, the number of test specimens described in S2.1 Number of roots, blank control samples For root cleaning, wipe the surface with isopropyl alcohol and then let it air dry.

[0083] It can reduce test bias caused by individual differences, and statistical reliability is significantly improved by calculating the average charge data of samples at the same node.

[0084] Furthermore, the preset aging temperature mentioned in S2.3 is... AC voltage is The preset time interval for nodes is sky.

[0085] The Joule heat generated by the load current during cable termination operation typically reaches a temperature of 60°C. 90℃, 90 An accelerated temperature of 110℃ can shorten the test cycle and the actual aging time without changing the aging mechanism of XLPE insulation.

[0086] Furthermore, S2.4 also includes:

[0087] S2.4.1. Using the testing platform of step S1, obtain the three-dimensional distribution data of space charge in the stress cone region of the test specimen, define a three-dimensional coordinate system, and introduce an electric field coupling model:

[0088] radial : (Radius of the inner semiconducting layer) (Outer semiconducting layer radius), electric field strength ,in The effective value of the applied voltage;

[0089] Axial The origin is the inner edge of the stress cone. The left side (outer side) is The right side (inner side) is axial coverage area of ​​stress cone electric field strength Pre-obtained through finite element simulation to meet the requirements ,in This is the attenuation coefficient, which is related to the stress cone structure;

[0090] Zhou Xiang : Define the circumferential weight coefficient ,when When in the area covered by aluminum foil The rest of the area ;

[0091] Space charge density matrix: ,in , , These represent the number of discrete grids in the radial, axial, and circumferential directions, respectively. Corresponding grid ; charge density;

[0092] S2.4.2 Introduce the space charge degradation risk index algorithm, the formula is as follows:

[0093]

[0094] in:

[0095] This is the degradation risk index, and its unit is: ;

[0096] For grid The resultant electric field strength at the location;

[0097] The aging sensitivity coefficient, ,in , These are the fitting coefficients. , , For aging time, The radius of the electric field peak is... ;

[0098] For the mesh volume element, ;

[0099] S2.4.3 Calculate the values ​​of all samples at the same node. The values ​​are then subjected to a Box-Cox transformation to eliminate data skewness, resulting in a standardized risk index. ,in , These represent the mean and standard deviation of the samples in the same batch.

[0100] Overcoming the limitations of traditional methods that only quantify the total charge amount, this method can distinguish the charge hazard between the high-field-strength region near the outer semiconducting layer and the low-field-strength region on the inner side. The same amount of charge in the high-field-strength region leads to more severe electric field distortion. By coupling the electric field, the actual degradation contribution of the charge can be accurately quantified. (Early aging stage) The values ​​are skewed, with most samples... The data is relatively small, but a few are significantly different due to individual variations. After transformation, the data conforms to a normal distribution, making... This improves comparability and provides a normalized data source for subsequent cluster analysis.

[0101] Furthermore, the finite element simulation described in S2.4.1 uses COMSOL Multiphysics software, and the electric field calculation error... Radial grid number axial grid number Circumferential grid number ;

[0102] The fitting coefficients described in S2.4.2 The data was obtained through least squares fitting, specifically: for the sample data in the initial aging stage, using... The objective is to minimize the sum of squared residuals between measured and theoretical values. The formula for the sum of squared residuals is: , The initial sample size; the formula for the Box-Cox transformation described in S2.4.3 is:

[0103]

[0104] in The transformation parameters are determined using the maximum likelihood estimation method. The maximum likelihood function is , This is the mean of the transformed samples.

[0105] Furthermore, S2.5 includes the following specific steps:

[0106] S2.5.1 Constructing the Feature Matrix ,in The rate of change of the risk index. For cumulative degradation;

[0107] S2.5.2. Clustering using the improved DBSCAN algorithm, defining a density threshold. The core point judgment condition: in The neighborhood contains There are 10 samples, and the similarity between samples is calculated using cosine distance:

[0108]

[0109] Clustering yields sample sets for three aging stages: early aging, middle aging, and late aging.

[0110] S2.5.3 For the late-stage aging set, the breakdown probability function is calculated using a survival analysis model:

[0111]

[0112] in Let be the risk rate function. As the benchmark risk rate, For regression coefficients; take the breakdown probability. corresponding value as critical value The actual breakdown threshold is obtained by combining the Box-Cox inverse transform: .

[0113] Advantages of the improved DBSCAN algorithm:

[0114] No need to preset the number of clusters, just use a density threshold. The core point is automatically divided into stages, avoiding the subjectivity of manually setting the K value;

[0115] Cosine distance measures similarity: feature matrix The dimensions of each dimension are different. Dimensionless With a value of 1 / day, cosine distance can eliminate the influence of dimensions, thus improving the consistency of clustering results compared to Euclidean distance;

[0116] Noise points are removed, and the number of neighboring samples is less than 5: Noise points are mostly test outliers, and removing them can make the purity of samples at each stage ≥95%.

[0117] Furthermore, the rate of change of the risk index described in S2.5.1 Calculated using the central difference method:

[0118]

[0119] in The time interval between adjacent aging duration nodes. , They are respectively Standardized risk index of adjacent nodes;

[0120] The clustering process of the improved DBSCAN algorithm described in S2.5.2 includes:

[0121] Randomly select unlabeled sample points Calculate its cosine distance to all other sample points. Statistical satisfaction The number of samples; if the number of samples Then mark Using this as the core, an initial cluster is established. ;

[0122] Traversing the initial clusters All core points, including each core point Unlabeled samples in the neighborhood are added This step is repeated recursively until no new samples can be added.

[0123] For all isolated samples that are not in the core, the number of samples in the neighborhood Mark them as noise points and remove them;

[0124] The survival analysis model described in S2.5.3 uses the Kaplan-Meier estimation method to calculate the baseline hazard rate. The estimation formula is: , For the first Number of breakdown samples at each time point For the first The number of surviving samples at each time point; the regression coefficients are obtained using Cox proportional hazards regression. The regression model is The significance level was set at 1. The effectiveness of the model was verified by the likelihood ratio test.

[0125] The calculation error of the central difference method is reduced, ensuring the accuracy of the aging rate feature in the feature matrix; the improved DBSCAN clustering process of random point selection and recursive merging of core points can avoid cluster splitting caused by initial point selection bias, making the clustering results highly repeatable.

[0126] Applications in operation and maintenance scenarios:

[0127] Before use, aging tests and breakdown thresholds were conducted. Three new, non-operating cable terminals of the same model were selected as blank control samples, and 15 retired terminals of the same model, which had been in operation for 8 years and confirmed to be aged, were used as test samples for accelerated aging tests.

[0128] Aging parameters: temperature 100℃, simulating summer high temperature and load temperature rise, AC voltage 26kV, aging nodes 75 days, 150 days, 225 days, 300 days, 375 days;

[0129] Quantization calculation: Calculating the values ​​of each node Value (deterioration risk index), Box-Cox transformation ( =0.4, after the maximum likelihood estimate is determined, the standardized risk index is obtained. Construct the feature matrix

[0130] ;

[0131] Clustering and Thresholding: An Improved DBSCAN Algorithm for Clustering =0.5, core sample size ≥5), dividing the late-aging sample set; survival analysis model calculated corresponding , and Ultimately based on Calculated .

[0132] At the operation and maintenance site (such as a substation), a platform is built according to the electroacoustic pulse method. The high-voltage AC power supply provides an AC electric field environment for the test sample through the AC matching circuit. The AC matching circuit undertakes the function of electrical signal matching to realize the synchronization of test timing and AC phase. The high-voltage pulse power supply applies a high-voltage pulse to the test sample to excite space charge vibration. The sensor is fixed to the stress cone area of ​​the cable terminal by a ring buckle. The space charge sensor, amplifier, and data acquisition and processing system are electrically connected in sequence to form a signal path.

[0133] Fifteen cable terminals at the maintenance site were tested one by one. A space charge testing platform was used to acquire three-dimensional charge distribution data in the stress cone region, and calculations were performed. , and then and In contrast, the early warning terminal was shut down for maintenance.

[0134] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for early warning of potential breakdown faults in cable terminations based on space charge measurements, characterized in that, Includes the following steps: S1. A space charge testing platform was built based on the electroacoustic pulse method; S2, design cable terminal aging test and determine breakdown threshold ; S3, obtaining the space charge data of the cable terminal to be detected: using the space charge test platform built according to S1, testing the cable terminal to be detected, and calculating the deterioration risk index ; S4, fault early warning determination: the following With In contrast, if , it is determined that the cable terminal to be detected enters the late aging stage, and a potential breakdown fault early warning signal is sent to the operation and maintenance department.

2. The method of claim 1, wherein the method is characterized by: The space charge testing platform described in S1 includes a high-voltage AC power supply, a high-voltage pulse power supply, an AC matching circuit, a portable space charge sensor, an amplifier, and a data acquisition and processing system. The AC matching circuit synchronizes the charge testing time with the AC phase, applies a high-voltage pulse to the cable terminal to excite the space charge vibration, captures the sound wave and converts it into an electrical signal, and then amplifies it and processes it through the amplifier and the data acquisition and processing system to obtain the three-dimensional distribution data of space charge at the installation position of the stress cone of the cable terminal.

3. The method for early warning of potential breakdown faults in cable terminals based on space charge measurement according to claim 2, characterized in that, The portable space charge sensor described in S1 includes a ring-shaped buckle and a detection element. The detection element consists of three parts: a measuring electrode, a piezoelectric film, and a sensor. The measuring electrode is on the top layer, the sensor is on the bottom layer, and the piezoelectric film is sandwiched between the two. The area of ​​the piezoelectric film is the same as the area of ​​the inner aluminum plate of the sensor and it only contacts the inner aluminum plate. The measuring electrode and the sensor are fixed together by screws. The contact surface of the measuring electrode has an arc-shaped structure. The sensor support structure is composed of four nested materials: the inner layer is polymethyl methacrylate, the second layer from the inside out is aluminum plate, the third layer is polytetrafluoroethylene, and the outermost layer is aluminum plate.

4. The method for early warning of potential breakdown faults in cable terminals based on space charge measurement according to claim 1, characterized in that, S2 includes the following steps: S2.1 Sample preparation: Select several complete cable terminals of the same specification, containing stress cone structure and with cross-linked polyethylene as insulation layer as test samples. At the same time, select new cable terminals of the same specification that have not been aged as blank control samples. Clean the surface of all samples. Prepare an accelerated aging device with stable temperature control and AC voltage output. S2.2 Initial test of the control sample: Using the space charge test platform built in step S1, several positions along the axial direction of the blank control sample were selected for testing, and the space charge distribution waveform was recorded to confirm that there was no space charge accumulation in the unaged state. S2.3 Accelerated aging test: Place the test sample in the accelerated aging device, set the preset aging temperature and AC voltage parameters for accelerated aging, stop the aging at least three preset time nodes, and take out a number of samples each time to cool to room temperature. S2.4, Space charge testing and multi-dimensional quantification of aging samples; S2.

5. Aging stage division and dynamic threshold determination based on the modified DBSCAN algorithm.

5. The method for early warning of potential breakdown faults in cable terminals based on space charge measurement according to claim 4, characterized in that, The number of test specimens described in S2.1 Number of roots, blank control samples For root cleaning, wipe the surface with isopropyl alcohol and then let it air dry.

6. The method for early warning of potential breakdown faults in cable terminals based on space charge measurement according to claim 4, characterized in that, The preset aging temperature mentioned in S2.3 is AC voltage is The preset time interval for nodes is sky.

7. The method for early warning of potential breakdown faults in cable terminals based on space charge measurement according to claim 4, characterized in that, S2.4 also includes: S2.4.

1. Using the testing platform of step S1, obtain the three-dimensional distribution data of space charge in the stress cone region of the test specimen, define a three-dimensional coordinate system, and introduce an electric field coupling model: radial : (Radius of the inner semiconducting layer) (Outer semiconducting layer radius), electric field strength ,in The effective value of the applied voltage; Axial The origin is the inner edge of the stress cone. The left side (outer side) is The right side (inner side) is axial coverage area of ​​stress cone electric field strength Pre-obtained through finite element simulation to meet the requirements ,in This is the attenuation coefficient, which is related to the stress cone structure; Zhou Xiang : Define the circumferential weight coefficient ,when When in the area covered by aluminum foil The rest of the area ; Space charge density matrix: ,in , , These represent the number of discrete grids in the radial, axial, and circumferential directions, respectively. Corresponding grid ; charge density; S2.4.2 Introduce the space charge degradation risk index algorithm, the formula is as follows: in: This is the risk index for degradation. For grid The resultant electric field strength at the location; The aging sensitivity coefficient, ,in , These are the fitting coefficients. , , For aging time, The radius of the electric field peak is... ; For the mesh volume element, ; S2.4.3 Calculate the values ​​of all samples at the same node. The values ​​are then subjected to a Box-Cox transformation to eliminate data skewness, resulting in a standardized risk index. ,in , These represent the mean and standard deviation of the samples in the same batch.

8. The method for early warning of potential breakdown faults in cable terminals based on space charge measurement according to claim 4, characterized in that, S2.5 includes the following specific steps: S2.5.1 Constructing the Feature Matrix ,in The rate of change of the risk index. For cumulative degradation; S2.5.

2. Clustering using the improved DBSCAN algorithm, defining a density threshold. The core point judgment condition: in The neighborhood contains There are 10 samples, and the similarity between samples is calculated using cosine distance: Clustering yields sample sets for three aging stages: early aging, middle aging, and late aging. S2.5.3 For the late-stage aging set, the breakdown probability function is calculated using a survival analysis model: in Let be the risk rate function. As the benchmark risk rate, For regression coefficients; take the breakdown probability. corresponding value as critical value The actual breakdown threshold is obtained by combining the Box-Cox inverse transform: .

9. The method for early warning of potential breakdown faults in cable terminals based on space charge measurement according to claim 7, characterized in that, The finite element simulation described in S2.4.1 uses COMSOL Multiphysics software, and the electric field calculation error... Radial grid number axial grid number Circumferential grid number ; The fitting coefficients described in S2.4.2 The data was obtained through least squares fitting, specifically: for the sample data in the initial aging stage, using... The objective is to minimize the sum of squared residuals between measured and theoretical values. The formula for the sum of squared residuals is: , The initial sample size; the formula for the Box-Cox transformation described in S2.4.3 is: in The transformation parameters are determined using the maximum likelihood estimation method. The maximum likelihood function is , This is the mean of the transformed samples.

10. The method for early warning of potential breakdown faults in cable terminals based on space charge measurement according to claim 8, characterized in that, The rate of change of the risk index described in S2.5.1 Calculated using the central difference method: in The time interval between adjacent aging duration nodes. , They are respectively Standardized risk index of adjacent nodes; The clustering process of the improved DBSCAN algorithm described in S2.5.2 includes: Randomly select unlabeled sample points Calculate its cosine distance to all other sample points. Statistical satisfaction The number of samples; if the number of samples Then mark Using this as the core, an initial cluster is established. ; Traversing the initial clusters All core points, including each core point Unlabeled samples in the neighborhood are added This step is repeated recursively until no new samples can be added. All isolated samples that are not core points are marked as noise points and removed. The survival analysis model described in S2.5.3 uses the Kaplan-Meier estimation method to calculate the baseline hazard rate. The estimation formula is: , For the first Number of breakdown samples at each time point For the first The number of surviving samples at each time point; the regression coefficients are obtained using Cox proportional hazards regression. The regression model is The significance level was set at 1. The effectiveness of the model was verified by the likelihood ratio test.

Images