Oil paper insulation aging state evaluation method and system

Abstract

The invention provides an oil paper insulation aging state evaluation method and system, and relates to the technical field of intelligent monitoring of power equipment, and the method comprises the steps: carrying out the frequency domain dielectric spectrum test of an oil paper insulation structure, and obtaining a dielectric response curve; inputting the dielectric response curve into a preset neural network model, and outputting a moisture content evaluation value of the insulation paper; directly measuring the actually measured moisture content of the insulating oil in the oil paper insulating structure through a chemical detection method; determining a sample point coordinate in a preset oil paper moisture balance curve coordinate system according to the moisture content evaluation value of the insulation paper and the actually measured moisture content of the insulation oil, the oil paper moisture balance curve coordinate system comprising a plurality of reference curves representing different aging states; according to the method, the frequency domain dielectric spectrum and chemical detection are combined, a lossless preliminary screening-precise quantification system is formed to guarantee data reliability, aging intervals are divided according to polymerization degree standardization, and the universality of the method is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent monitoring of electric power equipment, and in particular to a method and system for evaluating the aging state of oil-paper insulation. Background Art

[0002] The dielectric properties of oil-paper insulation are not only affected by moisture, but polar products produced by aging may also change the shape of the tanδ frequency domain spectrum, resulting in a strong coupling between the effects of moisture and aging on the dielectric response. Both may increase the tanδ value in the medium and low frequency bands. In addition, under high moisture conditions, oil-paper insulation with different degrees of aging may exhibit almost overlapping dielectric spectrum curves, making it difficult to distinguish the aging status through a single dielectric spectrum feature. Although some commercial equipment attempts to reduce interference by introducing aging correction factors, these can only optimize the accuracy of moisture assessment and cannot achieve effective evaluation of the aging status in reverse.

[0003] On the other hand, the equilibrium distribution characteristics of moisture in oil-paper insulation have been used for moisture assessment. Some traditional moisture balance curves achieve indirect estimation of moisture in paper by describing the effect of temperature on moisture distribution. However, most existing balance curves are based on unaged oil-paper samples and may not take into account the changes in insulation material performance caused by aging. The ability of cellulose paperboard to bind water molecules may be weakened after aging, and the increased polarity of the insulating oil causes the saturated moisture content to increase, which may lead to differences in the moisture distribution ratio in oil-paper insulation with different aging degrees at the same temperature. Therefore, traditional moisture balance curves may not reflect the impact of aging on moisture distribution and are difficult to use for correlation assessment of aging status. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for evaluating the aging status of oil-paper insulation, combining frequency domain dielectric spectroscopy with chemical testing to form a non-destructive initial screening-precision quantification system to ensure data reliability, divide the aging interval according to polymerization degree standardization, and enhance the versatility of the method.

[0005] In order to solve the above technical problems, the technical solutions of the present invention are as follows: In a first aspect, a method for evaluating the aging state of oil-paper insulation is provided, the method comprising: Step S1: Performing a frequency domain dielectric spectrum test on the oil-paper insulation structure to obtain a dielectric response curve; Step S2: Input the dielectric response curve into a preset neural network model and output an estimated value of the moisture content of the insulating paper; Step S3: directly measuring the actual moisture content of the insulating oil in the oil-paper insulation structure by a chemical detection method; Step S4: determining the coordinates of the sample points in a preset oil-paper moisture balance curve coordinate system based on the estimated moisture content of the insulating paper and the measured moisture content of the insulating oil. The oil-paper moisture balance curve coordinate system includes multiple reference curves representing different aging states. Step S5: Calculate the geometric distance between the sample point coordinates and each reference curve, and select the reference curve corresponding to the minimum distance; Step S6: outputting an aging status evaluation result of the oil-paper insulation structure according to the aging status interval associated with the reference curve.

[0006] Furthermore, step S1: performing a frequency domain dielectric spectrum test on the oil-paper insulation structure to obtain a dielectric response curve, including: Prepare an oil-paper insulation sample library based on preset aging gradients and moisture gradients; Under constant temperature conditions, an AC voltage with a frequency range of 1mHz-1kHz is applied to the oil-paper insulation structure in the oil-paper insulation sample library; The dielectric response data under AC voltage is collected by frequency domain dielectric spectrometer; A dielectric response curve is generated according to the dielectric response data, with frequency as the abscissa and dielectric loss factor tan δ as the ordinate.

[0007] Furthermore, step S2: inputting the dielectric response curve into a preset neural network model to output an estimated value of the moisture content of the insulating paper, including: Extract the dielectric loss factor tanδ values ​​of all test points within the preset frequency range from the dielectric response curve; The tanδ value set is used as the input feature quantity and input into the pre-trained BP neural network model; The BP neural network model outputs an evaluation value of the moisture content of the insulation paper.

[0008] Furthermore, step S3: directly measuring the actual moisture content of the insulating oil in the oil-paper insulation structure by a chemical detection method, comprising: After the oil-paper insulation structure reaches a moisture equilibrium state, extracting an oil sample from the insulating oil; The moisture content of the oil sample is detected by Karl Fischer titration to obtain the actual moisture content of the insulating oil.

[0009] Furthermore, step S4: based on the estimated moisture content of the insulating paper and the measured moisture content of the insulating oil, the coordinates of the sample points are determined in a preset oil-paper moisture balance curve coordinate system, wherein the oil-paper moisture balance curve coordinate system includes multiple reference curves representing different aging states, including: The measured moisture content of the insulating oil is used as the horizontal coordinate, and the estimated moisture content of the insulating paper is used as the vertical coordinate to form the coordinates of the sample points; Map the sample point coordinates to the preset oil-paper moisture balance curve coordinate system; Identify the position of the sample point coordinates relative to multiple reference curves in the oil-paper moisture balance curve coordinate system; The multiple reference curves are respectively associated with different aging state intervals, and each reference curve is pre-calibrated through an oil-paper moisture balance experiment corresponding to the aging state.

[0010] Furthermore, step S5: calculating the geometric distance between the sample point coordinates and each reference curve, and selecting the reference curve corresponding to the minimum distance, including: In the oil-paper moisture balance curve coordinate system, calculate the vertical distance from the sample point coordinate to each reference curve; Compare the calculated vertical distance values ​​to obtain a comparison result; Based on the comparison results, the reference curve corresponding to the vertical distance with the smallest value is selected.

[0011] Furthermore, step S6: outputting an aging status evaluation result of the oil-paper insulation structure according to the aging status interval associated with the reference curve, including: Associating the reference curve with a preset polymerization degree aging state interval; The polymerization degree aging state interval is output as an aging state evaluation result of the oil-paper insulation structure, wherein the polymerization degree aging state interval includes: unaged interval A1: polymerization degree 1000-1300, early aging interval A2: polymerization degree 700-1000, mid-aging interval A3: polymerization degree 400-700, and late aging interval A4: polymerization degree 0-400.

[0012] In a second aspect, a system for evaluating the aging state of oil-paper insulation is provided, comprising: Acquisition module, used to perform frequency domain dielectric spectrum test on oil-paper insulation structure and obtain dielectric response curve; A moisture assessment module is used to input the dielectric response curve into a preset neural network model and output an assessment value of the moisture content of the insulation paper; a detection module, configured to directly measure the actual moisture content of the insulating oil in the oil-paper insulation structure by a chemical detection method; an aging assessment module for determining sample point coordinates in a preset oil-paper moisture balance curve coordinate system based on the estimated moisture content of the insulating paper and the measured moisture content of the insulating oil, wherein the oil-paper moisture balance curve coordinate system includes multiple reference curves representing different aging states; The calculation module is used to calculate the geometric distance between the sample point coordinates and each reference curve, and select the reference curve corresponding to the minimum distance; The output module is used to output the aging status evaluation result of the oil-paper insulation structure according to the aging status interval associated with the reference curve.

[0013] According to a third aspect, a computing device includes: one or more processors; The storage device is used to store one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method.

[0014] In a fourth aspect, a computer-readable storage medium stores a program, which implements the method when executed by a processor.

[0015] The above solution of the present invention includes at least the following beneficial effects: By integrating frequency-domain dielectric spectrum testing, neural network fitting, chemical testing, and moisture balance curve analysis, the cross-influence of moisture and aging is decoupled from a physical mechanism perspective. Even under high-moisture conditions, different degrees of aging can still be distinguished by differences in moisture distribution ratios. By integrating the advantages of multiple technologies, the full-band dielectric spectrum from 1mHz to 1kHz is used to retain complete characteristics, the BP neural network is used to achieve high-precision assessment of moisture in paper, and the Karl Fischer titration method is used to ensure the actual measurement accuracy of moisture in oil, taking into account both laboratory accuracy and field applicability. The pre-calibrated reference curve fits the actual laws of aging, faithfully reflects the impact of aging on moisture distribution, and makes the evaluation results directly related to the degree of polymerization interval, with clear physical meaning. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The present invention provides a flow chart of an oil-paper insulation aging status assessment method.

[0017] Figure 2 This is a schematic diagram of an oil-paper insulation aging status assessment system provided by an embodiment of the present invention.

[0018] Figure 3 This is the result of moisture absorption by the oil-impregnated insulating paperboard of the present invention.

[0019] Figure 4 This is the tan δ frequency domain spectrum of the cardboard of the present invention aged for 50 days under different moisture conditions.

[0020] Figure 5 It is the tan δ frequency domain spectrum of the paperboard with a moisture content of 1.25% under different aging conditions of the present invention.

[0021] Figure 6 are the three neural network moisture content evaluation values ​​of the present invention.

[0022] Figure 7 It is a comparison between the neural network moisture content evaluation value of the present invention and the measured value.

[0023] Figure 8 This is the moisture balance curve of oil paper at different temperatures and different aging states of the present invention.

[0024] Figure 9 It is the fitting equation of the moisture balance curve of oil paper under five different aging states of the present invention.

[0025] Figure 10 It is the insulation state of the sample to be identified in the present invention.

[0026] Figure 11 It is a comparison between the measured value and the estimated value of the moisture content in the paper of the present invention.

[0027] Figure 12 It is the moisture balance curve between the sample point to be evaluated and the oil paper of the present invention.

[0028] Figure 13 It is the aging status evaluation result of the present invention. DETAILED DESCRIPTION

[0029] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0030] like Figure 1 As shown, an embodiment of the present invention provides a method for evaluating the aging state of oil-paper insulation, the method comprising the following steps: Step S1: Performing a frequency domain dielectric spectrum test on the oil-paper insulation structure to obtain a dielectric response curve; Step S2: Input the dielectric response curve into a preset neural network model and output an estimated value of the moisture content of the insulating paper; Step S3: directly measuring the actual moisture content of the insulating oil in the oil-paper insulation structure by a chemical detection method; Step S4: determining the coordinates of the sample points in a preset oil-paper moisture balance curve coordinate system based on the estimated moisture content of the insulating paper and the measured moisture content of the insulating oil. The oil-paper moisture balance curve coordinate system includes multiple reference curves representing different aging states. Step S5: Calculate the geometric distance between the sample point coordinates and each reference curve, and select the reference curve corresponding to the minimum distance; Step S6: outputting an aging status evaluation result of the oil-paper insulation structure according to the aging status interval associated with the reference curve.

[0031] In an embodiment of the present invention, a mapping relationship between the dielectric response curve and the moisture content of the insulating paper is fitted by a neural network, combined with the measured moisture content of the insulating oil, and the aging specificity of the moisture balance curve is utilized to realize the decoupling evaluation of the moisture and aging states, thereby improving the evaluation accuracy; the full-band dielectric loss tangent value is used as the input feature, combined with the BP neural network, a two-variable nonlinear mapping model is established to reduce the moisture evaluation error; the oil-paper moisture balance curve corresponding to different polymerization degrees is constructed, and the distance method is used to quantitatively match the aging state, and the aging interval is accurately divided in accordance with industry standards; non-destructive frequency domain dielectric spectrum testing and chemical testing are combined to form a non-destructive primary screening-precise quantification system to ensure data reliability; a balance curve database covering multiple temperatures is constructed, and the aging interval is divided according to the polymerization degree standardization to improve the accuracy.

[0032] In a preferred embodiment of the present invention, the above step S1: performing a frequency domain dielectric spectrum test on the oil-paper insulation structure to obtain a dielectric response curve may include: Step S1.1, preparing an oil-paper insulation sample library based on a preset aging gradient and moisture gradient; Step S1.2, under a constant temperature condition, applying an AC voltage in a frequency range of 1 mHz-1 kHz to the oil-paper insulation structure in the oil-paper insulation sample library; Step S1.3, collecting dielectric response data under the action of AC voltage using a frequency domain dielectric spectrometer; Step S1.4: generating a dielectric response curve with frequency as the abscissa and dielectric loss factor tan δ as the ordinate according to the dielectric response data.

[0033] In an embodiment of the present invention, a multi-dimensional sample library covering aging degree and moisture content is systematically constructed, and a standardized gradient design is used to ensure data repeatability and compliance with industry standards such as DL / T596-2005. At the same time, a 1mHz-1kHz broadband AC voltage scan is adopted under a constant temperature of 45°C, and the dielectric response data of high and low frequency bands are collected by MeggerIDAX300. This not only avoids temperature interference and improves the accuracy of moisture assessment, but also distinguishes the dielectric response differences of moisture and aging through the tanδ characteristics of the medium, low and high frequency bands. The final generated visualized tanδ frequency domain curve intuitively reflects the comprehensive impact of moisture and aging on the dielectric properties of the entire frequency band.

[0034] In the embodiments of the present invention, when applied specifically, it can be achieved through the following technical solutions, for example: In the above step S1.1, oil-paper insulation samples with different aging degrees and moisture contents are prepared through accelerated thermal aging tests and moisture absorption tests, and dielectric response tests are performed on them. A frequency domain dielectric spectrum database is constructed, and the moisture status of oil-paper insulation is evaluated based on this database.

[0035] According to the accelerated thermal aging test, 2mm thick ordinary cellulose insulation cardboard was processed into a 160mm diameter disc, then placed on a stainless steel metal rack and placed in a vacuum drying and oil immersion box to dry for 48 hours. The drying conditions were 105°C and the vacuum degree was less than 1mbar. The dried insulation cardboard was quickly injected into dry and degassed 25# cycloalkyl mineral insulating oil and vacuum immersed in the oil for 48 hours at 60°C and a vacuum degree of less than 1mbar.

[0036] To simulate the accelerated thermal aging test, metal aging cans numbered A1, A2, A3, A4, and A5 were placed with dry oil-impregnated insulation paperboard and insulating oil (the ratio of insulation paperboard to insulating oil was 10:1), and an appropriate amount of copper strips were added. The cans were sealed, vacuumed, and ventilated with nitrogen. A1 was preserved as the unaged sample, and the others were placed in an aging chamber at 120°C for accelerated thermal aging tests. After aging for 10 days, 25 days, 50 days, and 90 days, the aging cans were removed in turn and left to stand at room temperature for a period of time. After the thermal aging products of the oil-paper insulation reached equilibrium, the cans were opened for subsequent testing.

[0037] In order to obtain cardboard with different moisture contents under the same aging state, it is necessary to conduct moisture absorption experiments on the cardboard separately, setting eight different moisture gradients, that is, each aging state corresponds to eight different moisture states; the moisture absorption experiment steps are as follows: remove the cardboard from the aging tank, wipe the surface clean, and weigh the initial mass. Based on the initial moisture content, initial mass and expected moisture content of the cardboard, calculate the expected mass of the cardboard after moisture absorption. The cardboard is placed in the air to absorb moisture naturally, and the mass of the cardboard is measured in real time. Moisture absorption is stopped when the expected mass is reached.

[0038] The polymerization degree and moisture content of the oil-paper insulation samples prepared in different aging and moisture conditions are as follows: Figure 3 As shown in the figure, each aging state sample corresponds to 8 moisture gradients, distributed between 1.0% and 5.0%.

[0039] In step S1.2 above, the experiment conducted a frequency domain dielectric response test on 40 oil-paper insulation samples under five different aging conditions, corresponding to eight different moisture conditions. The insulation samples were subjected to a temperature test at 45°C and balanced for 24 hours to eliminate the influence of temperature fluctuations on the dielectric properties. A Megger IDAX300 instrument was used, with an AC voltage amplitude of 200 V and a frequency sweep range of 1 mHz-1 kHz. The frequency points were distributed logarithmically (e.g., 1 mHz, 10 mHz, 100 mHz, 1 Hz…1 kHz).

[0040] In step S1.3 above, the Megger IDAX300 instrument automatically records the voltage amplitude, current amplitude, and phase difference at each frequency point and calculates the dielectric loss factor (tanδ) and complex dielectric constant (ε*).

[0041] Step S1.4, plot the tanδ value of the frequency point on the logarithmic frequency axis to form a tanδ-f curve. The frequency domain dielectric spectra of the sample aged for 50 days under different moisture conditions are as follows: Figure 4 As shown, the time-frequency domain dielectric spectra of samples with a moisture content of around 1.25% at different aging levels are as follows Figure 5 As shown, they are used to analyze the effects of moisture-affected state and thermal aging state on dielectric properties.

[0042] from Figure 4 It can be seen from the figure that with the increase of moisture content, the tanδ dielectric loss spectrum gradually increases in the entire test frequency range. This is because water molecules are polar conductive molecules, which will increase the conductivity while strengthening the polarization strength; the dielectric loss can reflect the polarization and conductivity information inside the insulation, so the increase of moisture will cause the dielectric loss value to increase in the entire frequency band.

[0043] from Figure 5 It can be seen that with the increase of aging degree, the tanδ dielectric loss spectrum gradually increases in the medium and low frequency bands, and the overall change amplitude is uniform, while the change in the high frequency band is relatively small, which is mainly affected by the high moisture content; this is because oil-paper insulation will produce polar products such as moisture and organic acids during the high-temperature aging process, and the effect of these aging products on the dielectric loss spectrum is similar to that of moisture, so the change trend of the dielectric loss spectrum under different aging conditions is relatively similar to that under different moisture conditions.

[0044] Mark the tanδ peak value in the low- to mid-frequency band (10 MHz-100 Hz) and the tanδ platform value in the high-frequency band (1 kHz) as the key features of the neural network input.

[0045] In a preferred embodiment of the present invention, the above step S2: inputting the dielectric response curve into a preset neural network model to output an estimated value of the moisture content of the insulating paper may include: Step S2.1, extracting the dielectric loss factor tanδ values ​​of all test points within a preset frequency range from the dielectric response curve; Step S2.2: input the tanδ value set as input feature quantity into the pre-trained BP neural network model; Step S2.3: outputting the moisture content evaluation value of the insulation paper through the BP neural network model.

[0046] In an embodiment of the present invention, by adopting the 1mHz-1kHz full-band dielectric loss factor feature extraction, the sensitive frequency bands of moisture and aging are covered, the strong correlation between the tanδ feature quantity input to the BP neural network and the insulation state is ensured, and the moisture assessment error in the mid-term aging period is reduced; the nonlinear fitting ability of the BP neural network is utilized to decouple the cross-influence of aging and moisture, and its self-learning characteristics can also automatically extract implicit features such as the positive correlation between the rising slope of the medium and low frequency tanδ and the degree of aging, avoiding the subjectivity of artificial feature engineering; after calibration such as temperature correction, the evaluation value is highly consistent with the measured value, and the mid-term aging error is controlled within ±0.2%, meeting the DL / T984-2005 engineering accuracy requirements, and the automated evaluation of a single sample can be completed within 30 minutes.

[0047] In the embodiments of the present invention, when applied specifically, it can be achieved through the following technical solutions, for example: In step S2.1 above, 20 test points (e.g., 1mHz, 10mHz, 100mHz, 1Hz, 10Hz…1kHz) are evenly selected on the logarithmic frequency axis from 1mHz to 1kHz, corresponding to the dielectric loss factor tanδ value at each frequency point in the dielectric response curve.

[0048] In the above step S2.2, the neural network is a parallel processing connection network using mathematical methods, which has good self-organization, self-adaptation and fault tolerance, and has good applications in data fitting, pattern recognition, clustering, etc.; the working principle is that when the user gives sample input and output, the network continuously adjusts the connection of neuron structure and weights according to the excitation function, training algorithm, number of neurons, etc., so as to obtain an effective and appropriate neural network; after saving, the neural network is called, and the sample data to be identified is input, and the network can give a predicted output.

[0049] This evaluation uses a supervised BP neural network (20 neurons in the input layer correspond to tanδ values ​​at 20 frequency points, 40 neurons in the hidden layer, and 1 neuron in the output layer corresponds to moisture content). The input is extracted from the frequency domain dielectric spectrum, and the values ​​at each frequency point within the entire test frequency range of the dielectric spectrum (0.1mHz-10kHz) are used as dielectric characteristics. This processing can better fit the data and avoid data loss. Taking aging and moisture into consideration, the data from all test points in the dielectric spectrum can uniquely represent a dielectric spectrum curve corresponding to a unique insulation state, achieving the purpose of curve matching. The characteristic value T1 is extracted from the dielectric loss (tanδ) spectrum to construct the neural network Net1. The formula is as follows: ,in, The input of the neural network Net1 consists of 20 test point data with a sample dielectric loss spectrum frequency range of 1mHz to 1kHz. Since there are 40 sets of different sample data, the value of i ranges from 1 to 40. Indicates the frequency point of the dielectric loss spectrum curve, and the value of j ranges from 1 to 20.

[0050] Forty sets of sample data (five aging states × eight moisture gradients) were used and divided into training, validation, and test sets in a ratio of 7:2:1. The training was iteratively performed until the loss function (mean square error (MSE)) converged. The weight update strategy used a momentum factor (0.9) to accelerate convergence.

[0051] In the above step S2.3, the moisture state of the cardboard is evaluated. Through repeated training, the output results of Net1, namely the moisture content evaluation value, the measured value and the evaluation error value, are obtained. Figure 6 As shown in the figure; the evaluation error value is equal to the difference between the measured value and the evaluation value; in order to more intuitively analyze the accuracy of the moisture content evaluation value, a comparison chart between the evaluation value and the measured value is given, as shown in the figure Figure 7 shown.

[0052] Combine Figure 6 and Figure 7 It can be seen that the neural network evaluation is relatively accurate, and the error between the moisture content evaluation value and the measured value is small; before aging, the moisture content evaluation value is always too large; at the end of aging, the moisture content evaluation value is always too small, and the overall error presents a normal distribution. In the early and middle stages of aging, the evaluation error of the moisture state is very small.

[0053] In a preferred embodiment of the present invention, the above step S3: directly measuring the actual moisture content of the insulating oil in the oil-paper insulation structure by a chemical detection method may include: Step S3.1, extracting an oil sample from the insulating oil after the oil-paper insulation structure reaches a moisture equilibrium state; Step S3.2: Karl Fischer titration is used to detect the moisture content of the oil sample to obtain the actual moisture content of the insulating oil.

[0054] In an embodiment of the present invention, after determining that the oil-paper insulation structure has reached a moisture equilibrium state, a vacuum oil pump is used to extract an oil sample filtered through a 0.22μm filter membrane and the oil sample is sealed and stored with high-purity nitrogen. This process not only ensures that the moisture content of the oil sample reflects a steady-state distribution to avoid dynamic diffusion errors, but also prevents moisture absorption during sampling through vacuum extraction and nitrogen protection, which complies with the DL / T596-2005 operating specifications; the coulometric Karl Fischer titration is used to detect the moisture content of the oil sample with an accuracy of ±0.5mg / L, directly measuring the total amount of free water and dissolved water in the oil, and achieving quantitative positioning of the aging interval by matching the coordinates of the two on the equilibrium curve.

[0055] In the embodiments of the present invention, when applied specifically, it can be achieved through the following technical solutions, for example: In step S3.1 above, the insulating paperboard was subjected to accelerated thermal aging and moisture absorption tests. The paperboard was immersed in oil in a vacuum dry oil immersion tank, then placed in an aging tank to mix with the insulating oil. The aging tank was then filled with nitrogen, sealed, and placed in a 130°C aging chamber for accelerated thermal aging. After aging for 0, 10, 25, 50, and 80 days, the corresponding aging tanks were removed from the aging chamber and appropriate amounts of insulating paper samples were taken to test the degree of polymerization (DP). The measured DPs were 1098, 750, 433, 256, and 196, respectively.

[0056] Five cardboard samples with different aging degrees were placed in a humidity chamber with a temperature of 60℃ and a relative humidity of 60% for accelerated moisture absorption experiments. The moisture absorption test time was set to 0h, 0.25h, 0.5h, 1h, 1.5h, and 2h, respectively. The cardboard was taken out at regular intervals and placed in insulating oil of corresponding aging state. After being sealed and stored, it was placed in a constant temperature box and left to stand until the oil-paper moisture balance was achieved.

[0057] In order to study the influence of temperature on the moisture balance of oil-paper insulation, three moisture balance temperatures of 85°C, 70°C and 55°C were set, taking into account the possible operating temperatures during on-site transformer operation. At each temperature, there were five oil-paper insulation samples in different aging states. The moisture content of the insulating paper and insulating oil in dynamic equilibrium was measured regularly. When the moisture content remained unchanged, the moisture content in the insulating paper and insulating oil was tested, and the oil-paper moisture balance curves in different aging states and different temperatures were plotted.

[0058] In step S3.2 above, prepare the Karl Fischer reagent (a mixture of iodine, sulfur dioxide, pyridine, and methanol), calibrate it with pure water (10 μL) before use, and ensure that the titer is within the range of 3.5-4.5 mgH2O / mL; take 10 mL of the oil sample and inject it into the titration cell, start the coulometric Karl Fischer titrator, and monitor the end point with a platinum electrode. When the current suddenly jumps and stabilizes within 30 seconds, record the amount of charge consumed and convert it to water content (1 coulomb ≈ 0.036 mgH2O).

[0059] In a preferred embodiment of the present invention, in step S4, based on the estimated moisture content of the insulating paper and the measured moisture content of the insulating oil, the coordinates of the sample points are determined in a preset oil-paper moisture balance curve coordinate system. The oil-paper moisture balance curve coordinate system includes multiple reference curves representing different aging states, which may include: Step S4.1, using the measured moisture content of the insulating oil as the horizontal coordinate and the estimated moisture content of the insulating paper as the vertical coordinate to form the sample point coordinates; Step S4.2, mapping the sample point coordinates to the preset oil-paper moisture balance curve coordinate system; Step S4.3, identifying the positions of the sample point coordinates relative to the multiple reference curves in the oil-paper moisture balance curve coordinate system; Step S4.4: The multiple reference curves are respectively associated with different aging state intervals, and each reference curve is pre-calibrated through an oil-paper moisture balance experiment corresponding to the aging state.

[0060] In an embodiment of the present invention, by unifying the dimensions and temperature correction, the physical accuracy of the sample point coordinates is ensured, and a standardized two-dimensional data pair is provided for aging interval positioning based on the equilibrium curve, which is directly related to the moisture distribution state of the oil-paper structure; through multi-curve visual mapping, the relative positions of the sample points and different aging state curves are intuitively presented, and the standardized coordinate system is compatible with the DL / T596-2005 industry standard, which facilitates engineering personnel to understand the relationship between aging state and moisture distribution; combining spatial distance and slope matching, single distance calculation errors are avoided, and the reliability of aging interval judgment is improved, and visual position identification provides intuitive verification for distance method calculation; the experimentally calibrated reference curve is directly related to the degree of polymerization and the aging stage, and the pre-calibrated curve library supports on-site rapid matching, so that quantitative evaluation of the aging state can be achieved without repeated experiments.

[0061] In the embodiments of the present invention, when applied specifically, it can be achieved through the following technical solutions, for example: In the above step S4.1, the measured moisture content of the insulating oil (unit: %) and the estimated moisture content of the insulating paper (unit: %) are unified into mass percentages to ensure that the horizontal and vertical coordinates are consistent in dimension.

[0062] If the test temperature is inconsistent with the equilibrium curve temperature (such as 70℃), the moisture content is converted to temperature using the Arrhenius equation (for example, when converting from 45℃ to 70℃, it is corrected according to the moisture diffusion coefficient in the literature

[15] ).

[0063] According to the process of step S3, the oil paper moisture balance curves of five samples with different aging conditions at 85°C, 70°C and 55°C are obtained, as shown in FIG. Figure 8 As shown in the figure; it can be seen from the figure that at the same equilibrium temperature, except for the two curves of aging for 80 days and aging for 50 days, there are obvious differences in the four curves of aging for 0 days, 10 days, 25 days and 50 days. This is similar to the change law of the degree of polymerization of insulating paperboard. From aging for 0 days to aging for 50 days, the degree of polymerization changes greatly in each aging period, while the degree of polymerization after aging for 50 days and aging for 80 days is not much different; with the increase of aging, the moisture balance curve of oil-paper gradually moves from the y-axis to the x-axis, that is, when the moisture content in the oil is the same, the more severe the aging, the lower the moisture content in the paper. It can be understood that the increase of aging will cause more moisture in the paper to migrate into the oil.

[0064] In the above step S4.2, taking the equilibrium curve at 70℃ as an example, five reference curves (S1-S5) are drawn in the rectangular coordinate system. The horizontal axis is the moisture content in the oil (0-5%), and the vertical axis is the moisture content in the paper (0-6%). The curve equation is derived from Figure 9 The fitting result (such as S1:y=0.3228+0.0381x); mark the sample point (x oil ,y paper ), where x oil is the measured water content in oil, y paper Moisture content in paper evaluated for neural networks.

[0065] In step S4.3 above, observe the spatial distance between the sample point and each reference curve. For example, when the sample point is below the S2 curve and close to the S3 curve, it is preliminarily judged that the aging state is between the A2-A3 range. Analyze the consistency between the sample point and the slope of the curve. The more severe the aging, the smaller the slope of the curve (reference Figure 9 The slope of the curve decreases from 0.0381 to 0.0202). If the slope of the curve near the sample point is 0.0238, it matches the S3 curve (DP = 433, mid-aging).

[0066] In the above step S4.4, 15 oil-paper insulation samples with different aging and moisture conditions were prepared by accelerated thermal aging test and moisture balance test. The measured polymerization degree and moisture content were as follows: Figure 10 As shown in the figure, the neural network is used to evaluate the moisture state, and the comparison between the moisture content evaluation value of the insulation paper and the measured moisture content value is obtained. Figure 11 As shown in the figure, it can be seen that the moisture content evaluation results of the 15 oil-paper insulation samples are relatively accurate and the errors are within the allowable range.

[0067] Substitute the moisture content assessment value and the measured moisture content of the insulating oil into the 70℃ oil-paper moisture balance curve. The distribution of the sample points to be assessed in the oil-paper moisture balance curve is as follows: Figure 12 As shown; a moisture balance experiment was carried out at 70℃ on oil paper samples with different aging degrees (DP=1098, 756, 433, 256, 196), the moisture content of the oil and paper at equilibrium was measured, and the least squares method was used to fit the curve (for example, S1 corresponds to the unaged state of DP=1098); according to DL / T596-2005, curves S1-S5 were associated with the aging ranges A1-A4 respectively (for example, S1 corresponds to A1: DP≥1000).

[0068] In a preferred embodiment of the present invention, the above step S5: calculating the geometric distance from the sample point coordinates to each reference curve and selecting the reference curve corresponding to the minimum distance may include: Step S5.1, in the oil-paper moisture balance curve coordinate system, calculate the vertical distance from the sample point coordinate to each reference curve; Step S5.2, comparing the calculated vertical distance values ​​to obtain a comparison result; Step S5.3: Based on the comparison result, select the reference curve corresponding to the vertical distance with the smallest value.

[0069] In an embodiment of the present invention, standardized vertical distance calculation is used to ensure that the distances from sample points to each reference curve are comparable, thereby avoiding evaluation deviations caused by differences in the forms of curve equations. Data verification reduces extrapolation errors, making the distance values ​​more consistent with the curve characteristics of experimental calibration. Distance sorting and threshold verification are used to avoid the influence of accidental errors of a single distance value. The slope is combined with auxiliary judgment in the overlapping area of ​​the curves to improve the reliability of the comparison results and unify the aging status evaluation standards of different sample points. The reference curve corresponding to the minimum distance is selected and directly associated with the aging interval, so that the evaluation results clearly correspond to the degree of aggregation.

[0070] In the embodiments of the present invention, when applied specifically, it can be achieved through the following technical solutions, for example: In the above step S5.1, the fitting equation of each reference curve (such as S1: y = 0.3228 + 0.0381x) is arranged into a standard linear form (ax-y + b = 0), for example, S1 corresponds to "0.0381x-y + 0.3228 = 0".

[0071] Substituting the sample point coordinates (x0, y0) into the vertical distance calculation formula is as follows: ,in, =0 represents the curve equation, is the moisture content of the insulating oil of the sample, is the moisture content in the sample insulation paper, is the distance value, and a set of distance values ​​is obtained (such as the distance d1 from the sample point to S1, the distance d2 to S2, and the distance d5 to S5).

[0072] Check whether the sample point is within the valid data range of the curve (0-5% moisture in oil, 0-6% moisture in paper). If it is out of range, mark it as "extrapolated value" and use the trend of the adjacent curve to correct the distance (for example, when the moisture in oil is greater than 5%, refer to the S5 curve slope for extrapolation).

[0073] In step S5.2 above, the distance values ​​(d1-d5) from the sample point to the five curves are sorted by numerical value, retaining four decimal places to ensure accuracy (e.g., d1 = 0.0667, d2 = 0.2398, etc.). A distance difference threshold is set (e.g., 0.05). If the difference between the minimum distance and the next minimum distance is less than the threshold, it is determined to be a "curve overlap area" and further differentiation is required based on the curve slopes (e.g., if the slopes of S4 and S5 are close, the curve with a closer degree of convergence is matched first).

[0074] In the above step S5.3, the curve number corresponding to the minimum distance value is extracted from the sorting result (for example, if d1 is the smallest, it matches S1), and this curve is the membership curve of the sample point.

[0075] According to the preset curve-interval mapping relationship (S1→A1, S2→A2, S3→A3, S4→A4, S5→A4), the membership curve is converted into the corresponding aging interval (if it matches S1, then A1 is output).

[0076] In a preferred embodiment of the present invention, the above step S6: outputting the aging status evaluation result of the oil-paper insulation structure according to the aging status interval associated with the reference curve may include: Step S6.1, associating the reference curve with a preset polymerization degree aging state interval; Step S6.2, output the polymerization degree aging state interval as the aging state evaluation result of the oil-paper insulation structure, wherein the polymerization degree aging state interval includes: unaged interval A1: polymerization degree 1000-1300, early aging interval A2: polymerization degree 700-1000, mid-aging interval A3: polymerization degree 400-700, and late aging interval A4: polymerization degree 0-400.

[0077] In an embodiment of the present invention, through the key aging characteristic quantity of polymerization degree, a clear physical correlation is formed between the reference curve and the aging interval, which avoids the abstraction of the evaluation results and complies with the core judgment basis of DL / T596-2005 that the polymerization degree is associated with the life. Its preset non-aging interval, early aging interval, mid-aging interval, and late aging interval also cover the entire life cycle of insulation, providing a complete standard framework for evaluation; the output aging interval results are directly related to the insulation life stage, which can provide a clear decision-making basis for transformer condition maintenance, meeting the "condition maintenance" needs of the power industry.

[0078] In the embodiments of the present invention, when applied specifically, it can be achieved through the following technical solutions, for example: In the above step S6.1, based on the pre-calibrated reference curves (S1-S5), the degree of polymerization of the insulating paper corresponding to each curve is extracted (e.g., S1 corresponds to DP=1098, S2 corresponds to DP=756, S3 corresponds to DP=433, S4 corresponds to DP=256, and S5 corresponds to DP=196).

[0079] The degree of aggregation of the curve is compared with the preset aging interval (A1: 1000-1300, A2: 700-1000, A3: 400-700, A4: 0-400), and the mapping relationship is established: S1→A1 (DP=1098∈1000-1300), S2→A2 (DP=756∈700-1000), S3→A3 (DP=433∈400-700), S4→A4 (DP=256∈0-400), S5→A4 (DP=196∈0-400).

[0080] The corresponding relationship between the curve and the interval is stored in the evaluation system database to form a "curve number-aging interval" query table (for example, query S3 directly returns "A3") Figure 13 shown.

[0081] In the above step S6.2, based on the minimum distance reference curve selected in step S5 (e.g., sample TS7 matches S3), the aging interval associated with the curve (S3→A3) is retrieved from the database; the evaluation result is output in a standardized text format, including the interval name and the corresponding polymerization degree range, for example, "Aging status evaluation result: A3 (polymerization degree 400-700)".

[0082] If the sample point is located in the curve overlap area (for example, the distance between TS10 matching S3 and S4 is close), an "interval proximity prompt" will be added to the output result (for example, "close to the A4 interval, it is recommended to further confirm it in combination with the low-frequency characteristics in the dielectric loss spectrum").

[0083] like Figure 2 As shown, an embodiment of the present invention further provides an oil-paper insulation aging status assessment system, comprising: Acquisition module, used to perform frequency domain dielectric spectrum test on oil-paper insulation structure and obtain dielectric response curve; A moisture assessment module is used to input the dielectric response curve into a preset neural network model and output an assessment value of the moisture content of the insulation paper; a detection module, configured to directly measure the actual moisture content of the insulating oil in the oil-paper insulation structure by a chemical detection method; an aging assessment module for determining sample point coordinates in a preset oil-paper moisture balance curve coordinate system based on the estimated moisture content of the insulating paper and the measured moisture content of the insulating oil, wherein the oil-paper moisture balance curve coordinate system includes multiple reference curves representing different aging states; The calculation module is used to calculate the geometric distance between the sample point coordinates and each reference curve, and select the reference curve corresponding to the minimum distance; The output module is used to output the aging status evaluation result of the oil-paper insulation structure according to the aging status interval associated with the reference curve.

[0084] It should be noted that this system is a system corresponding to the above method, and all implementation methods in the above method embodiment are applicable to this embodiment and can achieve the same technical effects.

[0085] An embodiment of the present invention further provides a computing device comprising: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the above-described method. All implementations in the above-described method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0086] The embodiment of the present invention further provides a computer-readable storage medium storing instructions, which, when executed on a computer, causes the computer to execute the above-described method. All implementations in the above-described method embodiment are applicable to this embodiment and can achieve the same technical effects.

[0087] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for evaluating the aging state of oil-paper insulation, characterized in that: The method comprises: Step S1: Performing a frequency domain dielectric spectrum test on the oil-paper insulation structure to obtain a dielectric response curve; Step S2: Input the dielectric response curve into a preset neural network model and output an estimated value of the moisture content of the insulating paper; Step S3: directly measuring the actual moisture content of the insulating oil in the oil-paper insulation structure by a chemical detection method; Step S4: determining the coordinates of the sample points in a preset oil-paper moisture balance curve coordinate system based on the estimated moisture content of the insulating paper and the measured moisture content of the insulating oil. The oil-paper moisture balance curve coordinate system includes multiple reference curves representing different aging states. Step S5: Calculate the geometric distance between the sample point coordinates and each reference curve, and select the reference curve corresponding to the minimum distance; Step S6: outputting an aging status evaluation result of the oil-paper insulation structure according to the aging status interval associated with the reference curve.

2. The oil-paper insulation aging status assessment method according to claim 1, characterized in that: Step S1: Perform a frequency domain dielectric spectrum test on the oil-paper insulation structure to obtain a dielectric response curve, including: Prepare an oil-paper insulation sample library based on preset aging gradients and moisture gradients; Under constant temperature conditions, an AC voltage with a frequency range of 1mHz-1kHz is applied to the oil-paper insulation structure in the oil-paper insulation sample library; The dielectric response data under AC voltage is collected by frequency domain dielectric spectrometer; A dielectric response curve is generated according to the dielectric response data, with frequency as the abscissa and dielectric loss factor tan δ as the ordinate.

3. The oil-paper insulation aging status assessment method according to claim 2, characterized in that: Step S2: Input the dielectric response curve into a preset neural network model to output an estimated value of the moisture content of the insulating paper, including: Extract the dielectric loss factor tanδ values ​​of all test points within the preset frequency range from the dielectric response curve; The tanδ value set is used as the input feature quantity and input into the pre-trained BP neural network model; The BP neural network model outputs an evaluation value of the moisture content of the insulation paper.

4. The oil-paper insulation aging status assessment method according to claim 3, characterized in that: Step S3: directly measuring the actual moisture content of the insulating oil in the oil-paper insulation structure by a chemical detection method, including: After the oil-paper insulation structure reaches a moisture equilibrium state, extracting an oil sample from the insulating oil; The water content of the oil sample is detected by Karl Fischer titration to obtain the actual water content of the insulating oil.

5. The oil-paper insulation aging status assessment method according to claim 4, characterized in that: Step S4: Determine the coordinates of the sample points in a preset oil-paper moisture balance curve coordinate system based on the estimated moisture content of the insulating paper and the measured moisture content of the insulating oil. The oil-paper moisture balance curve coordinate system includes multiple reference curves representing different aging states, including: The measured moisture content of the insulating oil is used as the horizontal coordinate, and the estimated moisture content of the insulating paper is used as the vertical coordinate to form the coordinates of the sample points; Map the sample point coordinates to the preset oil-paper moisture balance curve coordinate system; Identify the position of the sample point coordinates relative to multiple reference curves in the oil-paper moisture balance curve coordinate system; The multiple reference curves are respectively associated with different aging state intervals, and each reference curve is pre-calibrated through an oil-paper moisture balance experiment corresponding to the aging state.

6. The oil-paper insulation aging status assessment method according to claim 5, characterized in that: Step S5: Calculate the geometric distance from the sample point coordinates to each reference curve, and select the reference curve corresponding to the minimum distance, including: In the oil-paper moisture balance curve coordinate system, calculate the vertical distance from the sample point coordinate to each reference curve; Compare the calculated vertical distance values ​​to obtain a comparison result; Based on the comparison results, the reference curve corresponding to the vertical distance with the smallest value is selected.

7. The oil-paper insulation aging status assessment method according to claim 6, characterized in that: Step S6: Outputting an aging status evaluation result of the oil-paper insulation structure according to the aging status interval associated with the reference curve, including: Associating the reference curve with a preset polymerization degree aging state interval; The polymerization degree aging state interval is output as an aging state evaluation result of the oil-paper insulation structure, wherein the polymerization degree aging state interval includes: unaged interval A1: polymerization degree 1000-1300, early aging interval A2: polymerization degree 700-1000, mid-aging interval A3: polymerization degree 400-700, and late aging interval A4: polymerization degree 0-400.

8. A system for evaluating the aging state of oil-paper insulation, the system implementing the method according to any one of claims 1 to 7, characterized in that: include: Acquisition module, used to perform frequency domain dielectric spectrum test on oil-paper insulation structure and obtain dielectric response curve; A moisture assessment module is used to input the dielectric response curve into a preset neural network model and output an assessment value of the moisture content of the insulation paper; a detection module, configured to directly measure the actual moisture content of the insulating oil in the oil-paper insulation structure by a chemical detection method; an aging assessment module for determining sample point coordinates in a preset oil-paper moisture balance curve coordinate system based on the estimated moisture content of the insulating paper and the measured moisture content of the insulating oil, wherein the oil-paper moisture balance curve coordinate system includes multiple reference curves representing different aging states; The calculation module is used to calculate the geometric distance between the sample point coordinates and each reference curve, and select the reference curve corresponding to the minimum distance; The output module is used to output the aging status evaluation result of the oil-paper insulation structure according to the aging status interval associated with the reference curve.

9. A computing device, characterized in that include: one or more processors; A storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a program, which, when executed by a processor, implements the method according to any one of claims 1 to 7.

Images