A method and device for early warning of cable joint insulation fault breakdown
By using hollow conductor arc surface modules and thin-film thermocouple arrays on cable joints, combined with differential input and spectrum analysis, the problem of inaccurate identification of insulation faults in cable joints is solved, enabling high-precision detection of minute temperature rises and early warning of faults.
Patent Information
- Application Number
- CN202611033787.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-13
- Publication Date
- 2026-11-10
AI Technical Summary
Existing technologies struggle to accurately identify the complete process of cable joint insulation faults from initial development to breakdown and short circuit, especially lacking accuracy for minute temperature rises within 0.1℃, making it difficult to accurately predict the critical point when discharge develops into a breakdown and short circuit.
The module is filled with heat-conducting fluid around a hollow conductor arc surface. Combined with a thin-film thermocouple array and differential input electrical signals, it can accurately identify and warn of insulation faults in cable joints by using temperature difference, spectrum analysis and location correlation determination.
It significantly improves the accuracy of identifying minute temperature rises within 0.1℃, reduces false alarms and missed alarms, and enables early prediction and graded alerts for cable joint insulation breakdown faults.
Smart Images

Figure CN123066493A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the technical field of electrical fault early warning and converter station testing instruments, and in particular to a method and device for early warning of cable joint insulation fault breakdown. Background Technology
[0002] High-voltage cable joints are a common site for insulation breakdown faults in cable lines. Currently, the industry widely uses technologies such as high-frequency discharge detection, ultrasonic signal monitoring, and partial discharge electrical signal acquisition and analysis to conduct insulation fault early warning. These methods mainly rely on discharge characteristic signals to determine anomalies and are widely used in engineering sites.
[0003] Relying solely on discharge characteristic signals makes it difficult to accurately assess the complete process of insulation defects from initial development to breakdown and short circuit. While some cable joints may exhibit minute, continuous discharge signals, they do not develop into short-circuit faults in the short term. Furthermore, the lack of a clear correlation between discharge amplitude, fault development rate, and remaining lifespan makes it difficult to accurately predict the critical point at which discharge develops into a breakdown and short circuit. Extensive research and engineering practice indicate that the core precursor to insulation breakdown is thermal collapse, meaning that abnormal heating occurs during the degradation of the insulating medium.
[0004] Existing temperature detection technologies have low efficiency in diagnosing minute heating, but are highly accurate in identifying early abnormal temperature rises above 0.1°C. However, before thermal collapse occurs in insulating materials, a minute temperature rise of less than 0.1°C will first appear. This signal is a key early characteristic of insulation degradation, but existing solutions have low accuracy in identifying minute temperature rises of less than 0.1°C. Summary of the Invention
[0005] To improve the accuracy of identifying cable joint temperature rises with a slight temperature rise below 0.1℃, this application provides a method and device for early warning of cable joint insulation fault breakdown.
[0006] Firstly, this application provides a method for early warning of insulation fault breakdown in cable joints, employing the following technical solution: A method for early warning of insulation failure breakdown in cable joints includes a module surrounded by multiple arc-shaped surfaces configured as hollow conductors, filled with a heat-conducting fluid. The multiple arc-shaped surfaces surround the grounded cable joint to be tested after being combined. Multiple thin-film thermocouple arrays that contact the outer surface of the cable joint are laid on the surface of the arc-shaped surfaces facing the cable joint. The surface of the thin-film thermocouple arrays is an insulating film and is arranged in an S-shape along the circumference of the cable joint. At least two thin-film thermocouple arrays are arranged in parallel. The thin-film thermocouple arrays are electrically connected to a data acquisition module, and the data acquisition module is electrically connected to an analysis module. The acquisition module acquires electrical signals from the thin-film thermocouple array, uses the electrical signals corresponding to the two parallel thin-film thermocouple arrays as differential input electrical signals, and outputs an interference value based on a preset interference strategy according to the differential input electrical signals. If the interference value is greater than the preset interference setting value, an interference signal prompt is issued; otherwise, the acquisition module converts the electrical signals into temperature data. The analysis module acquires temperature data from different locations in the thin-film thermocouple array, calculates the maximum and average values of multiple temperature data, and calculates the temperature difference between the maximum and average values. Based on the temperature data, a curve is fitted for a preset acquisition time. The curve is analyzed using a spectrum conversion algorithm to obtain spectrum data with corresponding frequency and energy dimensions. The energy value of the frequency dimension in the spectrum data within a preset frequency range is taken as the element energy value. If the temperature difference is greater than the preset set difference and the element energy value is greater than the preset energy set value, then it is determined whether there is a positional correlation between the thin film thermocouple array position corresponding to the maximum value and the thin film thermocouple array position corresponding to the element curve; if there is a positional correlation, a breakdown warning is issued and the corresponding thin film thermocouple array position is indicated; otherwise, an abnormal detection warning is issued.
[0007] By adopting the above technical solution, a hollow arc-shaped module carrying heat-conducting fluid is combined with a circumferentially S-shaped parallel array of thin-film thermocouples to achieve close contact with the cable joint for full-area temperature measurement. The dual-array differential input mechanism effectively shields the interference of the on-site environment, significantly improving the accuracy of identifying minute temperature rises within 0.1℃. At the same time, it integrates temperature difference calculation, time-series temperature curve spectrum analysis and characteristic energy screening, and combines the position correlation verification logic of abnormal high temperature points and temperature rise fluctuation characteristic points. This overcomes the technical limitations of traditional discharge monitoring, which cannot accurately judge the entire process of insulation degradation, and conventional temperature detection, which is difficult to capture the early characteristics of minute temperature rises in the early stage of thermal collapse. It solves the problems of no correlation between discharge signals and fault development rates and difficulty in predicting fault critical points. It can accurately identify early insulation degradation hazards of cable joints, accurately locate defect areas, effectively reduce false alarms and missed alarms, and realize early prediction and graded warning of cable joint insulation breakdown faults.
[0008] Furthermore, the step of identifying whether there is a positional correlation between the thin-film thermocouple array position corresponding to the maximum value and the thin-film thermocouple array position corresponding to the element curve also includes the following steps: The maximum value is obtained based on the temperature difference. The thermocouple number of the thin film thermocouple in the thin film thermocouple array corresponding to the maximum value is obtained. The corresponding position data is matched from the preset thermocouple arrangement library based on the thermocouple number as the position of the thin film thermocouple array corresponding to the maximum value and marked as the first array position. Obtain the corresponding element curve based on the element energy value, obtain the thermocouple number of the thin film thermocouple in the thin film thermocouple array corresponding to the element curve, match the corresponding position data from the preset thermocouple arrangement library based on the thermocouple number as the position of the thin film thermocouple array corresponding to the element curve, and mark it as the second array position. The quantity of elements in the first array position is calculated as the quantity of elements in the first position, and the quantity of elements in the second array position is calculated as the quantity of elements in the second position. If the difference between the quantity of elements in the first position and the quantity of elements in the second position is less than the preset quantity of elements in the position, then the elements in the first array position are paired one by one with the elements in the second array position that are closest to each other, and the number of successful pairings is calculated as the pairing quantity. The minimum value between the quantity of elements in the first position and the quantity of elements in the second position is used as the base quantity. The ratio of the number of pairs to the base number is the pairing percentage. If the pairing ratio is greater than the preset percentage, then there is a positional association; otherwise, there is no positional association.
[0009] By adopting the above technical solution, the problems of misjudgment caused by random temperature changes in the environment, single-point detection errors, and scattered interference signals can be effectively avoided. It can accurately distinguish between non-faulty scattered temperature anomalies and the same-source coupling temperature rise defects caused by insulation deterioration, and greatly improve the objectivity and accuracy of position correlation determination.
[0010] Furthermore, the method also includes the following steps: The distribution area of the elements in the first array position is obtained as the first distribution area, and the area of the first distribution area is calculated as the area of the first region; the distribution area of the elements in the second array position is obtained as the second distribution area, and the area of the second distribution area is calculated as the area of the second region. The shape similarity is calculated between the shape of the first distribution region and the shape of the second distribution region. If the shape similarity is greater than the preset similarity, the ratio of the shape similarity to the preset similarity is calculated as the similarity ratio. The preset percentage is then adjusted based on the negative correlation of the similarity ratio.
[0011] By adopting the above technical solutions, the dynamic adaptive adjustment of the correlation judgment criteria can be achieved; the location correlation judgment criteria can be supplemented from the dual dimensions of regional scope and spatial morphology; and a comprehensive judgment can be carried out by fully combining the actual diffusion morphology and distribution characteristics of the insulation deterioration heating area, so as to effectively adapt to the heating conditions of cable joints with different degrees of deterioration and different defect types.
[0012] Furthermore, the method also includes the following steps: The distribution area of the elements in the first array position is obtained as the first distribution area, and the area of the first distribution area is calculated as the area of the first region; the distribution area of the elements in the second array position is obtained as the second distribution area, and the area of the second distribution area is calculated as the area of the second region. The difference between the area of the first region and the area of the second region is calculated as the area difference. If the area difference is less than the preset area difference, the ratio of the area difference to the preset area difference is calculated as the area difference ratio. The set percentage is adjusted according to the positive correlation of the area difference ratio.
[0013] By adopting the above technical solution, and fully considering the actual characteristics of synchronous diffusion and high coupling of the heat generation area of cable joint insulation deterioration, the judgment threshold is adaptively optimized based on the degree of difference in area, thus making up for the one-sidedness of simply relying on point pairing for correlation judgment.
[0014] Furthermore, the method also includes the following steps: After the breakdown warning is issued, within a preset time period, multiple temperature difference values and multiple element energy values are obtained, and the time points corresponding to the temperature difference values and the time points corresponding to the element energy values are set to be out of sync. The mean of the temperature difference is calculated as the temperature difference mean, and the mean of the elemental energy values is calculated as the energy mean. If the average temperature difference is less than the preset set difference and the average energy is less than the preset energy set value, then the breakdown warning will be canceled.
[0015] By adopting the above technical solution, the drawbacks of one-sided judgment based on instantaneous data at a single moment are eliminated, and non-fault-related sudden abnormal interference such as instantaneous temperature changes, short-term signal disturbances, and occasional local temperature rises are effectively filtered out, thereby reducing false alarms caused by instantaneous fluctuation data.
[0016] Furthermore, the method also includes the following steps: Acquire the electrical signals of parallel thin-film thermocouple arrays; calculate the difference between the electrical signals of two adjacent thin-film thermocouple arrays to obtain the electrical difference signal; If an electrical differential signal exceeds the preset electrical setting signal, an abnormal electromagnetic interference warning will be issued.
[0017] By adopting the above technical solutions, real-time online monitoring and proactive early warning of complex electromagnetic interference at high-voltage sites can be achieved; external influencing factors such as electromagnetic coupling and electric field interference can be accurately identified from the signal acquisition source, and signal distortion problems can be detected in a timely manner to avoid electromagnetic interference causing distortion of temperature measurement data and deviation in subsequent algorithm analysis.
[0018] Furthermore, the method also includes the following steps: If the electrical differential signal is less than the electrical setting signal, and there is an electrical signal greater than the preset electrical setting signal, then the electrical signal is extracted as the analysis signal. Within the recently set analysis time, the signal of a single thermocouple in the thin-film thermocouple array corresponding to the analysis signal is taken as the conversion signal. The converted signal is subjected to spectral analysis to obtain the converted frequency domain data. The energy value within the set discharge frequency range in the converted frequency domain data is taken as the discharge energy value. If the discharge energy value is greater than the preset minimum energy value, a discharge abnormality prompt is issued.
[0019] By adopting the above technical solutions, the shortcomings of single temperature measurement can be overcome, early weak discharge anomalies can be accurately identified, and the comprehensiveness of cable joint insulation defect investigation and fault early warning can be improved.
[0020] Furthermore, the method also includes the following steps: The number of arc surfaces surrounding the module is even. Two adjacent arc surfaces surrounding the module form a group. The outer shell of each group of arc surfaces surrounding the module is electrically connected to a differential amplifier module. The differential amplifier module outputs the differential signal of each group of arc surfaces surrounding the module as an environmental interference signal. If the environmental interference signal is greater than the preset interference setting signal, an environmental interference warning will be issued.
[0021] By adopting the above technical solution, an even number of arc surfaces are grouped in pairs around the module. The differential signal of the shell is collected by the differential amplification module to monitor environmental interference in real time. This can accurately capture external stray interference and issue timely prompts, effectively reducing the impact of complex on-site environmental interference on the detection signal.
[0022] Furthermore, the method also includes the following steps: If there are multiple sets of arc surfaces surrounding the module, the difference in environmental interference signals between each set of arc surfaces surrounding the module is calculated as the environmental interference difference. The minimum value of the environmental interference signal in each group is extracted as the interference reference signal. If the environmental interference difference is greater than the environmental interference difference, an environmental interference alarm is triggered.
[0023] By adopting the above technical solutions, local abnormal interference can be accurately identified and alarms can be triggered, further enhancing the ability to suppress environmental interference, ensuring the purity of detection signals, and improving the anti-interference performance of the overall monitoring system.
[0024] Secondly, this application provides a cable joint insulation fault breakdown early warning device, which adopts the following technical solution: A cable joint insulation fault breakdown early warning device includes a processor, wherein the processor performs the steps of the cable joint insulation fault breakdown early warning method as described in any one of the above claims. Attached Figure Description
[0025] Figure 1 This is a step diagram of a cable joint insulation fault breakdown early warning method.
[0026] Figure 2 This is a schematic diagram of the integrated monitoring structure in Implementation Example 1.
[0027] Reference numerals: 1. Arc-shaped surface surrounding the module; 2. Thin-film thermocouple array. Detailed Implementation
[0028] The embodiments of this application are described in detail below, and examples of the embodiments are shown in the accompanying drawings.
[0029] This application discloses a method for early warning of insulation failure breakdown in cable joints. The method is mainly applied to the long-term online status monitoring of intermediate joints and terminal joints of 10kV~500kV high-voltage power cables. It targets progressive defects such as insulation aging, internal air gap discharge, local damage, and moisture deterioration in cable joints. It focuses on the early characteristic detection of trace temperature rise within 0.1℃ and low-frequency periodic thermal fluctuations in the early stage of insulation thermal collapse. It solves the industry pain points such as the inability of traditional partial discharge monitoring to predict the development process of faults, insufficient sensitivity of conventional temperature measurement, large interference from strong electromagnetic environments, and ambiguous defect location. It can achieve accurate identification and source tracing early warning in the incipient stage of insulation defects and is suitable for various complex engineering sites such as substations, cable tunnels, outdoor direct burial, and pipe corridors.
[0030] Reference Figure 1 and Figure 2 This method is based on an integrated monitoring structure, and the specific hardware configuration is as follows: The arc-shaped surfaces surrounding module 1 are arranged in 2, 4, or 6 groups, depending on the diameter of the cable joint. They are made of hollow conductor material, with a central angle of 90° or 60° for each module. The hollow conductor structure is encapsulated in high-strength epoxy resin, providing both mechanical support and electromagnetic shielding. The module's interior is a sealed hollow cavity, quantitatively filled with a specialized heat-conducting fluid such as liquid ethanol, which provides rapid heat transfer from the cable joint's surface, enabling efficient transmission of minute temperature rise signals. It also isolates the module from external temperature fluctuations, reducing temperature-related interference. Multiple arc-shaped surfaces surrounding module 1 are joined together using flanges and elastic gaskets to completely enclose the cable joint under test. The cable joint's metal shell is welded to a grounding terminal with a grounding wire cross-section of not less than 16mm², ensuring reliable grounding to shield against high-voltage alternating electric fields and stray induced currents, providing a stable electrical environment for weak signal acquisition.
[0031] Thin-film thermocouple array 2: Two or more thin-film thermocouple arrays 2 are neatly laid out on the inner surface of each arc surface surrounding module 1 facing the cable joint. A gapless bonding process is used, with a bonding pressure of 0.1MPa~0.3MPa, ensuring tight contact between the array and the outer insulation layer of the cable joint, eliminating heat conduction hysteresis caused by air gaps. The thin-film thermocouple array 2 uses ultra-thin polyimide-based thermocouples with a surface composite polytetrafluoroethylene insulating film, 10μm~20μm thick, and a breakdown voltage ≥10kV / mm, achieving high-voltage electrical isolation and component protection. All thin-film thermocouple arrays 2 are arranged continuously in an S-shaped meandering pattern along the circumference of the cable joint, with a meandering spacing of 5mm~10mm, covering 360° circumferentially without blind spots; and at least two arrays are arranged parallel along the axial direction, with a spacing of 20mm~30mm, providing hardware support for differential signal acquisition and common-mode interference cancellation.
[0032] Signal processing module: The thin-film thermocouple array 2 is electrically connected to the front-end acquisition module via a twisted-pair shielded cable (with the shielding layer grounded). The acquisition module and the analysis module exchange data via industrial Ethernet or 4G / 5G wireless communication to construct a closed-loop monitoring system. The acquisition module is equipped with an instrumentation amplifier, a 24-bit AD converter, and multi-stage anti-interference filtering circuits; the analysis module is a high-real-time embedded digital signal processor (DSP) with built-in simplified edge analysis algorithms and a threshold parameter library.
[0033] During the routine monitoring phase, the acquisition module and the analysis module work together to complete signal acquisition, interference identification, and data preprocessing. The specific steps are as follows: Electrical signal acquisition: The acquisition module continuously acquires the raw electrical signals output by all thin-film thermocouple arrays 2 in real time. The raw electrical signals are thermoelectric potential signals with an amplitude range of 10μV~100μV. The acquisition timestamp and the corresponding thermocouple number are recorded synchronously.
[0034] Differential interference identification: The acquisition module selects the real-time electrical signals of two parallel thin-film thermocouple arrays 2 as differential input electrical signals, and processes the differential signals based on a preset interference strategy, which includes a fusion of 50Hz power frequency filtering, spike pulse suppression, and a moving average filtering algorithm. The steps for processing the differential signals are as follows: power frequency and harmonic interference are filtered out by an IIR digital filter, random noise is suppressed by a 3-point moving average algorithm, and finally the peak value and RMS value of the differential signal are extracted, and the comprehensive interference value is calculated and output.
[0035] Interference detection and signal conversion: The acquisition module compares the real-time interference value with the preset interference setting value. The preset interference setting value is based on the actual measurement and calibration of the electromagnetic environment on site and is usually set to 10μV. If the interference value is greater than the interference set value: it is determined that there is an abnormality such as strong electromagnetic interference or drastic temperature change in the environment. The local sound and light prompts and the background remote interference signal prompts are immediately triggered. The time of interference occurrence, intensity and array number involved are recorded simultaneously to facilitate maintenance personnel to find the source of interference. If the interference value is less than or equal to the interference set value: the monitoring environment is determined to be stable and the signal is not distorted. The acquisition module starts the signal conversion program, and combines the graduation value of the thin-film thermocouple with the linear calibration coefficient of the multi-point calibration to convert the weak analog thermoelectric potential signal into standardized digital temperature data, which is then uploaded to the analysis module in real time in the format of "thermocouple number-acquisition time-temperature value".
[0036] The analysis module receives the temperature data from multiple points across the entire region uploaded by the acquisition module, and performs multi-dimensional in-depth calculations and analysis. The specific process is as follows: Basic temperature data calculations: The analysis module iterates through all valid temperature data within the current monitoring period. If the current monitoring period is 1 minute, it filters out the maximum temperature T_max and the average temperature across the entire region T_avg, and calculates the temperature difference ΔT=T_max-T_avg, which intuitively reflects the magnitude of local abnormal temperature rise. Using a preset acquisition duration as the period, such as 30 seconds, the least squares method is used to smooth and fit the continuous temperature data to generate a continuous, noise-free temperature time-series change curve, thus restoring the dynamic change law of temperature. The temperature time series curve is converted from time domain to frequency domain using Fast Fourier Transform to generate spectral data containing frequency dimension (0Hz~50Hz) and energy value dimension (unit: dBm). Spectral signals with frequencies within a preset frequency range (0.1Hz~10Hz, corresponding to low-frequency thermal fluctuation characteristics of insulation degradation) are selected, and the total energy value of this frequency band is calculated by integration and defined as the element energy value E.
[0037] Dual-threshold linkage judgment: The analysis module adopts a dual-threshold linkage mechanism of temperature amplitude and fluctuation characteristics. When both of the following conditions are met, it enters the location association verification stage: Temperature difference ΔT > preset difference. The preset difference is based on the thermal degradation test calibration of insulation material and is usually set to 0.05℃, which is the critical value of slight temperature rise in the early stage of insulation thermal collapse. The element energy value E is greater than the preset energy setting value. The preset energy setting value is based on the statistical data of normal cable joint monitoring and is usually set to -60dBm.
[0038] If any condition is not met, it is determined that there is no risk of abnormal failure and the system returns to continuous monitoring status.
[0039] Once the dual threshold criteria are met, the analysis module initiates location correlation identification to accurately distinguish between faults originating from the same source and random anomalies. The specific steps are as follows: Abnormal location marking and positioning: First array position marking: Based on the temperature difference ΔT exceeding the standard, the corresponding maximum temperature T_max is locked, and the unique number of the thin-film thermocouple to which T_max belongs is retrieved, such as: Module 1-Array 2-Point 36. Relying on the preset thermocouple arrangement library, which stores the circumferential angle coordinates θ (0°~360°), axial coordinates z (0mm~500mm), module number, and region division identifier of all thermocouples, the precise spatial position data is obtained by matching the numbers. The DBSCAN clustering algorithm (neighborhood radius 5mm, minimum number of points 3) is used to delineate the abnormal coverage area and mark it as the first array position, i.e., the high temperature abnormal area; Second array position marking: Based on the energy value E of the excessive element, the corresponding characteristic element curve is locked, that is, the frequency domain curve with the largest energy contribution in this frequency range. The thin film thermocouple number corresponding to the curve is located. Similarly, the position data is matched by the layout library, and the abnormal range of temperature rise fluctuation is delineated by the same clustering algorithm and marked as the second array position.
[0040] Position element quantity comparison: The number of valid abnormal thermocouple points within the first array position is counted and denoted as the element quantity N1 of the first position. Valid points must satisfy the condition that the temperature exceeds the threshold for 3 consecutive sampling cycles. The number of valid abnormal thermocouple points within the second array position is counted and denoted as the element quantity N2 of the second position. Calculate the element quantity difference ΔN=|N1-N2| and compare it with the preset element quantity N0 at the location. N0 is usually set to 20% of the larger value of N1 and N2. If ΔN>N0, it means that the size difference between the two abnormal areas is too large, and it is directly determined that there is no location correlation. If ΔN≤N0, proceed to the point matching stage.
[0041] Point pairing and association determination: Point pairing: Using the straight-line distance in space as the criterion, all abnormal points in the first array position and the second array position are paired one by one according to the principle of proximity. A pairing threshold is set, such as a straight-line distance ≤ 8mm. The number of successful pairings that meet the threshold requirement is counted M. Determining the base quantity: Select the minimum value between the quantity of the element at the first position N1 and the quantity of the element at the second position N2 as the base quantity N_base, that is, N_base=min(N1, N2); Pairing percentage calculation: P = (M / N_base) × 100%; Final determination: Compare the pairing percentage P with the preset percentage P0, which is initially set to 70%. If P > P0: It is determined that the positions of the first and second arrays are strongly correlated and belong to the same source fault caused by the same insulation defect. Immediately output the cable joint insulation breakdown warning prompt, and simultaneously mark the circumferential / axial coordinates of the abnormal area, the module and array number, and the defect coverage area on the background interface to guide the operation and maintenance personnel to carry out targeted repairs. If P≤P0: the two anomalies are determined to be independent of each other and caused by non-fault factors such as environmental disturbances or single-point detection errors. Only an anomaly detection prompt is output, and no fault warning is triggered, effectively reducing false alarms.
[0042] After extracting the contours and calculating the basic parameters of the two abnormal regions, this method further introduces quantitative evaluation of regional morphological similarity and dynamic threshold correction logic.
[0043] Specifically, based on the spatial coordinates and continuous distribution boundaries of each abnormal thermocouple location, the first distribution region corresponding to the first array position and the second distribution region corresponding to the second array position are delineated respectively. The areas of the first and second regions are calculated through geometric boundary fitting and regional gridding. Simultaneously, multi-dimensional morphological parameters such as contour curvature, circumferential extension shape, boundary orientation, and local concavity / convexity features of the two distribution regions are extracted. A contour feature matching algorithm is used for global comparison, and the shape similarity S between the first and second distribution regions is quantitatively calculated.
[0044] The real-time calculated shape similarity S is compared with the system's preset fixed similarity S0. Only when S > S0 does it indicate that the overall outline, diffusion pattern, and extension range of the high-temperature anomaly region and the temperature rise fluctuation anomaly region are highly similar, belonging to the coupled heating characteristics caused by the same insulation defect, thus meeting the prerequisite for threshold adjustment. Based on this, the similarity ratio K is calculated using the formula: K = S / S0. This similarity ratio K is used to characterize the degree of overlap between the shapes of the two anomaly regions. The larger the value of K, the higher the consistency of the region shape and the stronger the homology of the defects.
[0045] This scheme employs a quantitative adjustment mechanism that negatively correlates the similarity ratio with the corrected set percentage: Using the system's initial default set percentage P0 as the baseline, a fixed adjustment coefficient α is set. The final real-time set percentage P used for position association determination after correction satisfies the calculation formula: P = P0 - α × (K - 1); where the adjustment coefficient α is a field calibration constant, and its value range can be set to 5%–15%; simultaneously, gradient intervals are defined to constrain the adjustment amplitude and avoid excessive parameter deviation. When 1 < K ≤ 1.2, the morphological similarity is slightly high, so the set percentage is slightly lowered to reduce the threshold for association determination. When 1.2 < K ≤ 1.5, the morphology is highly consistent, and the setting percentage is adjusted downwards by a moderate amount; When K > 1.5, the outlines of the two regions are basically completely identical, so the set percentage is significantly reduced and the judgment conditions are relaxed. Conversely, the closer the similarity ratio is to 1, the greater the difference in regional morphology, the smaller the percentage reduction, and the stricter the judgment criteria.
[0046] After completing the initial calculation of the point pairing percentage, this method further introduces an abnormal area area matching degree analysis mechanism. By quantifying the difference in coverage scale between two abnormal areas, the set percentage threshold for location association determination is dynamically optimized, making up for the limitations of single point pairing or morphological analysis, and improving the accuracy of determining the same source fault under complex working conditions.
[0047] Distribution area extraction and area quantification calculation: Precise delineation of distribution areas: Based on the spatial coordinate data (circumferential angle θ, axial coordinate z) of all abnormal thermocouple points within the first and second array positions, the analysis module uses an α-shape algorithm (α value set to 5mm, adapted to thermocouple spacing) to fit the outer contour boundary of each abnormal point, forming a closed and continuous first and second distribution areas. This fitting method can accurately restore the actual boundary of irregular heating areas, avoiding area calculation deviations caused by rectangular selection, and is particularly suitable for the heating characteristics of irregular shape defects such as local air gaps and damage in cable joints.
[0048] Accurate area calculation: For the delineated first and second distribution areas, the actual coverage area is calculated using a polygon area integration algorithm. The closed contour boundary is discretized into several continuous tiny triangular units, with the included angle of each unit's vertex ≤ 5°; The area of each triangular unit is calculated sequentially based on the coordinate data, specifically using the vector cross product formula: S_Δ=1 / 2×|x_1(y_2-y_3)+x_2(y_3-y_1)+x_3(y_1-y_2)|, where (x_1, y_1), (x_2, y_2), and (x_3, y_3) are the coordinates of the triangle vertices; By summing the areas of all triangular units, we obtain the area of the first region A_1 and the area of the second region A_2. The calculation accuracy is kept to 0.1 mm² to ensure that subtle changes in area difference can be captured.
[0049] Determination of area difference and calculation of adjustment parameters: Area difference comparison: Calculate the absolute difference between the areas of the first and second regions, defined as area difference ΔA = |A_1 - A_2|. The system calls the preset area difference value A_0, which is based on the typical defect size of the cable joint and is usually set to 15%~25% of the maximum abnormal area area. For example, when the maximum area is 100mm², A_0 is set to 20mm². ΔA is then compared with A_0 in real time. If ΔA≥A_0: it indicates that the coverage scale of the two abnormal areas is significantly different, which does not meet the characteristics of synchronous diffusion of the same insulation defect. The set percentage adjustment is not activated, and the initial set percentage P_0 is maintained for judgment. If ΔA < A_0: This indicates that the two abnormal regions are highly matched in size and have the characteristics of the same source fault, and the threshold dynamic adjustment stage is entered.
[0050] Quantification of area difference ratio: Calculate the area difference ratio R_A using the formula: R_A=ΔA / A_0; This ratio represents the relative difference in area between two regions. The value of R_A ranges from [0, 1). The closer R_A is to 0, the higher the area consistency and the greater the probability of common source failure.
[0051] This scheme employs a quantitative adjustment logic that correlates the area difference ratio with the corrected set percentage: the smaller the area difference, i.e., the smaller R_A, the more consistent the coverage of the defect heating area, the stronger the homology, and the more appropriately the set percentage threshold can be lowered to avoid misjudgment due to minor area deviations; conversely, the larger the area difference, i.e., the larger R_A, the higher the set percentage threshold needs to be, and the judgment criteria need to be strictly enforced. The specific adjustment method is as follows: The adjustment formula is defined as follows: Based on the initial set percentage P_0 (e.g., 70%), an area adjustment coefficient β is introduced. β is calibrated based on field tests and has a value range of 8% to 12%, with a default of 10%. The corrected real-time set percentage P_A satisfies: P_A = P_0 - β × (1 - R_A). When R_A=0 (areas are completely identical), P_A=P_0-β, and the judgment threshold is reduced to the maximum extent; when R_A approaches 1 (area difference is close to the set difference), P_A approaches P_0, and the original judgment standard is maintained.
[0052] Gradient-based adjustment constraints: To avoid excessive threshold shift, adjustment gradients are defined based on actual engineering conditions. When 0≤R_A≤0.3 (area difference is very small): P_A=P_0-0.12×(1-R_A), the maximum reduction is 12%. For example, when P_0=70%, P_A can be as low as 58%, which greatly relaxes the judgment condition. When 0.3 < R_A ≤ 0.7 (moderate area difference): P_A = P_0 - 0.10 × (1 - R_A), moderate reduction, 3%~7%; When 0.7 < R_A < 1 (area difference is close to the set threshold): P_A = P_0 - 0.08 × (1 - R_A), slightly lowered, by 0% to 2.4%, close to the initial judgment standard.
[0053] Execution after adjustment: The analysis module uses the dynamically calculated P_A as the final position association determination threshold, replacing the initial percentage P_0, and compares it with the previously calculated pairing percentage P: If P > P_A: It is determined that the areas of the two abnormal regions match and the heights of the points overlap, which are the same source insulation defects, triggering a breakdown warning and precise location; If P ≤ P_A: it is determined to be a non-homogeneous random anomaly, and only a detection prompt is output.
[0054] To avoid false alarms caused by non-continuous anomalies such as transient environmental interference and occasional signal fluctuations, this method adds a closed-loop correction mechanism of time sequence verification and dual mean verification after triggering the breakdown warning. Through delayed acquisition, misaligned sampling, and statistical mean comparison, it accurately distinguishes between true insulation degradation and transient anomalies, further improving the reliability and engineering adaptability of the warning results.
[0055] Review cycle and sampling parameter settings: Preset time period definition: After the analysis module triggers a breakdown warning, it automatically starts the verification timer. The preset time period T_check is 3~5 minutes, based on the progressive characteristics of insulation degradation: the temperature rise and thermal fluctuations of true insulation defects will continue to exist, while transient interferences usually recover to stability within 3 minutes. During the verification period, the acquisition module and the analysis module continue to work uninterruptedly, maintaining the original monitoring frequency and signal processing logic.
[0056] Staggered sampling mechanism design: To avoid the synchronous impact of environmental interference at the same time, such as power frequency harmonics and instantaneous impacts from equipment start-up and shutdown, on the two types of data, a staggered sampling strategy is adopted: Temperature difference data acquisition: Continuous acquisition at a frequency of 10 seconds / time, i.e., the acquisition time points are t_0, t_0+10s, t_0+20s, ..., t_0+(n-1)×10s, where n is the number of acquisitions. 18 sets are collected within 3 minutes and 30 sets are collected within 5 minutes. Element energy value acquisition: The acquisition frequency is also 10 seconds / time, but the acquisition time points are staggered by 5 seconds, namely t_0+5s, t_0+15s, t_0+25s, ..., t_0+5s+(n-1)×10s, to ensure that the acquisition time of the two types of data does not overlap and to reduce the impact of synchronization interference on the mean calculation.
[0057] Data validity screening: During the data collection process, outliers that exceed the normal physical range are automatically removed. The 3σ criterion is adopted: values that deviate from the mean of the data type by ±3 times the standard deviation are removed, and valid data are retained for subsequent mean calculation, thereby reducing the distortion of statistical results by extreme outliers.
[0058] Double mean calculation and threshold comparison: Accurate calculation of the mean: Temperature difference mean calculation: For all valid temperature difference values collected within the review period, a weighted average algorithm is used to calculate the temperature difference mean. The weighting coefficient is positively correlated with the acquisition time, with later data having a higher weight, reflecting the dynamic changes in insulation status. The formula is: = / Where m is the number of effective temperature differences, ΔT_i is the temperature difference of the i-th acquisition, and w_i is the weight of the i-th acquisition, w_i=1+i×0.05, with a weight range of 1~2.5, to ensure that the later data have a more significant impact on the mean. Energy mean calculation: The energy mean is calculated using an arithmetic mean algorithm for all valid element energy values collected during the same period. The formula is: = Where k is the number of effective elemental energy values, E_j is the elemental energy value collected in the j-th time, and the arithmetic mean can objectively reflect the overall level of energy fluctuation.
[0059] Dual threshold linkage verification: The analysis module calls the preset setting difference ΔT_0 (e.g., 0.05℃) and energy setting value E_0 (e.g., -60dBm) consistent with the initial judgment, and calculates the... With ΔT_0, Perform a synchronous comparison with E_0: Validation logic: Only when <ΔT_0 and If <E_0 is satisfied simultaneously, the breakdown warning is determined to be a false trigger caused by a transient anomaly; Special case handling: If only a single mean is below the threshold, such as <ΔT_0 but If the value is ≥E_0, or vice versa, it is determined that there is still a potential risk in the insulation condition, the breakdown warning is maintained, and the "to be observed" status is marked, triggering the next round of short-cycle review, with the cycle shortened to 1 minute.
[0060] Warning cancellation and status feedback: Warning Cancellation: When the double mean verification meets the cancellation conditions, the analysis module immediately sends a command to simultaneously execute three operations: Turn off the local sound and light warning device and terminate the background system's pop-up windows, text messages and other warning notifications. Record key data such as the warning trigger time, cancellation time, average temperature difference during the review period, and average energy in the monitoring log to form a complete warning closed-loop record, which is convenient for later traceability and analysis; Send alert cancellation and instantaneous anomaly status reports to the operations and maintenance platform, explaining the reason for cancellation, to avoid incorrect responses from operations and maintenance personnel.
[0061] Subsequent monitoring and connection: After the warning is cancelled, the system will automatically resume the normal monitoring mode, maintain the original collection frequency and analysis logic, but highlight the monitoring data of the cable joint, and continuously track the temperature difference and element energy value change trend within 1 hour. If the double threshold exceeds the standard again, the review cycle will be shortened to 2 minutes to improve the response sensitivity.
[0062] To further enhance interference identification capabilities in complex electromagnetic environments under high voltage conditions and overcome the limitations of single differential input interference detection, this method also employs an electromagnetic interference monitoring mechanism that compares the signal differences between adjacent parallel arrays. By capturing signal imbalances between adjacent arrays, it accurately identifies hidden interferences such as local electromagnetic coupling and non-uniform electric field interference, ensuring data reliability from the signal acquisition source. The specific implementation process is as follows: Precise acquisition and synchronous control of electrical signals: Acquisition Objects and Scope: While acquiring all electrical signals of the thin-film thermocouple array 2 in real time, the acquisition module separately extracts the original electrical signals of all parallel thin-film thermocouple array 2, namely the thermoelectric potential analog signals, with an amplitude range of 10μV~100μV, and stores them according to the array number, such as "parallel array 1", "parallel array 2", etc., to ensure a one-to-one correspondence between the signals and the arrays.
[0063] Synchronous sampling guarantee: Relying on the high-precision synchronous clock built into the acquisition module, synchronous sampling is performed on all parallel thin-film thermocouple arrays 2. The sampling frequency is consistent with the conventional temperature signal acquisition frequency (100Hz), ensuring that the electrical signals of adjacent arrays have time homogeneity at the same time and avoiding the deviation in difference calculation caused by sampling time difference.
[0064] Signal preprocessing: The extracted parallel array electrical signal is first subjected to primary filtering using an RC low-pass filter circuit with a cutoff frequency of 1kHz to filter out high-frequency glitches and noise, while retaining the electromagnetic interference characteristic components in the original signal, without affecting the accuracy of subsequent temperature signal conversion.
[0065] Electrical differential signal calculation and feature extraction: Adjacent array pairing rules: Parallel arrays are paired in pairs according to their physical arrangement order. The physical arrangement order is adjacent along the axial or circumferential direction of the cable joint. For example, parallel array 1 is paired with parallel array 2, and parallel array 2 is paired with parallel array 3 (if there are multiple sets of parallel arrays). This ensures that the paired arrays are spatially adjacent to maximize the distinguishability of electromagnetic interference differences.
[0066] Difference signal operation logic: For each pair of adjacent arrays, at the same acquisition time, such as t_1, t_2, ..., t_n, perform point-by-point difference calculation of electrical signals to obtain the electrical difference signal ΔU_{i,j}(t). The calculation formula is: ΔU_{i,j}(t)=|U_i(t)-U_j(t)|; where U_i(t) is the instantaneous value of the electrical signal of the i-th parallel array at time t, and U_j(t) is the instantaneous value of the electrical signal of the j-th parallel array adjacent to i at the same time t. The absolute value operation ensures that the difference signal is non-negative, which is convenient for threshold comparison.
[0067] Difference signal feature quantization: To avoid misjudgment caused by signal fluctuations at a single moment, the electrical difference signal within 10 consecutive acquisition cycles (100ms) is statistically analyzed, and the peak value ΔU_peak and effective value ΔU_rms of each cycle are extracted. The effective value is used as the main criterion for judgment, as the effective value can better reflect the continuous intensity of the interference, while the peak value is used as an auxiliary reference.
[0068] Threshold comparison and interference alert triggering: Electrical setting signal calibration: The preset electrical setting signal U_set is determined based on the actual measurement and simulation test of the high voltage field electromagnetic environment. Combined with the signal noise level of the thin film thermocouple array 2, the normal effective value of noise is ≤2μV, and it is usually calibrated to 5μV~8μV. For example, 6μV is selected. This value can both shield normal signal fluctuations and accurately capture signal differences caused by electromagnetic interference.
[0069] Real-time comparison and judgment: The analysis module compares the effective value of the electrical difference signal ΔU_rms of each adjacent array pair with the electrical setting signal U_set in real time. If there exists any adjacent array pair with ΔU_rms > U_set, and this state lasts for more than 3 acquisition cycles (30ms), to avoid accidental triggering by instantaneous pulse interference, it is determined that there is local abnormal electromagnetic interference on site, and the interference intensity has affected the signal acquisition accuracy. If ΔU_rms≤U_set for all adjacent array pairs, then the current electromagnetic environment is considered stable and there is no local coupling interference.
[0070] Precise Interference Alert Output: When an abnormal electromagnetic interference alert is triggered, the system performs a triple feedback operation: Local notification: The monitoring device's audible and visual alarm is activated, emitting a high-frequency buzzer and flashing red light to alert on-site personnel; Background alerts: Push alerts to the operation and maintenance management platform for abnormal electromagnetic interference and signal imbalance between adjacent arrays, including key data such as the time of interference occurrence, the number of adjacent arrays involved, the effective value of the electrical difference signal, and the amplitude of the signal exceeding the set value; Location marking: On the array layout diagram of the monitoring interface, the locations of adjacent arrays affected by interference are highlighted to help maintenance personnel locate the source of interference. For example, arrays near a high-voltage device or busbar are more susceptible to interference.
[0071] Interference response and follow-up processing coordination: Signal acquisition adaptation adjustment: After the interference warning is triggered, the acquisition module automatically starts the anti-interference enhancement mode, adds a notch filter to the electrical signal of the affected array, specifically filters out the 50Hz power frequency and its 3rd and 5th harmonics, and increases the signal sampling rate to 200Hz to enhance signal redundancy and reduce the impact of interference on temperature data conversion.
[0072] Interference source tracing assistance: The system records data such as interference duration, intensity change trend, and array distribution involved. Combined with the layout of equipment around the cable joint, such as the location of surge arresters, switch cabinets, and transformers, it generates an interference source tracing analysis report, prompting maintenance personnel to focus on investigating potential interference sources such as high-voltage equipment discharge, poor grounding, and line coupling near adjacent arrays.
[0073] Monitoring logic linkage: If abnormal electromagnetic interference lasts for more than 5 minutes without relief, the system will automatically pause the fault determination in that area, only retaining interference monitoring and data recording. Normal analysis will resume after the interference is eliminated, thus avoiding false fault determinations under interference conditions.
[0074] After initial identification of abnormal electromagnetic interference, this method further improves the accuracy of identifying partial discharge anomalies in cable joints. The specific implementation process is as follows: Analysis of signal extraction and validity determination: Extraction Prerequisite Verification: The system first verifies two prerequisites to ensure the reliability of the extracted signal: Electromagnetic interference elimination: If the electrical difference signal ΔU_rms between adjacent parallel arrays is less than the preset electrical setting signal U_set, such as 6μV, it is determined that there is no significant electromagnetic interference on site and the signal is not distorted due to external coupling. Signal exceeding the limit determination: There is a real-time electrical signal U of the thin-film thermocouple array 2 that is greater than the preset electrical setting signal U_sig. This value is calibrated based on the normal temperature measurement signal range of the thermocouple and is usually set to 150μV. This is higher than the upper limit of the conventional temperature measurement thermoelectric potential signal and can also capture the weak electrical signal of partial discharge coupling.
[0075] If both conditions are met, the analysis signal extraction will be initiated; if either condition is not met, the normal temperature monitoring mode will be maintained and the discharge detection process will not be initiated.
[0076] Precise signal selection for analysis: The acquisition module traverses all electrical signals of the thin-film thermocouple array 2, locates all signal sources that satisfy U > U_sig, and classifies and stores them in the format of "array number - thermocouple number - signal amplitude - acquisition timestamp". These signals exceeding the threshold are extracted as analysis signals. To avoid false triggering by single-point instantaneous noise, a signal duration threshold is set: only when the electrical signal maintains U > U_sig for 3 consecutive sampling cycles (30μs, sampling rate 100kHz) is it determined to be a valid analysis signal; otherwise, it is regarded as random noise and discarded.
[0077] Signal filtering and preprocessing: Analysis duration calibration: The preset analysis duration T_ana is 3-5 seconds. Based on the periodic characteristics of partial discharge signals, a duration of 3 seconds or more can cover at least 3 discharge cycles, ensuring complete feature capture. The system extracts the most recent valid analysis signal within T_ana, focuses on the thin-film thermocouple array 2 corresponding to the signal, and preferentially selects the single thermocouple with the largest signal amplitude and longest duration as the target sensing unit. Its output continuous time-series electrical signal is the conversion signal. If multiple thermocouples in the same array output valid analysis signals, the group of thermocouples with the highest signal correlation (Pearson correlation coefficient ≥ 0.8) is selected, and its average signal is taken as the conversion signal to improve signal representativeness.
[0078] Conversion signal preprocessing optimization: The selected conversion signals undergo two-stage preprocessing to remove non-discharge characteristic interference. First stage: Use a band-stop filter (center frequency 50Hz, bandwidth 5Hz) to filter out power frequency and harmonic interference, and avoid power grid noise affecting the analysis of discharge signals; The second stage employs a wavelet thresholding denoising algorithm, selecting a db4 wavelet basis and decomposing the signal into three layers. The threshold is adaptively adjusted according to the Birgé-Massart criterion to suppress random noise and preserve the spike pulse characteristics of the discharge signal. The preprocessed data must meet a signal-to-noise ratio ≥20dB to ensure the accuracy of subsequent spectrum analysis.
[0079] Spectrum analysis and discharge energy calculation: Spectrum conversion parameter settings: Fast Fourier Transform is used to perform time-domain to frequency-domain conversion on the preprocessed converted signal. The number of sampling points is set to 8192, and the frequency resolution is ≤10Hz to ensure accurate identification of discharge characteristic frequencies. After conversion, converted frequency domain data containing frequency dimension (0Hz~500kHz) and energy value dimension (unit: dBm) are generated, fully covering the typical frequency range of partial discharge.
[0080] Discharge frequency range calibration: Based on the electrical characteristics of partial discharge in high-voltage cable joints, the discharge frequencies of different defect types vary, such as air gap discharge (10kHz~1MHz) and surface discharge (50kHz~500kHz). The preset discharge frequency range is 10kHz~1MHz, which avoids the power frequency interference band and fully covers the characteristic frequency range of common partial discharges.
[0081] Discharge energy value quantification calculation: Integrate the energy values of all frequency points falling within the 10kHz~1MHz frequency band in the converted frequency domain data to obtain the total energy value of that frequency band, defined as the discharge energy value E_dis. The calculation formula is: E_dis = Where f_1 = 10 kHz, f_2 = 1 MHz, and S(f) is the power spectral density function of the converted frequency domain data. The calculation results are rounded to 0.1 dBm to ensure that subtle changes in energy differences can be captured.
[0082] Discharge anomaly detection and prompting: Minimum Energy Value Calibration: The preset minimum energy value E_min is calibrated based on extensive experimental data: For healthy cable joints without discharge defects, the background energy value in the 10kHz~1MHz frequency band is monitored (typically ≤-65dBm); for cable joints with slight discharge defects, the measured discharge energy value is ≥-55dBm. Therefore, E_min is set to -60dBm, which can both shield background noise and identify weak discharge signals at an early stage.
[0083] Dual-judgment logic: The analysis module compares the calculated discharge energy value E_dis with E_min: If E_dis > E_min, and this condition is met for two consecutive analysis cycles (each cycle is 3-5 seconds), it is determined that there is a partial discharge abnormality in the cable joint, and the discharge intensity has reached a level that requires attention. If E_dis≤E_min, or only a single cycle satisfies E_dis>E_min, it is determined to be background noise or transient pulse, and no discharge abnormality prompt is triggered; only the relevant data is recorded in the monitoring log.
[0084] Multi-dimensional prompt output: When a discharge anomaly is triggered, the system constructs a feedback system of "tiered prompts + precise source tracing": Local notification: The monitoring device activates a special audible and visual notification, which is different from electromagnetic interference and breakdown warnings. It uses a flashing yellow light and intermittent buzzer to help on-site personnel quickly identify the type of fault. Background alert: Pushes "partial discharge anomaly" alert information to the operation and maintenance platform, including key parameters such as discharge occurrence time, array number and thermocouple coordinates involved, discharge energy value, and set discharge frequency range, and attaches a spectrum analysis chart to intuitively display the discharge characteristics; Linked source tracing: Combining previous temperature monitoring data, the location relationship (overlap, adjacent or unrelated) between the discharge anomaly area and the high temperature anomaly area is marked, providing a reference for the determination of defect type (such as air gap discharge, surface discharge).
[0085] To build a multi-layered, comprehensive interference protection system, environmental interference monitoring capabilities can also be strengthened at the equipment structure level. Specific methods are as follows: Hardware configuration and grouping design: Number of arc surfaces surrounding module 1: An even number of arc surfaces must be used to surround module 1, such as 2 groups, 4 groups, or 6 groups. The specific number is adapted to the diameter of the cable connector to be tested: 2 groups for diameters ≤150mm, 4 groups for diameters 150mm~300mm, and 6 groups for diameters >300mm. This ensures that all modules can be perfectly paired with each other, with no leftover units, guaranteeing the integrity of the grouped coverage. The central angle of each group of modules is consistent: 180° for 2 groups, 90° for 4 groups, and 60° for 6 groups. After splicing, the cable connector is seamlessly wrapped, ensuring both mechanical stability and uniform electromagnetic induction.
[0086] Grouping and pairing rules: According to the physical order of module installation, two adjacent arc surfaces are divided into a group around module 1, such as module 1 and module 2 as group 1, module 3 and module 4 as group 2, and so on. The splicing points of adjacent modules are connected by conductive metal clips to ensure the electrical continuity of the module shells within the group, while maintaining insulation isolation between groups. Through the built-in insulating pads of the clips, the breakdown voltage is ≥10kV to avoid signal crosstalk between groups.
[0087] Differential amplifier module configuration: Each set of arc surfaces is welded with signal lead terminals around the outer shell of module 1. The terminals are made of brass with a gold-plated surface to reduce contact resistance. They are electrically connected to the dedicated differential amplifier module via shielded signal wires (shield grounded, core wire cross-section 0.5mm²). The differential amplifier module uses high-voltage compatible chips, possessing strong anti-interference capabilities and weak signal amplification capabilities, making it suitable for harsh electrical environments in high-voltage fields.
[0088] Environmental interference signal acquisition and processing: Signal sensing principle: The arc-shaped surface surrounding module 1 is made of hollow conductor material. Its outer shell will induce a current / voltage signal that is positively correlated with the intensity of environmental interference in the complex electromagnetic environment of the high-voltage field. Since the two adjacent modules in the group are in similar electromagnetic environments, the induced signals tend to be consistent under normal circumstances. However, when there is external environmental interference, the induced signals of the two groups of modules will differ. The interference characteristic signal can be accurately extracted by differential amplification.
[0089] Differential signal amplification processing: The differential amplification module performs differential operations and amplifies the sensed signals from the outer casings of the two modules within each group. The specific process is as follows: First stage: The input buffer circuit isolates the module housing from the amplifier circuit to prevent the load from affecting the sensed signal; Second level: Perform differential operation, the calculation formula is U_{diff}=U_{shell1}-U_{shell2}, where U_{shell1} and U_{shell2} are the induced voltage signals of the shells of the two modules in the group, respectively. The third stage: linearly amplifies the differential signal according to the preset amplification factor. The preset amplification factor is based on the on-site interference intensity calibration and is usually set to 1000 times to 5000 times to ensure that weak interference signals can be effectively detected. At the same time, high-frequency glitches and noise are filtered out through the built-in low-pass filter (cutoff frequency 10kHz). Output: The amplified and filtered differential signal is the environmental interference signal U_env. The signal amplitude range is converted to 0~10V for easy subsequent threshold comparison.
[0090] Synchronous signal acquisition: The differential amplifier module and the front-end acquisition module use the same synchronous clock to synchronously acquire the environmental interference signal U_env of each group at a frequency of 50Hz, and attach the group number and acquisition timestamp information, and upload it to the analysis module for storage and analysis.
[0091] Threshold comparison and interference alert triggering: Interference setting signal calibration: The preset interference setting signal U_env-set is determined based on field measurements and simulation experiments: In a healthy environment without significant interference, the measured ambient interference signal amplitude is ≤2mV; when electromagnetic interference that affects temperature measurement accuracy is present, the signal amplitude is ≥8mV. Therefore, U_env-set is usually calibrated to 8mV, which can be adjusted according to the complexity of the field electromagnetic environment, with an adjustment range of 5mV~12mV. This threshold can both shield normal environmental background noise and accurately capture effective interference.
[0092] Real-time comparison and judgment logic: The analysis module performs real-time monitoring and threshold comparison of each group of uploaded environmental interference signals U_env. Single group determination: If U_env > U_env-set for any group, and this state lasts for more than 5 collection cycles (100ms), to avoid accidental triggering by instantaneous pulse interference, then it is determined that there is environmental interference in the area corresponding to that group; Global Interference Judgment: If more than 50% of the groups simultaneously detect U_env > U_env-set, it is judged as global environmental interference; if only 1 to 2 groups detect it, it is judged as regional environmental interference. The group number and corresponding physical area involved in the interference are marked simultaneously, such as "Group 2 - Area below the cable joint".
[0093] Multi-dimensional interference prompt execution: When an environmental interference prompt is triggered, the system outputs targeted feedback based on the interference type (regional / global): Local prompts: The sound and light alarms of the monitoring device will activate differentiated prompts, such as yellow light + intermittent buzzer for regional interference, and red light + continuous buzzer for global interference, so that on-site personnel can quickly distinguish the interference range; Background alert: Push "environmental interference" alert information to the operation and maintenance management platform, including key data such as interference type, occurrence time, involved group number, environmental interference signal amplitude, and amplitude exceeding the set signal, and simultaneously highlight the interference area on the module layout diagram; Data labeling: The temperature signals and discharge detection signals collected during this period are labeled with "interference status" to facilitate the removal of data affected by interference during subsequent data analysis, or to correct the analysis results based on the interference situation.
[0094] Interference response and monitoring logic linkage: Front-end data acquisition adaptation adjustment: Upon detecting environmental interference, the data acquisition module automatically activates anti-interference optimization strategies. An adaptive filtering algorithm is added to the electrical signal of the thin-film thermocouple array 2 to dynamically adjust the filter parameters according to the frequency of the interference signal and suppress the interference frequency band signal in a targeted manner. To increase the redundancy of temperature data acquisition, the sampling frequency was increased from 100Hz to 200Hz, and interference was reduced by smoothing through multiple sets of data. If there is strong interference across the entire area (U_env > 2 × U_env-set), temporarily suspend the discharge anomaly detection process, prioritize ensuring the stability of the temperature signal, and avoid misjudgment of the discharge signal caused by interference.
[0095] Interference Source Tracing Assistance: The system combines the distribution of interference areas and the trend of interference signal amplitude changes to generate an interference source tracing report. Regional interference: Maintenance personnel are advised to focus on checking high-voltage equipment (such as transformers, surge arresters, switchgear), grounding devices, cable accessories, etc. near the corresponding module area to see if there are any problems such as discharge, poor grounding, or line coupling. Global interference: This indicates that you should check whether there are external factors such as the start-up or shutdown of large equipment, lightning strikes, or power grid failures around the monitoring point, or whether there are any abnormalities in the grounding system of the monitoring device.
[0096] Continuous monitoring and recovery mechanism: The system continuously tracks changes in environmental interference signals. When the U_env of all groups is less than or equal to U_env-set and remains stable for more than 3 minutes, the environmental interference prompt will be automatically terminated, the normal monitoring mode will be restored, and the interference elimination time and recovery status will be recorded in the monitoring log.
[0097] For application scenarios where multiple sets of arc surfaces surround module 1 (e.g., 4 sets or 6 sets), this method further enhances the ability to finely identify environmental interference. Through inter-group interference signal difference analysis, it accurately captures local sudden strong interference and regional uneven interference characteristics, overcoming the limitation of single-set interference judgment in distinguishing between "globally uniform interference" and "locally abnormal interference." This provides a more accurate basis for interference source tracing and targeted protection. The specific implementation process is as follows: Preprocessing of multiple sets of environmental interference signals: Multiple group judgment and signal synchronization: The system first determines that the number of arc surfaces surrounding module 1 is ≥2 (i.e., meets the "multiple groups" condition), and then initiates the inter-group interference difference analysis process. All differential amplification modules and acquisition modules of all groups are strictly synchronized (clock error ≤1μs). Within the same time window, such as 100ms, including 5 acquisition cycles, the environmental interference signals U_{env-1}, U_{env-2}, ..., U_{env-n} output by each group are acquired respectively, where n≥2 is the number of groups, to ensure the temporal homogeneity of the signals between groups and avoid the deviation in difference calculation caused by sampling time difference.
[0098] Intra-group signal time-domain statistics: For each group, the "intra-group average interference signal" U_{env-avg-i}, i=1,2,…,n, is calculated using the arithmetic mean algorithm for the 5 environmental interference signal values collected within the time window. The formula is: U_{env-avg-i}=1 / 5× Where U_{env-ik} is the environmental interference signal value collected in the kth time for the i-th group. Instantaneous impulse noise within each group is filtered out by time-domain averaging to ensure signal stability for inter-group comparisons.
[0099] Calculation of environmental interference difference: Pairing rules between groups: The "full pairing" method is adopted to perform pairing calculations on all groups without omission. For example, the pairing combinations of 4 modules are (group 1, group 2), (group 1, group 3), (group 1, group 4), (group 2, group 3), (group 2, group 4), (group 3, group 4), ensuring that the interference differences between any two groups can be captured and avoiding omission of local strong interference areas.
[0100] Inter-group difference quantization: For each pair of combinations, calculate the absolute difference between the two groups of "intra-group average interference signals", defined as the environmental interference difference ΔU_{env-pq}, calculated as follows: ΔU_{env-pq}=|U_{env-avg-p}-U_{env-avg-q}|; where p and q are different group numbers, p≠q, and the absolute difference ensures that the interference difference is non-negative, facilitating unified threshold comparison. The interference differences of all pairs of combinations constitute the "inter-group interference difference set" {ΔU_{env-12}, ΔU_{env-13}, ..., ΔU_{env-(n-1)n}}.
[0101] Interference reference signal extraction and threshold calibration: Interference reference signal definition: Extract the minimum value of the "intra-group average interference signal" of all groups as the interference reference signal U_{ref}, with the formula: U_{ref}=min{U_{env-avg-1}, U_{env-avg-2}, ..., U_{env-avg-n}}; This reference signal represents the cleanest and weakest interference baseline level in the current environment. The smaller its value, the higher the probability of strong local interference. Using this as a baseline, the difference in interference between groups can be accurately amplified.
[0102] Environmental interference difference threshold calibration: The preset environmental interference difference threshold ΔU_{env-th} is calibrated based on multiple sets of module interference test data: Under normal uniform electromagnetic conditions, the interference difference between groups is ≤3mV; when a sudden interference source exists near a group, the difference between groups is ≥10mV. Therefore, ΔU_{env-th} is usually calibrated to 10mV, but can be adjusted according to the complexity of the on-site environment, with an adjustment range of 8mV~15mV. This threshold can both shield normal minor differences between groups and accurately identify abnormal interference differences.
[0103] Threshold comparison and alarm triggering: Multi-level judgment logic: The analysis module compares each ΔU_{env-pq} in the "inter-group interference difference set" one by one: Single pair determination: If ΔU_{env-pq} > ΔU_{env-th} for any pair combination, and this state lasts for more than 2 time windows (200ms), it is determined that there is a significant interference difference between the two groups of modules corresponding to the pair combination, that is, there is local strong interference in the region where one or both groups are located. Interference intensity classification: Calculate the ratio of ΔU_{env-pq} to ΔU_{env-th}, K_env=ΔU_{env-pq} / ΔU_{env-th}: when 1<K_env≤1.5, it is "slight local interference"; when 1.5<K_env≤2, it is "moderate local interference"; when K_env>2, it is "severe local interference".
[0104] Precise alarm output: When an environmental interference alarm is triggered, the system outputs multi-dimensional feedback consisting of "tiered alarm + area positioning": Local alarm: The monitoring device activates a graded audible and visual alarm. For mild interference, it is indicated by a yellow light and a slow-frequency buzzer; for moderate interference, it is indicated by an orange light and a medium-frequency buzzer; and for severe interference, it is indicated by a red light and a high-frequency buzzer. This is different from the usual environmental interference warning. Background alert: Push "local strong interference alarm" information to the operation and maintenance platform, including key data such as interference level, involved pairing combination (such as "group 2-group 3"), corresponding physical area (such as "left side of cable joint-lower area"), environmental interference difference, and interference reference signal. Simultaneously highlight the two groups of areas with the greatest difference on the module layout diagram. Data traceability: Automatically associates the time-series curves of environmental interference signals of each group during the period with the trend chart of inter-group difference changes, which facilitates the operation and maintenance personnel to analyze the occurrence and development patterns of interference.
[0105] Linkage processing and engineering adaptability: Enhanced front-end data acquisition protection: After an alarm is triggered, the system activates targeted anti-interference strategies for the affected area. For the signals of the thin-film thermocouple array 2 corresponding to the difference group, the "directional filtering" algorithm is enabled. Based on the frequency characteristics of the interference difference, a Notch filter is designed to suppress local interference frequency bands. Increase the signal sampling rate of the difference group to 300Hz, and reduce the impact of interference on temperature and discharge signal through multiple sets of data smoothing and redundancy verification; If the interference is severe and localized, suspend the detection of abnormal discharge in that area, prioritize the stability of temperature data, and resume detection once the interference subsides.
[0106] Precise guidance for interference tracing: Based on the regional distribution of paired combinations, it provides targeted tracing suggestions for operations and maintenance personnel: If the difference between "Group 2 and Group 3" is too large, it is suggested to focus on checking the area near the cable joints covered by the two groups of modules, such as the intermediate joint flange and accessory connection, to see if there are any problems such as loosening, discharge, or poor grounding. If the differences between multiple groups and a certain group (such as group 4) are all too large, the area where group 4 is located is determined to be the core area of the interference source, and it is suggested to prioritize the investigation of the coupling of high-voltage equipment and lines around this area.
[0107] Continuous monitoring and recovery mechanism: The system updates the inter-group interference difference every 50ms. When the ΔU_{env-pq} of all paired combinations is less than or equal to ΔU_{env-th} and remains stable for more than 2 minutes, the environmental interference alarm is automatically terminated, the normal monitoring mode is restored, and data such as alarm duration and interference peak value are recorded to form a complete traceability file.
[0108] This application also discloses a cable joint insulation fault breakdown early warning device, including a processor, wherein the processor executes the steps of the cable joint insulation fault breakdown early warning method as described in any of the above embodiments.
[0109] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.
Claims
1. A method for early warning of insulation fault breakdown in cable joints, characterized in that, The module (1) is surrounded by multiple arc surfaces that are set as hollow conductors and filled with heat-conducting fluid. The multiple arc surfaces surround the module (1) and then surround the grounded cable connector to be tested. A circular arc surface surrounding the surface of the module (1) facing the cable connector is covered with multiple thin-film thermocouple arrays (2) that contact the outer surface of the cable connector. The surface of the thin-film thermocouple arrays (2) is an insulating film and is placed in an S-shape along the circumference of the cable connector. At least two thin-film thermocouple arrays (2) are arranged in parallel. The thin-film thermocouple arrays (2) are electrically connected to the acquisition module, and the acquisition module is electrically connected to the analysis module. The acquisition module acquires the electrical signals on the thin film thermocouple array (2), takes the electrical signals corresponding to the two parallel thin film thermocouple arrays (2) as differential input electrical signals, outputs the interference value based on the differential input electrical signals and the preset interference strategy. If the interference value is greater than the preset interference setting value, the interference signal prompt is given. Otherwise, the acquisition module converts the electrical signal into temperature data; The analysis module acquires temperature data at different locations in the thin-film thermocouple array (2), calculates the maximum and average values of multiple temperature data, and calculates the difference between the maximum and average values as the temperature difference; fits a curve based on the temperature data for a preset acquisition time, analyzes the curve based on the spectrum conversion algorithm to obtain spectrum data with corresponding frequency and energy dimensions, and takes the energy value of the frequency dimension in the spectrum data within the preset set frequency range as the element energy value; If the temperature difference is greater than the preset set difference and the element energy value is greater than the preset energy set value, then identify whether the position of the thin film thermocouple array (2) corresponding to the maximum value and the position of the thin film thermocouple array (2) corresponding to the element curve are related; if they are related, then issue a breakdown warning and indicate the corresponding position of the thin film thermocouple array (2). Otherwise, an abnormal detection message will be displayed.
2. The cable joint insulation fault breakdown early warning method according to claim 1, characterized in that, The step of identifying whether there is a positional correlation between the position of the thin-film thermocouple array (2) corresponding to the maximum value and the position of the thin-film thermocouple array (2) corresponding to the element curve, also includes the following steps: The maximum value is obtained based on the temperature difference. The thermocouple number of the thin film thermocouple in the thin film thermocouple array (2) corresponding to the maximum value is obtained. The corresponding position data is matched from the preset thermocouple arrangement library according to the thermocouple number as the position of the thin film thermocouple array (2) corresponding to the maximum value, and marked as the first array position. Obtain the corresponding element curve based on the element energy value, obtain the thermocouple number of the thin film thermocouple array (2) corresponding to the element curve, match the corresponding position data from the preset thermocouple arrangement library based on the thermocouple number as the position of the thin film thermocouple array (2) corresponding to the element curve, and mark it as the second array position. The quantity of elements in the first array position is calculated as the quantity of elements in the first position, and the quantity of elements in the second array position is calculated as the quantity of elements in the second position. If the difference between the quantity of elements in the first position and the quantity of elements in the second position is less than the preset quantity of elements in the position, then the elements in the first array position are paired one by one with the elements in the second array position that are closest to each other, and the number of successful pairings is calculated as the pairing quantity. The minimum value between the quantity of elements in the first position and the quantity of elements in the second position is used as the base quantity. The ratio of the number of pairs to the base number is the pairing percentage. If the pairing ratio is greater than the preset percentage, then there is a positional association; otherwise, there is no positional association.
3. The cable joint insulation fault breakdown early warning method according to claim 2, characterized in that, The method also includes the following steps: The distribution area of the elements in the first array position is obtained as the first distribution area, and the area of the first distribution area is calculated as the area of the first region; the distribution area of the elements in the second array position is obtained as the second distribution area, and the area of the second distribution area is calculated as the area of the second region. The shape similarity is calculated between the shape of the first distribution region and the shape of the second distribution region. If the shape similarity is greater than the preset similarity, the ratio of the shape similarity to the preset similarity is calculated as the similarity ratio. The preset percentage is then adjusted based on the negative correlation of the similarity ratio.
4. The cable joint insulation fault breakdown early warning method according to claim 2, characterized in that, The method also includes the following steps: The distribution area of the elements in the first array position is obtained as the first distribution area, and the area of the first distribution area is calculated as the area of the first region; the distribution area of the elements in the second array position is obtained as the second distribution area, and the area of the second distribution area is calculated as the area of the second region. The difference between the area of the first region and the area of the second region is calculated as the area difference. If the area difference is less than the preset area difference, the ratio of the area difference to the preset area difference is calculated as the area difference ratio. The set percentage is adjusted according to the positive correlation of the area difference ratio.
5. The cable joint insulation fault breakdown early warning method according to claim 1, characterized in that, The method also includes the following steps: After the breakdown warning is issued, within a preset time period, multiple temperature difference values and multiple element energy values are obtained, and the time points corresponding to the temperature difference values and the time points corresponding to the element energy values are set to be out of sync. The mean of the temperature difference is calculated as the temperature difference mean, and the mean of the elemental energy values is calculated as the energy mean. If the average temperature difference is less than the preset set difference and the average energy is less than the preset energy set value, then the breakdown warning will be canceled.
6. The cable joint insulation fault breakdown early warning method according to claim 1, characterized in that, The method also includes the following steps: Obtain the electrical signals of the parallel thin-film thermocouple array (2); calculate the difference between the electrical signals of two adjacent thin-film thermocouple arrays (2) to obtain the electrical difference signal; If an electrical differential signal exceeds the preset electrical setting signal, an abnormal electromagnetic interference warning will be issued.
7. The cable joint insulation fault breakdown early warning method according to claim 6, characterized in that, The method also includes the following steps: If the electrical differential signal is less than the electrical setting signal, and there is an electrical signal greater than the preset electrical setting signal, then the electrical signal is extracted as the analysis signal. Within the recently set analysis time, the signal of a single thermocouple in the thin-film thermocouple array (2) corresponding to the analysis signal is taken as the conversion signal; The converted signal is subjected to spectral analysis to obtain the converted frequency domain data. The energy value within the set discharge frequency range in the converted frequency domain data is taken as the discharge energy value. If the discharge energy value is greater than the preset minimum energy value, a discharge abnormality prompt is issued.
8. The cable joint insulation fault breakdown early warning method according to claim 1, characterized in that, The method also includes the following steps: The number of arc surfaces surrounding module (1) is even. Two adjacent arc surfaces surrounding module (1) are grouped together. Each group of arc surfaces surrounding module (1) is electrically connected to a differential amplifier module. The differential amplifier module outputs the differential signal of each group of arc surfaces surrounding module (1) as an environmental interference signal. If the environmental interference signal is greater than the preset interference setting signal, an environmental interference warning will be issued.
9. The cable joint insulation fault breakdown early warning method according to claim 8, characterized in that, The method also includes the following steps: If there are multiple sets of arc surfaces surrounding module (1), the difference in environmental interference signals between each set of arc surfaces surrounding module (1) is calculated as the environmental interference difference. The minimum value of the environmental interference signal in each group is extracted as the interference reference signal. If the environmental interference difference is greater than the environmental interference difference, an environmental interference alarm is triggered.
10. A cable joint insulation fault breakdown early warning device, characterized in that, Includes a processor, wherein the steps of the cable joint insulation fault breakdown early warning method as described in any one of claims 1-9 are executed.
