Parameter identification method and system for oil paper insulation equivalent dielectric model

Abstract

The invention discloses a parameter identification method and system for an oil paper insulation equivalent dielectric model. The parameter identification method comprises the following steps: establishing the oil paper insulation equivalent dielectric model of target type electrical equipment; determining the type and number of undetermined parameters in the oil paper insulation equivalent dielectric model; performing a frequency domain dielectric spectrum actual measurement test on the target type of electrical equipment to obtain complex dielectric response data under different frequency points in a test frequency band; and substituting the complex dielectric response data into the oil paper insulation equivalent dielectric model, identifying the parameters to be determined based on the number of types of the parameters to be determined, and determining parameter identification values. According to the method, the parameters of the oil paper insulation equivalent dielectric model based on the frequency domain dielectric spectrum can be efficiently and accurately identified, the insulation state of electrical equipment can be better evaluated and predicted, and safe and stable operation of an equipment body and a power system is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical equipment, and more specifically, to a parameter identification method and system for an oil-paper insulation equivalent dielectric model. Background Art

[0003] Domestic and foreign studies on the basic parameters of oil-paper insulation in transformer main insulation under low temperature conditions, such as insulation resistance, dielectric loss factor and capacitance, generally believe that the phase change and exchange of the insulation material itself, the moisture contained in the material and other impurities under low temperature conditions will lead to nonlinear changes in dielectric properties, and the difficulty of distinguishing the measurement results will increase. Among them, the insulation resistance will reflect discreteness, and the correction method of its resistance will also fail, and the problem of dielectric properties is more obvious. Due to the effect of polarization relaxation, the dielectric constant will fluctuate first and then decrease; and the measuring instruments composed of different materials will have material interface offset due to thermal expansion and contraction, which may also cause continuous fluctuations in dielectric property measurement results. Therefore, under low temperature conditions, the traditional field test is faced with the problems of changes in the equipment state itself caused by environmental changes and instrument measurement errors. However, according to work needs, it is inevitable to carry out field tests in an environment below 5°C, but for low temperature areas such as Heilongjiang and Inner Mongolia, the field test equipment and test basis of coil equipment (power transformers and reactors) under low temperature conditions are still blank. At present, when the ambient temperature is below 5°C, the test will be carried out by on-site heating, and the objects heated include not only the equipment under test, but also the measuring instruments. However, heating the equipment before testing in the field test not only greatly increases the complexity of the operation, but also brings about problems such as unstable equipment operation, uneven temperature after heating of large equipment, and inaccurate test data due to heating of measuring instruments. The lack of accurate detection technology for the dielectric properties of transformers in low-temperature environments has brought great troubles to on-site test work, and may even have a great impact on the construction period and stable operation of the power grid. For this reason, it has great economic and social benefits to conduct research on dielectric properties in low-temperature environments to ensure the safe operation of transformers in low-temperature areas.

[0004] Parameter identification is a key step in the insulation status assessment of power equipment based on frequency domain dielectric spectroscopy. According to different measurement data types, parameter identification methods are mainly divided into three categories: methods based on reply voltage data, methods based on polarization / depolarization current data, and methods based on frequency domain dielectric spectroscopy (FDS).

[0005] In the application of frequency domain dielectric spectroscopy, some studies have established an extended Debye model (EDM) parameter equation group based on the reply voltage polarization spectrum and used the particle swarm optimization algorithm for parameter identification. Although this method has advantages in formula derivation and curve fitting, it lacks theoretical support when determining the number of polarization branches and the results are random. Other studies use the reply voltage, polarization current and depolarization current data to determine the number of polarization branches by identifying the peak points of the differential spectrum, and use the random point method to calculate the polarization branch parameters. Although this method improves the credibility of the identification results, the intervention of human factors may lead to inconsistency of the results. In addition, some literatures have proposed a second-order differential spectrum method for depolarization current to enhance the reliability of the identification results, but failed to fundamentally solve the problem of parameter uniqueness caused by the end point selection. There are also studies that use the third differential analytical method and the algorithm combining the stability diagram and the state space model to solve the polarization equivalent circuit parameters using the depolarization current data, providing a unique identification result, but this method cannot solve the geometric equivalent branch parameters. Some studies have tried to combine genetic algorithms and LM algorithms to solve EDM parameters, but the real and imaginary parts of the frequency domain response data of oil-paper insulation are quite different, and the weight coefficient setting of the objective function is subjective, resulting in random identification results. Another study proposed a method to determine the number of polarization branches based on the peak points of high-order differential spectra by differentiating the imaginary part of the complex dielectric constant and analyzing the peak point characteristics of the differential spectrum. However, this method is computationally complex, and because it includes the conductivity term, it is difficult to further calculate other parameters of the equivalent dielectric model.

[0006] In general, although there have been many attempts to identify the parameters of the oil-paper insulation equivalent dielectric model, there is still a lack of a method that can comprehensively and accurately identify the model parameters, which indicates that further research and innovation are needed in this field.

[0007] Therefore, a parameter identification method for the oil-paper insulation equivalent dielectric model is needed. Summary of the invention

[0008] The present invention provides a parameter identification method and system for an oil-paper insulation equivalent dielectric model, so as to solve the problem of how to identify the parameters in the oil-paper insulation equivalent dielectric model.

[0009] In order to solve the above problem, according to one aspect of the present invention, a parameter identification method of an oil-paper insulation equivalent dielectric model is provided, the method comprising:

[0010] Establish the oil-paper insulation equivalent dielectric model of the target type of electrical equipment;

[0011] Determining the type and quantity of undetermined parameters in the oil-paper insulation equivalent dielectric model;

[0012] Conducting frequency domain dielectric spectrum testing on the target type of electrical equipment to obtain complex dielectric response data at different frequency points within the test frequency band;

[0013] Substitute the complex dielectric response data into the oil-paper insulation equivalent dielectric model, identify the undetermined parameters based on the number of undetermined parameter types, and determine parameter identification values.

[0014] Preferably, the step of determining the type and quantity of undetermined parameters in the oil-paper insulation equivalent dielectric model comprises:

[0015] The oil-paper insulation equivalent dielectric model is simplified and improved to obtain an improved model: ε = ε Δ w+ε ∞ +σ, and determine the type and quantity of the parameters to be determined based on the improved model; where ε is the complex dielectric constant, which is a function of the complex frequency term w; ε Δ is the differential dielectric constant; ε ∞ is the optical frequency dielectric constant; σ is the error term; ε Δ , ε ∞ and σ are unknown parameters that need to be identified.

[0016] Preferably, the step of substituting the complex dielectric response data into the oil-paper insulation equivalent dielectric model, identifying the undetermined parameters based on the number of undetermined parameter types, and determining the parameter identification values ​​comprises:

[0017] The likelihood function is determined based on the improved model as:

[0018]

[0019] Determine the prior distributions that the undetermined parameters satisfy respectively:

[0020]

[0021] σ~Gamma(α,β),

[0022] The goal of determining parameter identification is to obtain the posterior distribution of the unknown parameters, which is:

[0023] p(ε ∞ ∣ε Δ ,σ,w,ε),

[0024] p(ε Δ |ε ∞ ,σ,w,ε),

[0025] p(σ|ε ∞ ,ε Δ ,w,ε),

[0026] Initialize the initial values ​​of each pending parameter;

[0027] The above variables are updated and iterated in turn using the posterior distribution expressions of the undetermined parameters. During the calculation process, other undetermined parameters are considered to be known and fixed to the values ​​obtained by the latest round of iterations.

[0028] When the preset number of iterations is reached or the change in the parameter value is less than the preset threshold, the update is stopped, and the mean of the posterior distribution of each parameter estimated when the final round of iteration is completed is used as the parameter identification value;

[0029] Among them, (ε1,ε2,ε3,...,ε N ) is the oil-paper insulation equivalent dielectric model at different excitation signal frequency terms w1, w2, w3, ..., w N The complex dielectric constant corresponding to the distribution below; N is the number of data pairs; ζ is the variance of the joint distribution i represents the i-th group of sample data; μ0 and ζ0 are ε ∞ The mean and variance of the normal distribution satisfied; μ1 and ζ1 are ε Δ The mean and variance of the normal distribution satisfied by the error term σ; α and β are the shape parameter and scale parameter of the Gamma distribution satisfied by the error term σ.

[0030] Preferably, the method

[0031] In progress ∞ When obtaining the posterior distribution of ∞ The distribution function of is:

[0032]

[0033] The method is carried out in Δ When obtaining the posterior distribution of Δ The distribution function of is:

[0034]

[0035] When the method is used to obtain the posterior distribution of σ, the distribution function of σ is:

[0036]

[0037] Among them, ε i 、w i are the values ​​of the complex dielectric constant and the complex frequency term in the i-th group of data samples respectively.

[0038] Preferably, the method further comprises:

[0039] The parameter identification value is substituted into the oil-paper insulation equivalent dielectric model to judge the insulation state of the target type electrical equipment based on the oil-paper insulation equivalent dielectric model including the parameter identification value.

[0040] According to another aspect of the present invention, a parameter identification system for an oil-paper insulation equivalent dielectric model is provided, the system comprising:

[0041] A model building unit, used to build an oil-paper insulation equivalent dielectric model of target type electrical equipment;

[0042] A parameter determination unit, used to determine the type and quantity of undetermined parameters in the oil-paper insulation equivalent dielectric model;

[0043] A complex dielectric response data acquisition unit, used to perform a frequency domain dielectric spectrum test on the target type electrical equipment to acquire complex dielectric response data at different frequency points within the test frequency band;

[0044] The parameter identification unit is used to substitute the complex dielectric response data into the oil-paper insulation equivalent dielectric model, identify the undetermined parameters based on the number of undetermined parameter types, and determine parameter identification values.

[0045] Preferably, the parameter determination unit determines the type and quantity of undetermined parameters in the oil-paper insulation equivalent dielectric model, including:

[0046] The oil-paper insulation equivalent dielectric model is simplified and improved to obtain an improved model: ε = ε Δ w+ε ∞ +σ, and determine the type and quantity of the parameters to be determined based on the improved model; where ε is the complex dielectric constant, which is a function of the complex frequency term w; ε Δ is the differential dielectric constant; ε ∞ is the optical frequency dielectric constant; σ is the error term; ε Δ , ε ∞ and σ are unknown parameters that need to be identified.

[0047] Preferably, the parameter identification unit substitutes the complex dielectric response data into the oil-paper insulation equivalent dielectric model, identifies the undetermined parameters based on the number of undetermined parameter types, and determines the parameter identification values, including:

[0048] The likelihood function is determined based on the improved model as:

[0049]

[0050] Determine the prior distributions that the undetermined parameters satisfy respectively:

[0051]

[0052] σ~Gamma(α,β),

[0053] The goal of determining parameter identification is to obtain the posterior distribution of the unknown parameters, which is:

[0054] p(ε ∞ ∣ε Δ ,σ,w,ε),

[0055] p(ε Δ |ε ∞ ,σ,w,ε),

[0056] p(σ|ε ∞ ,ε Δ ,w,ε),

[0057] Initialize the initial values ​​of each pending parameter;

[0058] The above variables are updated and iterated in turn using the posterior distribution expressions of the undetermined parameters. During the calculation process, other undetermined parameters are considered to be known and fixed to the values ​​obtained by the latest round of iterations.

[0059] When the preset number of iterations is reached or the change in the parameter value is less than the preset threshold, the update is stopped, and the mean of the posterior distribution of each parameter estimated when the final round of iteration is completed is used as the parameter identification value;

[0060] Among them, (ε1,ε2,ε3,…,ε N ) is the oil-paper insulation equivalent dielectric model at different excitation signal frequency terms w1, w2, w3, ..., w N The complex dielectric constant corresponding to the distribution below; N is the number of data pairs; ζ is the variance of the joint distribution i represents the i-th group of sample data; μ0 and ζ0 are ε ∞ The mean and variance of the normal distribution satisfied; μ1 and ζ1 are ε Δ The mean and variance of the normal distribution satisfied by the error term σ; α and β are the shape parameter and scale parameter of the Gamma distribution satisfied by the error term σ.

[0061] Preferably, the parameter identification unit,

[0062] In progress ∞ When obtaining the posterior distribution of ∞ The distribution function of is:

[0063]

[0064] The method is carried out in Δ When obtaining the posterior distribution of Δ The distribution function of is:

[0065]

[0066] When the method is used to obtain the posterior distribution of σ, the distribution function of σ is:

[0067]

[0068] Among them, ε i 、w i are the values ​​of the complex dielectric constant and the complex frequency term in the i-th group of data samples respectively.

[0069] Preferably, the system further comprises:

[0070] The state judgment unit is used to substitute the parameter identification value into the oil-paper insulation equivalent dielectric model to judge the insulation state of the target type electrical equipment based on the oil-paper insulation equivalent dielectric model containing the parameter identification value.

[0071] Based on another aspect of the present invention, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any step of a parameter identification method for an oil-paper insulation equivalent dielectric model.

[0072] According to another aspect of the present invention, the present invention provides an electronic device, including:

[0073] The computer-readable storage medium described above; and

[0074] One or more processors are used to execute the program in the computer-readable storage medium.

[0075] The present invention provides a parameter identification method and system for an oil-paper insulation equivalent dielectric model, including: establishing an oil-paper insulation equivalent dielectric model of a target type of electrical equipment; determining the type and number of undetermined parameters in the oil-paper insulation equivalent dielectric model; performing a frequency domain dielectric spectrum test on the target type of electrical equipment to obtain complex dielectric response data at different frequency points in the test frequency band; substituting the complex dielectric response data into the oil-paper insulation equivalent dielectric model, identifying the undetermined parameters based on the type and number of undetermined parameters, and determining parameter identification values. The method of the present invention can efficiently and accurately identify the parameters of the oil-paper insulation equivalent dielectric model based on the frequency domain dielectric spectrum, which helps to better evaluate and predict the insulation status of electrical equipment and ensure the safe and stable operation of the equipment body and the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] A more complete understanding of exemplary embodiments of the present invention may be obtained by referring to the following drawings:

[0077] Figure 1is a flow chart of a parameter identification method 100 of an oil-paper insulation equivalent dielectric model according to an embodiment of the present invention;

[0078] Figure 2 A schematic diagram of a process for determining a parameter identification value according to an embodiment of the present invention;

[0079] Figure 3 A schematic diagram of the real part of complex capacitance identified by different methods according to an embodiment of the present invention;

[0080] Figure 4 A schematic diagram of the imaginary part of the complex capacitance identified by different methods according to an embodiment of the present invention;

[0081] Figure 5 Schematic diagram of the structure of a parameter identification system 500 for an oil-paper insulation equivalent dielectric model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0082] Now, exemplary embodiments of the present invention are described with reference to the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to disclose the present invention in detail and completely and to fully convey the scope of the present invention to those skilled in the art. The terms used in the exemplary embodiments shown in the accompanying drawings are not intended to limit the present invention. In the accompanying drawings, the same units / elements are marked with the same reference numerals.

[0083] Unless otherwise specified, the terms (including technical terms) used herein have the commonly understood meanings to those skilled in the art. In addition, it is understood that the terms defined in commonly used dictionaries should be understood to have the same meanings as those in the context of the relevant fields, and should not be understood as idealized or overly formal meanings.

[0084] The present invention proposes a method for identifying parameters of an oil-paper insulation equivalent model based on frequency domain dielectric spectrum FDS, which is proposed to meet the needs of monitoring and evaluating the insulation performance of key oil-filled equipment in power systems under low temperature environments. Since the oil-paper insulation characteristics of oil-filled equipment will age over time, affecting its dielectric properties and operational safety, FDS technology, as a non-destructive detection method, can effectively evaluate the state of insulating materials, including the degree of aging and moisture content. The present invention establishes an equivalent dielectric model of oil-paper insulation and applies advanced statistical machine learning algorithms to identify model parameters, which can not only improve the accuracy of parameter estimation of the equivalent dielectric model of electrical equipment, but also provide an in-depth understanding of the physical properties of insulating materials under low temperature environments. This is of great significance for formulating maintenance strategies for electrical equipment, preventing faults, and improving the stability and reliability of power systems under low temperature environments. It is also in line with the development trend of smart grids and condition monitoring systems.

[0085] Figure 1 FIG. 1 is a flow chart of a parameter identification method 100 for an oil-paper insulation equivalent dielectric model according to an embodiment of the present invention. Figure 1 As shown, the parameter identification method of the oil-paper insulation equivalent dielectric model provided by the embodiment of the present invention can efficiently and accurately identify the parameters of the oil-paper insulation equivalent dielectric model based on the frequency domain dielectric spectrum, which is helpful to better evaluate and predict the insulation state of electrical equipment and ensure the safe and stable operation of the equipment body and the power system. The parameter identification method 100 of the oil-paper insulation equivalent dielectric model provided by the embodiment of the present invention starts from step 101. In step 101, the oil-paper insulation equivalent dielectric model of the target type electrical equipment is established.

[0086] In the present invention, it is necessary to first establish an equivalent dielectric model of the oil-paper insulation system of the electrical equipment. Common equivalent dielectric models include the Debye dielectric model, the Cole-Cole dielectric model, the Davidson-Cole dielectric model, and the Havriliak-Negami dielectric model. Several models are specifically introduced below:

[0087] 1) Debye dielectric model

[0088] The Debye equation is shown below, and its complex dielectric constant ε * The curve is a perfect semicircle in the complex dielectric constant plane. * Plotting the frequency, the imaginary part is about the characteristic frequency f p The loss peak is axially symmetrical, and the real part decreases monotonically with increasing frequency. p The slope maximum occurs at:

[0089]

[0090] Where: ε s Expressed as the dielectric constant of the dielectric under a constant electric field, ε ∞ It is expressed as the dielectric constant of the dielectric under the electric field at the optical frequency; τ is the polarization relaxation time of the dielectric.

[0091] However, in most cases, the complex dielectric constant curve actually measured deviates from the semicircle, and the half-peak width of the loss peak is wider than that calculated by the Debye model. At the same time, the typical Debye model uses the series connection of resistance and capacitance to equivalent the relaxation polarization process of the dielectric. In fact, the polarization process of the insulating material is not formed by a pair of dipoles but by multiple groups of dipoles. At this time, the polarization equivalent circuit can be represented by an extended Debye circuit model consisting of a series of resistors and capacitors in parallel.

[0092] 2) Cole-Cole dielectric model

[0093] The Cole-Cole model is an extension of the Debye model, describing a polarization process with a relaxation time distribution. The complex dielectric constant can be expressed as follows. The Cole-Cole model is a symmetrical arc in the complex dielectric constant plane. Because of the time distribution, the loss peak becomes wider compared to the Debye model, and the complex plane changes from a semicircle to an arc:

[0094]

[0095] In the formula, α has no actual physical meaning and is only a parameter, and its value range is 0<α<1.

[0096] Obviously, when α=1, the above equation degenerates into the Debye equation. The relaxation time of the Cole-Cole equation is the same as the relaxation time of the Debye function, which corresponds to the frequency of the vertex of the arc or semicircle in the complex dielectric constant plane. Although the specific physical meaning of the relaxation time distribution parameter α in the Cole-Cole equation is still unclear, this model can well describe some experimental systems that deviate from the Debye semicircle and is one of the important models for analyzing the frequency domain dielectric spectrum.

[0097] 3) Cole-Davidson dielectric model

[0098] In engineering applications, the actual test results not only do not conform to the Debye equation, but also have a large deviation from the Cole-Cole empirical equation. For example, for systems such as liquids or low-molecular-weight vitrified materials, the dielectric measurement results are an asymmetric arc in the Cole-Cole diagram. The complex dielectric constant can be described by the Cole-Davidson equation as follows:

[0099]

[0100] Where: β has no actual physical meaning and is only a parameter, and its value range is 0<β<1.

[0101] When the relaxation time distribution parameter β = 1, the equation is also reduced to the Debye equation. At the same time, the characteristic relaxation time of the asymmetric equation no longer corresponds to the vertex of the complex plane arc like the first two symmetric equations.

[0102] 4) Havriliak-Negami dielectric model

[0103] Havriliak and Negami proposed a more universal function to explain the dielectric relaxation and mechanical relaxation process of certain polymer systems, called the Havriliak-Negami equation (HN equation), as shown in equation (3-28). In fact, it is a combination of the Cole-Cole equation and the Cole-Davidson function. Obviously, when α=1 and β=1, the HN equation is also reduced to the Debye equation:

[0104]

[0105] The Havriliak-Negami relaxation model has wide applicability and can be used to analyze the dielectric relaxation process of various dielectrics. Therefore, it is often used to extract fitting parameters and analyze the curve of the frequency domain dielectric spectrum.

[0106] In step 102, the types and quantities of undetermined parameters in the oil-paper insulation equivalent dielectric model are determined.

[0107] Preferably, the step of determining the type and quantity of undetermined parameters in the oil-paper insulation equivalent dielectric model comprises:

[0108] The oil-paper insulation equivalent dielectric model is simplified and improved to obtain an improved model: ε = ε Δ w+ε ∞ +σ, and determine the type and quantity of the parameters to be determined based on the improved model; where ε is the complex dielectric constant, which is a function of the complex frequency term w; ε Δ is the differential dielectric constant; ε ∞ is the optical frequency dielectric constant; σ is the error term; ε Δ , ε ∞ and σ are unknown parameters that need to be identified.

[0109] In the present invention, after the oil-paper insulation equivalent dielectric model is established, it is necessary to clarify the type and number of undetermined parameters in the established model. Common model parameters mainly include the optical frequency dielectric constant ε ∞ 、Steady-state dielectric constant ε S , DC conductivity σ dc , average polarization relaxation time constant τ, arc asymmetry κ of the complex plane curve of the dielectric model, etc.

[0110] In the present invention, when determining the type and number of parameters to be determined, the model needs to be simplified. To the improved model: ε = ε Δ w+ε ∞ +σ, and determine the type and quantity of fixed parameters based on the improved model; where ε is the complex dielectric constant, which is a function of the complex frequency term w; ε Δ is the differential dielectric constant; ε ∞ is the optical frequency dielectric constant; σ is the error term; εΔ , ε ∞ and σ are unknown parameters that need to be identified.

[0111] In order to explain the proposed parameter identification algorithm process, the classic Debye dielectric model is taken as an example here to conduct frequency domain dielectric spectrum test on the oil-immersed transformer, obtain the complex dielectric response constants at different frequency points in the test frequency band, and substitute them into the constructed dielectric model to solve and identify the parameters to be determined. It should be noted that the proposed dielectric model parameter identification method is not limited to specific electrical equipment or specific equivalent dielectric models, and is essentially a parameter estimation method based on statistical machine learning.

[0112] In order to construct and describe the microscopic polarization relaxation process of the medium, Debye established the following dielectric model from the perspective of medium relaxation polarization:

[0113]

[0114] Where ε * is the complex dielectric constant, ε ∞ is the optical frequency dielectric constant, ε S is the steady-state dielectric constant, ω is the angular frequency, and i is the complex unit. The above formula gives the corresponding relationship between the dielectric constant and frequency, and is one of the most widely used dielectric models.

[0115] To facilitate mathematical description, some algebraic substitutions and symbol simplifications are now performed on the above formula:

[0116]

[0117] The simplified expression is as follows:

[0118] ε=ε Δ w+ε ∞ ,

[0119] Since noise and interference are unavoidable in the actual detection process, a detection error term σ needs to be introduced here. The improved model expression is as follows:

[0120] ε=ε Δ w+ε ∞ +σ,

[0121] It can be seen that the oil-paper insulation equivalent model based on frequency domain dielectric spectrum has at least three or more unknown parameters that need to be identified.

[0122] In step 103, a frequency domain dielectric spectrum test is performed on the target type electrical equipment to obtain complex dielectric response data at different frequency points within the test frequency band.

[0123] In step 104, the complex dielectric response data is substituted into the oil-paper insulation equivalent dielectric model, and the undetermined parameters are identified based on the number of the undetermined parameter types to determine parameter identification values.

[0124] Preferably, the step of substituting the complex dielectric response data into the oil-paper insulation equivalent dielectric model, identifying the undetermined parameters based on the number of undetermined parameter types, and determining parameter identification values ​​comprises:

[0125] The likelihood function is determined based on the improved model as:

[0126]

[0127] Determine the prior distributions that the undetermined parameters satisfy respectively:

[0128]

[0129] σ~Gamma(α,β),

[0130] The goal of determining parameter identification is to obtain the posterior distribution of the unknown parameters, which is:

[0131] p(ε ∞ ∣ε Δ ,σ,w,ε),

[0132] p(ε Δ |ε ∞ ,σ,w,ε),

[0133] p(σ|ε ∞ ,ε Δ ,w,ε),

[0134] Initialize the initial values ​​of each pending parameter;

[0135] The above variables are updated and iterated in turn using the posterior distribution expressions of the undetermined parameters. During the calculation process, other undetermined parameters are considered to be known and fixed to the values ​​obtained by the latest round of iterations.

[0136] When the preset number of iterations is reached or the change in the parameter value is less than the preset threshold, the update is stopped, and the mean of the posterior distribution of each parameter estimated when the final round of iteration is completed is used as the parameter identification value;

[0137] Among them, (ε1,ε2,ε3,...,ε N ) is the oil-paper insulation equivalent dielectric model at different excitation signal frequency terms w1, w2, w3, ..., w N The complex dielectric constant corresponding to the distribution below; N is the number of data pairs; ζ is the variance of the joint distribution i represents the i-th group of sample data; μ0 and ζ0 are ε ∞The mean and variance of the normal distribution satisfied; μ1 and ζ1 are ε Δ The mean and variance of the normal distribution satisfied by the error term σ; α and β are the shape parameter and scale parameter of the Gamma distribution satisfied by the error term σ.

[0138] Preferably, the method

[0139] In progress ∞ When obtaining the posterior distribution of ∞ The distribution function of is:

[0140]

[0141] The method is carried out in Δ When obtaining the posterior distribution of Δ The distribution function of is:

[0142]

[0143] When the method is used to obtain the posterior distribution of σ, the distribution function of σ is:

[0144]

[0145] Among them, ε i 、w i are the values ​​of the complex dielectric constant and the complex frequency term in the i-th group of data samples respectively.

[0146] In the present invention, after determining the parameters to be determined, the above model parameters are identified from the dielectric response data obtained from the actual measurement test, and the parameter identification values ​​are determined. Among them, by performing a frequency domain dielectric spectrum test on the target type electrical equipment, the complex dielectric response data at different frequency points in the test frequency band are obtained. The complex dielectric response data is usually a complex capacitance obtained based on the excitation complex voltage and the response complex current, or a complex dielectric constant calculated from the capacitance.

[0147] The determination of feasible solutions for each parameter belongs to a type of nonlinear optimization problem in high-dimensional space. Conventional optimization methods are often inaccurate and difficult to converge. Therefore, the present invention chooses to adopt a type of Bayesian estimation method based on statistical machine learning. Specifically, the sampling technology used here is based on Markov Chain Monte Carlo (MCMC), combined with the frequency-complex dielectric constant data obtained by the frequency domain dielectric spectrum test to effectively identify the undetermined parameters in the multidimensional parameter space in the constructed Debye model. The specific process is as follows:

[0148] Assume that different excitation signal frequencies w1, w2, w3, ..., w are actually measured. NThe complex permittivity sequence corresponding to the distribution is (ε1,ε2,ε3,...,ε N ), N is the number of data pairs. Assume that the error term σ satisfies Normal distribution, then for the improved model ε=ε Δ w+ε ∞ +σ, its likelihood function is as follows:

[0149]

[0150] Assume that the parameters to be determined satisfy the following prior distributions:

[0151]

[0152] σ~Gamma(α,β),

[0153] The goal of the proposed parameter identification method based on statistical machine learning theory is to obtain the posterior distribution of the above parameters:

[0154] p(ε ∞ |ε Δ ,σ,w,ε),

[0155] p(ε Δ |ε ∞ ,σ,w,ε),

[0156] p(σ|ε ∞ ,ε Δ ,w,ε),

[0157] Where w and ε represent the joint distribution of the test data samples.

[0158] The method of the present invention simplifies the sampling process by decomposing the joint distribution into the above conditional distributions and sampling one by one, and iterates in sequence until each conditional distribution is stable by updating each variable one by one; assuming that the distribution of other variables is known, only the conditional distribution of a single variable is considered each time, thereby achieving sampling of the entire joint distribution.

[0159] Among them, the process of obtaining the probability density of the posterior conditional distribution of each parameter is as follows:

[0160] 1.ε ∞ Obtaining the Posterior Distribution

[0161] According to the Bayesian formula, ε ∞ The conditional distribution of satisfies the following quantitative relationship:

[0162] p(ε ∞ ∣ε Δ ,σ,ε,w)∝p(ε,w|ε ∞ ,ε Δ ,σ)p(ε ∞ ),

[0163] Take the logarithm of the right-hand side and pay attention to the equation containing ε ∞ The partial derivatives of each item are taken, and the mean and variance of the conditional distribution are obtained by the correlation coefficient method, and then ε ∞ The distribution function of is as follows:

[0164]

[0165] Its mean is taken as parameter ε ∞ The identification value of is calculated in each iteration as follows:

[0166]

[0167] 2.ε Δ Obtaining the Posterior Distribution

[0168] Similarly, we can get ε Δ The distribution function of is as follows:

[0169]

[0170] Its mean is taken as parameter ε Δ The identification value of , that is, the calculation formula in each round of iteration is as follows:

[0171]

[0172] 3. Obtaining the posterior distribution of σ

[0173] Similarly, the distribution function of σ is as follows:

[0174]

[0175] Since the prior distribution of σ is Gamma distribution, the calculation formula for the parameter σ in each round of iteration is as follows:

[0176]

[0177] Among them, ε i 、w i are the values ​​of the complex dielectric constant and the complex frequency term in the i-th group of data samples respectively.

[0178] Combination Figure 2 As shown, the present invention first determines the selection of the prior distribution of the above-mentioned parameters and the expression of the probability density of the posterior distribution, and then begins to enter the loop iteration parameter. The specific process is as follows:

[0179] (1) Select appropriate prior distribution types for each unknown parameter in the model.

[0180] (2) Derive the explicit expression of the posterior conditional probability density of each parameter and enter the cyclic iterative parameter optimization process.

[0181] (3) The initial values ​​of the undetermined parameters are randomly initialized. Alternatively, appropriate initial values ​​can be set for the undetermined parameters based on experience to accelerate the convergence of the algorithm.

[0182] (4) The derived posterior distribution expressions of the unknown parameters are used to perform iterative update operations on the variables in turn. During the calculation process, other unknown parameters are considered to be known and fixed to the values ​​obtained from the latest round of iterations.

[0183] (5) Stop updating parameters after an appropriate number of iterations, or stop updating when the estimated parameter value does not change significantly. The mean of the distribution of each parameter obtained in the final iteration is used as the estimated value of the proposed algorithm, that is, to determine the identification value of each parameter.

[0184] Preferably, the method further comprises:

[0185] The parameter identification value is substituted into the oil-paper insulation equivalent dielectric model to judge the insulation state of the target type electrical equipment based on the oil-paper insulation equivalent dielectric model including the parameter identification value.

[0186] In the present invention, the parameter identification value is substituted into the oil-paper insulation equivalent dielectric model, so as to judge the insulation state of the target type electrical equipment based on the oil-paper insulation equivalent dielectric model including the parameter identification value.

[0187] In order to verify the effectiveness of the proposed parameter identification method, the present invention is based on the frequency domain dielectric spectrum data obtained from actual testing of a 220kV oil-immersed transformer product in a low temperature environment, and compares the parameter identification effects of the method of the present invention with the heuristic genetic algorithm commonly used in existing research.

[0188] After substituting the parameters identified by the two methods into the model equation, the complex dielectric constant ε is obtained. * As a result, the complex dielectric constant is converted to complex capacitance by combining the equivalent vacuum capacitance C0:

[0189] C * =σ * C0,

[0190] C * =C'+iC".

[0191] like Figure 3 and Figure 4As shown in the figure, the identification results of the real part C' and the imaginary part C" of the complex capacitance are compared and analyzed respectively. Intuitively, it can be seen that the identification accuracy of the proposed method in the higher frequency band is better than that of the comparison method in terms of the identification effect of the real and imaginary parts of the complex capacitance. The identification error is quantitatively analyzed as follows.

[0192] Establish an identification error function based on the normalized mean square error:

[0193]

[0194] Where N is the actual number of test samples, y i represents the true value of the i-th sample obtained by testing, Represents the predicted value of the i-th sample obtained by identification. max and min Represent the maximum and minimum values ​​in the test data respectively. The quantization error comparison results are shown in Table 1 below:

[0195] Table 1 Quantization error comparison results

[0196] Normalized mean square error Proposed parameter identification method Comparative parameter identification methods Real part of complex capacitance C' 0.0017 0.0691 The imaginary part of the complex capacitance C" 0.0023 0.0782

[0197] It can be seen that the performance of the proposed method is also significantly better than the comparison method in terms of quantization error.

[0198] This is because the determination of feasible solutions for each parameter in the oil-paper insulation equivalent dielectric model is essentially a nonlinear optimization problem in a high-dimensional space. Conventional optimization methods are often inaccurate and difficult to converge or obtain a better solution. The parameter identification method proposed in this invention is supported by statistical machine learning theory, which can ensure convergence and effectively avoid local optimality to achieve global optimal parameter optimization. This can not only improve the accuracy of parameter estimation, but also enable a deep understanding of the physical properties of materials; and it is of great significance for formulating maintenance strategies for electrical equipment, preventing faults, and improving the stability and reliability of power systems in low-temperature environments. It is also in line with the development trend of smart grids and condition monitoring systems.

[0199] Figure 5 FIG. 5 is a schematic diagram of a parameter identification system 500 for an oil-paper insulation equivalent dielectric model according to an embodiment of the present invention. Figure 5 As shown, a parameter identification system 500 of an oil-paper insulation equivalent dielectric model provided in an embodiment of the present invention includes: a model building unit 501 , a parameter determination unit 502 , a complex dielectric response data acquisition unit 503 and a parameter identification unit 504 .

[0200] Preferably, the model building unit 501 is used to build an oil-paper insulation equivalent dielectric model of a target type of electrical equipment.

[0201] Preferably, the parameter determination unit 502 is used to determine the type and quantity of undetermined parameters in the oil-paper insulation equivalent dielectric model.

[0202] Preferably, the parameter determination unit 502 determines the type and quantity of undetermined parameters in the oil-paper insulation equivalent dielectric model, including:

[0203] The oil-paper insulation equivalent dielectric model is simplified and improved to obtain an improved model: ε = ε Δ w+ε ∞ +σ, and determine the type and quantity of the parameters to be determined based on the improved model; where ε is the complex dielectric constant, which is a function of the complex frequency term w; ε Δ is the differential dielectric constant; ε ∞ is the optical frequency dielectric constant; σ is the error term; ε Δ , ε ∞ and σ are unknown parameters that need to be identified.

[0204] Preferably, the complex dielectric response data acquisition unit 503 is used to perform a frequency domain dielectric spectrum test on the target type electrical equipment to acquire complex dielectric response data at different frequency points within a test frequency band.

[0205] Preferably, the parameter identification unit 504 is used to substitute the complex dielectric response data into the oil-paper insulation equivalent dielectric model, identify the undetermined parameters based on the number of undetermined parameter types, and determine parameter identification values.

[0206] Preferably, the parameter identification unit 504 substitutes the complex dielectric response data into the oil-paper insulation equivalent dielectric model, identifies the undetermined parameters based on the number of undetermined parameter types, and determines the parameter identification values, including:

[0207] The likelihood function is determined based on the improved model as:

[0208]

[0209] Determine the prior distributions that the undetermined parameters satisfy respectively:

[0210]

[0211] σ~Gamma(α,β),

[0212] The goal of determining parameter identification is to obtain the posterior distribution of the unknown parameters, which is:

[0213]

[0214] Initialize the initial values ​​of each pending parameter;

[0215] The above variables are updated and iterated in turn using the posterior distribution expressions of the undetermined parameters. During the calculation process, other undetermined parameters are considered to be known and fixed to the values ​​obtained by the latest round of iterations.

[0216] When the preset number of iterations is reached or the change in the parameter value is less than the preset threshold, the update is stopped, and the mean of the posterior distribution of each parameter estimated when the final round of iteration is completed is used as the parameter identification value;

[0217] Among them, (ε1,ε2,ε3,…,ε N ) is the oil-paper insulation equivalent dielectric model at different excitation signal frequency terms w1, w2, w3, ..., w N The complex dielectric constant corresponding to the distribution below; N is the number of data pairs; ζ is the variance of the joint distribution i represents the i-th group of sample data; μ0 and ζ0 are ε ∞ The mean and variance of the normal distribution satisfied; μ1 and ζ1 are ε Δ The mean and variance of the normal distribution satisfied by the error term σ; α and β are the shape parameter and scale parameter of the Gamma distribution satisfied by the error term σ.

[0218] Preferably, the parameter identification unit 504,

[0219] In progress ∞ When obtaining the posterior distribution of ∞ The distribution function of is:

[0220]

[0221] The method is carried out in Δ When obtaining the posterior distribution of Δ The distribution function of is:

[0222]

[0223] When the method is used to obtain the posterior distribution of σ, the distribution function of σ is:

[0224]

[0225] Among them, ε i 、w i are the values ​​of the complex dielectric constant and the complex frequency term in the i-th group of data samples respectively.

[0226] Preferably, the system further comprises:

[0227] The state judgment unit is used to substitute the parameter identification value into the oil-paper insulation equivalent dielectric model to judge the insulation state of the target type electrical equipment based on the oil-paper insulation equivalent dielectric model containing the parameter identification value.

[0228] The parameter identification system 500 of the oil-paper insulation equivalent dielectric model of the embodiment of the present invention corresponds to the parameter identification method 100 of the oil-paper insulation equivalent dielectric model of another embodiment of the present invention, which will not be described in detail here.

[0229] Based on another aspect of the present invention, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any step of a parameter identification method for an oil-paper insulation equivalent dielectric model.

[0230] According to another aspect of the present invention, the present invention provides an electronic device, including:

[0231] The computer-readable storage medium described above; and

[0232] One or more processors are used to execute the program in the computer-readable storage medium.

[0233] The invention has been described above with reference to a few embodiments. However, it is readily apparent to a person skilled in the art that other embodiments than the ones disclosed above are equally within the scope of the invention, as defined by the appended patent claims.

[0234] Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise therein. All references to "a / said / the [means, components, etc.]" are to be openly interpreted as at least one instance of said means, components, etc., unless explicitly stated otherwise. The steps of any method disclosed herein do not necessarily have to be performed in the exact order disclosed, unless explicitly stated otherwise.

[0235] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0236] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0237] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0238] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0239] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A parameter identification method for an oil-paper insulation equivalent dielectric model, characterized in that: The method comprises: Establish the oil-paper insulation equivalent dielectric model of the target type of electrical equipment; Determining the type and quantity of undetermined parameters in the oil-paper insulation equivalent dielectric model; Conducting frequency domain dielectric spectrum testing on the target type of electrical equipment to obtain complex dielectric response data at different frequency points within the test frequency band; Substitute the complex dielectric response data into the oil-paper insulation equivalent dielectric model, identify the undetermined parameters based on the number of undetermined parameter types, and determine parameter identification values.

2. The method according to claim 1, characterized in that Determine the type and quantity of undetermined parameters in the oil-paper insulation equivalent dielectric model, including: The oil-paper insulation equivalent dielectric model is simplified and improved to obtain an improved model: ε = ε Δ w+ε ∞ +σ, and determine the type and quantity of the parameters to be determined based on the improved model; where ε is the complex dielectric constant, which is a function of the complex frequency term w; ε Δ is the differential dielectric constant; ε ∞ is the optical frequency dielectric constant; σ is the error term; ε Δ , ε ∞ and σ are unknown parameters that need to be identified.

3. The method according to claim 2, characterized in that Substituting the complex dielectric response data into the oil-paper insulation equivalent dielectric model, identifying the undetermined parameters based on the number of undetermined parameter types, and determining parameter identification values, comprises: The likelihood function is determined based on the improved model as: Determine the prior distributions that the undetermined parameters satisfy respectively: σ~Gamma(α,β), The goal of determining parameter identification is to obtain the joint posterior distribution of the unknown parameters, which is: p(e ∞ ∣e Δ ,σ,w,ε), p(e Δ |e ∞ ,σ,w,ε), p(s|e ∞ ,he Δ ,w,e), Initialize the initial values ​​of each pending parameter; The above variables are updated and iterated in turn using the posterior distribution expressions of the undetermined parameters. During the calculation process, other undetermined parameters are considered to be known and fixed to the values ​​obtained by the latest round of iterations. When the preset number of iterations is reached or the change in the parameter value is less than the preset threshold, the update is stopped, and the mean of the posterior distribution of each parameter estimated when the final round of iteration is completed is used as the parameter identification value; Among them, (ε1,ε2,ε3,...,ε N ) is the oil-paper insulation equivalent dielectric model at different excitation signal frequency terms w1, w2, w3, ..., w N The complex dielectric constant corresponding to the distribution below; N is the number of data pairs; ζ is the variance of the joint distribution; i represents the i-th group of sample data; μ0 and ζ0 are ε ∞ The mean and variance of the normal distribution satisfied; μ1 and ζ1 are ε Δ The mean and variance of the normal distribution satisfied by the error term σ; α and β are the shape parameter and scale parameter of the Gamma distribution satisfied by the error term σ.

4. The method according to claim 3, characterized in that The method is carried out in ∞ When obtaining the posterior distribution of ∞ The distribution function of is: The method is carried out in Δ When obtaining the posterior distribution of Δ The distribution function of is: When the method is used to obtain the posterior distribution of σ, the distribution function of σ is: Among them, ε i 、w i are the values ​​of the complex dielectric constant and the complex frequency term in the i-th group of data samples respectively.

5. The method according to claim 1, characterized in that The method further comprises: The parameter identification value is substituted into the oil-paper insulation equivalent dielectric model to judge the insulation state of the target type electrical equipment based on the oil-paper insulation equivalent dielectric model including the parameter identification value.

6. A parameter identification system for an oil-paper insulation equivalent dielectric model, characterized in that: The system comprises: A model building unit, used to build an oil-paper insulation equivalent dielectric model of target type electrical equipment; A parameter determination unit, used to determine the type and quantity of undetermined parameters in the oil-paper insulation equivalent dielectric model; A complex dielectric response data acquisition unit, used to perform a frequency domain dielectric spectrum test on the target type electrical equipment to acquire complex dielectric response data at different frequency points within the test frequency band; The parameter identification unit is used to substitute the complex dielectric response data into the oil-paper insulation equivalent dielectric model, identify the undetermined parameters based on the number of undetermined parameter types, and determine parameter identification values.

7. The system according to claim 6, characterized in that The parameter determination unit determines the type and quantity of undetermined parameters in the oil-paper insulation equivalent dielectric model, including: The oil-paper insulation equivalent dielectric model is simplified and improved to obtain an improved model: ε = ε Δ w+ε ∞ +σ, and determine the type and quantity of the parameters to be determined based on the improved model; where ε is the complex dielectric constant, which is a function of the complex frequency term w; ε Δ is the differential dielectric constant; ε ∞ is the optical frequency dielectric constant; σ is the error term; ε Δ , ε ∞ and σ are unknown parameters that need to be identified.

8. The system according to claim 7, characterized in that The parameter identification unit substitutes the complex dielectric response data into the oil-paper insulation equivalent dielectric model, identifies the undetermined parameters based on the number of undetermined parameter types, and determines the parameter identification values, including: The likelihood function is determined based on the improved model as: Determine the prior distributions that the parameters to be determined satisfy: σ~Gamma(α,β), The goal of determining parameter identification is to obtain the joint posterior distribution of the unknown parameters, which is: p(e ∞ ∣e Δ ,σ,w,ε), p(e Δ |e ∞ ,σ,w,ε), p(s|e ∞ ,he Δ ,w,e), Initialize the initial values ​​of each pending parameter; The above variables are updated and iterated in turn using the posterior distribution expressions of the undetermined parameters. During the calculation process, other undetermined parameters are considered to be known and fixed to the values ​​obtained by the latest round of iterations. When the preset number of iterations is reached or the change in the parameter value is less than the preset threshold, the update is stopped, and the mean of the posterior distribution of each parameter estimated when the final round of iteration is completed is used as the parameter identification value; Among them, (ε1,ε2,ε3,...,ε N ) is the oil-paper insulation equivalent dielectric model at different excitation signal frequency terms w1, w2, w3, ..., w N The complex dielectric constant corresponding to the distribution below; N is the number of data pairs; ζ is the variance of the joint distribution; i represents the i-th group of sample data; μ0 and ζ0 are ε ∞ The mean and variance of the normal distribution satisfied; μ1 and ζ1 are ε Δ The mean and variance of the normal distribution satisfied by the error term σ; α and β are the shape parameter and scale parameter of the Gamma distribution satisfied by the error term σ.

9. The system according to claim 8, characterized in that The parameter identification unit, In progress ∞ When obtaining the posterior distribution of ∞ The distribution function of is: The method is carried out in Δ When obtaining the posterior distribution of Δ The distribution function of is: When the method is used to obtain the posterior distribution of σ, the distribution function of σ is: Among them, ε i 、w i are the values ​​of the complex dielectric constant and the complex frequency term in the i-th group of data samples respectively.

10. The system according to claim 6, characterized in that The system further comprises: The state judgment unit is used to substitute the parameter identification value into the oil-paper insulation equivalent dielectric model to judge the insulation state of the target type electrical equipment based on the oil-paper insulation equivalent dielectric model containing the parameter identification value.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

12. An electronic device, characterized in that: include: The computer readable storage medium as claimed in claim 11; and one or more processors for executing the program in the computer-readable storage medium.

Images