A cable insulation partial defect diagnosis method based on differential spectrum composite features

Abstract

The application discloses a cable insulation partial defect diagnosis method based on differential frequency spectrum composite features, comprising the following steps: S1. wideband excitation injection and head input impedance acquisition; S2. differential impedance frequency spectrum construction: the absolute amplitudes of the measured complex input impedance spectrum Z in,aged ( f ) and the background baseline impedance spectrum Z in,health ( f ) are extracted and operated; S3. composite feature extraction under the dominant resonance mode: impedance differential maximum value, pole frequency offset and active admittance differential maximum value are extracted as the composite features; S4. cable insulation partial defect diagnosis based on the composite features extracted in the step S3; the application utilizes the composite features extracted by the differential frequency spectrum, relies on the change of the active admittance feature slope caused by the electrical tree carbonization channel, eliminates the feature aliasing of the partial water trees and high-risk electrical tree defects in the impedance amplitude under the same spatial size, and realizes the diagnosis of the water trees and the electrical trees and other cable insulation partial defects in the physical dimension.

Description

Technical Field

[0001] This invention relates to the field of power distribution cable insulation monitoring, and specifically to a method for diagnosing local defects in cable insulation based on differential spectrum composite characteristics. Background Technology

[0002] Cross-linked polyethylene (XLPE) insulated power cables are widely used in modern urban power distribution networks due to their excellent electrical insulation properties, high heat resistance, and good mechanical strength. However, power cables are usually laid in complex and harsh environments such as underground ducts, cable trenches, or damp soil. Under the combined stress of long-term humid and hot environments, high electric field stress, and chemical degradation, moisture intrusion and dielectric aging of their insulation layer are inevitable.

[0003] The evolution mechanism of moisture in XLPE insulation exhibits significant spatial scale differences and staged characteristics: initially, moisture extensively penetrates longitudinally through damage to the outer sheath, triggering macroscopic "global dampness" aging (volume effect) with uniform water absorption throughout the entire layer; while in areas of concentrated electric field or microscopic pores of the material, water molecules generate micron-sized local "water tree" defects under electrochemical reactions. As a latent degradation structure, water trees, if not intervened in time, will induce severe local discharge in their final stage and transform into "electric tree" defects accompanied by conductive carbonization channels under high field strength. The appearance of electric trees indicates that the insulating medium has entered an irreversible and severe pre-breakdown stage. According to statistics from power operation departments, single-phase grounding or phase-to-phase short-circuit faults caused by insulation moisture, water trees, and electric tree deterioration account for more than 40% of unplanned outages in distribution systems, seriously threatening the power supply reliability and resilience of smart distribution networks.

[0004] Currently, the main non-destructive condition assessment method for cable insulation defects includes broadband impedance spectroscopy (BIS). This method, by synchronously injecting a swept-frequency voltage signal at the beginning and measuring the current phasor, can obtain an input impedance curve containing rich frequency domain dielectric information, showing certain advantages in the uniform aging diagnosis of long-distance cables. However, the types of defects in actual distribution networks are extremely complex, and the signs of early local defects in insulation in the macroscopic electromagnetic response are extremely weak. This means that existing BIS detection technology still faces the following technical challenges when dealing with meter-level local micro-defects:

[0005] 1. Low sensitivity in detecting weak local defects in long cables: For long-distance cables, the original input impedance spectrum at the beginning of the cable exhibits a significant nonlinear fluctuation pattern, mainly determined by the distributed parameters (distributed inductance, distributed capacitance) of the long cable. When early local water tree or weak electrical tree defects appear inside the cable, due to the short length of the defect segment, the local impedance mismatch reflection signal caused by it and the healthy segment is very weak. Its distortion amplitude is usually only on the order of milliohms or even microohms. Directly extracting the absolute value characteristics (such as peak value or resonant frequency) of the cable impedance spectrum is prone to signal submersion, resulting in insufficient detection sensitivity and the existence of defect detection blind spots and the risk of missed detection.

[0006] 2. The aliasing of RC characteristics leads to a high misjudgment rate of local defect mechanisms: The physical essence of local water tree defects is the increase in local equivalent capacitance caused by the microscopic aggregation of polar water molecules, which is a purely capacitive characteristic and does not generate significant conductive leakage current. In contrast, electrical tree defects, due to local carbonization and depolymerization of the insulating material, form carbonized leakage channels with high conductivity, and are physically manifested as significant RC composite active power disturbances. However, traditional broadband impedance diagnostic systems often rely on the attenuation trend of absolute impedance amplitude or the frequency drift of a single resonant pole for one-dimensional evaluation. Since a one-meter-long water tree and an early electrical tree of the same size show a high degree of overlap in macroscopic impedance amplitude and frequency shift response, current technology lacks mathematical and signal processing methods to strictly separate the RC characteristics at the physical level. This aliasing of mechanistic characteristics easily leads to the misjudgment of high-risk "electrical tree defects" that are about to cause breakdown as low-risk "water tree defects" that can operate with the defect, posing a significant hidden danger to the safe operation of the distribution network.

[0007] Chinese invention patent application CN121091010A discloses a method and apparatus for diagnosing local aging conditions of cables based on the PDC branch parameter method. The apparatus includes: a PDC testing unit for performing PDC tests on the cable under test to obtain the polarization / depolarization current curve of the cable under test; a conductivity calculation unit for subtracting the steady-state value of the depolarization current curve from the steady-state value of the polarization current curve to obtain the steady-state conduction current, and calculating the DC conductivity γ based on the steady-state conduction current; and an asymmetry coefficient calculation unit for calculating the asymmetry coefficient. This invention uses the change in DC conductivity calculated from multiple consecutive PDC tests as a criterion to accurately identify water tree aging and other general defects such as thermal aging; at the same time, by calculating the asymmetry coefficient, it can judge the degree of local aging of the cable, and simultaneously realize the identification of local aging type and the diagnosis of aging degree through quantifiable diagnostic values. The invention is based on the PDC branch parameter method to extract parameters and fit models. However, different aging mechanisms may produce similar PDC responses, and it is difficult to distinguish between local defects and overall moisture, aging, and impurities. Furthermore, the diagnostic results often depend on the rationality of the fitted model.

[0008] Chinese invention patent application CN118628983A discloses a cable identification method and system based on image processing. The cable identification method based on image processing includes: acquiring and processing cross-sectional images of cable insulation materials; performing image segmentation and defect detection on the processed cross-sectional images, identifying and labeling defects in the images, and locating the starting point of water tree phenomenon based on the labeled defects; establishing a prediction model for the generation and development of water tree phenomenon based on the starting point of water tree phenomenon; dynamically adjusting imaging parameters in real time using adaptive optics compensation and image enhancement methods; dynamically adjusting image resolution using a multi-level resolution image acquisition strategy; acquiring raw cross-sectional image data and performing preprocessing; and monitoring and analyzing image data and electrical stress data in real time to identify potential water tree phenomena and insulation material defects. However, this invention relies on cross-sectional images, image segmentation, and defect recognition. Its limitations include the fact that uneven cutting surfaces can affect recognition, uneven lighting, insufficient lens clarity, and high noise levels can lead to significant misjudgments. Contamination, oxidation, dust, and reflections on the surface of insulating materials can also alter image features. Furthermore, this invention is based on image processing methods that typically require substantial training samples and parameter tuning. When there are significant differences in manufacturers, cable models, and structures, the model is prone to failure. Summary of the Invention

[0009] In view of this, the purpose of this invention is to provide a method for diagnosing local defects in cable insulation based on differential spectrum composite features. By utilizing the composite features extracted from the differential spectrum and relying on the change in the slope of the active admittance feature caused by the carbonization channel of the electrical tree, the feature overlap of local water tree and high-risk electrical tree defects in impedance amplitude under the same spatial size is eliminated, thereby realizing the diagnosis of local defects in cable insulation such as water tree and electrical tree from a physical dimension.

[0010] The technical solution adopted in this invention is as follows:

[0011] A method for diagnosing local defects in cable insulation based on differential spectrum composite characteristics includes the following steps:

[0012] S1. Wideband excitation injection and acquisition of the first-end input impedance: Calculate and obtain the measured complex input impedance spectrum Z of the cable under test. in,aged (f), and the complex input impedance spectrum of the healthy cable, which is used as the baseline impedance spectrum. Z in,health ( f );

[0013] S2. Constructing the differential impedance spectrum: Extract the measured complex input impedance spectrum Z. in,aged (f) Baseline impedance spectrum Z in,health The absolute magnitude of (f) is calculated and then the following operation is performed (i.e., the following operation is performed to calculate the absolute impedance differential spectrum ΔZ(f):

[0014] The dominant resonant mode frequency band of the locking cable;

[0015] S3. Extraction of composite features under the dominant resonant mode: In the frequency band of the dominant resonant mode of the cable, the impedance difference maxima, pole frequency offset and active admittance difference maxima are extracted simultaneously from the absolute impedance difference spectrum ΔZ(f) as composite features.

[0016] S4. Diagnose local defects in cable insulation based on the composite features extracted in step S3.

[0017] Preferably, in step S1, the measured complex input impedance spectrum Z of the cable under test is calculated and obtained. in,aged (f) The process includes: with the cable under test offline and its end open-circuited, injecting a wideband sweep excitation signal into the beginning of the cable under test, the frequency range of which covers the dominant resonant mode frequency band of the cable under test; acquiring the voltage phasor and current phasor at the beginning of the cable under test, and calculating the measured complex input impedance spectrum Z at the beginning of the cable under test. in,aged (f).

[0018] Preferably, in step S1, the wideband first-end input impedance Z in The formula for calculating (f) is as follows:

[0019] In the formula, U in (f) represents the wideband sweep excitation signal voltage; I in (f) represents the wideband sweep excitation current; during the calculation, the measured complex input impedance spectrum Z to be calculated is used. in,aged (f) Substitute the wideband first-end input impedance Z in (f).

[0020] Preferably, the wideband first-end input admittance Y in The formula for calculating (f) is as follows:

[0021] In the formula, G in (f) represents the real part of the admittance; jB in (f) represents the imaginary part of the admittance, where the real part of the admittance G in (f) As an equivalent input conductance, it is proportional to the active power absorbed by the cable network under high-frequency excitation and is used to identify local defects in cable insulation accompanied by carbonization leakage; in the calculation, the active admittance to be calculated is substituted into the real part of the admittance G. in (f).

[0022] Preferably, the frequency range of the sweep excitation signal is set to 5kHz to 1MHz in the test frequency band.

[0023] Preferably, in step S1, the healthy cable and the cable under test have the same specifications, and the calculated complex input impedance spectrum of the healthy cable and the calculated measured complex input impedance spectrum Z of the cable under test are obtained. in,aged The calculation process for (f) is the same.

[0024] Preferably, in step S3, the calculation process for the impedance difference maxima is as follows:

[0025] In the formula, ΔZ max As the maximum value of the impedance difference;

[0026] The calculation process for pole frequency offset is as follows:

[0027] In the formula, Δf res As the dominant resonant pole frequency offset, it is the pole frequency offset; The center frequency between adjacent steady-state poles of the dominant resonance of the cable under test; The dominant resonant center frequency between adjacent steady-state poles of a healthy cable;

[0028] The calculation process for the active admittance differential maximum is as follows:

[0029] In the formula, ΔG max As the active admittance differential maximum The active admittance of the cable under test is given. The active admittance of a healthy cable.

[0030] Preferably, the diagnosis of local defects in cable insulation in step S4 includes the following process:

[0031] The impedance difference maximum ΔZ max Compared with the preset macroscopic coarse screening threshold T h1 A comparison is performed, where the impedance difference maximum value ΔZ max Greater than the macroscopic coarse screening threshold T h1 If the test cable is found to have undergone moisture aging dominated by volume, it is determined that the cable under test has undergone moisture aging.

[0032] If the impedance difference maximum value ΔZ max Equal to or less than the macroscopic coarse screening threshold T h1 If the test cable has a local defect, it is determined that the cable under test has a local defect.

[0033] Preferably, the diagnosis of local defects in cable insulation in step S4 includes the following process:

[0034] With the dominant resonant pole frequency offset Δf res With the horizontal axis as the finite difference maximum value ΔG of the active admittance, the value is...max Using the vertical axis as the ordinate, a two-dimensional feature space is mapped out. The slope K of the eigenvector of the measured feature point in this two-dimensional feature space is calculated using the following formula:

[0035] ;

[0036] Based on the feature vector slope K and the preset mechanism decoupling threshold T h2 The comparison is performed, and the type of local defect in the insulation of the tested cable is determined based on the comparison results.

[0037] Preferably, when performing the feature vector slope K and the preset mechanism decoupling threshold T... h2 During the comparison process:

[0038] If the slope K of the feature vector is greater than the decoupling threshold T of the mechanism h2 The local defect type of the tested cable insulation is determined to be a water tree defect dominated by pure water soluble polarization; if the slope K of the eigenvector is equal to or less than the mechanical decoupling threshold T... h2 The local defect type of the tested cable insulation was determined to be an electrical tree defect with a high risk of breakdown and an accompanying carbonization leakage mechanism.

[0039] This invention proposes a method for diagnosing local defects in cable insulation based on differential spectrum composite characteristics. This method is based on an active power loss feature extraction method that decouples the real part of complex admittance. By converting the complex impedance measured at the beginning of the cable into a complex admittance, the real part of the admittance (i.e., the equivalent input conductance) is directly extracted. The real part of the admittance physically corresponds directly to the high-frequency active power loss of the insulation medium. This conversion method successfully extracts the "active power leakage" feature of the defect independently without requiring complex inversion algorithms, providing a direct basis for distinguishing defects with different physical mechanisms. This invention further proposes… A two-dimensional diagnostic criterion based on the ratio of frequency offset to admittance difference is proposed: Addressing the technical challenge of distinguishing between local water trees and local electrical trees when their sizes are similar due to highly similar impedance amplitude changes, this invention changes the traditional approach of judging based on a single feature. It utilizes the essential difference between water trees and electrical trees, extracts the pole frequency offset and the maximum value of the active admittance difference, and calculates the ratio between the two (i.e., the slope K in the feature space). This ratio amplifies the conductivity characteristics of the electrical tree, causing numerical differences between the originally overlapping features of water trees and electrical trees, thereby achieving accurate classification and identification of local defects in cable insulation.

[0040] The present invention has the following main advantages:

[0041] 1) Significantly improved the detection sensitivity of early-stage minute local defects.

[0042] By constructing the absolute impedance differential spectrum, this invention effectively filters out the huge standing wave background interference inherent in long cables, enabling early weak local defect signals (such as the micro-ohmic impedance change caused by a water tree only 1 meter long) that were originally completely submerged to be clearly captured, improving the detection signal-to-noise ratio and effectively preventing the missed detection of early cable defects.

[0043] 2) Effectively avoids misjudgment of high-risk electrical tree defects, improving power grid security.

[0044] This invention utilizes active admittance characteristics and slope as the basis for judgment, providing a clear and feasible physical basis for classifying cable aging defects: Among them, electrical trees containing carbonized channels will exhibit a ratio different from that of water trees, thus being accurately identified by the system, completely solving the problem of easy confusion between these two types of defects in traditional methods, and providing reliable decision support for on-site operation and maintenance. Attached Figure Description

[0045] Figure 1 This is a flowchart illustrating the steps of the cable insulation local defect diagnosis method based on differential spectrum composite characteristics according to a specific embodiment of this application.

[0046] Figure 2 This is a schematic diagram of the system simulation model (35kV XLPE cable) in a specific embodiment of this application;

[0047] Figure 3 This is the input impedance spectrum of the healthy cable in a specific embodiment of this application;

[0048] Figure 4 This is a schematic diagram of the finite element modeling structure of a healthy cable in a specific embodiment of this application;

[0049] Figure 5 This is a schematic diagram of the finite element modeling structure of a cable in a damp state in a specific embodiment of this application;

[0050] Figure 6 This is a schematic diagram of the finite element modeling structure of a cable with water tree aging of the XLPE insulation layer in a specific embodiment of this application;

[0051] Figure 7 This is a schematic diagram of the finite element modeling structure of the cable with electrical tree aging of the XLPE insulation layer in a specific embodiment of this application;

[0052] Figure 8 This is the input impedance differential spectrum diagram in a specific embodiment of this application;

[0053] Figure 9 This is a diagram showing the maximum impedance difference amplitude in a specific embodiment of this application;

[0054] Figure 10 This is a comparison diagram of the resonant pole frequency offset in a specific embodiment of this application;

[0055] Figure 11 This is a comparison diagram of the high-frequency active admittance differential values ​​in a specific embodiment of this application;

[0056] Figure 12 This refers to the two-dimensional feature space constructed in the specific embodiments of this application (based on the dominant resonant pole frequency offset Δf). res With the horizontal axis as the finite difference maximum value ΔG of the active admittance, the value is... max (Using the vertical axis) Comparison chart. Detailed Implementation

[0057] Please refer to the above. Figure 1 As shown, this embodiment provides a method for diagnosing local defects in cable insulation based on differential spectrum composite characteristics, including the following steps:

[0058] S1. Wideband excitation injection and acquisition of the first-end input impedance: Calculate and obtain the measured complex input impedance spectrum Z of the cable under test. in,aged (f), and the complex input impedance spectrum of the healthy cable, which is used as the baseline impedance spectrum. Z in,health ( f );

[0059] Preferably, in step S1, the measured complex input impedance spectrum Z of the cable under test is calculated and obtained. in,aged (f) The process includes: with the cable under test offline and its end open-circuited, injecting a wideband sweep excitation signal into the beginning of the cable under test (preferably, in this embodiment, the frequency range of the sweep excitation signal is set to the test frequency band from 5kHz to 1MHz), the frequency range of the sweep excitation signal covering the dominant resonant mode frequency band of the cable under test; acquiring the voltage phasor and current phasor at the beginning of the cable under test, and calculating the measured complex input impedance spectrum Z at the beginning of the cable under test. in,aged (f); It should be noted that the “open-circuit at the end” boundary condition involved in this application is used to cause the high-frequency electromagnetic wave (i.e., the broadband sweep frequency excitation signal) to undergo total reflection at the end of the cable. The total reflection wave interferes with and superimposes with the local insulation defects along the line, so as to map the longitudinal micro-insulation state of the cable to the macro-input impedance response at the beginning end.

[0060] More preferably, in step S1, the wideband first-end input impedance Z in The formula for calculating (f) is as follows:

[0061] In the formula, U in (f) represents the wideband sweep excitation signal voltage; I in (f) represents the wideband sweep excitation current; during the calculation, the measured complex input impedance spectrum to be calculated is substituted into the wideband first-end input impedance Z.in (f).

[0062] Preferably, in this embodiment, the wideband first-end input admittance Y in The formula for calculating (f) is as follows:

[0063] In the formula, G in (f) represents the real part of the admittance; jB in (f) represents the imaginary part of the admittance, where the real part of the admittance is G. in (f) As the equivalent input conductance, it is proportional to the active power absorbed by the cable network under high-frequency excitation and is used to identify local defects in cable insulation accompanied by carbonization leakage. In the calculation, the active admittance to be calculated is substituted into the real part of the admittance G. in (f).

[0064] Preferably, in step S1, the specifications of the healthy cable and the cable under test are the same, and the calculated complex input impedance spectrum of the healthy cable is the same as the calculated measured complex input impedance spectrum Z of the cable under test. in,aged The calculation process for (f) is the same.

[0065] S2. Constructing the differential impedance spectrum: Extract the measured complex input impedance spectrum Z. in,aged (f) Baseline impedance spectrum Z in,health The absolute magnitude of (f) is calculated as follows:

[0066] The dominant resonant mode frequency band of the locking cable;

[0067] The difference operation proposed in step S2 is used to filter out the weak local insulation defect features caused by the macroscopic length of the cable in the frequency domain, and to extract the weak high-frequency reflection signal caused by the meter-level small impedance mismatch to the vicinity of the zero axis baseline, thereby greatly improving the signal-to-noise ratio of signal detection.

[0068] S3. Extraction of composite features under the dominant resonant mode: In the frequency band of the dominant resonant mode of the cable, the impedance difference maxima, pole frequency offset and active admittance difference maxima are extracted simultaneously from the absolute impedance difference spectrum ΔZ(f) as composite features. These three parameters are mutually independent orthogonal quantitative features, which are the composite features extracted in step S3.

[0069] Preferably, in step S3, the calculation process for the impedance difference maxima is as follows:

[0070] In the formula, ΔZ max As the maximum value of the impedance difference;

[0071] The calculation process for pole frequency offset is as follows:

[0072] In the formula, Δf res The frequency shift of the dominant resonant pole is the center frequency shift between the dominant resonant pole (also known as the steady-state pole) before and after aging. The center frequency between adjacent steady-state poles of the dominant resonance of the cable under test; The dominant resonant center frequency between adjacent steady-state poles of a healthy cable;

[0073] It should be noted that the center frequency between adjacent steady-state poles of the dominant resonance of the cable involved in this application can be measured by instruments such as network analyzers, frequency response analyzers, and impedance analyzers. These are common knowledge in the field, and the embodiments of this application will not be described in detail.

[0074] The calculation process for the active admittance differential maximum is as follows:

[0075] In the formula, ΔG max As the active admittance differential maximum The active admittance of the cable under test is given. The active admittance of a healthy cable.

[0076] S4. Based on the composite features extracted in step S3, perform local defect diagnosis of cable insulation, i.e. Figure 1 The marked "fault diagnosis based on composite features":

[0077] Preferably, the diagnosis of local defects in cable insulation in step S4 includes the following process:

[0078] The impedance difference maximum ΔZ max Compared with the preset macroscopic coarse screening threshold T h1 A comparison is performed, where the impedance difference maximum value ΔZ max Greater than the macroscopic coarse screening threshold T h1 If the cable under test is found to have undergone moisture aging dominated by volume (i.e., ...), then it is determined that the cable under test has undergone moisture aging dominated by volume. Figure 1 (The label indicates "diagnosed as damp aging"); if the impedance difference maximum value ΔZ max Equal to or less than the macroscopic coarse screening threshold T h1 If the cable under test is found to have a local defect, the following procedure can be used to further diagnose the local defects in the cable insulation.

[0079] Preferably, the diagnosis of local defects in cable insulation in step S4 includes the following process:

[0080] With the dominant resonant pole frequency offset Δf res With the horizontal axis as the finite difference maximum value ΔG of the active admittance, the value is... maxUsing the vertical axis as the ordinate, a two-dimensional feature space is mapped out. The slope K of the eigenvector of the measured feature point in this two-dimensional feature space is calculated using the following formula:

[0081] ;

[0082] Based on the feature vector slope K and the preset mechanism decoupling threshold T h2 The comparison is performed, and the type of local defect in the insulation of the tested cable is determined based on the comparison results.

[0083] Preferably, in this embodiment, the feature vector slope K is decoupled from a preset mechanism decoupling threshold T. h2 During the comparison process: If the slope K of the feature vector is greater than the machine decoupling threshold T h2 The local defect type of the tested cable insulation was determined to be a water tree defect dominated by pure water soluble polarization (i.e., Figure 1 (The label "diagnosed as water tree aging"), if the feature vector slope K is equal to or less than the mechanism decoupling threshold T h2 The local defect type of the tested cable insulation was determined to be an electrical tree defect with a high risk of breakdown and an accompanying carbonization leakage mechanism (i.e., Figure 1 The diagnosis is marked as "electrical tree aging".

[0084] To enable those skilled in the art to better understand the technical solutions of this invention, based on the above implementation schemes, the following specific embodiments will be proposed in conjunction with the accompanying drawings of the embodiments of this invention. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this invention.

[0085] In the following specific embodiment, a 35kV XLPE cable is used as an example for simulation modeling. Please refer to the system simulation model for details. Figure 2 As shown: For local defects, the entire cable is divided into three cascaded sections: a 500m healthy section, a 1m locally aged section in the middle, and a subsequent 500m healthy section. By changing the RLGC parameters of the equivalent model corresponding to the 1m locally aged section, the simulation of various types of local defects can be achieved.

[0086] Specifically, in this embodiment, the total cable length is set to 1001m, and the distribution parameters under healthy conditions are taken from the standard 35kV XLPE cable.

[0087] For localized defects, the defect length is set at 1m, located 500m from the middle of the cable.

[0088] Frequency sweep parameters: Set the test frequency band to 10kHz to 1MHz, and the number of sweep points to 1000.

[0089] Preset threshold settings: Based on cable length and specifications, the system presets a macroscopic coarse screening threshold Th1 = 5.0Ω for screening overall moisture; and a preset slope threshold (i.e., the preset mechanistic decoupling threshold) Th2 = 5.0 × 10⁻⁶ for distinguishing between water trees and electrical trees. -5 S / Hz, where S represents the admittance unit and Hz represents the frequency unit. Before all tests begin, the reference head-end input impedance spectrum Z is obtained under the condition of open circuit at the end by running a healthy cable model. in,health (f) (i.e., the baseline impedance spectrum) and store it in system memory. Please refer to [link to relevant documentation]. Figure 3 As shown, Figure 3 The horizontal axis corresponds to the test frequency band, and the vertical axis, "input impedance," corresponds to the reference input impedance spectrum Z. in,health The absolute magnitude of (f).

[0090] To accurately locate and simulate local insulation defects, the entire cable is equivalently divided into three cascaded two-port networks: a healthy segment at the beginning (length l1 = 500m), a segment with local defects (length l2 = 1m), and a healthy segment at the end (length l3 = 500m).

[0091] Based on the above implementation process provided in this application, this specific embodiment is as follows: Figure 2 The simulation model shown serves as a platform for finite element modeling and simulation of three different types of aging defects—moisture, water treeing, and electrical treeing—in the local aging section to obtain realistic cable insulation state parameters, thereby enabling the construction of the differential impedance spectrum.

[0092] A two-dimensional cross-sectional model of a typical 35kV medium-voltage XLPE cable was constructed. The physical field is governed by the quasi-static electromagnetic field equations. The modeling considers the skin effect of the semiconducting layer and the dielectric polarization of the insulation layer under alternating high-frequency electric fields. To accurately reflect the degradation mechanism, specific dielectric properties were assigned to the defect regions, as detailed in the following process.

[0093] Please refer to the above. Figure 4 As shown in Table 1 below, the electromagnetic parameters of each layer of the healthy cable are as follows. The dielectric loss tangent tanδ of the XLPE insulation layer (i.e., the "insulation layer" in Table 1) is 0.0044 under dry and healthy conditions, and the imaginary part j of the corresponding complex permittivity is approximately 0.01 in the range of 10kHz to 1MHz.

[0094] Table 1

[0095]

[0096] When XLPE cables operate in humid environments, moisture gradually penetrates the insulation layer due to damage to the outer sheath or failure of the intermediate joint seal. Unlike localized water treeing aging, overall dampness is characterized by uniform water absorption of the insulation layer, resulting in an overall increase in the real part of the dielectric constant and the tangent of the dielectric loss angle.

[0097] Please refer to the above. Figure 5 As shown in Table 2 below, the simulation electrical parameters of the XLPE insulation layer under wet cable conditions are set as follows. The simulation of the wet part takes the final wet state as the test object. Under the condition of sufficient absorption, the weight gain rate is about 2%-2.5%, and the relative permittivity εr increases from 2.3 to about 3.22-3.45, with a change rate of 40-50%.

[0098] Table 2

[0099]

[0100] The "water tree" (also known as water tree) structure involved in this application embodiment is a dendritic degradation structure formed by the long-term combined action of moisture and electric field on the XLPE insulation layer. In general simulations, it is described as a "pearl string"-like defect composed of micron-sized water-filled pores and nano-sized connecting channels. The finite element modeling used in this application embodiment is based on this structure. After the material degrades in the water tree region, the dielectric properties change significantly - the relative permittivity of water (about 80) is much higher than that of XLPE (about 2.3), and ions are dissolved in the water tree channels, and the conductivity is also much higher than that of unaged insulation.

[0101] Please refer to the above for further details. Figure 6 As shown, this model uses six cylindrical water channels and water droplets in an alternating combination to form a water tree. Taking the final development stage of the water tree as the modeling object, a total of three water trees are set. The angle between adjacent trees is 30°. The water trees in the water tree area are interconnected to form a complex three-dimensional network structure with a significantly increased density. This area is considered equivalent to a water tree. The diameter of the main channel is 4 micrometers, the length of the water tree is 600 micrometers, and the lateral width is 270 micrometers.

[0102] The equivalent electrical parameters of the material in the water tree region are achieved by changing the properties of the original XLPE insulation layer. The simulation electrical parameters for water tree aging of the XLPE cable layer are shown in Table 3 below.

[0103] Table 3

[0104]

[0105] The "electric tree" (also known as an electrical tree) structure involved in this application is a dendritic degradation phenomenon of the XLPE insulation layer caused by partial discharge under local high electric field strength. Under the action of a strong electric field, partial discharge occurs at gas cavities or material defects, and high-energy electrons bombard the XLPE molecular chains to break and form gas channels. The high temperature generated by the discharge causes carbonization and decomposition of the channel walls, generating conductive carbon particles that are deposited on the inner wall of the channels, further aggravating the local electric field distortion and discharge intensity.

[0106] Please refer to the above for further details. Figure 7 As shown, the needle tip is set to penetrate the XLPE insulation layer to a depth of 5mm; the initial stage of electrical tree formation starts from the needle tip, forming a long and narrow main channel:

[0107] The main channel length is set to 430μm and the radius to 2μm.

[0108] Carbon particles, as the core conductive component of the electric tree channel, have a significant impact on the electric field distribution due to their position, size, and density. Five carbon particles are uniformly distributed, with semi-minor axes of 2µm and semi-major axes of 4µm, and the channel is filled with a gaseous medium. The overall longitudinal length of the electric tree is about 600-800μm, and the transverse width is about 380-450μm.

[0109] The equivalent electrical parameters of the material in the electrical tree region are achieved by changing the properties of the original XLPE insulation layer. The simulation electrical parameters for electrical tree aging of the XLPE cable layer are shown in Table 4 below.

[0110] Table 4

[0111]

[0112] The parameters of the cable insulation aging defects obtained from the finite element simulation of the cable, as described above, are substituted into the 1m defect model (i.e., a local defect segment with a length of 1m) in the circuit simulation. The input impedance spectra obtained from the various cable insulation layer defects (including wet aging, water tree aging, and electrical tree aging) are then compared with the input impedance spectra of the healthy state to obtain the following result: Figure 8 The differential impedance spectrum is shown.

[0113] This embodiment extracts composite features under the dominant resonant mode according to the following process:

[0114] Based on the physical nature of different insulation aging processes, this embodiment extracts the following three independent quantitative features from the differential spectrum:

[0115] (1) Maximum impedance difference magnitude (i.e., the maximum impedance difference ΔZ) max Extensive moisture intrusion can cause a global surge in the equivalent capacitance of a cable, leading to severe resonance mismatch in the differential spectrum. Extracting extreme amplitude values ​​within a specific frequency band:

[0116] ;

[0117] This feature can serve as a coarse screening threshold, effectively separating moisture aging (volume effect) from localized micro-defects (water treeing / electrical treeing). Furthermore, the resonant point with the largest maximum impedance difference amplitude is defined as the dominant resonant mode; please refer to [further details]. Figure 9 As shown in this embodiment, the ΔZ corresponding to the humid state is... max The ΔZ value is 8.4143Ω, corresponding to the aging state of the water tree. max The ΔZ is 0.7103Ω, corresponding to the electrical tree aging state. max It is 1.2495Ω.

[0118] (2) Resonant pole frequency offset (i.e., dominant resonant pole frequency offset Δf) res Local capacitance increments lead to a decrease in local phase velocity, which in turn causes the high-frequency resonant poles to drift to lower frequencies; the center frequency offset between the steady-state poles (i.e., the dominant resonant poles) of the defective cable and the healthy cable is extracted respectively:

[0119] ;

[0120] This capacitive characteristic can quantify the degree of polarization enhancement caused by water droplet accumulation within the cable; please refer to [reference needed]. Figure 10 As shown in this embodiment, the Δf corresponding to the humid state is... res The corresponding Δf at 80.15Hz under water tree aging conditions res The corresponding Δf under the electrical tree aging state is 6.68Hz. res It is 13.36Hz.

[0121] (3) High-frequency active admittance differential value (i.e., the maximum value of the active admittance differential ΔG) max —RC decoupling characteristics: Since the impedance amplitude characteristics of a one-meter-long water tree and an electrical tree highly overlap, this application specifically introduces the active dimension described in this item: Water tree aging manifests as purely capacitive reactive disturbances; while electrical trees, due to the carbonization channels accompanying breakdown precursors, exhibit significant RC composite active disturbances (sudden increase in insulation conductance G); constructing the differential extremum characteristics of the real part of admittance:

[0122] ;

[0123] Please refer to the above for further details. Figure 11 As shown in this embodiment, ΔG corresponds to the humid aging state. max The value is 5.9409e-03S, corresponding to ΔG under the aging state of the water tree. maxThe value is 5.0153e-04S, corresponding to ΔG under the aging state of electrical tree. max The value is 5.9172e-04S. The results demonstrate that the ΔG corresponding to water tree aging is... max The noise floor is close to zero, while the active power loss generated by the electrical tree in the high-frequency resonant band is higher than that of the water tree, which confirms the influence of carbon particles (carbonized channels) on the series resistance of the cable.

[0124] This embodiment also performs local defect diagnosis of cable insulation according to the following process:

[0125] Based on the extracted ΔZ max , Δf res and ΔG max Perform the following step-by-step physical classification and discrimination:

[0126] First, based on the system's preset macroscopic coarse screening threshold Th1 = 5.0Ω for screening global moisture, the moisture status of the cable can be diagnosed. The ΔZ corresponding to the moisture state obtained in the simulation of the embodiment is shown below. max The ΔZ value is 8.4143Ω, corresponding to the aging state of the water tree. max The ΔZ is 0.7103Ω, corresponding to the electrical tree aging state. max It is 1.2495Ω; where, when ΔZ max When the resistance is 8.4143Ω, it is significantly greater than Th1, which indicates that the cable is in a damp condition. The simulation in the example yielded the ΔZ corresponding to the water tree aging state. max The ΔZ is 0.7103Ω, corresponding to the electrical tree aging state. max The Ω value is 1.2495Ω, which is less than Th1, so we proceed to the next stage of fault diagnosis – the orthogonal decoupling classification process.

[0127] Secondly, in the orthogonal decoupling classification process, the dominant resonant pole frequency offset Δf is used. res With the horizontal axis as the finite difference maximum value ΔG of the active admittance, the value is... max Construct an active dimension for the vertical axis to jointly map out a two-dimensional feature space; calculate the slope K of the feature vector of the measured feature point in this two-dimensional feature space.

[0128] .

[0129] Please refer to the above for further details. Figure 12 As shown, the K value obtained in the simulation of the embodiment under humid conditions is 7.4121 × 10⁻⁶. -5 The K value corresponding to the aging state of the water tree is 7.5088 × 10⁻⁶. -5 The K value corresponding to the aging state of the electrical tree is 4.4296 × 10⁻⁶. -5 .

[0130] K is compared with the preset slope threshold used to distinguish between water trees and electric trees (i.e., the preset mechanistic decoupling threshold Th2) = 5.0 × 10 -5 S / Hz comparison allows for precise diagnosis of water tree and electrical tree defects. Specifically, global humidity and local microscopic water trees exhibit high linear coplanarity in the two-dimensional feature space, with extremely similar slopes. This confirms that both are physically homologous and dominated by the intrusion of polar water molecules (pure capacitive polarization). In contrast, electrical tree defects, due to the introduction of a violent ohmic conductivity mechanism by the carbonization channels preceding breakdown, cause a surge in high-frequency active absorption in the system, resulting in a significant shift in the slope of its eigenvector. This allows for a complete physical orthogonal separation from water trees in the two-dimensional geometric space (i.e., the two-dimensional feature space).

[0131] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0132] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for diagnosing local defects in cable insulation based on differential spectrum composite characteristics, characterized in that, The steps include the following: S1. Wideband excitation injection and acquisition of the first-end input impedance: Calculate and obtain the measured complex input impedance spectrum Z of the cable under test. in,aged (f), and the complex input impedance spectrum of the healthy cable, which is used as the baseline impedance spectrum. Z in,health ( f ); S2. Constructing the differential impedance spectrum: Perform the following calculations for the absolute impedance differential spectrum ΔZ(f): The dominant resonant mode frequency band of the locked cable: S3. Extraction of composite features under dominant resonant mode: Simultaneously extract the impedance difference maxima, pole frequency offset, and active admittance difference maxima of composite features from the absolute impedance difference spectrum ΔZ(f). S4. Diagnose local defects in cable insulation based on the composite features extracted in step S3.

2. The method for diagnosing local defects in cable insulation based on differential spectrum composite characteristics according to claim 1, characterized in that, In step S1, the measured complex input impedance spectrum Z of the cable under test is calculated and obtained. in,aged (f) The process includes: with the cable under test offline and its end open-circuited, injecting a wideband sweep excitation signal into the beginning of the cable under test, the frequency range of which covers the dominant resonant mode frequency band of the cable under test; acquiring the voltage phasor and current phasor at the beginning of the cable under test, and calculating the measured complex input impedance spectrum Z at the beginning of the cable under test. in,aged (f).

3. The method for diagnosing local defects in cable insulation based on differential spectrum composite characteristics according to claim 2, characterized in that, In step S1, the wideband first-end input impedance Z in The formula for calculating (f) is as follows: In the formula, U in (f) represents the wideband sweep excitation signal voltage; I in (f) represents the wideband sweep excitation current; during the calculation, the measured complex input impedance spectrum Z to be calculated is used. in,aged (f) Substitute the wideband first-end input impedance Z in (f).

4. The method for diagnosing local defects in cable insulation based on differential spectrum composite characteristics according to claim 3, characterized in that, Wideband first-end input admittance Y in The formula for calculating (f) is as follows: In the formula, G in (f) represents the real part of the admittance; jB in (f) represents the imaginary part of the admittance, where the real part of the admittance G in (f) As an equivalent input conductance, it is proportional to the active power absorbed by the cable network under high-frequency excitation and is used to identify local defects in cable insulation accompanied by carbonization leakage; in the calculation, the active admittance to be calculated is substituted into the real part of the admittance G. in (f).

5. The method for diagnosing local defects in cable insulation based on differential spectrum composite characteristics according to claim 2, characterized in that, The frequency range of the sweep excitation signal is set to 5kHz to 1MHz in the test frequency band.

6. The method for diagnosing local defects in cable insulation based on differential spectrum composite characteristics according to claim 1 or 2, characterized in that, In step S1, the healthy cable and the cable under test have the same specifications, and the complex input impedance spectrum of the healthy cable and the measured complex input impedance spectrum Z of the cable under test are calculated and obtained. in,aged The calculation process for (f) is the same.

7. The method for diagnosing local defects in cable insulation based on differential spectrum composite characteristics according to claim 1, characterized in that, In step S3, the calculation process for the impedance difference maxima is as follows: In the formula, ΔZ max As the maximum value of the impedance difference; The calculation process for pole frequency offset is as follows: In the formula, Δf res As the frequency offset of the dominant resonant pole; The center frequency between adjacent steady-state poles of the dominant resonance of the cable under test; The dominant resonant center frequency between adjacent steady-state poles of a healthy cable; The calculation process for the active admittance differential maximum is as follows: In the formula, ΔG max As the active admittance differential maximum The active admittance of the cable under test is given. The active admittance of a healthy cable.

8. The method for diagnosing local defects in cable insulation based on differential spectrum composite characteristics according to claim 1, characterized in that, The diagnosis of local defects in cable insulation in step S4 includes the following process: The impedance difference maximum ΔZ max Compared with the preset macroscopic coarse screening threshold T h1 A comparison is performed, where the impedance difference maximum value ΔZ max Greater than the macroscopic coarse screening threshold T h1 If the test cable is found to have undergone moisture aging dominated by volume, it is determined that the cable under test has undergone moisture aging. If the impedance difference maximum value ΔZ max Equal to or less than the macroscopic coarse screening threshold T h1 If the test cable has a local defect, it is determined that the cable under test has a local defect.

9. The method for diagnosing local defects in cable insulation based on differential spectrum composite characteristics according to claim 8, characterized in that, The local defect diagnosis of cable insulation in step S4 also includes the following process: With the dominant resonant pole frequency offset Δf res With the horizontal axis as the finite difference maximum value ΔG of the active admittance, the value is... max Using the vertical axis as the ordinate, a two-dimensional feature space is mapped out. The slope K of the eigenvector of the measured feature point in this two-dimensional feature space is calculated using the following formula: ； Based on the feature vector slope K and the preset mechanism decoupling threshold T h2 The comparison is performed, and the type of local defect in the insulation of the tested cable is determined based on the comparison results.

10. The method for diagnosing local defects in cable insulation based on differential spectrum composite characteristics according to claim 9, characterized in that, The feature vector slope K is decoupled from the preset mechanism decoupling threshold T. h2 During the comparison process: If the slope K of the feature vector is greater than the decoupling threshold T of the mechanism h2 The local defect type of the tested cable insulation is determined to be a water tree defect dominated by pure water soluble polarization; if the slope K of the eigenvector is equal to or less than the mechanical decoupling threshold T... h2 The local defect type of the tested cable insulation was determined to be an electrical tree defect with a high risk of breakdown and an accompanying carbonization leakage mechanism.

Images