Transformer monitoring method based on fusion of mechanism model and data model

Abstract

The invention discloses a transformer monitoring method based on fusion of a mechanism model and a data model, relates to the technical field of fault identification, and solves the problems of single monitoring dimension, insufficient data fusion and poor adaptability to a complex operation environment due to dependence on a single monitoring means or a simple data processing method in the prior art. The mechanism model established by the invention provides a physical background and an interpretation basis for the fusion model, so that the fault diagnosis result is more interpretable. When a fault is diagnosed by the fusion model, the reason of the fault can be analyzed through the mechanism model, for example, insulation aging is caused by winding hot-spot temperature rise, and then an overheating fault is generated. In addition, the introduction of the data model enhances the adaptability of the fusion model to different operating environments and transformer types. Different transformer winding structures and operation conditions may have differences, and the data model can learn the differences, so that the fusion model can better adapt to various conditions, and the generalization ability of fault diagnosis is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fault identification, and in particular to a transformer monitoring method based on the fusion of a mechanism model and a data model. Background Art

[0002] Power transformers, as core equipment in power systems, undertake the crucial task of converting and distributing electrical energy. Their healthy operation is directly related to the stability and reliability of power systems. However, transformers may face various faults during long-term operation, such as winding overheating, insulation aging, and partial discharge. If these faults are not promptly detected and addressed, they can cause equipment damage or even widespread power outages. Therefore, transformer fault identification technology is crucial. Traditional fault diagnosis methods primarily rely on single monitoring methods, such as dissolved gas analysis (DGA) or vibration signal detection, but these methods have limitations. While DGA can identify internal fault characteristics, it has a long data acquisition cycle and poor real-time performance. While vibration signal detection offers a fast response, it lacks precision in fault type identification. Furthermore, existing technologies lack data fusion and multi-dimensional analysis, making it difficult to comprehensively and accurately reflect the operating status of the transformer.

[0003] Existing transformer fault diagnosis technologies suffer from numerous shortcomings. First, a single monitoring method cannot fully reflect the complex fault conditions of a transformer. For example, the DGA method cannot directly detect winding overheating faults, while vibration signal detection struggles to distinguish fault types. Second, existing methods suffer from shortcomings in data fusion and fail to fully leverage the advantages of multi-source data. Furthermore, traditional diagnostic models place high demands on data quality and sample integrity, making them difficult to adapt to the fault diagnosis needs of complex operating environments.

[0004] In view of this, a transformer monitoring method based on the fusion of mechanism model and data model is needed. Summary of the Invention

[0005] In response to the problems in the existing technology that rely on a single monitoring method or a simple data processing method, there are problems such as a single monitoring dimension, insufficient data fusion, and poor adaptability to complex operating environments. The present invention provides a transformer monitoring method based on the fusion of a mechanism model and a data model. It can calculate the hot spot temperature of the winding through the mechanism model and extract the characteristics of the dissolved gas content in the oil in combination with the data model, thereby realizing the fusion of multi-dimensional information. This solution not only makes up for the shortcomings of a single monitoring method, but also improves the accuracy and real-time performance of fault diagnosis through data fusion, can more comprehensively reflect the operating status of the transformer, and effectively solve the limitations of the existing technology. The specific technical solution is as follows:

[0006] A transformer monitoring method based on the fusion of mechanism model and data model includes the following steps:

[0007] Establish a mechanism model: Establish a winding temperature distribution model based on the heat conduction equation, consider the Joule heat generated by the current as the heat source, solve the temperature distribution using the finite difference method, and determine the winding hotspot temperature;

[0008] Establish a data model: collect monitoring data on the dissolved gas content in transformer oil, perform data preprocessing, and extract statistical features, rate of change features, and cumulative features;

[0009] The winding hotspot temperature calculated by the mechanism model is fused with the gas content characteristics extracted by the data model to form a comprehensive feature vector. The fusion model is constructed using a machine learning algorithm, and the model is trained and optimized to achieve high-precision diagnosis of transformer faults.

[0010] Preferably, boundary conditions are further provided for the mechanism model. Specifically, the boundary conditions are as follows: assuming that both ends of the winding are in good contact with the cooling medium, a convection heat transfer boundary condition is adopted.

[0011] Preferably, for the mechanism model, an initial condition is also provided: assuming that the initial temperature of the winding is the ambient temperature T0.

[0012] Preferably, a correlation analysis method is used for the features extracted from the data model to screen gas content features with a high correlation with transformer faults. For example, the Pearson correlation coefficient between each gas content feature and the fault label is calculated, and features with an absolute value of the correlation coefficient greater than a certain threshold are selected as valid features.

[0013] Preferably, the transformer mechanism model is expressed as follows:

[0014] Assuming that the transformer winding is a one-dimensional uniform cylinder, its temperature distribution can be described by the heat conduction equation:

[0015]

[0016] Where T is temperature, t is time, α is thermal diffusivity, and Q(x,t) is the heat source term, which represents the amount of heat generated per unit volume per unit time. For transformer windings, the heat source mainly comes from the Joule heat generated by the current passing through the windings, and its expression is:

[0017] Q(x,t)=I 2 (t)R(x)

[0018] Where I(t) is the winding current and R(x) is the distribution of winding resistance along the length direction.

[0019] The finite difference method is used to discretize and solve the above heat conduction equation. The winding is divided into several small segments along the length direction, and the temperature of each small segment is calculated using the following iterative formula:

[0020]

[0021] Among them, T i n represents the temperature of the i-th segment at the n-th time step, Δt is the time step, Δx is the segment length, ρ is the winding material density, and c is the specific heat capacity.

[0022] Through iterative calculation, the temperature distribution of the winding at different time points can be obtained, and then the hot spot temperature position and temperature value of the winding can be determined.

[0023] Preferably, for the mechanism model, boundary conditions are also set, and the boundary conditions are specifically:

[0024]

[0025] Where k is the thermal conductivity of the winding material, h is the convection heat transfer coefficient, T 油 The temperature of the cooling oil.

[0026] Preferably, the monitoring data of the dissolved gas content in the transformer oil includes the content of hydrogen (H2), methane (CH4), ethane (C2H6), ethylene (C2H4), acetylene (C2H2), carbon monoxide (CO) and carbon dioxide (CO2).

[0027] Preferably, the comprehensive characteristic vector is obtained by calculating the winding hot spot temperature T based on the mechanism model. 热点 As one of the features, it is fused with the feature of dissolved gas content in oil extracted by the data model, and is expressed as follows:

[0028] X=[T 热点 , μ H ,σ H ,ΔCH4,C C2H2 ,…].

[0029] A computer-readable storage medium includes a stored program, wherein when the program is run, the device where the computer-readable storage medium is located is controlled to execute the transformer monitoring method based on the fusion of mechanism model and data model as described above.

[0030] A processor is used to run a program, wherein when the program is run, the transformer monitoring method based on the fusion of a mechanism model and a data model as described above is executed.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] The mechanism model established in this invention provides a physical context and explanatory basis for the fusion model, making fault diagnosis results more interpretable. When the fusion model diagnoses a fault, the mechanism model can be used to analyze the cause of the fault, such as increased winding hotspot temperature leading to insulation aging, which in turn causes an overheating fault. Furthermore, the introduction of the data model enhances the fusion model's adaptability to different operating environments and transformer types. The winding structures and operating conditions of different transformers may vary. The data model can learn from these differences, enabling the fusion model to better adapt to various situations and improving the generalization of fault diagnosis. Furthermore, the winding hotspot temperature reflects the thermal state of the transformer, while the dissolved gas content in the oil reflects the chemical state of the transformer. Fusion of these two dimensions of information allows for a more comprehensive assessment of the transformer's operating status. Increased winding hotspot temperature may cause changes in the dissolved gas content in the oil. By fusing these two features, the type and severity of the fault can be more accurately determined. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly describes the drawings required for the specific embodiments or the description of the prior art. Similar elements or parts are generally identified by similar reference numerals throughout the drawings. Elements or parts in the drawings are not necessarily drawn to scale.

[0034] Figure 1 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0035] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0036] It will be understood that when used in this specification and the appended claims, the terms “comprises” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0037] It should also be understood that the terms used in the present specification are only for the purpose of describing particular embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0038] It should be further understood that the term "and / or" used in the present description and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0039] In one embodiment of the present invention, a transformer monitoring method based on the fusion of mechanism model and data model is provided. Figure 1 As shown, the following steps are included:

[0040] Step 1: Construct the transformer mechanism model and data model;

[0041] Among them, the transformer mechanism model is expressed as follows:

[0042] Assuming that the transformer winding is a one-dimensional uniform cylinder, its temperature distribution can be described by the heat conduction equation:

[0043]

[0044] Where T is temperature, t is time, α is thermal diffusivity, and Q(x,t) is the heat source term, which represents the amount of heat generated per unit volume per unit time. For transformer windings, the heat source mainly comes from the Joule heat generated by the current passing through the windings, and its expression is:

[0045] Q(x,t)=I 2 (t)R(x)

[0046] Where I(t) is the winding current and R(x) is the distribution of the winding resistance along the length direction.

[0047] The finite difference method is used to discretize and solve the above heat conduction equation. The winding is divided into several small segments along the length direction, and the temperature of each small segment is calculated using the following iterative formula:

[0048]

[0049] Among them, T i n represents the temperature of the i-th segment at the n-th time step, Δt is the time step, Δx is the segment length, ρ is the winding material density, and c is the specific heat capacity.

[0050] Through iterative calculation, the temperature distribution of the winding at different time points can be obtained, and then the hot spot temperature position and temperature value of the winding can be determined.

[0051] Furthermore, boundary conditions are set:

[0052] Assuming that both ends of the winding are in good contact with the cooling medium (such as transformer oil), the convection heat transfer boundary condition is adopted:

[0053]

[0054] Where k is the thermal conductivity of the winding material, h is the convection heat transfer coefficient, T 油 The temperature of the cooling oil.

[0055] Furthermore, an initial condition is set: it is assumed that the initial temperature of the winding is the ambient temperature T0.

[0056] The mechanism model provides the physical context and explanatory basis for the fusion model, making the fault diagnosis results more interpretable. For example, when the fusion model diagnoses a fault, the mechanism model can be used to analyze the cause of the fault, such as the increase in the hot spot temperature of the winding leading to insulation aging and then an overheating fault.

[0057] The transformer data model is represented as follows:

[0058] S1: Collect monitoring data on the dissolved gas content in transformer oil, including the content of gases such as hydrogen (H2), methane (CH4), ethane (C2H6), ethylene (C2H4), acetylene (C2H2), carbon monoxide (CO) and carbon dioxide (CO2).

[0059] Furthermore, the data is cleaned to remove outliers and noise data. For example, gas content data that exceeds the normal range can be detected and removed through box plots.

[0060] S2: Feature extraction, including:

[0061] Calculate the statistical characteristics of gas content, such as mean, variance, standard deviation, skewness, kurtosis, etc. Taking hydrogen as an example, for the hydrogen content H, calculate its mean μ H and standard deviation σ H :

[0062]

[0063]

[0064] Among them, N is the number of data samples, H i is the hydrogen content of the i-th sample.

[0065] Furthermore, the change characteristics of the gas content are calculated. Taking methane as an example, for methane CH4, the change rate ΔCH4 is calculated:

[0066]

[0067] Among them, CH 4，当前 is the current methane content, CH 4，前一时刻 is the methane content at the previous moment.

[0068] Furthermore, the cumulative characteristics of the gas content are extracted. Taking acetylene as an example, the cumulative amount of acetylene content C2H2 is calculated. C2H2 :

[0069]

[0070] Where N is the number of data samples, C C2H2,i is the acetylene content of the i-th sample.

[0071] S3: Feature Selection: Use correlation analysis to select gas content features that are highly correlated with transformer faults. For example, calculate the Pearson correlation coefficient between each gas content feature and the fault label, and select features with an absolute value of the correlation coefficient greater than a certain threshold (such as 0.5) as valid features.

[0072] The introduction of the data model enhances the fusion model's adaptability to different operating environments and transformer types. For example, different transformers may have different winding structures and operating conditions. By learning these differences, the data model enables the fusion model to better adapt to various situations and improve the generalization ability of fault diagnosis.

[0073] Step 2: Fault diagnosis is performed by integrating the mechanism model and the data model. This includes the following steps:

[0074] Feature fusion: The winding hot spot temperature T calculated based on the mechanism model 热点 As one of the features, it is combined with the dissolved gas content features in oil extracted by the data model (including mean, standard deviation, rate of change, and cumulative amount) to form a comprehensive vector:

[0075] X=[T 热点 , μ H ,σ H ,ΔCH4,C C2H2 ,…]

[0076] Build the fusion model:

[0077] Select an appropriate machine learning algorithm (such as a support vector machine, random forest, or neural network) as the framework for the fusion model. Taking the support vector machine as an example, its basic principle is to classify sample data with different fault states by finding an optimal hyperplane. Normalize the fused feature vector X to meet the input requirements of the machine learning algorithm. Train the fusion model using sample data with labeled fault states, optimizing the model by adjusting model parameters (such as the support vector machine penalty parameter C and kernel function parameter γ).

[0078] To troubleshoot:

[0079] The comprehensive feature vector of the transformer to be diagnosed is input into the trained fusion model, and the model outputs the fault diagnosis results (such as normal, overheating fault, discharge fault, etc.).

[0080] In summary, the present invention accurately describes the changing patterns of transformer winding hotspot temperatures based on physical principles, reflecting the underlying causes of faults. For example, an increase in winding hotspot temperature may be due to faults such as winding insulation aging or local short circuits. The mechanism model can provide early warning of these potential faults. By analyzing and mining large amounts of monitoring data, it can discover implicit patterns and features within the data, demonstrating strong generalization and adaptability. For example, the changing characteristics of dissolved gas content in oil can reflect faults such as overheating and discharge within the transformer. The data model can automatically learn the relationship between these features and the fault. Finally, combining the mechanism model with the data model not only considers the physical nature of the fault but also fully utilizes the information in the data, thus overcoming the shortcomings of a single model. For example, relying solely on the mechanism model may lead to bias due to simplified model assumptions, while the data model may be limited by data quality and quantity. The fusion of the two models allows for mutual correction and complementation, improving the accuracy and reliability of fault diagnosis.

[0081] Furthermore, the winding hotspot temperature reflects the transformer's thermal state, while the dissolved gas content in the oil reflects the transformer's chemical state. Fusion of these two dimensions of information allows for a more comprehensive assessment of the transformer's operating status. For example, an increase in the winding hotspot temperature may cause a change in the dissolved gas content in the oil. By fusing these two features, the type and severity of the fault can be more accurately determined.

[0082] Those skilled in the art will appreciate that the units of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition of each example has been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0083] In the embodiments provided by the present invention, it should be understood that the division of units is merely a logical function division, and there may be other division methods in actual implementation, for example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.

[0084] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0085] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-0nly Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk, etc., various media that can store program code.

[0086] 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, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and description of the present invention.

Claims

1. A transformer monitoring method based on the fusion of mechanism model and data model, characterized in that: The steps include: Establish a mechanism model: Establish a winding temperature distribution model based on the heat conduction equation, consider the Joule heat generated by the current as the heat source, solve the temperature distribution using the finite difference method, and determine the winding hotspot temperature; Establish a data model: collect monitoring data on the dissolved gas content in transformer oil, perform data preprocessing, and extract statistical features, rate of change features, and cumulative features; The winding hotspot temperature calculated by the mechanism model is fused with the gas content characteristics extracted by the data model to form a comprehensive feature vector. The fusion model is constructed using a machine learning algorithm, and the model is trained and optimized to achieve high-precision diagnosis of transformer faults.

2. A transformer monitoring method based on the fusion of mechanism model and data model according to claim 1, characterized in that: Boundary conditions are also set for the mechanism model. The specific boundary conditions are: assuming that both ends of the winding are in good contact with the cooling medium, convection heat transfer boundary conditions are adopted.

3. A transformer monitoring method based on the fusion of mechanism model and data model according to claim 2, characterized in that: For the mechanism model, initial conditions are also set: it is assumed that the initial temperature of the winding is the ambient temperature T0.

4. The transformer monitoring method based on the fusion of mechanism model and data model according to claim 1 is characterized in that: Based on the features extracted from the data model, the correlation analysis method is used to screen the gas content features with high correlation with transformer faults.

5. The transformer monitoring method based on the fusion of mechanism model and data model according to claim 1 is characterized in that: The transformer mechanism model is expressed as follows: Assuming the transformer winding is a one-dimensional uniform cylinder, its temperature distribution is described by the following heat conduction equation: Where T is temperature, t is time, α is thermal diffusivity, and Q(x,t) is the heat source term, which represents the heat generated per unit volume per unit time. For transformer windings, the heat source comes from the Joule heating generated by the current passing through the windings, and its expression is: Q(x,t)=I 2 (t)R(x) Where I(t) is the winding current, and R(x) is the distribution of winding resistance along the length direction; The finite difference method is used to discretize and solve the above heat conduction equation. The winding is divided into several small segments along the length direction. The temperature of each small segment is calculated using the following iterative formula: Among them, T i n represents the temperature of the i-th segment at the n-th time step, Δt is the time step, Δx is the segment length, ρ is the winding material density, and c is the specific heat capacity; Through iterative calculation, the temperature distribution of the winding at different time points is obtained, and then the hot spot temperature position and temperature value of the winding are determined.

6. A transformer monitoring method based on the fusion of mechanism model and data model according to claim 5, characterized in that: For the mechanism model, boundary conditions are also set. The specific boundary conditions are: Where k is the thermal conductivity of the winding material, h is the convection heat transfer coefficient, T 油 The temperature of the cooling oil.

7. The transformer monitoring method based on the fusion of mechanism model and data model according to claim 1 is characterized in that: The monitoring data of dissolved gas content in transformer oil include the content of hydrogen, methane, ethane, ethylene, acetylene, carbon monoxide and carbon dioxide.

8. The transformer monitoring method based on the fusion of mechanism model and data model according to claim 1 is characterized in that: The comprehensive eigenvector is obtained by converting the winding hot spot temperature T calculated based on the mechanism model into 热点 As one of the features, it is fused with the feature of dissolved gas content in oil extracted by the data model, and is expressed as follows: X=[T 热点 ,m H ,s H ,ΔCH4,C C2H2 ,…].

9. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the transformer monitoring method based on the fusion of mechanism model and data model according to any one of claims 1 to 8.

10. A processor, characterized in that: The processor is used to run a program, wherein the program, when running, executes the transformer monitoring method based on the fusion of mechanism model and data model according to any one of claims 1 to 8.

Images