Cable insulation aging evaluation method based on K-means algorithm

Abstract

The invention relates to a cable insulation aging evaluation method based on a K-means algorithm, and the method comprises the steps: preparing slice samples in different thermal aging states, carrying out the rapid ultra-low frequency dielectric loss test of the aged slice samples through an ultra-low frequency dielectric loss detector (iFDS), and obtaining the dielectric constant of the aged slice at each frequency. And calculating layered contrast ratios in various aging states, and determining layered contrast ratio ranges corresponding to different aging degrees by using a K-means clustering algorithm. An ultra-low frequency dielectric loss detector (iFDS) is used for conducting rapid ultra-low frequency dielectric loss testing on the predicted cable, the layering contrast ratio is calculated, and the aging degree of the cable to be detected is determined according to the corresponding layering contrast ratio range.

Description

Technical Field

[0001] The present invention relates to the field of cable safety detection, and in particular to a cable insulation aging evaluation method based on a K-means algorithm. Background Art

[0002] 10kV cross-linked polyethylene (XLPE) power cables play a vital role in power transmission and distribution due to their excellent electrical and mechanical properties. With the continuous growth of power demand and the acceleration of urbanization, the utilization rate and operating load of these cables are increasing. However, the insulation performance of cables may deteriorate due to a variety of factors during long-term operation. Among them, thermal aging is a common defect in cable insulation. Overload operation, poor heat dissipation, improper installation, etc. of cables can cause thermal aging of cables and degradation of insulation material performance. Therefore, the evaluation of cable insulation status is key to ensuring the reliability and safety of power systems. By conducting a series of tests and monitoring on cables, their insulation performance can be evaluated to predict their remaining service life and maintenance needs.

[0003] At present, the application of dielectric response theory in insulation medium aging diagnosis has become a research hotspot. The core of this theory is that when voltage is applied to the insulation medium, a series of dielectric response parameters can be observed. By analyzing these parameters, the performance changes of the insulation medium can be evaluated, thereby predicting its potential failure risk.

[0004] At present, the insulation status of XLPE cables is often evaluated by dielectric loss tangent (tanδ) test under 0.1Hz sinusoidal voltage excitation to detect the overall insulation thermal aging, moisture and water tree degradation of the cable. However, although ultra-low frequency dielectric loss measurement below 0.1Hz can provide more in-depth aging information, the testing process is time-consuming. In addition, when the aging degree of the cable is not serious, the tanδ value at 0.1Hz may not accurately reflect the actual aging degree of the insulation material. Therefore, test results relying on a single frequency may not fully and accurately reveal the true state of the insulation material. At the same time, the test results are only used to determine whether the cable has defects, and no systematic method for evaluating the degree of cable defects has been proposed.

[0005] In summary, the existing technology has major deficiencies in cable insulation evaluation. The long detection time and inaccuracy affect the evaluation of power cable insulation. At the same time, no evaluation method for cable insulation degree is proposed. In order to solve the above problems, it is necessary to study and develop a 10kV XLPE cable insulation evaluation method to solve the above problems. Summary of the invention

[0006] In view of the above problems existing in the prior art, the present invention relates to a cable insulation aging evaluation method based on K-means algorithm, the main purpose of which is to solve the problems of long time consumption and inaccuracy under single frequency testing.

[0007] In order to solve the above technical problems, the present invention adopts the following technical solution: a cable insulation aging evaluation method based on K-means algorithm, comprising the following steps:

[0008] S100 prepares cable slice specimens in different aging states;

[0009] S200 uses an ultra-low frequency dielectric loss tester to perform dielectric loss tests on cable slice samples in different aging states to obtain the dielectric constant of the test frequency.

[0010] S300 performs FTIR spectrum analysis and DSC analysis on cable slice samples to obtain the carbonyl index CI value and crystallinity X of cable slice samples in different aging states. c Value, and process the data:

[0011] ΔCI i =CI i -CI0

[0012] ΔX ci =X c0 -X ci

[0013] Among them, CI0 and X c0 Respectively represent the carbonyl index and crystallinity of the unaged cable slice, CI i and X ci They respectively represent the carbonyl index and crystallinity of the cable slice sample in the ith aging state.

[0014] S400 calculates the delamination contrast of cable slice specimens under different aging conditions. Several carbonyl indices and crystallinity obtained by S300 are used as variables for clustering using the K-means clustering method. The delamination contrast range corresponding to different aging degrees is determined based on the clustering results.

[0015] Furthermore, the process of preparing the aged slice specimen in S100 is as follows:

[0016] S110: Cut a section of 10kV XLPE insulated cable to make a cable section sample, and use the slicing equipment J / Q sliver cable slicing machine to obtain the cable slice sample.

[0017] S120: The cable slice samples are subjected to heat aging treatment in a hot air circulation aging box. The set temperatures are 90°C, 120°C, and 150°C, and the heating time is set at 240h, 480h, and 720h at each temperature to obtain cable slice samples in different aging states.

[0018] Furthermore, the steps of calculating the layered contrast ratio under different aging conditions in S400 are as follows:

[0019] S410: The layered contrast calculation formula is as follows:

[0020]

[0021] Among them, D i is the layered contrast ratio of the dielectric constant spectrum curve under the i-th aging state, n is the number of detection frequency points, and f k is the detection frequency. ε′0 represents the dielectric constant of the unaged cable slice, and ε′ represents the dielectric constant of the aged cable slice.

[0022] S420 determines the layered contrast range corresponding to different aging conditions:

[0023] S421: taking several carbonyl indices and crystallinity obtained in steps S300 and S300 as variables, clustering is performed using the K-means clustering method, and classifying the slice specimens under different aging conditions using the software spss;

[0024] S422: Determine the stratified contrast ranges corresponding to different aging degrees based on the clustering results, and divide the aging degrees into the following four stages: initial aging, moderate aging, severe aging, and unusable. The stratified contrast D ranges corresponding to the four stages are: 0<D≤0.2, 0.2<D≤1, 1<D≤5, D>5.

[0025] Compared with the prior art, the present invention has at least the following advantages:

[0026] Compared with the single frequency 0.1Hz dielectric loss detection, the frequency test range of the ultra-low frequency dielectric loss detector (iFDS) used in the present invention is wider, usually 1mHz~1kHz, and the dielectric constant measured at each frequency is fully utilized when calculating the layered contrast, which can effectively avoid the errors or errors caused by the single frequency point dielectric loss test due to field factors. At the same time, the detector can establish an equivalent mathematical model based on the obtained voltage and current signals to describe the insulation system regarded as a two-port network. The system replaces the traditional sweep mode, shortens the test time, and realizes the rapid acquisition of the dielectric response parameters of the insulating medium in the ultra-low frequency band. Combined with the K-means clustering algorithm proposed in the present invention, it shows a significant efficiency advantage. Compared with other clustering algorithms, the K-means algorithm is due to its simple iterative process and low computational complexity. When processing data sets of the same size, the calculation time is greatly reduced. At the same time, the present invention selects the difference between the CI index and the degree of crystallinity, which can effectively avoid the error caused by the different shape and structure of the cable itself, so that the layered contrast range determined after cluster analysis is more accurate and more applicable. The present invention performs simple data processing on the dielectric loss test results - calculating the layer contrast, can preliminarily judge the aging degree of the cable, make corresponding countermeasures in time, and provide certain guidance for the maintenance and repair of the cable. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 The figure is a flow chart of a 10kV XLPE cable insulation testing method.

[0028] Figure 2 This is the K-means clustering analysis algorithm process.

[0029] Figure 3 ΔCI ​​corresponding to different aging states.

[0030] Figure 4 ΔX corresponding to different aging states c .

[0031] Figure 5 The classification result of aging status.

[0032] Figure 6 is the aging degree corresponding to different aging states. DETAILED DESCRIPTION

[0033] The present invention is described in further detail below.

[0034] A cable insulation aging assessment method based on the K-means algorithm uses an ultra-low frequency dielectric loss detector (iFDS) to perform rapid ultra-low frequency dielectric loss test on aged slice samples, which can reduce the test time and avoid measurement errors caused by a single frequency. It can reveal the true state of the insulating material more comprehensively and accurately. At the same time, a method for determining the degree of aging using the K-means algorithm is proposed. Compared with other clustering algorithms, it has higher computational efficiency and is easy to implement. By performing simple data processing on the dielectric loss test results - calculating its layered contrast, the degree of cable aging can be preliminarily judged, and corresponding countermeasures can be taken in a timely manner, which can provide certain guidance for the maintenance and repair of the cable.

[0035] A cable insulation aging assessment method based on K-means algorithm includes the following steps:

[0036] S100 prepares cable slice specimens in different aging states;

[0037] S200 uses an ultra-low frequency dielectric loss tester to perform dielectric loss tests on cable slice samples in different aging states to obtain the dielectric constant of the test frequency.

[0038] S300 performs FTIR spectrum analysis and DSC analysis on cable slice samples to obtain the carbonyl index CI value and crystallinity X of cable slice samples in different aging states. c Value, and process the data:

[0039] ΔCI i =CI i -CI0

[0040] ΔX ci =X c0 -X ci

[0041] Among them, CI0 and X c0 Respectively represent the carbonyl index and crystallinity of the unaged cable slice, CI i and X ci They respectively represent the carbonyl index and crystallinity of the cable slice sample in the ith aging state.

[0042] S400 calculates the delamination contrast of cable slice samples under different aging conditions. The delamination contrast characterizes the degree to which the cable slice samples under different aging conditions deviate from the normal cable dielectric constant spectrum curve. Several carbonyl indices and crystallinity obtained by S300 are used as variables for clustering using the K-means clustering method, and the delamination contrast range corresponding to different aging degrees is determined based on the clustering results.

[0043] Furthermore, the process of preparing the aged slice specimen in S100 is as follows:

[0044] S110: Cut a section of 10kV XLPE insulated cable to make a cable section sample, and use the slicing equipment J / Q sliver cable slicing machine to obtain the cable slice sample.

[0045] S120: The cable slice samples are subjected to heat aging treatment in a hot air circulation aging box. The set temperatures are 90°C, 120°C, and 150°C, and the heating time is set at 240h, 480h, and 720h at each temperature to obtain cable slice samples in different aging states.

[0046] Furthermore, the steps of calculating the layered contrast ratio under different aging conditions in S400 are as follows:

[0047] S410: The layered contrast calculation formula is as follows:

[0048]

[0049] Among them, D i is the layered contrast ratio of the dielectric constant spectrum curve under the i-th aging state, n is the number of detection frequency points, and f k is the detection frequency. ε′0 represents the dielectric constant of the unaged cable slice, and ε′ represents the dielectric constant of the aged cable slice.

[0050] S420 determines the layered contrast range corresponding to different aging conditions:

[0051] S421: The carbonyl index and crystallinity obtained in steps S300 and S300 were used as variables and clustered using the K-means clustering method. The sliced ​​specimens under different aging conditions were classified using the software spss. Because the carbonyl index CI can quantitatively characterize the degree of degradation of XLPE thermal aging, the crystallinity can also reflect the insulation aging behavior of the material to a certain extent. Materials with high crystallinity usually have higher thermal stability, because the molecular arrangement in the crystalline region is more orderly and the intermolecular interaction is stronger, which can better ensure the integrity of its structure under high temperature conditions and is not prone to thermal degradation.

[0052] The K-means clustering analysis algorithm process is as follows Figure 2As shown. First, select K initial cluster centers. For each data point in the data set, calculate its distance from each cluster center. For each cluster, recalculate the cluster center. If it converges or reaches the preset number of iterations, output the classification result, otherwise continue to repeat the above steps. The cluster center selection steps are as follows: randomly select a point from the data set as the starting cluster center. Then, for each point, calculate the minimum distance between it and the selected cluster center. The greater the distance, the greater the chance of becoming a new cluster center. Subsequently, a probabilistic method, such as the roulette wheel selection method, is used to determine the next cluster center. This process will be repeated until the required K cluster centers are finally determined.

[0053] S422: Determine the stratified contrast ranges corresponding to different aging degrees based on the clustering results, and divide the aging degrees into the following four stages: initial aging, moderate aging, severe aging, and unusable. The stratified contrast D ranges corresponding to the four stages are: 0<D≤0.2, 0.2<D≤1, 1<D≤5, D>5.

[0054] Initial aging: Under normal use conditions, the insulation layer of the cable may change slightly in color or hardness after a period of time, but it will not affect its electrical performance.

[0055] Moderate Aging: The cable begins to show some physical damage, such as minor cracks, hardening, and brittleness, but can still continue to be used.

[0056] Severe aging: Physical damage to the cable becomes apparent, with severe cracks, crazing, peeling of the insulation, and even exposed conductors.

[0057] Unusable: The cable is no longer functioning properly and there may be a serious electrical fault, such as insulation breakdown or conductor breakage.

[0058] Example: FTIR spectrum analysis and DSC analysis were performed on the cable slice samples to obtain the carbonyl index CI value and crystallinity X of the cable slice samples in different aging states. c , and perform data processing on it, subtracting the corresponding value of the cable slice from the corresponding value of the aging slice, as shown in Table 1. Figure 3 and Figure 4 shown.

[0059] Table 1: ΔCI and Δcrystallinity corresponding to aged slices under different aging conditions

[0060]

[0061]

[0062] It can be seen from Table 1 that the ΔCI carbonyl index and Δcrystallinity are positively correlated with the degree of aging. When the heating temperature is as high as 150 degrees Celsius and the heating time is as long as 720 hours, the ΔCI carbonyl index is as high as 0.5 and the Δcrystallinity is as high as 0.23. At this time, the cable has been seriously aged. The combination of the CI carbonyl index and the crystallinity can well characterize the insulation aging state. Therefore, the processed ΔCI carbonyl index and Δcrystallinity are used as cluster analysis variables of the K-means algorithm to make the classification results more accurate. At the same time, the difference between the CI carbonyl index and the crystallinity selected in the present invention can effectively avoid the error caused by the different shapes and structures of the cables themselves, so that the stratified contrast range determined after the cluster analysis is more accurate and more applicable.

[0063] SPSS software was used to classify the aging state by taking CΔI and XΔc as variables and setting the number of iterations to n = 10. The classification results are shown in the figure. Figure 5 shown.

[0064] Then, the stratified contrast ranges corresponding to different aging degrees are determined according to the classification results, as shown in Table 2.

[0065]

[0066] Figure 6 The aging degree corresponding to different aging states, combined with Figure 3 and Figure 4 It can be seen that the degree of aging determined according to the stratified contrast range corresponds to the aging state corresponding to the above CΔI and XΔc, indicating that the stratified contrast range determined by the above method is reliable. The K-means clustering algorithm proposed in the present invention shows a significant efficiency advantage. Compared with other clustering algorithms, the K-means algorithm has a simple iterative process and low computational complexity. When processing data sets of the same size, the calculation time is greatly reduced. In addition, the use of ultra-low frequency dielectric loss detectors also greatly reduces the evaluation time of insulation aging. At the same time, the present invention selects the difference between the CI carbonyl index and the crystallinity to effectively avoid the error caused by the different shapes and structures of the cable itself, so that the stratified contrast range determined after clustering analysis is more accurate and more applicable.

[0067] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A cable insulation aging assessment method based on K-means algorithm, characterized by: The steps include: S100 prepares cable slice specimens in different aging states; S200 uses an ultra-low frequency dielectric loss tester to perform dielectric loss tests on cable slice samples in different aging states to obtain the dielectric constant of the test frequency. S300 performs FTIR spectrum analysis and DSC analysis on cable slice samples to obtain the carbonyl index CI value and crystallinity X of cable slice samples in different aging states. c Value, and process the data: ΔCI i =CI i -CI0 ΔX ci =X c0 -X ci Among them, CI0 and X c0 Respectively represent the carbonyl index and crystallinity of the unaged cable slice, CI i and X ci They represent the carbonyl index and crystallinity of the cable slice sample in the i-th aging state respectively; S400 calculates the delamination contrast of cable slice specimens under different aging conditions. Several carbonyl indices and crystallinity obtained by S300 are used as variables for clustering using the K-means clustering method. The delamination contrast range corresponding to different aging degrees is determined based on the clustering results.

2. A cable insulation aging evaluation method based on K-means algorithm as claimed in claim 1, characterized in that: The S100 process for preparing the aged slice specimen is as follows: S110: Cut a section of 10kV XLPE insulated cable to make a cable section sample, and use the slicing equipment J / Q sliver cable slicing machine to obtain the cable slice sample; S120: The cable slice samples are subjected to heat aging treatment in a hot air circulation aging box. The set temperatures are 90°C, 120°C, and 150°C, and the heating time is set at 240h, 480h, and 720h at each temperature to obtain cable slice samples in different aging states.

3. A cable insulation aging evaluation method based on K-means algorithm as claimed in claim 2, characterized in that: The steps for S400 to calculate the layer contrast ratio under different aging conditions are as follows: S410: The layered contrast calculation formula is as follows: Among them, D i is the layered contrast ratio of the dielectric constant spectrum curve under the i-th aging state, n is the number of detection frequency points, and f k is the detection frequency; ε′0 represents the dielectric constant of the unaged cable slice, and ε′ represents the dielectric constant of the aged cable slice; S420 determines the layered contrast range corresponding to different aging conditions: S421: taking several carbonyl indices and crystallinity obtained in steps S300 and S300 as variables, clustering is performed using the K-means clustering method, and classifying the slice specimens under different aging conditions using the software spss; S422: Determine the stratified contrast ranges corresponding to different aging degrees based on the clustering results, and divide the aging degrees into the following four stages: initial aging, moderate aging, severe aging, and unusable. The stratified contrast D ranges corresponding to the four stages are: 0<D≤0.2, 0.2<D≤1, 1<D≤5, D>5.

Images