Transformer insulation system equivalent circuit parameter identification method

Through the multi-objective particle swarm optimization algorithm and the mixed time frequency domain dielectric spectrum feature quantity, the long test time and data inaccuracy in the insulating aging state evaluation of transformer oil paper are solved, and the accurate evaluation of the aging state of insulating medium is achieved and the circuit is realized. Parameter identification improves diagnostic accuracy.

CN119087147APending Publication Date: 2024-12-06XIAMEN UNIV OF TECH
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411129856.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-16
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

When evaluating the insulating aging state of transformer oil paper, the prior art test time is long and susceptible to environmental interference, which affects data accuracy, limits the application and promotion of equivalent circuit parameters in the field of insulating aging diagnosis.

Method used

The multi-objective particle swarm optimization algorithm is used to combine the reply voltage and the frequency domain dielectric spectrum feature quantity to construct the objective function of the mixed frequency domain dielectric spectrum feature quantity, and the measurement is performed through the reply voltage tester and the spectrum dielectric analyzer to shorten the test time And improve parameter identification accuracy.

Benefits of technology

By shortening the test time and improving data accuracy, the accurate evaluation of the aging state of the transformer insulating medium is achieved, and the application value of equivalent circuit parameter identification is enhanced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119087147A_ABST
    Figure CN119087147A_ABST
Patent Text Reader

Abstract

The invention relates to a transformer insulation system equivalent circuit parameter identification method, which comprises the following steps of: performing return voltage measurement on a transformer at different charging time and discharging time to obtain a return voltage characteristic quantity corresponding to a return voltage curve; performing frequency domain dielectric spectrum measurement on the transformer at different frequencies to obtain frequency domain dielectric spectrum characteristic quantities; based on the obtained return voltage characteristic quantity, constructing a first objective function for solving equivalent circuit parameters; based on the obtained frequency domain dielectric spectrum characteristic quantity, constructing a second objective function and a third objective function for solving equivalent circuit parameters; and after the three objective functions are used for particle fitness calculation, identifying parameters in an extended Terabye model corresponding to the transformer through a multi-objective particle swarm optimization algorithm. On the premise of ensuring the identification accuracy of the equivalent circuit parameters, the time-frequency domain dielectric spectrum field measurement time can be greatly shortened, and the identification efficiency of the equivalent circuit parameters is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of transformers, and in particular to a method for identifying equivalent circuit parameters of a transformer insulation system. Background Art

[0002] Accurately evaluating the aging state of transformer oil-paper insulation is of great significance to reducing the transformer insulation failure rate, extending the transformer service life and reducing maintenance costs. At present, most domestic and foreign scholars use the extended Debye equivalent circuit model to study the aging of transformer oil-paper insulation. The various polarization branches in the extended Debye equivalent circuit model can effectively characterize the corresponding process of polarization relaxation of the insulating medium. Its circuit parameters have a good functional relationship with the recovery voltage polarization spectrum, polarization-depolarization current curve and frequency domain dielectric spectrum aging characteristic quantity. Through the analysis of the extended Debye equivalent circuit parameters, the aging of the transformer insulation medium can be reliably evaluated. However, the existing equivalent circuit parameter identification basically uses a single type of time domain dielectric spectrum or frequency domain dielectric spectrum characteristic quantity. On the one hand, due to the test principle of the test instrument itself, some spectral line tests take a long time; on the other hand, due to the long test time, it is easy to be disturbed by environmental factors and affect the data accuracy, which is not conducive to the application and promotion of equivalent circuit parameters in the field of transformer insulation aging diagnosis. Summary of the invention

[0003] In order to solve the above problems, the present invention proposes a method for identifying equivalent circuit parameters of a transformer insulation system.

[0004] The specific plan is as follows:

[0005] A method for identifying equivalent circuit parameters of a transformer insulation system comprises the following steps:

[0006] S1: measuring the recovery voltage of the transformer at different charging and discharging times to obtain the recovery voltage characteristic quantity corresponding to the recovery voltage curve; measuring the frequency domain dielectric spectrum of the transformer at different frequencies to obtain the frequency domain dielectric spectrum characteristic quantity;

[0007] S2: constructing a first objective function for solving the equivalent circuit parameters based on the obtained recovery voltage characteristic quantity; constructing a second objective function and a third objective function for solving the equivalent circuit parameters based on the obtained frequency domain dielectric spectrum characteristic quantity;

[0008] The first objective function is expressed as: Wherein, F1 represents the first target value, min represents the minimum value, n represents the number of polarization branches included in the extended Debye model, i represents the serial number of the polarization branches included in the extended Debye model, and k represents the serial number of the number of reply voltage measurements. represents the initial slope obtained when the kth recovery voltage is measured, U rk(t) represents the recovery voltage measured at time t during the kth recovery voltage measurement, C g represents the geometric capacitance in the extended Debye model, U0 represents the power supply voltage during the recovery voltage measurement, τ i =R pi C pi represents the time constant of polarization branch i, R pi represents the resistance of polarization branch i, C pi represents the capacitance of polarization branch i, t ck represents the charging time when the kth recovery voltage is measured, t dk Indicates the discharge time when the kth recovery voltage is measured;

[0009] The second objective function is expressed as: F2 = min[∑(c' 测量 (ω)-c'(ω)) 2 ], where F2 represents the second target value, ω represents the angular frequency, c' 测量 (ω) represents the real part of the complex capacitance obtained by direct measurement, and C'(ω) represents the real part of the complex capacitance obtained by calculation. The calculation formula for the real part of the complex capacitance is:

[0010] The third objective function is expressed as: F3 = min[∑(c" 测量 (ω)-c"(ω)) 2 ], where F3 represents the third target value, c" 测量 (ω) represents the imaginary part of the complex capacitance obtained by direct measurement, and c"(ω) represents the imaginary part of the complex capacitance obtained by calculation. The calculation formula for the imaginary part of the complex capacitance is: R g represents the insulation resistance in the extended Debye model;

[0011] S3: After using the above three objective functions for the fitness calculation of particles, the parameters in the extended Debye model corresponding to the transformer are identified through the multi-objective particle swarm optimization algorithm.

[0012] Furthermore, step S3 specifically includes the following steps:

[0013] S301: Initialize and set the particle dimension to the number of parameters contained in all polarization branches in the extended Debye model, and initialize the particle population and the number of iterations;

[0014] S302: Initialize the positions and velocities of all particles in the population;

[0015] S303: Calculate the fitness of each particle in the population, and use the fitness of each particle in the initial state as the current optimal fitness of each particle, and use the position of each particle in the initial state as the current optimal position of each particle;

[0016] S304: Screening all non-inferior solutions from all particles included in the population to form a non-inferior solution set;

[0017] S305: randomly select a particle from the non-inferior solution set at the current moment as the global optimal solution, and update the position of each particle in the population at the next moment based on the particle position corresponding to the global optimal solution;

[0018] S306: Based on the position of each particle in the population at the next moment, the fitness of each particle at the next moment is calculated, and the fitness of each particle at the next moment is compared with the fitness of the current moment, and the current optimal fitness of each particle at the next moment is updated based on the comparison result, and the current optimal position of each particle at the next moment is obtained;

[0019] S307: based on the fitness of each particle at the next moment, all non-inferior solutions are screened out, and all the screened non-inferior solutions are merged with the current optimal position of each particle in the population at the next moment, and all non-inferior solutions are re-screened from the merged result to form a non-inferior solution set at the next moment, and the same particles are deleted;

[0020] S308: Repeat steps S305-S307 until the iteration ends, and output the non-inferior solution set at the next moment obtained by the final iteration;

[0021] S309: normalizing each non-inferior solution in the non-inferior solution set according to the objective function category corresponding to the fitness, and adding three categories of normalized results as the global fitness of each non-inferior solution;

[0022] S310: taking the non-inferior solution corresponding to the global fitness minimum as the optimal solution, and obtaining the parameter value of each polarization branch in the extended Debye model based on the optimal solution.

[0023] Furthermore, the method for selecting non-inferior solutions from all particles included in the population is: taking the fitness of the particle corresponding to the first objective function as the first type of fitness, taking the fitness corresponding to the second objective function as the second type of fitness, and taking the fitness corresponding to the third objective function as the third type of fitness; if the first type of fitness of the remaining particles in the population except the detected particle is smaller than the first type of fitness of the detected particle, or if the second type of fitness and the third type of fitness of the remaining particles are simultaneously smaller than the second type of fitness and the third type of fitness of the detected particle, then it is determined that the detected particle is dominated, and the dominated particle is defined as an inferior solution, and the particle that is not dominated is a non-inferior solution.

[0024] Furthermore, the recovery voltage is measured using a recovery voltage tester RVM5461.

[0025] Furthermore, the frequency domain dielectric spectrum was measured using the spectrum dielectric analyzer DIRANA.

[0026] Furthermore, the charging time used in measuring the recovery voltage is twice the discharging time.

[0027] Furthermore, the recovery voltage characteristic quantity includes an initial slope corresponding to the recovery voltage curve.

[0028] Furthermore, the frequency domain dielectric spectrum feature includes a real part of complex capacitance and an imaginary part of complex capacitance.

[0029] The present invention adopts the above technical scheme to construct an objective function for equivalent circuit parameter identification based on hybrid time-frequency domain dielectric spectrum feature quantities, and adopts a multi-objective particle swarm optimization algorithm for parameter identification to improve the accuracy of equivalent circuit parameter identification based on different types of feature quantities. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 Shown is a flow chart of a method according to an embodiment of the present invention.

[0031] Figure 2 Shown is a schematic diagram of the extended Debye model in this embodiment.

[0032] Figure 3 Shown is a comparison diagram of the actual measurement and simulation of the recovery voltage polarization spectrum in this embodiment.

[0033] Figure 4 The figure shows the comparison between the actual measurement and simulation of the complex capacitance of the transformer T1 in this embodiment.

[0034] Figure 5 The figure shows the comparison between the actual measurement and simulation of the complex capacitance of the transformer T2 in this embodiment. DETAILED DESCRIPTION

[0035] To further illustrate various embodiments, the present invention provides drawings. These drawings are part of the disclosure of the present invention, which are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in this field should be able to understand other possible implementations and advantages of the present invention.

[0036] The present invention will now be further described with reference to the accompanying drawings and specific implementation methods.

[0037] The embodiment of the present invention provides a method for identifying equivalent circuit parameters of a transformer insulation system. Figure 1 As shown, the method comprises the following steps:

[0038] S1: The recovery voltage of the transformer is measured at different charging times and discharging times to obtain the recovery voltage characteristic quantity corresponding to the recovery voltage curve; the frequency domain dielectric spectrum of the transformer is measured at different frequencies to obtain the frequency domain dielectric spectrum characteristic quantity.

[0039] In this embodiment, the recovery voltage of the transformer is measured by using the recovery voltage tester RVM5461. c and discharge time t d , and obtain the recovery voltage curve (i.e., the recovery voltage variation curve relative to time t), and then based on the recovery voltage curve, the corresponding initial slope dU r / dt, maximum recovery voltage U rmax and its corresponding recovery voltage peak time t peak The characteristic quantity of recovery voltage is obtained by calculating the recovery voltage polarization spectrum (the voltage polarization spectrum is the spectrum line of the maximum value of the recovery voltage changing with the charging time). In this embodiment, the charging time is preferably twice the discharge time, that is, t c / t d =2.

[0040] In this embodiment, a spectrum dielectric analyzer DIRANA is used to measure the frequency domain dielectric spectrum of the transformer. The frequency domain dielectric spectrum characteristics such as the real part, imaginary part and dielectric loss of the complex capacitance are obtained by changing the frequency f, thereby obtaining the real part spectrum line of the complex capacitance, the imaginary part spectrum line of the complex capacitance and the dielectric loss spectrum line that change with the frequency.

[0041] Due to the testing principle of the testing instrument itself, the testing time of the recovery voltage polarization spectrum and the frequency domain dielectric spectrum are relatively long, averaging more than 3 hours. At the same time, the measurement of the recovery voltage polarization spectrum consumes a lot of time under long charging time, and this section of the recovery voltage polarization spectrum mainly reflects the aging of the insulating paper, while the frequency domain dielectric spectrum measurement consumes a lot of time in the low frequency band, and this section of the frequency domain dielectric spectrum mainly reflects the aging of the insulating paper. Moreover, the longer the measurement time, the greater the probability of being interfered by external environmental factors. In order to reduce the measurement time and accurately reflect the aging of each insulating medium, this embodiment preferably selects a large number of recovery voltage polarization spectrum feature quantities under short charging time (that is, charging time t c <50s), a small amount of recovery voltage polarization spectrum characteristics under long charging time (such as charging time of 200s and 1000s). When performing frequency domain dielectric spectrum measurement, it is preferred to select a small amount of low-frequency frequency domain dielectric spectrum characteristics (such as frequency f=0.001Hz, 0.01Hz, 0.1Hz and 1Hz).

[0042] S2: Based on the obtained recovery voltage characteristic quantity, a first objective function for solving the equivalent circuit parameters is constructed; based on the obtained frequency domain dielectric spectrum characteristic quantity, a second objective function and a third objective function for solving the equivalent circuit parameters are constructed.

[0043] The extended Debye model is an equivalent circuit model commonly used to study the response characteristics of dielectrics. It mainly includes geometric equivalent circuit and polarization equivalent circuit. g, Insulation resistance R g The polarization resistance R on multiple polarization branches pi , polarization capacitance C pi Its basic circuit structure is as follows Figure 2 In this model, C g Characterizes the total capacitance of the insulating body of a certain structure and lossless polarization, R g Characterize the conductivity characteristics inside the medium. Both parameters can be obtained by on-site measurement of instruments. This embodiment mainly solves the identification of polarization equivalent circuit parameters.

[0044] (1) In order to construct the first objective function, in this embodiment, Kirchhoff's theorem is first used to Figure 2 The extended Debye model in is analyzed:

[0045]

[0046]

[0047]

[0048] Among them, U0 represents the power supply voltage when measuring by the recovery voltage method, U r (t), U cpi (t) respectively represent the recovery voltage and polarization capacitance C at both ends of the insulating medium (i.e., equivalent circuit) at time t pi The voltage across the two ends, τ i =R pi C pi represents the time constant of polarization branch i.

[0049] By combining formulas (1)-(3), we can get the following formula:

[0050]

[0051] For formula (4), it is necessary to assume that when t = 0, U r When (t) = 0, multiple sets of measured data of the recovery voltage obtained by multiple measurements are substituted into formula (4) to obtain the following equations:

[0052]

[0053] in, t cm and t dm They represent the initial slope, charging time and discharging time during the mth recovery voltage measurement respectively, and m represents the total number of measurements during the recovery voltage measurement.

[0054] In order to solve the multiple unknowns in the equation, multiple equations in formula (5) are combined to construct their root mean square values, which are used as the first objective function for solving the equivalent circuit:

[0055]

[0056] Wherein, F1 represents the first target value, min represents the minimum value, n represents the number of polarization branches included in the extended Debye model, i represents the serial number of the polarization branches included in the extended Debye model, and k represents the serial number of the number of reply voltage measurements. represents the initial slope obtained when the kth recovery voltage is measured, U rk (t) represents the recovery voltage measured at time t during the kth recovery voltage measurement, t ck represents the charging time when the kth recovery voltage is measured, t dk Indicates the discharge time when the kth recovery voltage is measured.

[0057] (2) In order to construct the second objective function and the third objective function, this embodiment performs the following derivation.

[0058] Figure 2 The port admittance of the extended Debye model is:

[0059]

[0060] Then, according to the definition of complex capacitance, the formula of complex capacitance is derived:

[0061]

[0062] The relationship between the real part C'(ω) and the imaginary part C″(ω) of the complex capacitance is derived by the complex capacitance formula:

[0063]

[0064]

[0065] Based on the measured and calculated values ​​of the real and imaginary parts of the complex capacitance, the corresponding second and third objective functions are constructed:

[0066] F2=min[∑(c' 测量 (ω)-c'(ω)) 2 ] (11)

[0067] F3=min[∑(c" 测量 (ω)-c"(ω)) 2 ] (12)

[0068] Among them, c' 测量 (ω), c" 测量(ω) represents the measured values ​​of the real and imaginary parts of the complex capacitance obtained by the field test of the frequency domain dielectric spectrum meter; C'(ω) and c"(ω) are the calculated results of the real and imaginary parts of the complex capacitance obtained by solving formulas (9) and (10) respectively, and ω represents the angular frequency ω = 2πf.

[0069] S3: After using the above three objective functions for the fitness calculation of particles, the parameters in the extended Debye model corresponding to the transformer are identified through the multi-objective particle swarm optimization algorithm.

[0070] According to the above content, there is a good functional relationship between the characteristic quantity of the reply voltage, the characteristic quantity of the frequency domain dielectric spectrum and the parameters of the extended Debye model, and there are corresponding objective functions. In order to comprehensively utilize the characteristic quantity of the reply voltage and the characteristic quantity of the frequency domain dielectric spectrum, in this embodiment, a multi-objective particle swarm algorithm is used to identify the parameters in the extended Debye model, and the steps are as follows:

[0071] S301: Initialize and set the particle dimension to the number of parameters contained in all polarization branches in the extended Debye model, and initialize the particle population size and the number of iterations.

[0072] Since the parameter C in the extended Debye model g and R g All of them can be obtained through on-site measurement by instruments. Therefore, this embodiment mainly solves the identification of parameters in the polarization branches, so the particle dimension is set to the number of parameters contained in all polarization branches.

[0073] In this embodiment, the number of initialized sub-populations is 50 and the number of iterations is 100.

[0074] S302: Initialize the positions and velocities of all particles in the population.

[0075] In this embodiment, a chaotic random method is used to distribute the initial positions of particles, and upper and lower limits of the particle positions are set to prevent large differences or negative values ​​of particles.

[0076] x(i,j)=rand(i,j)×(upper_limit(i,j)-lower_limit(i,j))+lower_limit(i,j) (13)

[0077] S303: Calculate the fitness of each particle in the population, and use the fitness of each particle in the initial state as the current optimal fitness of each particle, and use the position of each particle in the initial state as the current optimal position of each particle.

[0078] Based on the three objective function calculation formulas obtained in step S2, the fitness of the particle corresponding to the first objective function is taken as the first type fitness P x, the fitness corresponding to the second objective function is taken as the second type of fitness R x , the fitness corresponding to the third objective function is taken as the third type of fitness C x , where P x Corresponding to the recovery voltage related fitness, R x Corresponding to the fitness related to the real part of complex capacitance, C x Corresponding to the fitness related to the imaginary part of the complex capacitance.

[0079] Save the fitness of the initialized particles to P x , R x , C x In the matrix, set the initial P x , R x , C x is the initial current optimal fitness P xbest , R xbest , C xbest , where P xbest , R xbest , C xbest They are all 1*50 matrices.

[0080] S304: Filter all non-inferior solutions from all particles included in the population to form a non-inferior solution set.

[0081] In this embodiment, the method for selecting non-inferior solutions from all particles included in the population is as follows: if the particles in the population other than the detected particle have the first type of fitness P x Less than the first fitness P of the particle being tested x , or if the remaining particles have the second type of fitness R x And the third type of fitness C x At the same time, it is smaller than the second fitness R of the detected particle x And the third type of fitness C x , the detected particle is determined to be dominated, and the dominated particles are defined as inferior solutions, and the particles that are not dominated are non-inferior solutions. Save the non-inferior solution fitness vector F j (i,j) and particle position F j,x (i,j). Where i = 1, 2, 3…N, N is the number of non-inferior solutions; j = 1, 2, 3, represents P x , R x , C x Three fitness values.

[0082] S305: Randomly select a particle from the non-inferior solution set at the current time (time t) as the global optimal solution G best , and based on the global optimal solution G best The corresponding particle position updates the position of each particle in the population at the next moment (time t+1).

[0083] The position update formulas in this embodiment are as shown in formulas (14) and (15):

[0084]

[0085] x i (t+1)=x i (t)+v i (t+1) (15)

[0086] Among them, δ represents the inertia weight, and are learning factors, X best Indicates the current optimal position of each individual particle. The updated particles need to be judged one by one whether they exceed the upper and lower limits of the particles. If they exceed the upper limits, the particles are randomly assigned.

[0087] S306: Based on the position of each particle in the population at time t+1 (the next time), calculate the fitness of each particle at time t+1, and compare the fitness of each particle at time t+1 with the fitness at time t (the current time), update the current optimal fitness of each particle at time t+1 based on the comparison result, and obtain the current optimal position of each particle at time t+1.

[0088] The current optimal position of each particle at time t+1 is the particle position corresponding to the current optimal fitness at time t+1.

[0089] When updating the current optimal fitness of each particle at time t+1 based on the comparison results, if the fitness at time t+1 is greater than the current optimal fitness at time t, the fitness at time t+1 is used as the current optimal fitness at time t+1; if the fitness at time t+1 is less than or equal to the current optimal fitness at time t, the current optimal fitness at time t is used as the current optimal fitness at time t+1.

[0090] S307: Based on the fitness of each particle at time t+1, all non-inferior solutions are screened out, and all the screened non-inferior solutions are merged with the current optimal position of each particle in the population at time t+1. All non-inferior solutions are re-screened from the merged results to form a non-inferior solution set at time t+1, and the same particles are deleted.

[0091] The method of screening non-inferior solutions and re-screening non-inferior solutions from the combined results adopts the same method as in S304.

[0092] For example, if the non-inferior solutions obtained by fitness screening at time t+1 include 30 particles, and the current optimal position of each particle in the population at time t+1 includes 50 particles, the merging result is 80 particles, and then the non-inferior solutions are screened from the 80 particles.

[0093] S308: Repeat steps S305-S307 until the iteration ends, and output the non-inferior solution set obtained by the final iteration;

[0094] S309: normalize each non-inferior solution in the non-inferior solution set according to the objective function category corresponding to the fitness, and add up the three categories of normalization results as the global fitness of each non-inferior solution.

[0095] There are three types of objective functions corresponding to fitness, namely the first type of fitness P x , the second type of fitness R x And the third type of fitness C x , After normalizing each category into a dimensionless constant, the normalized results of the three categories are added together as the global fitness.

[0096] The normalization formula is as follows:

[0097]

[0098] Among them, F j '(i,j) represents the normalized parameters, F j (i, j) represents the parameters before normalization, x jmax 、x jmin They represent the maximum and minimum values ​​of the same type of fitness respectively.

[0099] S310: taking the non-inferior solution corresponding to the global fitness minimum as the optimal solution, and obtaining the parameter value of each polarization branch in the extended Debye model based on the optimal solution.

[0100] Example verification

[0101] The partial recovery voltage characteristic quantities and partial frequency domain dielectric spectrum characteristic quantities obtained from the field tests of transformers T1 (240MVA, 220kV, put into operation for 14 years) and T2 (31.5MVA, 110kV, put into operation for 4 years) are shown in Table 1 and Table 2 respectively. The extended Debye equivalent circuit parameters of the two transformers are identified by the method of this embodiment, and the identification results are shown in Table 3.

[0102] Table 1

[0103]

[0104]

[0105] Table 2

[0106]

[0107] Table 3

[0108]

[0109] (1) Comparative analysis of recovery voltage polarization spectra

[0110] According to the extended Debye equivalent circuit parameters of the two transformers obtained by the identification method of this embodiment, the corresponding reply voltage polarization spectrum is calculated and compared with the actual measured reply voltage polarization spectrum. The comparison results are shown as follows: Figure 3 shown.

[0111] In order to verify the reliability of the simulation point, the fit between the measured value and the calculated value of the recovery voltage polarization spectrum is calculated using formula (17):

[0112]

[0113] Among them, N represents the degree of fit, y is the measured value, and y * Indicates the calculated value. When N is closer to 1, the calculated value has a higher fit with the measured value. Figure 3 From the parameter calculation, it can be seen that the fitting degrees of the recovery voltage polarization spectra of transformers T1 and T2 are 94.38% and 96.64% respectively. This shows that the hybrid time-frequency domain dielectric spectrum feature quantity composed of the recovery voltage feature quantity under short charging time and a small amount of low-frequency frequency domain dielectric spectrum feature quantity and the recovery voltage feature quantity under long charging time, combined with the extended Debye equivalent circuit identified by the multi-objective particle swarm optimization algorithm, can also restore the complete recovery voltage polarization spectrum that is highly consistent with the field measurement, that is, the extended Debye equivalent circuit can accurately reflect the time-domain dielectric properties of the insulating medium inside the transformer. The reliability of the parameter identification method based on the hybrid time-frequency domain dielectric spectrum feature quantity proposed in this embodiment is verified.

[0114] (2) Comparative analysis of frequency domain dielectric spectra

[0115] Similarly, combining formula (17) and Figure 4 , Figure 5 From the parameter calculation, it can be seen that the fitting degrees of the real and imaginary parts of the complex capacitance of transformer T1 are 99.93% and 98.68% respectively, and the fitting degrees of the real and imaginary parts of the complex capacitance of T2 are 99.67% and 98.70% respectively. This shows that the hybrid time-frequency domain dielectric spectrum feature quantity composed of a small number of low-frequency frequency-domain dielectric spectrum feature quantities and the reply voltage feature quantity, combined with the extended Debye equivalent circuit obtained by multi-objective particle swarm optimization algorithm, can also restore the complete frequency-domain dielectric spectrum that is highly consistent with the field measurement. That is, the parameter identification method based on the hybrid time-frequency domain dielectric spectrum feature quantity proposed in this embodiment can accurately reflect the frequency-domain dielectric characteristics of the insulating medium inside the transformer. The reliability of this method is further verified.

[0116] Although the present invention has been specifically shown and described in conjunction with the preferred embodiments, it should be understood by those skilled in the art that various changes may be made to the present invention in form and details without departing from the spirit and scope of the present invention as defined by the appended claims, all of which are within the scope of protection of the present invention.

Claims

1. A method for identifying equivalent circuit parameters of transformer insulation system. It is characterized in that The following steps are involved: S1: measuring the recovery voltage of the transformer at different charging and discharging times to obtain the recovery voltage characteristic quantity corresponding to the recovery voltage curve; The frequency domain dielectric spectrum of the transformer is measured at different frequencies to obtain the characteristic quantity of the frequency domain dielectric spectrum; S2: constructing a first objective function for solving the equivalent circuit parameters based on the obtained recovery voltage characteristic quantity; constructing a second objective function and a third objective function for solving the equivalent circuit parameters based on the obtained frequency domain dielectric spectrum characteristic quantity; The first objective function is expressed as: Among them, F 1 represents the first target value, min represents the minimum value, n represents the number of polarization branches included in the extended Debye model, i represents the serial number of the polarization branches included in the extended Debye model, k represents the serial number of the number of recovery voltage measurements, represents the initial slope obtained when the kth recovery voltage is measured, U rk (t) represents the recovery voltage measured at time t during the kth recovery voltage measurement, C g represents the geometric capacitance in the extended Debye model, U 0 Indicates the power supply voltage when the recovery voltage is measured, τ i =R pi C pi represents the time constant of polarization branch i, R pi represents the resistance of polarization branch i, C pi represents the capacitance of polarization branch i, t ck represents the charging time when the kth recovery voltage is measured, t dk Indicates the discharge time when the kth recovery voltage is measured; The second objective function is expressed as: F 2 =min[∑(c' 测量 (ω)-c'(ω)) 2 ], where F 2 represents the second target value, ω represents the angular frequency, c' 测量 (ω) represents the real part of the complex capacitance obtained by direct measurement, and C'(ω) represents the real part of the complex capacitance obtained by calculation. The calculation formula for the real part of the complex capacitance is: The third objective function is expressed as: F 3 =min[∑(c" 测量 (ω)-c"(ω)) 2 ], where F 3 represents the third target value, c" 测量 (ω) represents the imaginary part of the complex capacitance obtained by direct measurement, and c"(ω) represents the imaginary part of the complex capacitance obtained by calculation. The calculation formula for the imaginary part of the complex capacitance is: R g represents the insulation resistance in the extended Debye model; S3: After using the above three objective functions for the fitness calculation of particles, the parameters in the extended Debye model corresponding to the transformer are identified through the multi-objective particle swarm optimization algorithm.

2. The method for identifying equivalent circuit parameters of a transformer insulation system according to claim 1, Features: Step S3 specifically includes the following steps: S301: Initialize and set the particle dimension to the number of parameters contained in all polarization branches in the extended Debye model, and initialize the particle population and the number of iterations; S302: Initialize the positions and velocities of all particles in the population; S303: Calculate the fitness of each particle in the population, and use the fitness of each particle in the initial state as the current optimal fitness of each particle, and use the position of each particle in the initial state as the current optimal position of each particle; S304: Screening all non-inferior solutions from all particles included in the population to form a non-inferior solution set; S305: randomly select a particle from the non-inferior solution set at the current moment as the global optimal solution, and update the position of each particle in the population at the next moment based on the particle position corresponding to the global optimal solution; S306: Based on the position of each particle in the population at the next moment, the fitness of each particle at the next moment is calculated, and the fitness of each particle at the next moment is compared with the fitness of the current moment, and the current optimal fitness of each particle at the next moment is updated based on the comparison result, and the current optimal position of each particle at the next moment is obtained; S307: based on the fitness of each particle at the next moment, all non-inferior solutions are screened out, and all the screened non-inferior solutions are merged with the current optimal position of each particle in the population at the next moment, and all non-inferior solutions are re-screened from the merged result to form a non-inferior solution set at the next moment, and the same particles are deleted; S308: Repeat steps S305-S307 until the iteration ends, and output the non-inferior solution set at the next moment obtained by the final iteration; S309: normalizing each non-inferior solution in the non-inferior solution set according to the objective function category corresponding to the fitness, and adding three categories of normalized results as the global fitness of each non-inferior solution; S310: taking the non-inferior solution corresponding to the global fitness minimum as the optimal solution, and obtaining the parameter value of each polarization branch in the extended Debye model based on the optimal solution.

3. The method for identifying equivalent circuit parameters of a transformer insulation system according to claim 2, Features: The method for selecting non-inferior solutions from all particles included in the population is as follows: the fitness of the particle corresponding to the first objective function is taken as the first type of fitness, the fitness corresponding to the second objective function is taken as the second type of fitness, and the fitness corresponding to the third objective function is taken as the third type of fitness; if the first type of fitness of the remaining particles in the population except the detected particle is smaller than the first type of fitness of the detected particle, or if the second type of fitness and the third type of fitness of the remaining particles are both smaller than the second type of fitness and the third type of fitness of the detected particle, then the detected particle is determined to be dominated, and the dominated particle is defined as an inferior solution, and the particle that is not dominated is a non-inferior solution.

4. The method for identifying equivalent circuit parameters of a transformer insulation system according to claim 1, Features: The recovery voltage is measured using the recovery voltage tester RVM5461.

5. The method for identifying equivalent circuit parameters of a transformer insulation system according to claim 1, Features: The frequency domain dielectric spectrum was measured using a spectrum dielectric analyzer DIRANA.

6. The method for identifying equivalent circuit parameters of a transformer insulation system according to claim 1, Features: The recovery voltage measurement is performed using a charge time that is twice the discharge time.

7. The method for identifying equivalent circuit parameters of a transformer insulation system according to claim 1, Features: The recovery voltage characteristic quantity includes an initial slope corresponding to the recovery voltage curve.

8. The method for identifying equivalent circuit parameters of a transformer insulation system according to claim 1, Features: The frequency domain dielectric spectrum characteristics include the real part of complex capacitance and the imaginary part of complex capacitance.

Citation Information

Cited By

  • Method for charge analysis and insulation state evaluation of active-service cable based on potential test

    CN120370094A

  • State evaluation method, device and equipment based on frequency domain dielectric spectrum and surface potential

    CN120908621A