Transformer oil paper insulation moisture distribution test and balance curve correction method based on electric field and temperature coupling effect

Abstract

The invention discloses a transformer oil paper insulation moisture distribution test and balance curve correction method based on an electric field and temperature coupling effect, and aims at overcoming the defects of a traditional method in the aspects of temperature field calculation, moisture detection and insulation evaluation. The defects of a traditional temperature field, moisture detection and insulation evaluation are overcome, a simulation device and multiple sets of samples are constructed in experiment preparation, moisture testing is accurate through a specific instrument method, data collection and analysis are scientific, parameters are set, and algorithm architecture processing is applied. An electric field temperature coupling model is established and optimized according to a large amount of data, and moisture is evaluated and parameters can be adjusted according to operating parameters during application. The method can accurately calculate a temperature field, accurately detect moisture, effectively evaluate insulation, provide reliable guidance for operation and maintenance of a transformer and guarantee stability of an electric power system in multi-scene detection of urban center substations, wind power plant booster stations, factory distribution rooms and the like, and is high in practical value and wide in prospect.

Description

Technical Field

[0001] The present invention relates to the field of oil-paper insulation detection and analysis of power transformers, and in particular to a method for testing the moisture distribution of transformer oil-paper insulation based on the coupled action of electric field and temperature and correcting its equilibrium curve. Background Art

[0002] In the research and application of power transformers, the oil-paper insulation system plays an irreplaceable and crucial role in maintaining the stable and reliable operation of transformers. Among them, the factors of temperature and moisture have a significant impact on the performance of oil-paper insulation, which are the key points of focus in this field.

[0003] Regarding the calculation of the temperature field of transformers, a relatively mature method framework has been established for traditional calculation methods. When calculating the winding loss, multiple key parameters need to be comprehensively considered, such as the current intensity in the winding, the resistance value, the width of the wire, the operating frequency, the peak value of the magnetic flux density, and the current density of the wire. Based on this, the result is obtained through a specific calculation method. The core loss is usually approximately taken as the no-load loss value, which can be directly obtained from the nameplate parameters of the transformer. When calculating the temperature field subsequently, the basic laws such as mass conservation, momentum conservation, and energy conservation need to be followed, and the temperature field and the flow field are coupled and calculated to deduce the temperature distribution inside the transformer.

[0004] However, traditional calculation methods have revealed many obvious defects in actual application scenarios. When facing the complex environment in which transformers actually operate, their coping ability is rather insufficient: transformers are subjected to the combined action of multiple physical fields such as electromagnetic fields, temperature fields, and flow fields during actual operation. Although in the calculation of load loss, by ignoring the stray losses of structural components and the box body, the problems that may occur when the unit loss in the magnetic field calculation is loaded into the temperature field calculation are avoided, creating favorable conditions for the subsequent visualization evaluation of the degree of polymerization. However, this also sacrifices the accuracy of the temperature distribution calculation to a certain extent. Especially in the prediction of the temperature at local hot spots inside the transformer, due to the failure to fully consider the influence of stray losses, it is very likely that the prediction results will deviate greatly. Taking large transformers as an example, although the stray losses of structural components and the box body are at a relatively small level compared to the winding loss, their influence on temperature in local specific areas cannot be ignored, and traditional methods are difficult to accurately capture the subtle changes in temperature in this case.

[0005] In the development process of oil-paper insulation moisture detection technology, there are still many problems to be solved in the current technical status quo. The existing sensing technologies are still unable to achieve non-invasive and high-precision detection of the moisture content inside the solid insulation of transformers in operation. In practical engineering applications, the oil-paper moisture equilibrium curve is mostly used to indirectly estimate the moisture content. The early Oommen curve can only be used to describe the equilibrium relationship with the moisture in the paper when the moisture in the oil is within the range of 50 ppm. Although the Griffin curve and the MIT curve have extended the applicable range of the equilibrium curve to a certain extent, they all have a common limitation: that is, only the single factor of temperature is considered for the influence on moisture equilibrium.

[0006] In the actual operating environment of transformers, the electric field is an important factor that cannot be ignored. As a polar molecule, water will have a significant change in its diffusion and equilibrium processes under the action of the electric field. For example, the electric field will cause the polarization of water molecules, which will then lead to changes in their motion states, and ultimately the moisture content and distribution in the oil and paper will also change accordingly. However, the existing oil-paper moisture equilibrium curves do not fully consider the influence of this key factor of the electric field during the construction process, which seriously affects the accuracy when applying these curves in practice and cannot truly and accurately reflect the actual moisture distribution inside the transformer.

[0007] From the application of algorithms in the analysis of transformer insulation status, although some algorithms have been applied to this field currently, their performance is unsatisfactory when dealing with problems such as the moisture distribution of oil-paper insulation and the change of insulation field strength under the complex multi-physical field coupling effect inside the transformer. Facing the non-linear relationship data generated by the complex physical processes inside the transformer, the constructed models have significant deficiencies in terms of accuracy and generalization ability.

[0008] In the actual operation process of transformers, there are extremely complex non-linear correlation relationships among factors such as moisture, temperature, and electric field strength: the change of temperature will not only affect the diffusion rate of moisture, but also change the physical properties of oil-paper insulation and thus affect the distribution of the electric field; and the change of electric field strength will in turn act on the migration and equilibrium processes of moisture. Traditional algorithms often have difficulty fully exploring the deep information hidden in the data when dealing with these complex interrelationships and cannot accurately construct a dynamic relationship model among moisture, temperature, electric field strength, and insulation performance. This leads to the situation that it is easy to have misjudgments or inaccurate evaluation results when using these algorithms for insulation status assessment, and it is difficult to meet the actual needs of providing a reliable basis for transformer operation and maintenance work.

[0009] In the context of large-scale access of new energy power generation to the power grid, the operating conditions of the power grid have become increasingly complex and diverse. The intermittent and fluctuating characteristics of wind power generation and photovoltaic power generation have significantly increased the load change frequency of transformers, which undoubtedly poses a higher challenge to the insulation performance of transformers. Traditional transformer insulation detection and evaluation methods designed based on steady-state operating conditions are no longer able to meet the operating requirements of modern power grids.

[0010] In the continuous process of promoting the construction of smart grids, higher requirements are put forward for the intelligent operation and maintenance of transformers. It is necessary to be able to monitor the insulation status of transformers in real time and accurately, and give early warnings of potential fault risks in order to take effective maintenance measures in a timely manner. However, there is still a large gap in the existing technologies to achieve this goal.

[0011] Internationally, many research institutions and enterprises are also actively engaged in related research work. In some developed countries, power companies and research institutions have invested a large amount of resources in the research and development of transformer insulation technologies. For example, some power research institutions in the United States have carried out in-depth explorations on the multi-physical field coupling phenomena inside transformers with the help of advanced experimental equipment and simulation technologies, and are committed to building more accurate temperature field and moisture distribution models. Some European enterprises have focused on sensor technology innovation and actively developed new sensors to improve the monitoring accuracy of internal parameters of transformers. However, there is still room for further breakthroughs in comprehensively considering the coupling effects of temperature and electric field.

[0012] In China, with the rapid development of the power industry, the research on transformer insulation technologies has also been increasingly emphasized. Some universities and research institutes have carried out research on related topics and achieved certain phased results in theoretical analysis and experimental research. However, compared with the international advanced level, there is still a certain gap in the application and industrialization process of the technologies. Therefore, it is necessary to further strengthen the cooperation between industry, academia and research, accelerate the promotion and application of related technologies, and promote the improvement of domestic transformer insulation technology levels.

[0013] In the future, with the continuous innovation and development of related fields such as materials science, sensor technology, algorithms and artificial intelligence, new breakthrough opportunities are expected to be brought to the detection and analysis technologies of transformer oil-paper insulation. For example, new nanomaterials may be applied to the transformer oil-paper insulation system to effectively improve its insulation performance and moisture resistance; high-precision sensor technology is expected to achieve real-time and accurate monitoring of the moisture and temperature inside transformers; advanced artificial intelligence algorithms may more accurately predict the insulation status and fault risks of transformers, providing a more reliable guarantee for the safe and stable operation of the power system.

[0014] In summary, in the current field of transformer oil-paper insulation detection and analysis, there are many deficiencies in temperature field calculation, moisture detection technology, and algorithm application. Therefore, there is an urgent need to develop an innovative technical method that can comprehensively and accurately consider the coupling effect of temperature and electric field, achieve precise testing and in-depth analysis of the moisture distribution in oil-paper insulation, and effectively evaluate the insulation safety margin. By applying this method, the gaps in the existing technology can be filled, the scientificity and effectiveness of transformer operation and maintenance work can be improved, and thus a solid guarantee can be provided for the safe and stable operation of the power system. Summary of the Invention

[0015] The technical problem to be solved by the present invention is a method for testing the moisture distribution in transformer oil-paper insulation based on the coupling effect of electric field and temperature and correcting its equilibrium curve. This method solves a series of key problems in the research of transformer oil-paper insulation through multiple steps. Due to the simplified treatment of the stray losses of structural components and the box body in the traditional method, the temperature field calculation is inaccurate. The existing sensing technology cannot non-invasively and accurately sense the moisture content of solid insulation, and the oil-paper moisture equilibrium curve does not consider the influence of the electric field, resulting in incorrect evaluation of the moisture distribution. In addition, due to the lack of accurate understanding of the moisture distribution under the coupling of temperature and electric field, it is difficult to accurately analyze the dynamic changes of the insulation field strength and safety margin.

[0016] To solve the above technical problems, the technical solutions adopted by the present invention are as follows:

[0017] A method for testing the moisture distribution in transformer oil-paper insulation based on the coupling effect of electric field and temperature and correcting its equilibrium curve, characterized in that the testing of the moisture distribution in transformer oil-paper insulation includes the following steps: A1 Experimental preparation, A2 Moisture content testing, A3 Data collection and analysis.

[0018] As a preferred technical solution of the present invention, the specific content of A1 Experimental preparation is:

[0019] A1-1 Construct a test device for simulating the internal environment of the transformer, which is equipped with facilities for precisely controlling the temperature and applying electric fields of different intensities. The temperature control range covers the temperature range that actually appears during the operation of the transformer, and the adjustment accuracy of the electric field intensity reaches the level of effectively distinguishing the influence of different electric fields on the moisture distribution and can stably maintain the set temperature and electric field conditions to ensure the accuracy and repeatability of the experiment;

[0020] A1-2 Prepare multiple groups of oil-paper insulation samples. The initial moisture content of each group of samples has a certain gradient distribution and covers the moisture range specifically contained in the transformer oil-paper insulation. At the same time, the other physical and chemical properties of each group of samples are the same to exclude the interference of other factors on the test results of the moisture distribution.

[0021] Furthermore, the test device constructed by A1-1 for simulating the internal environment of the transformer is specifically as follows:

[0022] For temperature control, a platinum resistance temperature sensor with a precision of not less than ±1°C or other similar sensors is used, and a heating / cooling system composed of an electric heating wire and a circulating cooling water pipe based on PID control or other similar control methods is adopted. The measurement precision of the temperature sensor is not less than ±X°C, where X is a smaller value determined according to the actual operating temperature precision of the transformer. The heating / cooling system responds quickly and stably maintains the set temperature, controlling the temperature fluctuation during the test within a very small range, so that the influence of temperature factors on the moisture distribution can be accurately measured;

[0023] For electric field application, a high-stability DC power supply and an electrode device with precisely adjustable spacing are used. The output voltage stability of the DC power supply is not less than ±Y%, where Y is a smaller value determined according to the electric field strength precision. The spacing adjustment precision of the electrode device is not less than ±Z mm, where Z is a smaller value precisely set according to different electric field strengths, so as to make the electric field strength accurate and stable for studying the law of moisture distribution under different electric field strengths.

[0024] As a preferred technical solution of the present invention, the A2 moisture content test is specifically as follows:

[0025] For the moisture test in oil in A2-1, a water analyzer with high sensitivity and accuracy or other similar devices is used, and its measurement error does not exceed ±Appm, where A is a smaller value determined according to the precision requirement of the moisture content in oil, and the measurement range covers the moisture content range that appears in transformer oil, so as to accurately measure the change of moisture in oil under different conditions;

[0026] For the moisture test of cardboard in A2-2, a method combining a moisture analyzer and spectroscopy is adopted. The titration precision of the moisture analyzer reaches ±B μg, where B is a smaller value determined according to the precision requirement of the moisture content in cardboard. The spectroscopic device performs high-resolution scanning on the cardboard, and the scanning point spacing does not exceed C mm, where C is a smaller value of the moisture imaging precision. Through the combination of the two, the accurate determination of the moisture content of the cardboard and the visualization of the moisture distribution are realized, improving the comprehensiveness and reliability of the test.

[0027] Furthermore, when using the spectroscopic device to test the moisture of cardboard:

[0028] Special sample treatment and test environment settings are adopted, and sample cell materials with extremely small influence on the spectral absorption of the spectroscopic device and stable chemical properties are used, including but not limited to specially treated quartz and specific polymer materials, so that the spectral signal is not disturbed during the test, improving the measurement accuracy;

[0029] An accurate calibration model between the spectral signal and the moisture content of the cardboard is established. The goodness of fit of this calibration model is not less than D%, where D is a relatively high value determined according to the actual accuracy. The model is optimized through a large number of tests on standard samples and data analysis, so that the moisture content of the cardboard can be accurately retrieved based on the spectral signal.

[0030] As a preferred technical solution of the present invention, the A3 data acquisition and analysis are specifically as follows: When obtaining the oil-paper moisture equilibrium point data under different conditions: Set the test time period and data acquisition frequency according to the actual situation. The test time length should enable the moisture to reach a stable equilibrium state, and the data acquisition frequency density should include the intervals that capture the key nodes of the moisture content change, especially in stages including but not limited to the initial, middle, and near-equilibrium stages of moisture change for intensive acquisition, so that the obtained equilibrium point data is accurate and reliable;

[0031] Adopt methods including but not limited to multiple parallel tests and data statistical analysis. The number of each group of parallel tests is not less than E groups, where E is a specific value determined according to statistical requirements. Conduct statistical analysis on the data of multiple groups of tests, including but not limited to calculating statistical parameters such as the mean value and standard deviation to reduce test errors and uncertainties and improve the credibility of the equilibrium point data;

[0032] Use algorithms including but not limited to multiple linear regression to conduct modeling analysis on multiple factors including but not limited to temperature, electric field strength, and initial moisture content that affect the oil-paper moisture balance, determine the quantitative relationship between each factor and the moisture equilibrium point, and the determination coefficient R of the regression model 2 is not less than H%, where H is a relatively high value determined according to the actual data fitting accuracy requirements to improve the accuracy and scientificity of the prediction of the moisture equilibrium point;

[0033] Using a deep neural network architecture with powerful learning capabilities to deeply mine and extract features from spectral images and related test data, specifically: This architecture consists of a series of specially designed layers. Some layers are responsible for extracting key features from the original data, some layers are used to streamline the data scale, and the remaining layers are tasked with integrating information and making judgments; Through such an architecture, it automatically captures unique data feature patterns under different combinations of moisture content, electric field strength, and temperature, thereby significantly enhancing the accuracy of discriminating and classifying the moisture distribution state of oil-paper insulation; When training this neural network model, its accuracy must reach more than I%, where I is determined according to the actual application; Before actually applying this technology, preprocess the data: Use a standardized method to normalize the moisture content in oil and the moisture content in cardboard data to the range of 0 to 1, and perform transformation operations on the spectral image including but not limited to rotation, mirroring, and scaling to expand the data volume, so that the number of data samples after these processes increases by at least W%, where W is a proportion determined according to the actual situation to improve the training efficiency of this neural network model and its adaptability to different situations. As a preferred technical solution of the present invention, the specific steps of S6 spatio-temporal collaborative regulation are as follows: On the time scale, based on the power generation and consumption characteristics and response speeds of different controllable resources, reasonably regulate resources on the day-ahead, intra-day, and real-time time scales to maximize benefits, including but not limited to energy storage and hydrogen production during peak new energy generation periods, and charging energy storage using large grid power during low load periods. During the regulation process, use the same algorithm as in the S4 regulation strategy formulation step to dynamically adjust according to the time scale and resource status; On the spatial dimension, follow the principles of in-situ, nearby, same voltage level balance and consumption to achieve hierarchical and zonal coordination of resources, including hierarchical aggregation of adjustable capabilities, hierarchical decomposition of regulation targets, and autonomous and mutual assistance within the same layer among the high-voltage transmission layer, medium-voltage distribution network layer, and low-voltage distribution area layer.

[0034] As a preferred technical solution of the present invention, the specific method for correcting the balance curve is as follows:

[0035] Based on a large amount of experimental data and in-depth theoretical analysis, establish a mathematical model considering the coupled effect of electric field and temperature. This model accurately describes the diffusion, migration, and equilibrium processes of moisture in oil-paper insulation. The parameters in the model are determined by fitting and optimizing experimental data and the model has been strictly verified and tested to accurately reflect the actual situation;

[0036] Use the finite element analysis algorithm to simulate and verify the corrected balance curve, compare the simulation results with the actual experimental data, and control the error within ±F%, where F is a relatively small value determined according to the accuracy. Continuously optimize the model and curve to accurately predict the moisture distribution of oil-paper insulation under different electric field and temperature conditions.

[0037] Develop an algorithm to improve the performance of the previously constructed mathematical model. Specifically, this algorithm works as follows: First, randomly scatter many "explorers" in this space. These "explorers" simulate a group of small bugs groping in the dark. Each "bug" carries its own position information and moving speed. They will continuously move and change their positions in this space. During their movement, they will find a set of optimal "password combinations". This set of "password combinations" is the key parameter that makes the model fit the actual experimental data most perfectly. There is a target accuracy for this searching process, not less than J%, where J is a relatively high standard determined according to actual requirements. During the entire exploration process, according to the characteristics presented by the data and the state during model training, automatically adjust some important "exploration rules", including but not limited to the influence on the inertia of the "bugs" when moving and their speed of learning new positions. Moreover, record in detail all the relevant information when the "bugs" change their positions during the exploration process to form a record log. This log clearly records when the adjustment was made, the position information before and after the adjustment, and how these adjustments affected the model performance. And the recording form of this log is convenient for viewing and analysis at any time.

[0038] As a preferred technical solution of the present invention, when applying this method to evaluate the actual moisture distribution of transformer oil-paper insulation: According to the actual operating parameters of the transformer, including but not limited to the load rate, operating time, and oil temperature, quickly and accurately determine the temperature field and electric field distribution inside the transformer. By matching with the established database or model, predict the moisture distribution state of the oil-paper insulation, providing timely and effective guidance for the operation and maintenance of the transformer.

[0039] Make adaptive adjustments for transformers of different types and specifications. By analyzing the structure and material characteristics of the transformer, adjust the relevant parameters in the model to make the method generally and accurately applicable to the moisture distribution evaluation and insulation state monitoring of various actual transformers.

[0040] As a preferred technical solution of the present invention, in terms of data management and storage:

[0041] Establish a perfect data management system to classify, store, and manage all the data generated during the test process, including but not limited to sample information, test conditions, test data, and analysis results. The data storage format conforms to international common standards, facilitating data query, sharing, and further analysis.

[0042] Adopt data encryption and backup technologies to keep the data secure and complete, preventing data loss or illegal tampering. The data backup frequency is not less than G times per month, where G is a value determined according to the importance of the data to cope with possible hardware failures and other unexpected situations.

[0043] As a preferred technical solution of the present invention, in terms of the operation specifications and standard formulation of the method:

[0044] Formulate a detailed operation manual and standard operating procedures, clarify the operation requirements, precautions and quality control key points of each step. The operators undergo professional training and strictly conduct tests and data analysis in accordance with the operating procedures to ensure the implementation quality of the method and the reliability of the results.

[0045] The beneficial effects produced by adopting the above technical solutions are as follows: In the calculation of the transformer temperature field, it breaks through the limitations of the traditional treatment of stray losses, comprehensively considers various factors, and uses advanced model algorithms to achieve high-precision calculation, accurately determine the hot spot and temperature trend, and provide a key decision-making basis for operation and maintenance. Based on this, the operation and maintenance personnel can take effective measures such as optimizing heat dissipation and adjusting the load in advance to prevent overheating faults, enhance power supply stability, and extend the life of the transformer. In the detection of oil-paper insulation moisture and the correction of the equilibrium curve, it overcomes the existing sensing technology problems. When correcting the equilibrium curve, it fully considers the coupling of electric field and temperature, accurately evaluates the moisture distribution, and timely discovers abnormal accumulation sites. The power personnel can quickly implement drying maintenance and other means to prevent the decline of insulation performance, reduce the risks of short circuit and breakdown, and improve operation reliability. In the evaluation of insulation field strength and safety margin, with the help of advanced mathematical models and optimization algorithms, it accurately analyzes the changes in insulation characteristics under different working conditions, provides scientific decision-making support for operation and maintenance management. The operation and maintenance personnel can formulate a reasonable maintenance plan based on this, accurately replace the problem components, ensure the stable operation of the transformer, greatly reduce the operation and maintenance costs, and improve the economic benefits and operation efficiency of the power system. At the same time, the perfect data management and operation specifications of this technology ensure the security, integrity and traceability of data, improve the consistency and reliability of technology implementation, promote the improvement of industry technology, and lay a solid foundation for the safe and stable operation of the power system. Brief Description of the Drawings

[0046] Figure 1 : Temperature field distribution automatically calculated considering the actual operating load;

[0047] Figure 2 : Pressure test system;

[0048] Figure 3 : Calibration of terahertz spectroscopy and Karl Fischer;

[0049] Figure 4 : Schematic diagram of the moisture matrix for spectral scanning;

[0050] Figure 5 : Moisture distribution of samples under 5 kV / mm pressure and without pressure at different times;

[0051] Figure 6 : Changes in moisture in paper under different electric field strengths;

[0052] Figure 7 :Variation of moisture content in oil under different electric field strengths. Detailed implementation manners

[0053] Now, the principles of the present disclosure will be described with reference to several exemplary embodiments shown in the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the description of these embodiments is only for facilitating those skilled in the art to better understand and thereby implement the present disclosure, rather than limiting the scope of the present disclosure in any way.

[0054] In the description of the following embodiments, specific details such as specific system structures and technologies are presented for illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, the detailed descriptions of well-known systems, devices, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0055] It should be understood that when used in the specification of the present application and the appended claims, the term "comprising" indicates the presence of the described features, wholes, steps, operations, components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, components.

[0056] Embodiment 1: Overview of overall steps

[0057] Taking a transformer in a certain substation as an example, an operation overview of the measurement of the moisture distribution in oil-paper insulation and the correction of the equilibrium curve is given.

[0058] Refer to Attachment Figure 1 and 2 , in the A1 experimental preparation stage, a test device for simulating the internal environment of the transformer is built. For temperature control, a platinum resistance temperature sensor with an accuracy of not less than ±1°C and a thermal / cooling system composed of an electric heating wire and a circulating cooling water pipeline based on PID control are selected. The measurement accuracy of the sensor meets the accuracy requirements of the actual operating temperature of the transformer. The thermal / cooling system can respond quickly and stably maintain the set temperature, with extremely small temperature fluctuations, ensuring that the influence of temperature on the moisture distribution can be accurately measured. The temperature control range covers the actual operating temperature range of the transformer. For electric field application, a high-stability DC power supply and an electrode device with accurately adjustable spacing are used. The output voltage stability of the DC power supply is not less than ±Y%, and the spacing adjustment accuracy of the electrode device is not less than ±Z mm, ensuring accurate and stable electric field strength and facilitating the study of the moisture distribution law under different electric field strengths. At the same time, multiple groups of oil-paper insulation samples are prepared, with their initial moisture contents distributed in a gradient and covering the common moisture range of transformer oil-paper insulation. The other physical and chemical properties of each group of samples are the same to avoid interfering with the test results.

[0059] Refer to AttachmentFigure 3 and 5 , enter the A2 water content test session. For the water content test in oil, a water analyzer with high sensitivity and accuracy is used, and the measurement error does not exceed ±A ppm. The measurement range covers the water content range in transformer oil, and it can accurately measure the change of water content in oil. For the cardboard water content test, a method combining a water analyzer and spectroscopy is adopted. The titration accuracy of the water analyzer reaches ±B ug, and the spectroscopy equipment scans the cardboard with high resolution, and the scanning point spacing does not exceed C mm. It can not only accurately measure the cardboard water content but also visualize the water distribution, improving the test reliability. When using the spectroscopy equipment to test the cardboard water content, sample cell materials such as specially treated quartz or specific polymers with stable chemical properties and minimal influence on spectral absorption are used to ensure that the spectral signal is not interfered during the test process. Also, an accurate calibration model of the spectral signal and the cardboard water content is established through a large number of standard sample tests and data analysis, and the goodness of fit is not less than D%, ensuring that the cardboard water content can be accurately retrieved based on the spectral signal.

[0060] Refer to Appendix Figure 4 and 6 -7. In the A3 data acquisition and analysis process, set the test time period and data acquisition frequency according to the actual situation to ensure that the test time is sufficient for the water to reach a stable equilibrium state. The data acquisition frequency is densely collected at key stages such as the initial, middle, and near-equilibrium stages of water content change to obtain accurate and reliable oil-paper water equilibrium point data. Adopt multiple sets of parallel tests and data statistical analysis methods. The number of each set of parallel tests is not less than [E] groups. By calculating statistical parameters such as the average value and standard deviation, reduce the test error and uncertainty, and improve the credibility of the equilibrium point data. Use multiple linear regression algorithms, etc. to model and analyze the factors affecting the oil-paper water equilibrium, such as temperature, electric field strength, and initial water content, and establish a regression model with a determination coefficient R 2 not less than H%, determine the quantitative relationship between each factor and the water equilibrium point, and improve the accuracy and scientificity of predicting the water equilibrium point. Use a deep neural network architecture to deeply mine and extract features from terahertz spectral images and related test data. This architecture consists of specially designed layers, and different layers are responsible for extracting key features, reducing the data scale, and integrating information for judgment respectively. Through training, it can automatically capture unique data feature patterns under different combinations of water content, electric field strength, and temperature, enhancing the discrimination and classification accuracy of the oil-paper insulation water distribution state. The accuracy rate is required to reach more than I% during training, and the data is standardized before actual application. The data of the water content in oil and the cardboard water content are normalized to the range of 0 to 1, and operations such as rotation, mirroring, and scaling are performed on the spectral images to expand the data volume, so that the number of data samples increases by at least W%, improving the training efficiency and adaptability of the neural network model.

[0061] Next, the equilibrium curve is corrected. Based on a large amount of experimental data and in-depth theoretical analysis, a mathematical model considering the coupled effect of electric field and temperature is established. This model accurately describes the diffusion, migration, and equilibrium process of moisture in oil-paper insulation. The model parameters are determined by fitting and optimizing experimental data, and are strictly verified and tested to ensure that they can reflect the actual situation. The finite element analysis algorithm is used to simulate and verify the corrected equilibrium curve. By comparing the simulation results with the actual experimental data, the error is controlled within ±F%. The model and curve are continuously optimized to accurately predict the moisture distribution in oil-paper insulation under different electric field and temperature conditions. An algorithm is also developed to place elements similar to "explorers" in a space. They carry position information and moving speeds and continuously move and change positions to find the best "password combination" (key parameters) that makes the model fit the actual experimental data. The target accuracy of this search process is not less than J%, and important rules such as the moving inertia of the "explorers" and the speed of learning new positions are automatically adjusted according to the data characteristics and the training status of the model. At the same time, the relevant information of each position change is detailedly recorded to form a record log for convenient viewing and analysis, including the adjustment time, the position information before and after adjustment, and the impact on the model effect, etc.

[0062] When applied to the assessment of the moisture distribution in the oil-paper insulation of actual transformers, based on the actual operating parameters of the transformers such as load factor, operating time, and oil temperature, the internal temperature field and electric field distribution are quickly and accurately determined, matched with the established database or model, and the moisture distribution state of the oil-paper insulation is predicted to provide guidance for the operation and maintenance of transformers. For transformers of different types and specifications, the relevant parameters of the model are adjusted by analyzing their structural and material characteristics, so that the method can be generally and accurately applied to the moisture distribution assessment and insulation status monitoring of various transformers.

[0063] In terms of data management and storage, a perfect data management system is established to classify and store the sample information, test conditions, test data, analysis results, etc. generated by the tests. The data storage format conforms to international general standards, which is convenient for querying, sharing, and further analysis. Data encryption and backup technologies are adopted to ensure the security and integrity of the data, prevent loss or illegal tampering. The data backup frequency is not less than G times per month to cope with hardware failures and unexpected situations.

[0064] In terms of formulating the operation specifications and standards of the method, detailed operation manuals and standard operating procedures are formulated to clarify the operation requirements, precautions, and quality control key points of each step. After professional training, the operators strictly conduct tests and data analysis according to the procedures to ensure the quality of method implementation and the reliability of results.

[0065] Example 2: Insulation Detection of the Main Transformer in the Central Substation of the City

[0066] In an important substation in the center of a city, there is a main transformer with a capacity of 500 MVA.

[0067] In the experimental preparation stage, the temperature control of the constructed test device adopts a platinum resistance temperature sensor with an accuracy of ±0.5°C, combined with an advanced PID-controlled electric heating wire and a circulating cooling water pipeline system. The measurement accuracy of the temperature sensor can reach ±0.3°C, which can ensure that the temperature fluctuation is within a very small range, covering the temperature range of -20°C to 100°C for the actual operation of the transformer. For the electric field application part, a high-stability DC power supply with an output voltage stability of ±0.5% and an electrode device with a spacing adjustment accuracy of ±0.5 mm are used. Five groups of oil-paper insulation samples are prepared, and their initial moisture contents are 0.5%, 1%, 1.5%, 2%, and 2.5% respectively, covering the common moisture range of the oil-paper insulation of this transformer, and the other physical and chemical properties of each group of samples are the same.

[0068] When testing the moisture content, for the moisture in oil, a coulomb micro water analyzer with a measurement error not exceeding ±2 ppm is selected, and its measurement range is 1 μg to 80 mg. For the moisture in cardboard, a Karl Fischer moisture analyzer with a titration accuracy of ±5 μg is combined with terahertz spectroscopy. The scanning point spacing of the spectral equipment for the cardboard is 2 mm. A specially treated quartz sample cell is used for testing, and the goodness of fit of the calibration model between the spectral signal and the cardboard moisture content established through a large number of standard sample tests and data analysis reaches 99.3%.

[0069] During the data acquisition and analysis process, the test time period is set to 300 hours. Data is collected every 5 hours in the initial stage of moisture change (0 - 50 hours), every 10 hours in the middle stage (50 - 200 hours), and every 20 hours near the equilibrium stage (200 - 300 hours). Eight groups of parallel tests are carried out, and errors are reduced by calculating statistical parameters such as the average value and standard deviation. The determination coefficient R of the regression model established by using the multiple linear regression algorithm 2 reaches 0.92. A deep neural network architecture is used to process the terahertz spectral images and test data. The accuracy during training reaches more than 90%, and through data standardization and spectral image transformation operations, the number of data samples is increased by 30%.

[0070] Regarding the balance curve correction, a mathematical model considering the coupled action of the electric field and temperature is established based on a large amount of test data. After simulation and verification by the finite element analysis algorithm, the error between the simulation results and the actual test data is controlled within ±3%. During the process of developing the algorithm to find the optimal parameters, the target accuracy is set to 95%. By continuously adjusting rules such as the moving inertia and learning speed of the "explorer", the model and curve are successfully optimized, and the moisture distribution of the oil-paper insulation under different electric field and temperature conditions is accurately predicted.

[0071] When applied to this transformer, based on parameters such as the real-time monitored load rate of 80%, operating time of 5 years, and oil temperature of 70°C, the internal temperature field and electric field distribution are quickly determined. After matching with the established database, the moisture distribution state of the oil-paper insulation is predicted. It is found that the moisture content in local areas is on the high side, and suggestions for replacing insulation components are provided to the operation and maintenance personnel in a timely manner, ensuring the stable power supply of the substation.

[0072] Example 3: Insulation Assessment of a Step-up Transformer in a Wind Farm

[0073] An insulation condition assessment is required for a 200MVA step-up transformer in a certain wind farm.

[0074] In the experimental preparation, a platinum resistance temperature sensor with an accuracy of ±0.8°C and a thermal / cooling system based on PID control are selected for temperature control of the test device. The measurement accuracy of the temperature sensor is ±0.4°C, and the temperature control range is from -10°C to 90°C, which can quickly respond and stably maintain the set temperature. A DC power supply with an output voltage stability of ±0.8% and an electrode device with precisely adjustable spacing are used for electric field application. The adjustment accuracy of the electrode spacing is ±0.8mm. Four groups of oil-paper insulation samples are prepared, with initial moisture contents of 0.8%, 1.2%, 1.6%, and 2% respectively.

[0075] In the moisture content test section, a water analyzer with a measurement error not exceeding ±3ppm is used for testing the moisture in oil, and the measurement range is 2μg to 90mg. For testing the moisture in cardboard, a moisture analyzer with a titration accuracy of ±6μg is used in combination with terahertz spectroscopy, and the spacing between spectral scanning points does not exceed 3mm. A specific polymer sample cell is used for testing, and the goodness of fit of the calibration model established between the spectral signal and the cardboard moisture content reaches 99%.

[0076] When collecting and analyzing data, the test time is set to 250 hours. The data collection frequency is once every 4 hours in the initial stage of moisture change (0 - 40 hours), once every 8 hours in the middle stage (40 - 180 hours), and once every 16 hours close to the equilibrium stage (180 - 250 hours). Six groups of parallel tests are carried out to improve the credibility of the equilibrium point data through statistical analysis. The coefficient of determination R of the model established using the multiple linear regression algorithm 2 is 0.9. The deep neural network architecture is used to mine and extract features from the data, and the training accuracy requirement is above 88%. The data sample quantity is increased by 25% through pre-processing of the data.

[0077] In the balance curve correction, the established mathematical model accurately describes the process of moisture in the oil-paper insulation. After verification by the finite element analysis algorithm, the error is controlled within ±4%. The target accuracy of the developed algorithm is 93%. By recording the position information of the "Explorer" and adjusting the log, the model is effectively optimized to accurately predict the moisture distribution.

[0078] Based on the actual operating parameters of the transformer, such as a load factor of 60%, an operating time of 3 years, and an oil temperature of 65°C, the internal temperature field and electric field distribution are determined. After matching with the model, the moisture distribution state is predicted. It is found that the moisture in some parts of the insulating paper is abnormal. Based on this, the operation and maintenance personnel adjust the operation mode and strengthen the monitoring, ensuring the reliable operation of the wind farm.

[0079] Example 4: Insulation Monitoring of Transformers in Factory Power Distribution Rooms

[0080] A 100 MVA transformer in the factory power distribution room is taken as the research object.

[0081] In terms of experimental preparation, the temperature control of the test device uses a platinum resistance temperature sensor with an accuracy of ±1°C and a combined system of high-efficiency electric heating wires and circulating cooling water pipelines. The measurement accuracy of the temperature sensor reaches ±0.6°C, and the temperature control range covers -5°C to 85°C. The electric field is applied using a DC power supply with an output voltage stability of ±1% and an electrode device with an electrode spacing adjustment accuracy of ±1 mm. Six groups of oil-paper insulation samples are prepared, with initial moisture contents of 0.6%, 0.9%, 1.2%, 1.5%, 1.8%, and 2.1% respectively.

[0082] When testing the moisture content, a high-sensitivity water analyzer with a measurement error not exceeding ±4 ppm is selected for testing the moisture in the oil, and the measurement range is 3 μg to 70 mg. The moisture content of the cardboard is tested by combining a moisture analyzer with a titration accuracy of ±7 μg and terahertz spectroscopy, and the spectral scanning point spacing is 2.5 mm. A specially treated quartz sample cell is used to ensure the test accuracy, and the goodness of fit of the calibration model established between the spectral signal and the cardboard moisture content is 98.8%.

[0083] During the data acquisition and analysis process, the test time is set to 280 hours. The data acquisition frequency is once every 6 hours in the initial stage of moisture change (0 - 60 hours), once every 12 hours in the middle stage (60 - 220 hours), and once every 24 hours close to the equilibrium stage (220 - 280 hours). Seven groups of parallel tests are carried out, and the test error is reduced by calculating statistical parameters. The coefficient of determination R of the regression model established using the multiple linear regression algorithm 2 reaches 0.88. The data is processed using a deep neural network architecture. The accuracy during training needs to reach more than 85%, and the data sample size is increased by 35% through pre-processing of the data.

[0084] When correcting the equilibrium curve, the mathematical model established based on the test data is strictly verified, and the error controlled by simulation and verification using the finite element analysis algorithm is within ±5%. The target accuracy of the developed algorithm is set to 90%. By continuously adjusting and recording, the model and curve are optimized to achieve accurate prediction of the moisture distribution under different electric fields and temperatures.

[0085] Combined with the operating parameters such as the load factor of 70% of the transformer, the operating time of 4 years, and the oil temperature of 60°C, the internal temperature field and electric field distribution are determined. After matching with the database or model, the moisture distribution state of the oil-paper insulation is predicted, providing timely and effective guidance for the operation and maintenance of the transformer and ensuring the stable power supply for factory production.

[0086] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent substitutions or changes, shall be covered by the protection scope of the present invention.

Claims

1. A method for testing the moisture distribution of transformer oil-paper insulation based on the coupled action of electric field and temperature and correcting its equilibrium curve, characterized in that The test of the moisture distribution in the transformer oil-paper insulation includes the following steps: A1 Experimental preparation, A2 Moisture content test, and A3 Data collection and analysis.

2. A method for testing the moisture distribution in transformer oil-paper insulation based on the coupled action of electric field and temperature and correcting its equilibrium curve, characterized in that, The specific content of the A1 experimental preparation is as follows: A1-1 Construct a test device for simulating the internal environment of the transformer. This device is equipped with facilities for precisely controlling the temperature and applying electric fields of different intensities. The temperature control range covers the temperature range that actually appears during the operation of the transformer. The adjustment accuracy of the electric field intensity can effectively distinguish the influence of different electric fields on the moisture distribution and can stably maintain the set temperature and electric field conditions to ensure the accuracy and repeatability of the test. A1-2 Prepare multiple groups of oil-paper insulation samples. The initial moisture content of each group of samples has a certain gradient distribution and covers the moisture range specifically contained in the transformer oil-paper insulation. At the same time, the other physical and chemical properties of each group of samples are the same to exclude the interference of other factors on the test results of the moisture distribution.

3. A method for testing the moisture distribution and correcting the equilibrium curve of transformer oil-paper insulation based on the coupled action of electric field and temperature, characterized in that, The test device constructed in A1-1 for simulating the internal environment of the transformer is specifically as follows: For temperature control, it adopts, including but not limited to, a platinum resistance temperature sensor with an accuracy of not less than ±1°C and a heat / cooling system, including but not limited to a combination of an electric heating wire and a circulating cooling water pipeline based on PID control. The measurement accuracy of the temperature sensor is not less than ±X°C, where X is a value determined according to the temperature accuracy of the actual transformer operation. The heat / cooling system responds quickly and stably maintains the set temperature, so that the temperature fluctuation during the test process is controlled within a very small range, and the influence of temperature factors on the moisture distribution can be accurately measured. For electric field application, use a high-stability DC power supply and an electrode device with precisely adjustable spacing. The output voltage stability of the DC power supply is not less than ±Y%, where Y is a value determined according to the electric field intensity accuracy. The spacing adjustment accuracy of the electrode device is not less than ±Z mm, where Z is a smaller value precisely set according to different electric field intensities, so that the electric field intensity is accurate and stable to study the law of moisture distribution under different electric field intensities.

4. A method for testing the moisture distribution of transformer oil-paper insulation based on the coupling effect of electric field and temperature and correcting its equilibrium curve, characterized in that, The specific content of the A2 moisture content test is as follows: A2-1 For the test of moisture in oil, use, including but not limited to, a water analyzer with high sensitivity and accuracy. Its measurement error does not exceed ±A ppm, where A is a value determined according to the accuracy requirement of the moisture content in the oil, and the measurement range covers the moisture content range that appears in the transformer oil, so as to accurately measure the change of moisture in the oil under different conditions. A2-2 For the test of moisture in cardboard, adopt a method combining a moisture analyzer and a spectrum. The titration accuracy of the moisture analyzer reaches ±B ug, where B is a value determined according to the accuracy requirement of the moisture content in the cardboard. The spectrum device performs a high-resolution scan on the cardboard, and the scan point spacing does not exceed C mm, where C is the value of the moisture imaging accuracy. By combining the two, the accurate determination of the moisture content in the cardboard and the visualization of the moisture distribution are realized, improving the comprehensiveness and reliability of the test.

5. A method for testing the moisture distribution of transformer oil-paper insulation based on the coupling effect of electric field and temperature and correcting its equilibrium curve, characterized in that, When using the spectrum device to test the moisture in the cardboard: By adopting special sample treatment and test environment settings, and using sample cell materials that have minimal impact on the spectral absorption of spectral devices and are chemically stable, including but not limited to specially treated quartz and specific polymer materials, the spectral signal during the test is not disturbed, improving the measurement accuracy. An accurate calibration model between the spectral signal and the moisture content of the cardboard is established. The goodness of fit of this calibration model is not less than D%, where D is a value determined according to the actual accuracy. The model is optimized through a large number of standard sample tests and data analysis, enabling the accurate inversion of the cardboard moisture content based on the spectral signal.

6. A method for testing the moisture distribution of transformer oil-paper insulation based on the coupled action of electric field and temperature and correcting its equilibrium curve, characterized in that, The specific A3 data acquisition and analysis are as follows: When obtaining the moisture equilibrium point data of oiled paper under different conditions: Set the test time period and data acquisition frequency according to the actual situation. The test time length enables the moisture to reach a stable equilibrium state, and the data acquisition frequency density includes intervals that capture the key nodes of moisture content changes, especially during stages including but not limited to the initial, middle, and near-equilibrium periods of moisture change, to make the obtained equilibrium point data accurate and reliable. Adopt methods including but not limited to multiple groups of parallel tests and data statistical analysis. The number of each group of parallel tests is not less than E groups, where E is a specific value determined according to statistical requirements. Statistical analysis is performed on the data of multiple groups of tests, including but not limited to calculating statistical parameters such as the mean and standard deviation to reduce test errors and uncertainties and improve the credibility of the equilibrium point data. Using algorithms including but not limited to multiple linear regression to perform modeling analysis on multiple factors including but not limited to temperature, electric field strength, and initial moisture content that affect the moisture balance of oil-paper, determining the quantitative relationship between each factor and the moisture balance point, and the coefficient of determination R of the regression model 2 is not less than H%, where H is a value determined according to the fitting accuracy requirements of actual data to improve the accuracy and scientificity of the prediction of the moisture balance point; Use a deep neural network architecture with strong learning ability to deeply mine and extract features from terahertz spectral images and related test data. Specifically: This architecture consists of a series of specially designed layers. Some layers are responsible for extracting key features from the original data, some layers are used to reduce the data scale, and the remaining layers are tasked with integrating information and making judgments. Through such an architecture, it automatically captures unique data feature patterns under different combinations of moisture content, electric field strength, and temperature, thereby significantly enhancing the discrimination and classification accuracy of the moisture distribution state of oiled paper insulation. When training this neural network model, its accuracy must reach above I%, where I is determined according to the actual application. Before actually applying this technology, preprocess the data: Use a standardization method to standardize the data of the moisture content in oil and the moisture content of cardboard to the range of 0 to 1, and perform transformation operations on the spectral images including but not limited to rotation, mirroring, and scaling to expand the data volume, so that the number of data samples after these treatments increases by at least W%, where W is a ratio determined according to the actual situation, thereby improving the training efficiency of this neural network model and its adaptability to different situations.

7. A method for testing the moisture distribution of transformer oil-paper insulation based on the coupled action of electric field and temperature and correcting its equilibrium curve, characterized in that, The specific balance curve correction method is as follows: Based on a large amount of test data and in-depth theoretical analysis, establish a mathematical model considering the coupled action of electric field and temperature. This model accurately describes the diffusion, migration, and equilibrium processes of moisture in oiled paper insulation. The parameters in the model are determined by fitting and optimizing the test data, and the model has been strictly verified and tested to accurately reflect the actual situation. The modified equilibrium curve is simulated and verified using the finite element analysis algorithm, and the simulation results are compared and analyzed with the actual test data, with the error controlled within ±F%, where F is a value determined according to the accuracy. The model and curve are continuously optimized to accurately predict the moisture distribution of oil-paper insulation under different electric field and temperature conditions. Develop an algorithm to improve the effect of the previously constructed mathematical model. Specifically, first randomly scatter many "explorers" in this space. These "explorers" simulate a group of small bugs groping in the dark. Each "bug" carries its own position information and moving speed. They will continuously move and change their positions in this space and find a set of optimal "password combinations" during their movement. This set of "password combinations" is the key parameter that makes the model fit the actual experimental data most perfectly. There is a target accuracy for this searching process, not less than J%, where J is a standard determined according to the actual requirements. During the entire exploration process, some important "exploration rules" are automatically adjusted according to the characteristics presented by the data and the state during model training, including but not limited to the influence on the inertia of the "bugs" when moving and their speed of learning new positions. Moreover, all relevant information when the "bugs" change their positions during the exploration process is recorded in detail to form a record log. This log clearly records when the adjustment was made, the position information before and after the adjustment, and how these adjustments affected the model effect. And the recording form of this log is convenient for viewing and analysis at any time.

8. A method for testing the moisture distribution in transformer oil-paper insulation based on the coupled action of electric field and temperature and correcting its equilibrium curve, characterized in that, When applying this method to evaluate the moisture distribution of the actual transformer oil-paper insulation: According to the actual operating parameters of the transformer, including but not limited to the load factor, operating time, and oil temperature, quickly and accurately determine the temperature field and electric field distribution inside the transformer. By matching with the established database or model, predict the moisture distribution state of the oil-paper insulation, providing timely and effective guidance for the operation and maintenance of the transformer. Make adaptive adjustments to transformers of different types and specifications. By analyzing the structure and material characteristics of the transformer, adjust the relevant parameters in the model to make the method generally and accurately applicable to the moisture distribution evaluation and insulation status monitoring of various actual transformers.

9. A method for testing the moisture distribution in transformer oil-paper insulation based on the coupled action of electric field and temperature and correcting the equilibrium curve, characterized in that, In terms of data management and storage: Establish a perfect data management system to classify, store, and manage all data generated during the test process, including but not limited to sample information, test conditions, test data, and analysis results. The data storage format conforms to international general standards, facilitating data query, sharing, and further analysis. Adopt data encryption and backup technologies to ensure the security and integrity of the data, prevent data loss or illegal tampering. The data backup frequency is not less than G times per month, where G is a value determined according to the importance of the data, to cope with possible hardware failures and other unexpected situations.

10. A method for testing the moisture distribution in transformer oil-paper insulation based on the coupled action of electric field and temperature and correcting its equilibrium curve, characterized in that, In terms of formulating operation specifications and standards for the method: Develop a detailed operation manual and standard operating procedures, clarify the operation requirements, precautions and quality control key points for each step, and ensure that the operators have received professional training and conduct tests and data analysis in strict accordance with the operating procedures, so as to maintain the implementation quality of the method and the reliability of the results.

Images