Frequency domain dielectric spectrum analysis method and system for monitoring insulation aging of electrical transformer

Abstract

The application discloses a frequency domain dielectric spectrum analysis method and system for monitoring insulation aging of an electrical transformer, relates to the technical field of electrical equipment state monitoring and frequency domain modeling analysis, and comprises the following steps: collecting complex impedance data of the electrical transformer under different frequencies, and constructing an equivalent circuit model comprising a resistance unit and two constant phase elements; introducing a structural constraint relationship to parameters of the equivalent circuit model, and obtaining a parameter combination satisfying a uniqueness condition based on constraint optimization; matching the optimized parameter combination with an aging mechanism dictionary, and judging an insulation state grade and a dominant aging mechanism by using a classifier. The method constructs a fitting process and a scoring mechanism under constraint driving, and improves the interpretability, stability and cross-sample universality of the parameters. The physical modeling result is formally introduced into a data-driven state and mechanism identification framework, and the practicability and vitality of the diagnostic system are improved.

Description

Technical Field

[0001] This invention relates to the field of electrical equipment condition monitoring and frequency domain modeling analysis technology, specifically to a frequency domain dielectric spectrum analysis method and system for monitoring the insulation aging of electrical transformers. Background Technology

[0002] With the continuous improvement of power system requirements for equipment reliability, instrument transformers, as key primary equipment in substations, have received widespread attention for their insulation condition monitoring technology. In recent years, frequency domain dielectric spectroscopy analysis has been increasingly applied to the insulation aging assessment of electrical equipment due to its advantages such as non-destructive nature, high sensitivity, and ability to quantify the state. This method measures the complex impedance or dielectric response of the equipment at different frequencies to uncover its internal dielectric polarization characteristics and charge migration behavior. It also utilizes equivalent circuit modeling to transform the complex electrical response into interpretable physical parameters, making it one of the important means for online insulation condition monitoring and degradation mechanism identification.

[0003] However, existing insulation monitoring technologies based on frequency domain modeling still have many limitations in parameter extraction and mechanism interpretation. First, most common parameter fitting methods ignore the physical constraints of the model structure, resulting in multiple feasible solutions for the same spectrum, making it difficult to achieve physical interpretability of parameters and consistency of mechanisms. Second, the lack of a systematic mapping relationship between parameters and aging mechanisms means that the diagnosis of insulation aging often only stays at the trend identification level, making it difficult to accurately attribute the dominant mechanism. Third, most technical solutions lack the ability to model the logical relationships between parameters, and cannot handle parameter mismatch or redundant modeling problems caused by the coupling between aging mechanisms, further affecting the robustness and repeatability of the diagnosis. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by this invention is that existing methods for analyzing the insulation status of electrical transformers have problems such as non-unique solutions for equivalent circuit model parameters, lack of mapping mechanism between parameters and aging mechanism, difficulty in handling mechanism coupling and structural constraint logic in diagnostic models, and how to achieve joint discrimination of insulation status level and dominant aging mechanism based on frequency domain parameters.

[0006] To address the aforementioned technical problems, this invention provides the following technical solution: a frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers, comprising: collecting complex impedance data of electrical transformers at different frequencies and constructing an equivalent circuit model containing resistive units and two constant phase elements; introducing structural constraint relationships into the parameters of the equivalent circuit model and obtaining parameter combinations that satisfy the uniqueness condition based on constraint optimization; matching the optimized parameter combinations with an aging mechanism dictionary and using a classifier to determine the insulation state level and the dominant aging mechanism.

[0007] As a preferred embodiment of the frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers according to the present invention, the method for acquiring complex impedance data of electrical transformers at multiple frequencies includes: applying an excitation signal with a frequency range of 10μHz to 1MHz to the target transformer; measuring the frequency domain response of the target transformer using an impedance analyzer; acquiring complex impedance data composed of the real and imaginary parts or the converted complex dielectric constant data; performing complex domain filtering and background noise suppression on the sampled data; and using discrete Fourier transform to interpolate non-equidistant frequency points into equidistant spectral curves.

[0008] As a preferred embodiment of the frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers according to the present invention, the equivalent circuit model includes establishing a resistor... Two constant phase elements connected in series and The equivalent circuit model of the CPE element, and its impedance function. Represented as: in, These are the fitting coefficients. For phase factor, The imaginary unit, ω is the angular frequency.

[0009] As a preferred embodiment of the frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers according to the present invention, the structural constraint relationship includes setting physical mechanism relationship constraints, resistance With the second constant phase element Fit coefficient They satisfy an inverse proportional constraint relationship constant; Set the response rate level limit and phase factor. and satisfy , The first constant phase element phase factor, Second constant phase element Phase factor; Set aging trend structure matching constraints to make and The rates of change are in the same direction, satisfying... , Second constant phase element The fitting coefficient, For time; Structural constraints are embedded in the fitted optimization objective function in the form of hard constraints, and the optimal parameter set is solved by penalty term optimization algorithm or constraint programming method.

[0010] As a preferred embodiment of the frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers according to the present invention, the method for obtaining parameter combinations that satisfy the uniqueness condition includes scoring different parameter combination solutions using multiple indicators, including at least fitting error, physical consistency score, trend stability score, and frequency response explanatory power score. Finally, a comprehensive scoring rule is used to determine the global optimal solution, and models with parameter exchange, numerical degradation, or error masking behavior are screened out.

[0011] As a preferred embodiment of the frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers described in this invention, the step of matching the optimized parameter combination with the aging mechanism dictionary includes comparing each set of parameters with a pre-established mechanism fingerprint reference library. The aging mechanism dictionary stores multiple sets of typical aging mechanism labels corresponding to resistance, capacitance, CPE coefficient and phase index. The matching process adopts a threshold matching and similarity scoring mechanism. When the matching score of a certain set of parameters in the dictionary exceeds a set threshold, the corresponding set of parameters is mapped to the corresponding dominant aging mechanism. If no match is found, the anomaly classifier module is activated to mark unknown mechanisms and expand the mechanism model.

[0012] As a preferred embodiment of the frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers according to the present invention, the method of using a classifier to determine the insulation state level and the dominant aging mechanism includes, The final selected parameter set is used as a feature vector and input into the trained classification model. The classification model includes support vector machine, random forest or convolutional neural network model, which is used to output the current insulation aging level and the dominant aging mechanism label. Cross-validation and dynamic update mechanism are used during model training.

[0013] As a preferred embodiment of the frequency domain dielectric spectrum analysis system for monitoring the insulation aging of electrical transformers according to the present invention, the system includes: an equivalent circuit construction module, a constraint optimization module, and an aging monitoring module; the equivalent circuit construction module is used to collect complex impedance data of the electrical transformer at different frequencies and construct an equivalent circuit model containing resistive elements and two constant phase elements; the constraint optimization module is used to introduce structural constraint relationships into the parameters of the equivalent circuit model and obtain parameter combinations that satisfy the uniqueness condition based on constraint optimization; the aging monitoring module is used to match the optimized parameter combinations with an aging mechanism dictionary and use a classifier to determine the insulation state level and the dominant aging mechanism.

[0014] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement a frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers.

[0015] A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of a frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers are disclosed.

[0016] The beneficial effects of this invention are as follows: The frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers provided by this invention ensures the integrity of the spectral signal and the physical separation of the modeling parameters through a standardized modeling structure and high-quality frequency domain data input. This provides a stable and physically traceable parameter space for subsequent mechanism classification and state judgment. It breaks away from the traditional logic of judging parameter quality solely based on "minimum error" in fitting, and constructs a constraint-driven fitting process and scoring mechanism. This realizes the transformation of the parameter space from "optimal fit" to "physical uniqueness," improving the interpretability, stability, and cross-sample universality of the parameters. By formally introducing the physical modeling results into a data-driven state and mechanism identification framework, it not only achieves intelligent and labeling of the diagnostic process but also possesses the ability to learn new mechanisms. This ensures that the system still has the ability to adapt and upgrade when facing unknown aging scenarios, significantly improving the practicality and vitality of the diagnostic system. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is an overall flowchart of the frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers provided in Embodiment 1 of the present invention. Detailed Implementation

[0019] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0020] Example 1, referring to Figure 1 As an embodiment of the present invention, a frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers is provided, comprising: S1: Collect complex impedance data of electrical transformers at different frequencies and construct an equivalent circuit model containing resistive elements and two constant phase elements.

[0021] Furthermore, the acquisition of complex impedance data of electrical transformers at multiple frequencies includes applying an excitation signal with a frequency range of 10μHz to 1MHz to the target transformer, measuring the frequency domain response of the target transformer using an impedance analyzer, acquiring complex impedance data composed of the real and imaginary parts or the converted complex permittivity data, performing complex domain filtering and background noise suppression on the sampled data, and using discrete Fourier transform to interpolate non-equidistant frequency points into equidistant spectral curves.

[0022] It should be noted that the equivalent circuit model includes establishing a circuit composed of resistors. Two constant phase elements connected in series and The equivalent circuit model of the CPE element, and its impedance function. Represented as: in, These are the fitting coefficients. For phase factor, The imaginary unit, ω is the angular frequency.

[0023] It should also be noted that an equivalent circuit model Specifically, it is expressed as follows: First constant phase element The fitting coefficients and phase factors are used to fit polarization hysteresis or interface polarization, and the second constant phase element. The fitting coefficients and phase factors are used to fit space charge accumulation or hygroscopic properties.

[0024] It should also be noted that by collecting complex impedance data of electrical transformers at different frequencies and constructing an equivalent circuit model containing resistive elements and two constant-phase elements, the complex frequency-domain dielectric response behavior was transformed into a structured set of parameters with physical meaning. Furthermore, the sampling process introduced wide-band excitation (covering μHz to MHz), combined with complex-domain filtering, noise suppression, and equidistant interpolation operations, effectively improving the frequency coverage integrity and spectral smoothness of the data, and enhancing the stability and resolution of subsequent modeling. The established equivalent circuit model maps polarization behavior and charge accumulation to two constant-phase elements, respectively, which helps to separate the influence of different aging mechanisms on the spectrum, achieving a preliminary correspondence between modeling parameters and physical aging phenomena, and laying the foundation for parameter interpretability.

[0025] S2: Introduce structural constraints on the parameters of the equivalent circuit model, and obtain parameter combinations that satisfy the uniqueness condition based on constraint optimization.

[0026] Furthermore, structural constraints include setting physical mechanism constraints, such as resistance. With the second constant phase element Fit coefficient They satisfy an inverse proportional constraint relationship constant( ); Set the response rate level limit and phase factor. and satisfy , The first constant phase element The phase factor typically reflects the "rapid polarization" process. Second constant phase element The phase factor usually reflects "slow polarization" or "space charge effect". Generally, interface polarization is faster and has a higher response frequency, while space charge polarization is slower and has a lower response frequency. Therefore, the phase response order should satisfy this relationship. Set aging trend structure matching constraints to make and The rates of change are in the same direction, satisfying... , Second constant phase element The fitting coefficient, For time; Structural constraints are embedded as hard constraints in the fitted optimization objective function. The optimal parameter set is solved using penalty term optimization algorithms or constraint programming methods. For constraints related to physical mechanisms, penalty terms can be added to the optimization objective function. : in, For the sum of squared fitting errors, This is a penalty factor used to adjust the constraint strength. This is an empirical constant, which can be determined through experimental statistics (e.g., by averaging multiple samples). For mechanism response rate level constraints, this inequality can be transformed into a penalty term. : in, The basic loss term.

[0027] It should also be noted that in frequency domain dielectric spectrum analysis, traditional equivalent circuit modeling aims to minimize spectral error by independently optimizing the parameters of each circuit element (such as resistance R, capacitance C, and constant phase elements Q and n). While this approach performs well in terms of fitting error, it suffers from the following serious problems: multiple different parameter combinations may result in very similar fitting results, leading to parameter "exchange" or "aliasing" issues; the lack of physical coupling constraints between parameters may cause physically unreasonable deviations in some parameter values; and the model exhibits poor stability and reconfigurability across data from different devices or time periods. To address these issues, this invention proactively introduces structural constraints on parameters during the modeling stage, establishing functional relationships between parameters to prevent physically meaningless optimal solutions, thereby improving the interpretability and uniqueness of the parameters.

[0028] It should be noted that obtaining parameter combinations that satisfy the uniqueness condition includes using multiple indicators to score different parameter combination solutions. These indicators include at least fitting error, physical consistency score, trend stability score, and frequency response explanatory power score. Finally, a comprehensive scoring rule is used to determine the global optimal solution, and models with parameter exchange, numerical degradation, or error masking behavior are screened out.

[0029] It should also be noted that a preferred approach for scoring using multiple indicators specifically includes, in part, the fitting error. Represented as: in, This represents the total number of frequency sampling points. For the first Measured complex impedance values ​​at each frequency point; In the parameter group The complex impedance value calculated by the following model; Physical consistency score Represented as: Trend stability score Represented as: in, This represents the total number of time series points (i.e., the number of monitoring time periods). Frequency response explanatory power score Represented as: in, Frequency point upper model impedance versus parameters The partial derivative of the parameter represents the "influence" of the parameter at that frequency point.

[0030] Overall score Represented as: in, , , , Let be the weight coefficients for each scoring item, satisfying... ; The weighting coefficients can be set according to task requirements: When precision is prioritized: When interpretability takes precedence: When stability is the priority: If no candidate satisfies the minimum threshold (e.g., Score < 0.8), the model is re-initialized or the structure is switched. Finally, the parameter combination with the highest score is selected as the "unique and reliable solution".

[0031] It should also be noted that, based on structural constraints, this invention further introduces a multi-dimensional, multi-factor parameter screening and scoring mechanism. This mechanism evaluates the "reliability," "uniqueness," "trend rationality," and "spectral segment explanatory power" of candidate parameter combinations from multiple dimensions, ultimately selecting a unique optimal solution through a total score. This effectively eliminates "pseudo-optimal solutions" that appear well-fitted but are fundamentally unreliable. By introducing structural constraints during the modeling process, the physical relationships between parameters in the equivalent circuit model are actively restricted, thus solving problems such as parameter "commutativity," "degradation," and "pseudo-optimal solutions" present in traditional minimum error optimization. More specifically, three types of structural constraints are introduced: first, the inverse relationship between resistance and slow polarization terms is used to control the consistency of the aging path; second, the magnitude relationship of phase factors is used to maintain the physical time order of the polarization mechanism; and third, the consistency of the temporal change direction of the fitting coefficients is used to maintain the logical rationality of the aging trend. All constraints are embedded in the fitting optimization objective function through penalty functions or constraint programming methods, ensuring that physical consistency is satisfied while minimizing the fitting error. By introducing a multi-dimensional scoring system (including fitting error, physical consistency, trend stability, and frequency response resolution), the credibility and uniqueness of the final output parameter set are further ensured, and pseudo-solutions that perform well but are physically unreasonable are eliminated.

[0032] S3: Match the optimized parameter combination with the aging mechanism dictionary, and use a classifier to determine the insulation state level and the dominant aging mechanism.

[0033] Furthermore, the optimized parameter combination is matched with the aging mechanism dictionary. This includes comparing each set of parameters with a pre-established mechanism fingerprint reference library. The aging mechanism dictionary stores multiple sets of typical aging mechanism labels corresponding to resistance, capacitance, CPE coefficient and phase index. The matching process adopts a threshold matching and similarity scoring mechanism. When the matching score of a set of parameters in the dictionary exceeds the set threshold, the corresponding set of parameters is mapped to the corresponding dominant aging mechanism. If no match is found, the anomaly classifier module is activated to mark unknown mechanisms and expand the mechanism model.

[0034] It should be noted that using a classifier to determine the insulation condition level and dominant aging mechanism involves inputting the final selected parameter set as a feature vector into the trained classification model. The classification model includes a support vector machine, random forest, or convolutional neural network model, which is used to output the current insulation aging level and dominant aging mechanism label. Cross-validation and dynamic update mechanisms are used during model training.

[0035] It should also be noted that a preferred embodiment of a typical aging mechanism label is specifically represented as follows: in, The label is a string indicating the dominant aging mechanism represented by the sample. The mechanism label adopts a well-structured and extensible naming system, currently including but not limited to the following examples: interface polarization dominant, space charge accumulation dominant, moisture-induced polarization, temperature rise-induced dielectric degradation, and nonlinear breakdown indication mechanism.

[0036] The matching algorithm logic and execution flow in obtaining the optimal parameter set Then, the mechanism identification process begins. The identification process includes the following steps: Normalization preprocessing: Normalize the parameter group to be identified and all sample parameter vectors in the mechanism library to the [0,1] interval; Similarity calculation: Cosine similarity is used to match each group of mechanism samples. The calculation formula is as follows: in: Represents the dot product of vectors; This represents the Euclidean norm (L2 norm). For the first in the mechanism library The parameter vector of each sample.

[0037] Judgment and Output: Setting a similarity threshold (e.g., 0.85), if there exist samples that satisfy: If the corresponding mechanism label is output as the recognition result, then if no mechanism sample meets the above conditions, it is determined to be an "unknown mechanism" and proceeds to the following processing flow. When the parameter set cannot match any known mechanism sample, the system will execute the following processing mechanism: The current parameter set is added to the "sample pool to be confirmed" as a potential candidate for a new mechanism; The labeling of this sample requires manual review, and experts will classify the mechanism based on the experimental results. If it is verified to be a new mechanism, the mechanism dictionary can be expanded to form new sample entries; The new mechanism labels and their parameter features are persistently stored in the system for subsequent matching.

[0038] It should also be noted that by matching the optimized parameter combination with a pre-built aging mechanism dictionary and using a classifier to determine the insulation state level and the dominant aging mechanism, an automatic mapping from the parameter space to the mechanism label space is achieved. The mechanism dictionary is constructed using expert knowledge, measured samples, and modeling data, possessing a clear label structure and scalability. The matching method employs normalization processing and cosine similarity scoring to ensure the numerical stability and accuracy robustness of the mechanism identification process. The classifier module constructs support vector machines, random forests, or convolutional neural networks based on training samples, achieving multi-dimensional feature discrimination of state levels. Notably, this step introduces "abnormal mechanism identification" and a "sample pool to be reviewed" when matching fails, enabling the entire identification system to have self-expanding knowledge capabilities and forming a closed-loop logic of "parameters → labels → library update".

[0039] Example 2, an embodiment of the present invention, provides a frequency domain dielectric spectrum analysis system for monitoring the insulation aging of electrical transformers, including an equivalent circuit construction module, a constraint optimization module, and an aging monitoring module.

[0040] The equivalent circuit construction module is used to collect the complex impedance data of electrical transformers at different frequencies and construct an equivalent circuit model containing resistive units and two constant phase elements; the constraint optimization module is used to introduce structural constraint relationships into the parameters of the equivalent circuit model and obtain parameter combinations that satisfy the uniqueness condition based on constraint optimization; the aging monitoring module is used to match the optimized parameter combinations with the aging mechanism dictionary and use a classifier to determine the insulation state level and the dominant aging mechanism.

Claims

1. A frequency domain dielectric spectrum analysis method for monitoring insulation aging of electrical transformers, characterized in that, include: Collect complex impedance data of electrical transformers at different frequencies and construct an equivalent circuit model containing resistive elements and two constant phase elements. Structural constraints are introduced into the parameters of the equivalent circuit model, and parameter combinations that satisfy the uniqueness condition are obtained based on constraint optimization. The optimized parameter combination is matched with the aging mechanism dictionary, and a classifier is used to determine the insulation state level and the dominant aging mechanism. The equivalent circuit model includes, Establish a resistor Two constant phase elements connected in series and The equivalent circuit model of the CPE element, and its impedance function. Represented as: in, These are the fitting coefficients. For phase factor, The imaginary unit, Angular frequency; Structural constraints include, Set physical mechanism relationship constraints, resistance With the second constant phase element Fit coefficient They satisfy an inverse proportional constraint relationship constant; Set the mechanism response rate level limit, phase factor and satisfy , The first constant phase element phase factor, Second constant phase element Phase factor; Set aging trend structure matching constraints to make and The rates of change are in the same direction, satisfying the condition that... , Second constant phase element The fitting coefficient, For time; Structural constraints are embedded in the fitted optimization objective function in the form of hard constraints, and the optimal parameter set is solved by penalty term optimization algorithm or constraint programming method.

2. The frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers as described in claim 1, characterized in that: The acquired complex impedance data of the electrical transformer at multiple frequencies includes, An excitation signal with a frequency range of 10μHz to 1MHz is applied to the target transformer. The frequency domain response of the target transformer is measured by an impedance analyzer. Complex impedance data composed of real and imaginary parts or complex permittivity data after conversion are collected. Complex domain filtering and background noise suppression are performed on the sampled data. Discrete Fourier transform is used to interpolate non-equidistant frequency points into equidistant spectrum curves.

3. The frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers as described in claim 1, characterized in that: The method for obtaining parameter combinations that satisfy the uniqueness condition includes, Multiple indicators are used to score solutions with different parameter combinations. These indicators include at least fitting error, physical consistency score, trend stability score, and frequency response explanatory power score. Finally, a comprehensive scoring rule is used to determine the global optimal solution and to filter out models that exhibit parameter swapping, numerical degradation, or error masking behavior.

4. The frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers as described in claim 3, characterized in that: The step of matching the optimized parameter combination with the aging mechanism dictionary includes... Each set of parameters is compared with a pre-established mechanism fingerprint reference library. The aging mechanism dictionary stores multiple sets of typical aging mechanism labels corresponding to resistance, capacitance, CPE coefficient and phase index. The matching process adopts threshold matching and similarity scoring mechanism. When the matching score of a set of parameters in the dictionary exceeds the set threshold, the corresponding set of parameters is mapped to the corresponding dominant aging mechanism. If no match is found, the anomaly classifier module is activated to mark unknown mechanisms and expand the mechanism model.

5. The frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers as described in claim 4, characterized in that: The method of using a classifier to determine the insulation condition level and dominant aging mechanism includes, The final selected parameter set is used as a feature vector and input into the trained classification model. The classification model includes support vector machine, random forest or convolutional neural network model, which is used to output the current insulation aging level and the dominant aging mechanism label. Cross-validation and dynamic update mechanism are used during model training.

6. A frequency domain dielectric spectrum analysis system for monitoring the insulation aging of electrical transformers, employing the frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers as described in any one of claims 1 to 5, characterized in that: Includes an equivalent circuit construction module, a constraint optimization module, and an aging monitoring module; The equivalent circuit construction module is used to collect the complex impedance data of the electrical transformer at different frequencies and construct an equivalent circuit model containing a resistive unit and two constant phase elements. The constraint optimization module is used to introduce structural constraint relationships into the parameters of the equivalent circuit model, and obtain parameter combinations that satisfy the uniqueness condition based on constraint optimization. The aging monitoring module is used to match the optimized parameter combination with the aging mechanism dictionary, and to use a classifier to determine the insulation state level and the dominant aging mechanism.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers as described in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the frequency domain dielectric spectrum analysis method for monitoring the insulation aging of electrical transformers as described in any one of claims 1 to 5.

Images