Oil paper insulation hot spot area inversion and aging state evaluation method and system

Abstract

The invention provides an oil paper insulation hot spot area inversion and aging state evaluation method and system. The method comprises the following steps: field FDS data acquisition and basic parameter measurement; establishing an XY model dielectric reference; calibrating an actual measurement FDS curve; modeling of an extended Debye model is carried out; inverting a hot spot area FDS by using an improved non-dominated sorting genetic algorithm; and state evaluation. The technical problems that the state of the transformer can only be judged according to the average evaluation result of the main insulation of the transformer, serious aging of insulation at a hot spot is covered, and the evaluation accuracy of the insulation state of the transformer is affected are solved.

Description

Technical Field

[0001] This invention relates to the field of power system equipment condition monitoring and insulation material performance evaluation, specifically to a method and system for hot spot area inversion and aging condition assessment of oil-paper insulation. Background Technology

[0002] Converter transformers are core equipment in ultra-high voltage direct current (UHVDC) transmission systems, enabling the conversion of alternating current (AC) to direct current (DC). Compared to ordinary transformers, they withstand combined DC and AC electric fields, as well as polarity reversal voltages, resulting in a more complex internal environment. Oil-paper insulation is a major component of the transformer's internal insulation, and its performance determines the transformer's service life. Therefore, accurately understanding the condition of transformer oil-paper insulation is crucial for preventing power transmission interruptions and even major blackouts caused by insulation faults, and has significant economic and academic value.

[0003] The insulation system is crucial for ensuring the normal and stable operation of a transformer, and insulating paperboard and insulating oil are both essential insulating materials. Insulating paperboard, as a fixed internal insulation structure of the transformer, is the main insulating material that distinguishes and protects the high-voltage and low-voltage windings. Insulating oil, which fills the entire transformer tank, is the liquid insulation of the power transformer, serving both insulation and heat dissipation functions. The combination of these two materials effectively improves the overall insulation performance. Insulating paperboard that is fully impregnated with oil exhibits significantly improved heat resistance, breakdown field strength, and mechanical strength compared to unimpregnated paperboard. Furthermore, the insulating oil keeps the internal metal components of the transformer in an environment where they cannot come into contact with the external environment, reducing the possibility of oxidation. During long-term operation, influenced by load fluctuations, ambient temperature, and electrical stress, hot spots can easily form inside the insulation. Increased temperature leads to a significantly accelerated aging rate of the material, which may then cause insulation breakdown accidents.

[0004] Current research on the condition assessment of oil-paper insulation primarily employs time-frequency domain dielectric response testing techniques, including the recovery voltage method, polarization-depolarization current method, and frequency-domain dielectric response testing. Compared to traditional time-domain testing methods, frequency-domain dielectric response testing is more suitable for complex field environments, accurately obtaining the response signal within the transformer's main insulation to test excitation. Furthermore, the polarization information within the dielectric under different frequency excitations can effectively distinguish between insulation aging and moisture conditions. However, FDS-based transformer oil-paper insulation assessment methods can only obtain the model's average aging parameters and cannot assess the actual aging state of the transformer's main insulation. Judging the condition solely based on the average assessment results of the transformer's main insulation will mask the severe aging at hot spots, affecting the accuracy of the transformer insulation condition assessment.

[0005] The existing invention patent application document CN111948502A, entitled "An Evaluation Method for the Aging State of Oil-Paper Insulation of Converter Transformers," includes the following steps: obtaining initial data of oil-paper insulation samples through a micro-current sampling system, calculating the interface polarization decay current and obtaining the time-domain dielectric response spectrum through differential transformation, extracting the master relaxation time and fitting it with the degree of polymerization of the oil-paper, and finally evaluating its aging state based on the master relaxation time of the actual equipment. This existing scheme is based on time-domain dielectric response technology, and the obtained dielectric response information is the integral of the broadband domain response. The analysis results reflect the overall and average aging state of the insulation system. However, for the complex structure of the converter transformer valve-side outgoing line device, the internal electric field distribution is uneven, which easily forms local hot spots and leads to non-uniform aging of the insulation. The time-domain method is difficult to effectively isolate and identify the micro-regional aging characteristics caused by local hot spots, and the evaluation conclusion may mask the severe aging in local areas, thus affecting the accuracy of the condition assessment.

[0006] The existing invention patent application document CN111983403A, entitled "Analysis Method of Dielectric Properties of Composite Insulation Structure Using Reverse Finite Element Technology," includes the following steps: S2: Using a dielectric response analysis and testing instrument to perform frequency domain dielectric spectrum measurement on a composite oil-paper insulation structure containing multiple dielectrics in high-voltage electrical equipment, obtaining the overall measured frequency domain dielectric response data of the composite insulation part; S3: Based on the equipment structure diagram and dimensional information provided by the manufacturer of the equipment under test in S1, using finite element software to establish a full-size finite element simulation model of the equipment; S4: Performing quasi-static field sweep frequency analysis under the finite element model in S2; S5: Using the data post-processing function of the finite element software under the sweep frequency analysis, establishing the mapping relationship between the dielectric variable x of each part of the material and the dielectric response of the measurement port according to the actual measurement port in S1; S6: Using finite element analysis... The method combines software with numerical calculation software, establishing a data exchange system between the finite element analysis environment and the numerical calculation environment. In the numerical calculation software, a multi-parameter nonlinear optimization model based on the measured dielectric response in S1 and the Cole-Cole model used in S3 is established. A calculation system is constructed whereby the finite element software transmits the calculated values ​​of the dielectric response at the measurement port to the numerical analysis software, and the numerical analysis software controls the finite element software to update the dielectric variable values. In S6, a fusion method combining the Genetic Algorithm and the Levenberg-Marquardt algorithm (GA-LMA) is used. Through the data exchange system between the finite element environment and the numerical calculation environment in S5, the multi-parameter nonlinear optimization model in S5 is solved in the numerical calculation software, and the dielectric variable vector x0 is obtained by inversion. This existing method can separate the dielectric variables of oil and paper, but it requires precise matching of equipment structural dimensions, and the modified Cole-Cole model mainly addresses moisture and polarization characteristics, failing to fully consider the differential impact of non-uniform aging caused by local hot spots on dielectric parameters. For scenarios like converter transformer valve side outgoing line devices with complex internal structures that are prone to local hot spots due to load fluctuations, the inverted dielectric variables still tend to have regional average characteristics. This makes it difficult to capture the micro-scale aging parameter anomalies in the hot spot area, and it cannot effectively isolate the aging characteristics caused by local hot spots. The assessment conclusions may easily mask serious local aging hazards, affecting the refinement and accuracy of insulation condition diagnosis.

[0007] In summary, the existing technology has a technical problem: it can only judge the condition of the transformer's main insulation based on the average assessment results, which leads to the masking of severe aging of the insulation at hot spots and affects the accuracy of transformer insulation condition assessment. Summary of the Invention

[0008] The technical problem to be solved by the present invention is: how can the existing technology solve the problem that the condition of the transformer insulation is only judged based on the average evaluation result of the main insulation of the transformer, which leads to the masking of the serious aging of the insulation at hot spots and affects the accuracy of the transformer insulation condition evaluation.

[0009] This invention solves the above-mentioned technical problems by employing the following technical solution: A method for inverting hot spot areas and assessing aging status of oil-paper insulation includes:

[0010] S1. Collect on-site FDS data, process it to obtain the measured FDS curve, and measure the basic parameters of the sample.

[0011] S2. Simplify the main insulation structure, integrate all paper tubes, oil channels and support bars separately, establish the XY model dielectric reference, and obtain the XY theoretical curve; among them, obtain the geometric changes and map the set changes to the superposition relationship of the total complex dielectric constant;

[0012] S3. Compare the measured FDS curve with the theoretical XY curve, and eliminate the dielectric response deviation through the difference correction operation to obtain the calibrated FDS curve;

[0013] S4. By connecting R and C devices in series and parallel, the Debye model is extended to represent the electrical characteristics of the transformer oil-paper insulation. Based on the real and imaginary part curves of the complex capacitance, the loss factor curve within the test frequency band is obtained.

[0014] S5. Improve the non-dominated sorting genetic algorithm, invert the hotspot region FDS, establish the objective function y, and calculate the goodness-of-fit R. 2 The error between the calculated fitting curve and the measured data is characterized. An elite strategy is implemented to save individuals with fitness at level 1 in the population. The input population is sorted according to non-dominance relationship. Crowding degree is used to sort individuals at the same level in the non-dominance level to obtain the optimal solution set.

[0015] S6. Based on the optimal solution set, evaluate the oil-paper insulation status of the high-voltage converter transformer side outgoing line device.

[0016] This invention performs information inversion and aging status assessment of hotspot areas in oil-paper insulation. It uses Debye-XY to model the equivalent circuit structure of oil-paper insulation at different aging levels, and then employs an improved non-dominated sorting genetic algorithm (NSGA) for inversion calculations based on domain information. This inversion technique enables the stripping and extraction of aging parameters from microscopic regions within the insulation structure, thereby more accurately assessing its true state.

[0017] In a more specific technical solution, in S2, the frequency domain dielectric spectrum of the transformer's main insulation system at temperature T is expressed using the following logic:

[0018]

[0019] In the formula, X is the ratio of the total thickness of the paper tube to the thickness of the main insulation between the high-voltage and low-voltage windings, and Y is the ratio of the total width of the support bars to the average perimeter of the main insulation between the high-voltage and low-voltage windings. The frequency domain spectrum of the total complex permittivity of the oil-paper insulation system; The frequency domain spectrum of the complex permittivity of mineral insulating oil; ε0 is the frequency domain spectrum of the complex dielectric constant of the insulating paperboard; σ(T) is the DC conductivity of the insulating oil at temperature T; ε0 is the vacuum dielectric constant.

[0020] In a more specific technical solution, within S3, the calibration FDS curve is expressed using the following logic:

[0021] C m (ω)=[C′ m (ω), C″ m (ω)]

[0022] In the formula, C m (ω) represents the frequency domain dielectric spectrum obtained after calibration, C′ m (ω) represents the real part of the quasi-followed FDS curve, C″ m (ω) represents the imaginary part of the calibrated FDS curve, where ω refers to the angular frequency.

[0023] This invention addresses the issue of uneven aging of the oil-paper insulation on the valve-side outgoing lines of high-voltage converter transformers. It employs an XY model to create an equivalent circuit model of the oil-paper insulation, and then corrects the FDS curve using XY theoretical curves. This avoids errors caused by structural differences and provides assurance for subsequent evaluation of the insulation condition by extracting R and C parameters.

[0024] In a more specific technical solution, in S4, the number of polarization branches n of the Debye model is expanded using the following logic:

[0025]

[0026] In the formula: Z is the equivalent impedance of the tested sample; C i R is the equivalent capacitance of the i-th relaxed branch; i ω is the equivalent resistance of the i-th branch; ω is the angular frequency of the test excitation; n is the number of polarization branches in the extended Debye model.

[0027] This invention addresses the limitation of FDS-based transformer oil-paper insulation assessment methods, which only provide average aging parameters and fail to assess the actual aging state of the transformer's main insulation. By extending the Debye model and employing an improved non-dominated sorting genetic algorithm to find the optimal solution, the numerical values ​​of oil-paper insulation aging-related parameters are obtained. This allows for precise calculation of the aging degree in different hotspot areas, improving the accuracy of oil-paper aging assessment.

[0028] In a more specific technical solution, the complex capacitor C is expressed using the following logic. * :

[0029]

[0030] In the formula, C′ is the real part of the complex capacitor, and C″ is the imaginary part of the complex capacitor.

[0031] In a more specific technical solution, the real part y of the complex capacitor is expressed using the following logic:

[0032]

[0033] Express the imaginary part C″ of the complex capacitor using the following logic:

[0034]

[0035] In a more specific technical solution, in S5, the objective function y is established using the least squares method, following the principle of minimizing the sum of squared relative errors:

[0036]

[0037] In the formula, C″ 测量 (ω), C″ 测量 (ω) represent the measured data of the real and imaginary parts of the complex capacitance of the oil-paper insulation sample, respectively; C′ 拟合 (ω), C″ 拟合 (ω) represents the real and imaginary parts of the complex capacitance calculated using the extended Debye model, respectively; m is the number of test frequency points.

[0038] In a more specific technical solution, in S5, the following logic is used to determine the technical fit goodness R. 2 :

[0039]

[0040] In the formula: y i These are the measured data points; Calculate data for the fitting results; y mean This is the arithmetic mean of the original data.

[0041] In a more specific technical solution, in S5, the non-dominant relation ordering includes:

[0042] Population initialization: Form an initial population of size N, defined as the parent population P1 of the genetic algorithm;

[0043] Calculate the fitness function; solve for the non-dominated ranking and crowding distance i. distance The fitness function is obtained, and the objective function of the R and C parameters in the extended Debye model is calculated using the objective function expression. The non-dominated ranking method is introduced to calculate the crowding distance i. distance Determine the optimal individual;

[0044] Select, crossover, and mutation operations are performed on the parent population P1 to obtain a new population Q1;

[0045] Implement an elite strategy; merge the parent population P1 with the new population Q1 to form a new population R1 with a size of 2N individuals, and rank them according to their non-dominant order and crowding distance i. distance Select N superior individuals and continue to perform crossover and mutation to obtain the next generation of new population P2;

[0046] When the preset iteration termination condition is met, the optimal solution set is output as the parameters of the extended Debye model.

[0047] In a more specific technical solution, the system for inverting hot spots and assessing aging status of oil-paper insulation includes:

[0048] The data acquisition and measurement module is used to acquire on-site FDS data, process it to obtain the measured FDS curve, and measure the basic parameters of the sample.

[0049] The reference establishment module is used to simplify the main insulation structure, integrate and process all paper tubes, oil channels and support bars separately, establish the XY model dielectric reference, and obtain the XY theoretical curve; among them, the geometric changes are acquired and the aggregate changes are mapped to the superposition relationship of the total complex dielectric constant. The reference establishment module is connected to the acquisition and measurement module.

[0050] The measured FDS curve calibration module is used to compare the measured FDS curve with the theoretical XY curve. Through the difference correction operation, the dielectric response deviation is eliminated and the calibrated FDS curve is obtained. The measured FDS curve calibration module is connected to the acquisition and measurement module.

[0051] The extended Debye modeling module is used to extend the Debye model through the series and parallel connection of R and C devices to represent the electrical characteristics of transformer oil-paper insulation. Based on the real and imaginary part curves of the complex capacitance, the loss factor curve in the test frequency band is obtained. The extended Debye modeling module is connected to the measured FDS curve calibration module.

[0052] The non-dominated sorting genetic inversion module is used to improve the non-dominated sorting genetic algorithm, invert the hotspot region FDS, establish the objective function y, and calculate the goodness-of-fit R. 2 The error between the fitted curve calculation results and the measured data is characterized. An elite strategy is implemented to save individuals with fitness at level 1 in the population. The input population is sorted according to non-dominance relationship. Crowding degree is used to sort individuals at the same level in the non-dominance level to obtain the optimal solution set. The non-dominance sorting genetic inversion module is connected to the extended Debye model modeling module.

[0053] The state assessment module is used to assess the oil-paper insulation state of the high-voltage converter transformer side outgoing line device based on the optimal solution set. The state assessment module is connected to the non-dominated sorting genetic inversion module.

[0054] The present invention has the following advantages over the prior art:

[0055] This invention performs information inversion and aging status assessment of hotspot areas in oil-paper insulation. It uses Debye-XY to model the equivalent circuit structure of oil-paper insulation at different aging levels, and then employs an improved non-dominated sorting genetic algorithm (NSGA) for inversion calculations based on domain information. This inversion technique enables the stripping and extraction of aging parameters from microscopic regions within the insulation structure, thereby more accurately assessing its true state.

[0056] This invention addresses the issue of uneven aging of the oil-paper insulation on the valve-side outgoing lines of high-voltage converter transformers. It employs an XY model to create an equivalent circuit model of the oil-paper insulation, and then corrects the FDS curve using XY theoretical curves. This avoids errors caused by structural differences and provides assurance for subsequent evaluation of the insulation condition by extracting R and C parameters.

[0057] This invention addresses the limitation of FDS-based transformer oil-paper insulation assessment methods, which only provide average aging parameters and fail to assess the actual aging state of the transformer's main insulation. By extending the Debye model and employing an improved non-dominated sorting genetic algorithm to find the optimal solution, the numerical values ​​of oil-paper insulation aging-related parameters are obtained. This allows for precise calculation of the aging degree in different hotspot areas, improving the accuracy of oil-paper aging assessment.

[0058] This invention solves the technical problem in the prior art that the condition of transformer insulation can only be judged based on the average evaluation results of the main insulation of the transformer, which leads to the masking of severe aging of insulation at hot spots and affects the accuracy of transformer insulation condition evaluation. Attached Figure Description

[0059] Figure 1 This is a schematic diagram of the basic steps of the method for inverting hot spots and assessing aging status of oil-paper insulation according to Embodiment 1 of the present invention;

[0060] Figure 2 These are structural diagrams of the oil-paper insulation at different aging levels in Embodiment 1 of the present invention;

[0061] Figure 3 This is a schematic diagram illustrating the specific steps of the inversion calculation using the multi-constraint non-dominated sorting genetic algorithm in Embodiment 1 of the present invention. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] Example 1

[0064] like Figure 1 As shown, the method for inverting hot spots and assessing aging status of oil-paper insulation provided by the present invention includes the following basic steps:

[0065] S1. On-site FDS data acquisition and basic parameter measurement;

[0066] In this embodiment, an IDAX-300 insulation diagnostic analyzer manufactured by Megger was used for FDS testing. The test voltage amplitude was 0–200V, and the test frequency was 10. -3 ~10 4 Hz. The sample was placed in a temperature-controlled aging test oven to achieve frequency domain dielectric response testing of the oil-paper insulation model within a temperature range of 10–100℃.

[0067] S2 and XY model dielectric references are established;

[0068] In this embodiment, the XY model can reflect the dielectric response characteristics of oil-paper insulation and can be directly linked to the main insulation structure of the transformer. It is widely used to diagnose the condition of the transformer's main insulation system.

[0069] like Figure 2 As shown, in this embodiment, for ease of analysis, the main insulation structure is often simplified, integrating all paper tubes, oil channels, and support bars separately. Here, X is the ratio of the total thickness of the paper tubes to the thickness of the main insulation between the high and low voltage windings, and Y is the ratio of the total width of the support bars to the average perimeter of the main insulation between the high and low voltage windings. The geometric changes are directly mapped to the superposition relationship of the total complex dielectric constant. The frequency domain dielectric spectrum of the transformer main insulation system at temperature T is as follows:

[0070]

[0071] In the formula, The frequency domain spectrum of the total complex permittivity of the oil-paper insulation system; The frequency domain spectrum of the complex permittivity of mineral insulating oil; σ(T) is the frequency domain spectrum of the complex dielectric constant of the insulating paperboard; σ(T) is the DC conductivity of the insulating oil at temperature T; ε0 is the vacuum dielectric constant, and ε0 = 8.85 × 10⁻⁶. -12 F / m.

[0072] S3. Calibration of measured FDS curve;

[0073] In this embodiment, the measured FDS curve is compared with the theoretical XY curve. Dielectric response deviations introduced by differences in equipment structure are eliminated through difference correction to obtain the calibrated FDS curve C. m (ω)=[C′ m (ω), C″ m (ω)。

[0074] In the formula, C m (ω) represents the frequency domain dielectric spectrum obtained after calibration, C′ m (ω) represents the real part of the quasi-followed FDS curve, C″ m (ω) represents the imaginary part of the calibrated FDS curve, where w refers to the angular frequency.

[0075] S4. Extend the Debye model modeling;

[0076] In this embodiment, the electrical characteristics of transformer oil-paper insulation are represented by extending the Debye model through the series and parallel connection of R and C devices. The number of polarization branches n in the extended Debye model can be selected according to the oil-paper insulation medium. In this patent, the number of branches is selected as 6.

[0077]

[0078] In the formula, Z is the equivalent impedance of the tested sample; C i R is the equivalent capacitance of the i-th relaxed branch; i ω is the equivalent resistance of the i-th branch. w is the angular frequency of the test excitation; n is the number of polarization branches in the extended Debye model.

[0079] Complex capacitor C * for

[0080]

[0081] The real part of the complex capacitor C′ is

[0082]

[0083] The imaginary part C″ of the complex capacitor is

[0084]

[0085] The loss factor curve within the test frequency band can be calculated based on the real and imaginary part curves of the complex capacitor.

[0086] S5. Improved non-dominated sorting genetic algorithm for inverting hotspot regions FDS;

[0087] In this embodiment, if the least squares method is chosen to minimize the sum of squared absolute errors for all data points, the fitting effect for high-frequency data may be poor, failing to meet the requirements for full-band fitting accuracy. Therefore, the data in this patent will follow the method of minimizing the sum of squared relative errors, and the objective function y will be established using the least squares method.

[0088]

[0089] In the formula: C′ 测量 (ω), C″ 测量 (ω) represents the measured data of the real and imaginary parts of the complex capacitance of the oil-paper insulation sample, respectively; C″ 拟合 (ω), C″ 拟合 (ω) represents the real and imaginary parts of the complex capacitance calculated using the extended Debye model, respectively; m is the number of test frequency points.

[0090] In this embodiment, the goodness of fit R 2 It can be used to characterize the error between the calculated results of the fitted curve and the measured data. The specific expression is:

[0091]

[0092] In the formula: y i These are the measured data points; Calculate data for the fitting results; y mean R is the arithmetic mean of the original data. 2 The value range is (0,1), and the larger the value, the better the fit.

[0093] In this embodiment, an elite strategy is added to the improved NSGA algorithm to preserve individuals with fitness at level 1 in the population. A method is proposed to sort the input population according to non-dominance relationship, and the concept of crowding is used to sort individuals at the same level in the non-dominance hierarchy, thereby shortening the computation time.

[0094] like Figure 3 As shown, in this embodiment, the specific steps of the inversion calculation using the multi-constraint non-dominated sorting genetic algorithm include:

[0095] S51. Population initialization: Randomly generate N sets of Debye model parameters as the parent population P1.

[0096] S52. Calculate the fitness function; where P1 is sorted non-dominated and the crowding distance is calculated.

[0097] S53. Genetic operations: selection, crossover, and mutation to generate offspring population Q1;

[0098] S54, Implement an elite strategy;

[0099] S55, Merge populations: R1 = P1UQ1 (size: 2N);

[0100] S56. Perform non-dominated sorting and crowding distance calculation on the new population R1;

[0101] S57. Determine whether the termination condition is met; if not, proceed to step S53.

[0102] S58. Output the optimal solution.

[0103] In this embodiment, population initialization is performed during the sorting of the input population according to the non-dominance relationship. The R and C parameters of the Debye model are extended under the initial input state, and an initial population of size N is randomly formed, which is defined as the first generation (i.e., the parent population P1) in the genetic algorithm;

[0104] Calculate the fitness function. The improved NSGA algorithm mainly employs the solution of non-dominated sorting and crowding distance i. distance The fitness function is obtained through this method. The objective function is then used to calculate the objective optimization function for the R and C parameters in the extended Debye model. Finally, a non-dominated ranking method is introduced to calculate i. distance Thus, the optimal individual is determined;

[0105] Selection, crossover, and mutation. Individuals are selected based on their objective function value, with those having smaller values ​​serving as parents to form the next generation. Two individuals from the parent generation are randomly selected, each containing the same gene locus, and their chromosomes are swapped. Following the corresponding mutation probability, some new individuals resulting from the crossover are selected from the population, and a gene locus within each new individual is randomly replaced. By performing these three operations on the parent generation, a new population Q1 can be obtained.

[0106] In this embodiment, an elite strategy is employed. Specifically, P1 and Q1 are merged into a new population R1 with a population size of 2N, and individuals are ranked according to their non-dominant order and i. distance The N superior individuals are selected to undergo crossover and mutation to obtain the next generation population P2.

[0107] In this embodiment, the termination condition is determined by whether a better fitness value exists or the maximum number of iterations is reached. If so, the process terminates immediately, and the optimal solution set is the extended Debye model parameters; if not, the aforementioned elite strategy is executed again.

[0108] S6. Status assessment;

[0109] In this embodiment, the insulation state is evaluated based on the extracted parameters:

[0110] A decrease in R0 indicates an increase in the conductivity of the oil or paper, which may be caused by increased moisture or aging; changes in C0 or polarization branch parameters reflect enhanced interfacial polarization, indicating an accelerated degree of aging.

[0111] In summary, this invention performs information inversion and aging status assessment of hotspot areas in oil-paper insulation. It uses Debye-XY to model the equivalent circuit structure of oil-paper insulation at different aging levels, and then employs an improved non-dominated sorting genetic algorithm (NSGA) for inversion calculations based on domain information. This inversion technique enables the stripping and extraction of aging parameters from microscopic regions within the insulation structure, thereby providing a more accurate assessment of its true condition.

[0112] This invention addresses the issue of uneven aging of the oil-paper insulation on the valve-side outgoing lines of high-voltage converter transformers. It employs an XY model to create an equivalent circuit model of the oil-paper insulation, and then corrects the FDS curve using XY theoretical curves. This avoids errors caused by structural differences and provides assurance for subsequent evaluation of the insulation condition by extracting R and C parameters.

[0113] This invention addresses the limitation of FDS-based transformer oil-paper insulation assessment methods, which only provide average aging parameters and fail to assess the actual aging state of the transformer's main insulation. By extending the Debye model and employing an improved non-dominated sorting genetic algorithm to find the optimal solution, the numerical values ​​of oil-paper insulation aging-related parameters are obtained. This allows for precise calculation of the aging degree in different hotspot areas, improving the accuracy of oil-paper aging assessment.

[0114] This invention solves the technical problem in the prior art that the condition of transformer insulation can only be judged based on the average evaluation results of the main insulation of the transformer, which leads to the masking of severe aging of insulation at hot spots and affects the accuracy of transformer insulation condition evaluation.

[0115] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for oil-paper insulation hotspot area inversion and aging condition assessment, characterized in that, The method comprises: S1, collecting field FDS data, processing to obtain measured FDS curve, measuring basic parameters of the sample; S2, simplifying the main insulation structure, integrating all paper tubes, oil channels and stay bars respectively, establishing XY model dielectric reference, obtaining XY theoretical curve; wherein, the geometric change is obtained, and the set change is mapped into the superposition relationship of the total complex permittivity; S3, comparing the measured FDS curve with the XY theoretical curve, eliminating dielectric response deviation through difference correction operation, and obtaining calibrated FDS curve; S4, expanding Debye model through series and parallel connection of R and C devices, representing electrical characteristics of transformer oil paper insulation, and calculating loss factor curve in the test frequency band according to the real part and imaginary part curves of complex capacitance; S5, improve non-dominated sorting genetic algorithm, inverse hotspot area FDS, establish target function y, calculate goodness of fit R 2 , error between fitting curve calculation result and measured data, execute elite strategy, save individual with fitness in first level in population, sort input population according to non-dominated relationship, use crowding degree, sort individuals in same level in non-dominated level, obtain optimal solution set; S6, evaluating the oil paper insulation state of the high-voltage converter valve side outgoing device according to the optimal solution set.

2. The oil-paper insulation hotspot region inversion and aging condition assessment method according to claim 1, characterized in that, In S2, the following logic is used to express the frequency domain dielectric spectrum of the transformer main insulation system at temperature T: where X is the ratio of the total thickness of the paper cylinder to the thickness of the main insulation between the high and low voltage windings, and Y is the ratio of the total width of the spacers to the average circumference of the main insulation between the high and low voltage windings, is the frequency domain spectrum of the complex permittivity of the oil-paper insulation system; is the frequency domain spectrum of the complex permittivity of the mineral insulating oil; is the frequency domain spectrum of the complex permittivity of the insulating paperboard; σ(T) is the DC conductivity of the insulating oil at temperature T; ε0 is the vacuum permittivity.

3. The oil-paper insulation hotspot region inversion and aging condition assessment method according to claim 1, characterized in that, In S3, the following logic is used to express the calibrated FDS curve: C m (ω) = [C' m (ω),C" m (ω)] where C m (w) represents the real part of the calibrated FDS curve, C′ m (ω) represents the imaginary part of the calibrated FDS curve, C″ m (ω) represents the imaginary part of the calibrated FDS curve, ω refers to the angular frequency.

4. The oil-paper insulation hotspot region inversion and aging condition assessment method according to claim 1, characterized in that, In S4, the following logic is used to expand the polarization branch number n of the Debye model: where: Z is the equivalent impedance of the sample under test; C i Ci is the equivalent capacitance of the ith relaxation branch; R i Ri is the equivalent resistance of the ith branch; ω is the angular frequency of the test excitation; n is the number of polarization branches of the extended Debye model.

5. The oil-paper insulation hotspot region inversion and aging condition assessment method according to claim 4, characterized in that, With the following logic, the complex capacitance C is expressed * : In the formula, C' is the real part of the complex capacitance, and C'' is the imaginary part of the complex capacitance.

6. The oil-paper insulation hotspot region inversion and aging condition assessment method according to claim 5, characterized in that, The following logic is used to express the real part C' of the complex capacitance: The following logic is used to express the imaginary part C'' of the complex capacitance:

7. The oil-paper insulation hotspot region inversion and aging condition assessment method according to claim 1, characterized in that, In S5, the following logic is used to establish the objective function y by using the least square method in the way of minimizing the sum of squares of relative errors: In the formula, C′ 测量 (ω) and C″ 测量 (ω) are respectively the measured data of the real part and the imaginary part of the complex capacitance of the oil-paper insulation sample; C′ 拟合 (ω) and C″ 拟合 (ω) are respectively the real part and the imaginary part of the complex capacitance calculated by using the extended Debye model; and m is the number of test frequency points.

8. The oil-paper insulation hotspot region inversion and aging condition assessment method according to claim 1, characterized in that, In the S5, the goodness of fit R is calculated using the following logic: 2 : where: y i is the measured data point; is the calculated data from the fit; y mean is the arithmetic mean of the original data.

9. The oil-paper insulation hotspot region inversion and aging condition assessment method of claim 1, wherein, In S5, the non-dominated relation sorting comprises: Population initialization; form an initial population with a quantity of N, defined as the parent population P1 of genetic algorithm; Computing fitness function; solving non-dominated sorting and crowded distance i distance , get fitness function, use objective function formula to calculate the objective optimization function of R, C parameters in extended Debye model, introduce non-dominated sorting method to calculate crowded distance i distance , determine the optimal individual; Perform selection, crossover and mutation operations on the parent population P1 to obtain a new population Q1; Performing elitist strategy; merging the parent population P1 and the new population Q1 into a new population R1 with 2N individuals, according to the non-dominated sorting and the crowding distance i distance Determining the better N individuals, and continuing to perform crossover and mutation to obtain a new population P2 of offspring; When the preset iteration termination condition is met, output the optimal solution set as the parameter of the expanded Debye model.

10. An oil-paper insulation hotspot area inversion and aging condition assessment system, characterized in that, The system comprises: A collection and measurement module is configured to collect field FDS data, process to obtain measured FDS curve, and measure basic parameters of the sample; A reference establishing module is configured to simplify the main insulation structure, integrate all paper tubes, oil channels and stay bars respectively, establish XY model dielectric reference, and obtain XY theoretical curve; wherein, the geometric change is obtained, and the set change is mapped into the superposition relationship of the total complex permittivity, and the reference establishing module is connected with the collection and measurement module; A measured FDS curve calibration module is configured to compare the measured FDS curve with the XY theoretical curve, eliminate dielectric response deviation through difference correction operation, and obtain calibrated FDS curve, and the measured FDS curve calibration module is connected with the collection and measurement module; An expanded Debye model modeling module is configured to expand Debye model through series and parallel connection of R and C devices, represent electrical characteristics of transformer oil paper insulation, and calculate loss factor curve in the test frequency band according to the real part and imaginary part curves of complex capacitance; and the expanded Debye model modeling module is connected with the measured FDS curve calibration module; The non-dominated sorting genetic inversion module is used to improve the non-dominated sorting genetic algorithm, to perform inversion on the hotspot area FDS, to establish a target function y, and to calculate a goodness of fit R 2 , to represent the error between the fitting curve calculation result and the measured data, to execute an elite strategy, to save the individuals with the fitness in the first level in the population, to sort the input population according to the non-dominated relationship, to use the crowding degree to sort the individuals in the same level in the non-dominated level, and to obtain an optimal solution set. The non-dominated sorting genetic inversion module is connected with the extended Debye model modeling module. A state evaluation module is configured to evaluate the oil-paper insulation state of the high-voltage converter transformer valve-side outgoing line device according to the optimal solution set, and the state evaluation module is connected with the non-dominant sorting genetic inversion module.

Images