Insulating oil aging failure early warning method, device, equipment and storage medium

Abstract

The present application relates to the technical field of failure prediction, and particularly relates to an insulating oil aging failure early warning method, device, equipment and storage medium, the method realizes insulating oil aging failure early warning through four stages of model construction, data acquisition and prediction, curve analysis and early warning generation; the constructed breakdown voltage prediction model has strong generalization ability, can deeply analyze the nonlinear relationship of the absorption spectrum characteristics, is suitable for multiple insulating oil types at the same time, and has high prediction precision; through regular collection of absorption spectrum information, combined with real-time calculation of the breakdown voltage prediction model, the insulating oil aging state is continuously monitored, the hysteresis limitation of traditional offline detection is effectively broken through, dynamic tracking of the aging state is realized, and the failure risk of the insulating oil can be found in advance; through construction and analysis of the breakdown voltage change curve, the aging inflection point of the insulating oil can be recognized in advance, the failure risk is timely warned, the judgment result is intuitive and accurate, and the error caused by subjective experience is effectively avoided.

Description

Technical Field

[0001] This invention relates to the field of failure prediction technology, and in particular to a method, device, equipment and storage medium for early warning of aging failure of insulating oil. Background Technology

[0002] The insulating oil filling the inside of a transformer has two key functions: First, the insulating oil can effectively dissipate heat from the coils, ensuring that the coils will not be affected by excessive temperature during operation, or even be damaged. Second, the insulating oil can effectively prevent discharge between coils in high-voltage environments. It should be noted that once a discharge occurs between coils, it will cause extremely serious damage to the entire power network, and may lead to a series of serious problems such as power transmission interruption and equipment failure.

[0003] However, insulating oil cannot always maintain good performance. As the usage time increases, the insulating oil will be continuously corroded by water and oxygen. Under these circumstances, the insulating oil will gradually be oxidized and then aging will occur. After aging, the insulating oil will have a significant decrease in insulation performance, eventually leading to discharge between coils.

[0004] Furthermore, when the transformer tank ages due to long-term use, and cracks and leaks appear, the intrusion of water and oxygen will increase significantly. Under these circumstances, the aging rate of the insulating oil will be significantly accelerated, further threatening the normal operation of the transformer and the stability of the power grid.

[0005] Currently, there is a lack of corresponding technical means for in-situ online monitoring of the breakdown voltage, early warning of insulation oil breakdown, and judgment of aging trends during the aging process of insulating oil, and this technical gap urgently needs to be filled. Summary of the Invention

[0006] To fill the gaps in the existing technology, the purpose of this invention is to provide a method for early warning of aging failure of insulating oil. This method achieves early warning of aging failure of insulating oil through four stages: model construction, data acquisition and prediction, curve analysis and early warning generation. It solves the technical problem that traditional methods cannot monitor the aging process in situ online.

[0007] The first aspect of this invention provides a method for early warning of aging failure of insulating oil, comprising: pre-constructing a breakdown voltage prediction model; periodically collecting absorption spectrum information of insulating oil and recording the sampling time points corresponding to the absorption spectrum information; inputting the absorption spectrum information into the pre-constructed breakdown voltage prediction model to obtain the breakdown voltage corresponding to each sampling time point; constructing a breakdown voltage change curve based on the sampling time and its corresponding breakdown voltage; when a preset detection time point is reached, determining whether there is a failure problem and confirming aging information based on the constructed breakdown voltage change curve; and generating an early warning report based on the judgment result and aging information.

[0008] Optionally, in a first implementation of the first aspect of the present invention, the pre-constructed breakdown voltage prediction model includes: measuring the voltage information of various transformer oil samples throughout their entire life cycle to obtain sample voltage information, wherein the sample voltage information includes the initial breakdown voltage and the first breakdown voltage; measuring the absorption characteristics of various transformer oil samples in a set spectral region based on a preset wavelength selection rule to obtain sample absorption spectrum information; constructing a sample dataset using the sample absorption spectrum information as features and the sample voltage information as labels, and dividing the dataset into a training set and a test set based on a preset partitioning ratio; and training and testing a machine learning model using the training set and the test set with the goal of minimizing the error between the predicted breakdown voltage and the true value to obtain the breakdown voltage prediction model.

[0009] Optionally, in a second implementation of the first aspect of the present invention, the step of periodically collecting the absorption spectrum information of the insulating oil and recording the sampling time points corresponding to the absorption spectrum information, and inputting the absorption spectrum information into a pre-constructed breakdown voltage prediction model to obtain the breakdown voltage corresponding to each sampling time point, includes: periodically collecting the absorption spectrum information of the insulating oil in a set spectral region based on a preset wavelength selection rule, and recording the sampling time points corresponding to the absorption spectrum information; and inputting the periodically collected absorption spectrum information into the pre-constructed breakdown voltage prediction model to obtain the breakdown voltage corresponding to each sampling time point.

[0010] Optionally, in a third implementation of the first aspect of the present invention, the step of constructing a breakdown voltage variation curve based on the sampling time and its corresponding breakdown voltage, and determining whether a failure problem exists and confirming aging information based on the constructed breakdown voltage variation curve when a preset detection time point is reached, includes: connecting the breakdown voltages corresponding to each sampling time point sequentially with the sampling time as the horizontal axis and the breakdown voltage as the vertical axis to obtain a breakdown voltage variation curve; calculating the failure progress based on the breakdown voltage corresponding to the sampling time closest to the detection time point when the preset detection time point is reached; determining whether a failure problem exists based on the calculated failure progress, and confirming aging information based on the calculated failure progress and the constructed breakdown voltage variation curve.

[0011] Optionally, in a fourth implementation of the first aspect of the present invention, the step of confirming aging information based on the calculated failure progress and the constructed breakdown voltage change curve includes: calculating the aging rate at each sampling time in the breakdown voltage change curve; confirming inflection point information based on the calculated aging rate, wherein the inflection point information includes one or more accelerating inflection points and / or one or more decelerating inflection points; segmenting the breakdown voltage change curve based on the confirmed inflection point information to obtain multiple curve segments; and fitting the multiple curve segments using a pre-constructed natural aging model and an enhanced aging model to obtain fitting information, and confirming the aging information based on the fitting information.

[0012] Optionally, in a fifth implementation of the first aspect of the present invention, the step of using a pre-constructed natural aging model and an enhanced aging model to fit multiple curve segments to obtain fitting information, and confirming aging information based on the fitting information, includes: pre-constructing a natural aging model based on a first-order kinetic equation and a enhanced aging model based on a zero-order kinetic equation; using the pre-constructed natural aging model to fit multiple curve segments to obtain a first goodness of fit and a first rate coefficient corresponding to each curve segment; using the pre-constructed enhanced aging model to fit multiple curve segments to obtain a second goodness of fit and a second rate coefficient corresponding to each curve segment; comparing the first goodness of fit and the second goodness of fit, and determining the aging type of each curve segment based on the comparison result; evaluating the aging status based on the changing trends of the first rate coefficient and the second rate coefficient of each curve segment; and integrating the aging type and the aging status to obtain aging information.

[0013] Optionally, in the sixth implementation of the first aspect of the present invention, the step of generating an early warning report based on the judgment result and aging information includes: when the judgment result indicates that there is a risk of failure, integrating the judgment result and aging information to generate an alarm instruction; when the judgment result indicates that there is no risk of failure, obtaining a preset early warning template, filling the judgment result and aging information into the obtained early warning template respectively, and obtaining an early warning report.

[0014] A second aspect of the present invention provides an insulating oil aging failure early warning device, comprising: a construction module for pre-constructing a breakdown voltage prediction model; a acquisition module for periodically acquiring absorption spectral information of the insulating oil and recording the sampling time points corresponding to the absorption spectral information, inputting the absorption spectral information into the pre-constructed breakdown voltage prediction model to obtain the breakdown voltage corresponding to each sampling time point; a judgment module for constructing a breakdown voltage change curve based on the sampling time and its corresponding breakdown voltage, and when a preset detection time point is reached, judging whether a failure problem exists and confirming aging information based on the constructed breakdown voltage change curve; and an early warning module for generating an early warning report based on the judgment result and aging information.

[0015] A third aspect of the present invention provides an insulating oil aging failure early warning device, the insulating oil aging failure early warning device comprising: a memory and at least one processor, the memory storing instructions; at least one processor calling the instructions in the memory to cause the insulating oil aging failure early warning device to execute each step of the insulating oil aging failure early warning method described above.

[0016] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the steps of the insulating oil aging failure early warning method described in any of the preceding claims.

[0017] The technical solution of this invention achieves early warning of aging failure of insulating oil through four stages: model construction, data acquisition and prediction, curve analysis, and early warning generation. It can provide early warning before the insulating oil first undergoes electrical breakdown. Specifically, firstly, a breakdown voltage model is pre-constructed. This model has strong generalization ability, can deeply analyze the nonlinear relationship of absorption spectrum characteristics, and is adaptable to various types of insulating oil with high prediction accuracy. Then, by periodically collecting absorption spectrum information and combining it with the real-time calculation of the breakdown voltage prediction model, continuous monitoring of the aging state of insulating oil is achieved, effectively overcoming the lag limitation of traditional offline detection and realizing dynamic tracking of the aging state, enabling early detection of the failure risk of insulating oil. Furthermore, by constructing and analyzing the breakdown voltage change curve, the aging inflection point of the insulating oil can be identified in advance, and the failure risk can be warned in a timely manner. The judgment result confirmed by the breakdown voltage change curve is intuitive and accurate, effectively avoiding errors caused by subjective experience. Attached Figure Description

[0018] Figure 1 A logic flowchart of the insulating oil aging failure early warning method provided in an embodiment of the present invention;

[0019] Figure 2 This is a schematic diagram of the structure of the insulating oil aging failure early warning device provided in an embodiment of the present invention;

[0020] Figure 3 This is a schematic diagram of the structure of the insulating oil aging failure early warning device provided in an embodiment of the present invention. Detailed Implementation

[0021] This invention provides a method, apparatus, device, and storage medium for early warning of insulating oil aging failure. In this invention, the terms "first," "second," "third," "fourth," etc. (if applicable) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0022] This application discloses a method for early warning of aging failure of insulating oil. For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the insulating oil aging failure early warning method of the present invention includes:

[0023] 101. Construct a breakdown voltage prediction model;

[0024] In this embodiment, based on the correlation between the absorption spectrum of insulating oil and the breakdown voltage, a machine learning model is used to establish a mapping relationship, enabling the direct prediction of the breakdown voltage from the absorption spectrum information.

[0025] 102. Periodically collect the absorption spectrum information of insulating oil and record the sampling time points corresponding to the absorption spectrum information. Input the absorption spectrum information into the pre-constructed breakdown voltage prediction model to obtain the breakdown voltage corresponding to each sampling time point.

[0026] In this embodiment, the absorption spectrum information of insulating oil is collected periodically based on a preset cycle. The preset cycle can be one day, one week, or one month. For example, in the early stage of use of insulating oil, the preset cycle can be set to one month. As the aging process of insulating oil accelerates, the system can dynamically adjust the sampling cycle according to the aging progress, gradually shortening it from the initial 30 days to 7 days, so as to more accurately monitor the aging state of insulating oil. The collected absorption spectrum information can reflect the concentration changes of various chemical components in insulating oil, and these chemical components are closely related to the breakdown voltage of insulating oil. The collected absorption spectrum information is input into a pre-constructed breakdown voltage prediction model. The model can quickly calculate the breakdown voltage value corresponding to each sampling time point and store it to provide data support for subsequent analysis and early warning.

[0027] 103. Based on the sampling time and its corresponding breakdown voltage, construct a breakdown voltage change curve. When the preset detection time point is reached, determine whether there is a failure problem and confirm the aging information based on the constructed breakdown voltage change curve.

[0028] In this embodiment, the preset detection time point has flexible setting options, such as detection every 3 months or 6 months, which is intended to be adjusted in a timely manner according to actual usage and maintenance needs. For example, in the initial stage of use of insulating oil, given its relatively stable performance and small initial changes, the detection time point can be set to once every 6 months. By reasonably setting the detection time point, not only can the state of the insulating oil be effectively monitored, but the workload and unnecessary cost of frequent detection can also be reduced to a certain extent, thereby ensuring the safe operation of the equipment while achieving reasonable allocation of maintenance resources.

[0029] 104. Generate early warning reports based on the judgment results and aging information.

[0030] This application discloses a method for early warning of aging failure of insulating oil. It achieves early warning of aging failure of insulating oil through four stages: model construction, data acquisition and prediction, curve analysis, and early warning generation. This method can provide early warning before the first electrical breakdown of the insulating oil. Specifically, firstly, a breakdown voltage model is pre-constructed. This model has strong generalization ability, can deeply analyze the nonlinear relationship of absorption spectral characteristics, and is adaptable to various types of insulating oil with high prediction accuracy. Then, by periodically collecting absorption spectral information and combining it with real-time calculations of the breakdown voltage prediction model, continuous monitoring of the aging state of the insulating oil is achieved. This effectively overcomes the lag limitations of traditional offline detection, enabling dynamic tracking of the aging state and early detection of the failure risk of the insulating oil. Furthermore, by constructing and analyzing the breakdown voltage change curve, the aging inflection point of the insulating oil can be identified in advance, providing timely early warning of failure risk. The judgment results confirmed by the breakdown voltage change curve are intuitive and accurate, effectively avoiding errors caused by subjective experience.

[0031] Furthermore, in this embodiment of the invention, the pre-built breakdown voltage prediction model includes:

[0032] 201. Measure the voltage information of various transformer oil samples throughout their entire life cycle to obtain sample voltage information, including initial breakdown voltage and first breakdown voltage;

[0033] In this embodiment, at least three typical insulating oils were selected, such as mineral oil, vegetable oil, and synthetic oil. For each type of insulating oil, samples were collected at different stages of its entire life cycle, specifically including brand new oil, oil used for 1 year, oil used for 3 years, oil used for 5 years, and failed oil. According to the GB / T 507-2002 standard, an automatic breakdown voltage tester was used to measure the breakdown voltage of various samples at different life cycles. During the measurement process, two key data were recorded: the initial breakdown voltage of brand new oil and the first breakdown voltage of failed oil when it reached the failure threshold. To ensure the accuracy and reliability of the measurement results, each sample was measured 5 times, and the average value of these 5 measurements was taken as the final breakdown voltage data of the sample, i.e., the sample breakdown voltage information.

[0034] In this embodiment, the model is trained based on multi-type, full life-cycle sample data, which has a strong generalization ability and can adapt to various types of insulating oil, thereby achieving high-precision prediction and providing strong support for the quality control and maintenance of insulating oil.

[0035] 202. Based on the preset wavelength selection rules, the absorption characteristics of various transformer oil samples in the set spectral region are measured to obtain the sample absorption spectrum information;

[0036] In this embodiment, the wavelength range is divided according to a pre-set wavelength selection rule. Specifically, the spectral region is set to 400-750nm, with 5 wavelengths set in the 400-550nm range and 3 wavelengths set in the 550-750nm range, thus determining a total of 8 uniformly distributed wavelengths. Subsequently, the absorption spectrum of each sample is measured using a UV-Vis spectrometer. To ensure the accuracy and reliability of the measurement results, each selected wavelength is measured 5 times, and the average value of the data obtained from these 5 measurements is taken. Finally, these processed data are integrated to form the absorption spectrum information of the sample.

[0037] In this embodiment, by averaging multiple measurements, the accuracy and reliability of the acquired data are effectively ensured. This not only minimizes the impact of random errors but also significantly improves the credibility of the data, providing a high-quality data foundation for subsequent model training. This ensures that the model receives accurate and reliable input during training, further improving the model's performance and prediction accuracy.

[0038] 203. Using sample absorption spectrum information as features and sample voltage information as labels, construct a sample dataset and divide the dataset into training set and test set based on a preset division ratio;

[0039] In this embodiment, the required dataset is constructed using absorption spectral information as input features and the corresponding breakdown voltage as output label. To ensure the representativeness and sufficiency of the dataset, the sample size for each type of oil is no less than 300. After the dataset is constructed, it is randomly divided according to a preset division ratio of 8:2 to form a training set and a test set. The training set contains 80% of the data and is used for the model training process. The test set contains the remaining 20% ​​of the data and is used to evaluate the prediction accuracy of the model.

[0040] 204. With the goal of minimizing the error between the predicted breakdown voltage and the true value, the machine learning model is trained and tested using training and testing sets to obtain the breakdown voltage prediction model.

[0041] In this embodiment, the machine learning model is the LightGBM model. The core objective of using the LightGBM model is to minimize the mean square error (MSE) between the predicted breakdown voltage and the actual value. During the model training phase, the model parameters are iteratively adjusted using training set data. Specifically, the learning rate is set to 0.01, and the tree depth is set to 5. Through continuous iteration, the model gradually learns the data features in the training set to achieve the goal of minimizing MSE. After model training is completed, the model is validated using a test set, focusing on calculating prediction accuracy (measured by the bias rate) and efficiency (measured by computation speed), and requiring the prediction error to be strictly controlled within 5%. To ensure that the model can effectively predict different types of insulating oil, the model is repeatedly trained and optimized for different types of insulating oil. During this process, the model parameters are continuously adjusted so that the model can learn the characteristics of different types of insulating oil, thereby improving the model's generalization ability and providing strong support for the quality assessment of insulating oil and the safe operation of power equipment.

[0042] Further, in this embodiment of the invention, the step of periodically collecting the absorption spectrum information of the insulating oil, recording the sampling time points corresponding to the absorption spectrum information, and inputting the absorption spectrum information into a pre-constructed breakdown voltage prediction model to obtain the breakdown voltage corresponding to each sampling time point includes:

[0043] 301. Based on the preset wavelength selection rules, periodically collect the absorption spectrum information of insulating oil in the set spectral region, and record the sampling time points corresponding to the absorption spectrum information;

[0044] In this embodiment, a fiber optic spectrometer is used to collect the absorption spectrum information of insulating oil in a set spectral region (consistent with 400-750nm in step 202) according to a preset cycle (e.g., once a month). During equipment installation, the signal acquisition head needs to be inserted into the insulating oil through the sampling valve of the transformer. The insertion depth should be no less than 5cm, and it should be kept away from the coil to avoid electromagnetic interference. The fiber optic cable needs to be fixed, the bending radius should be no less than 5cm, and the connection should be properly sealed. The initial sampling cycle is set to once every 720 hours (i.e., 30 days), and the sampling time is arranged between 3-5 am. Before each collection, the equipment needs to be calibrated using a standard whiteboard. At the same time, the sampling time point should be recorded synchronously, accurate to the day, for example, recorded as "May 10, 2024, usage duration 180 days". Through regular collection, the temporal continuity of the data can be ensured, the aging process of the insulating oil can be dynamically tracked, and the random errors that may be generated by a single detection can be avoided. In addition, combining timed sampling with triggered sampling can not only ensure the continuity of data, but also increase the monitoring density when abnormal operating conditions occur.

[0045] 302. Input the periodically collected absorption spectrum information into the pre-constructed breakdown voltage prediction model to obtain the breakdown voltage corresponding to the sampling time point;

[0046] In this embodiment, the absorption spectrum information collected each time is preprocessed, including noise removal and normalization, and then input into a pre-constructed breakdown voltage prediction model. The model will output the breakdown voltage prediction value corresponding to the sampling time point and store the output breakdown voltage corresponding to the sampling time point in the database.

[0047] Furthermore, in this embodiment of the invention, the step of constructing a breakdown voltage change curve based on the sampling time and its corresponding breakdown voltage, and determining whether a failure problem exists and confirming aging information based on the constructed breakdown voltage change curve when a preset detection time point is reached, includes:

[0048] 401. With sampling time as the x-axis and breakdown voltage as the y-axis, connect the breakdown voltages corresponding to each sampling time point in sequence to obtain the breakdown voltage variation curve.

[0049] In this embodiment, a data visualization tool, such as Matplotlib, is used to connect the breakdown voltage values ​​at each sampling time point in sequence, with the sampling time (unit: days) as the horizontal axis and the predicted breakdown voltage (unit: kV) as the vertical axis, to form a continuous curve and obtain the breakdown voltage change curve.

[0050] In this embodiment, by constructing a breakdown voltage change curve, the trend of breakdown voltage change over time can be presented intuitively, which facilitates the rapid identification of its aging pattern, such as whether the voltage decreases at a constant rate or at an accelerated rate. Compared with discrete data points, the curve analysis method allows technicians to detect signs of accelerated aging in advance.

[0051] 402. When the preset detection time point is reached, the failure progress is calculated based on the breakdown voltage corresponding to the sampling time closest to the detection time point;

[0052] In this embodiment, a pre-set detection time point is established, such as the end of each quarter or year, and manual adjustment is supported. When the preset detection time point is reached, the sampling time closest to that detection time point is selected from the curve. For example, if the detection time is set to June 30, 2024, the closest sampling time might be June 25, 2024. Subsequently, the breakdown voltage corresponding to that sampling time is extracted. The calculation is then performed according to the failure progress calculation formula, specifically:

[0053] Failure progress = |Extracted breakdown voltage - Initial breakdown voltage| / |Initial breakdown voltage - First breakdown voltage| × 100%;

[0054] For example, if the initial breakdown voltage is 38kV, the currently extracted breakdown voltage is 33kV, and the first breakdown voltage is 25kV, then the failure progress = |33-38| / |38-25|×100%=5 / 13≈38.5%.

[0055] 403. Determine whether there is a failure problem based on the calculated failure progress, and confirm the aging information based on the calculated failure progress and the constructed breakdown voltage change curve.

[0056] In this embodiment, when the failure progress is less than or equal to a preset threshold 'a', it is determined that there is no failure risk; if the failure progress is greater than the preset threshold 'a', it is determined that there is a failure risk. At this time, the system will generate an alarm command to remind relevant personnel to perform replacement operations as soon as possible. The value of the threshold 'a' can be determined according to the specific scenario. For important substations, 'a' is set to 0.5 for strict warning; for general substations, 'a' is set to 0.8 for moderate warning; and for temporary substations, 'a' is set to 0.95 for lenient warning. Setting the threshold 'a' according to the specific scenario can adapt to the needs of different scenarios, greatly improve the flexibility of the warning mechanism, ensure that technicians can receive alarm information and respond in a timely manner, thereby effectively improving the fault prevention rate.

[0057] Furthermore, in this embodiment of the invention, confirming aging information based on the calculated failure progress and the constructed breakdown voltage change curve includes:

[0058] 501. Calculate the aging rate at each sampling time in the breakdown voltage change curve;

[0059] In this embodiment, the aging rate is calculated for each sampling time point in the breakdown voltage variation curve using the following formula:

[0060] ;

[0061] in, U AI, n and U AI, n+1 It is the nth point time. t n and the (n+1)th point time t n+1 The corresponding breakdown voltage.

[0062] 502. Based on the calculated aging rate, confirm the inflection point information, which includes one or more accelerating inflection points and / or one or more decelerating inflection points;

[0063] When natural aging progresses to a certain stage, if the transformer enclosure is damaged by external factors such as sudden cracks, causing a large amount of impurities such as metal particles, insoluble substances, and water to seep into the insulating oil in a short period of time, the aging rate coefficient will increase sharply, triggering accelerated aging. This results in a significant increase in the aging rate coefficient, and the breakdown voltage change curve will show a clear inflection point. On the other hand, if the transformer enclosure is repaired, the aging speed will slow down, the aging rate coefficient will decrease significantly, and a clear inflection point will also be formed on the breakdown voltage change curve.

[0064] In this embodiment, the inflection point type is determined by comparing the aging rates at adjacent time points. If the aging rate increases significantly for three consecutive time points after the determination time point n, such as when the rate reaches more than three times the previous rate, then the determination time point n is an acceleration inflection point. If the aging rate decreases significantly for three consecutive time points after the determination time point n, such as when the rate drops to less than one-third of the previous rate, then the determination time point n is a deceleration inflection point.

[0065] In this embodiment, the aging rate and inflection point identification can capture sudden changes during the aging process, such as accelerated aging caused by cracks in the transformer tank, thereby avoiding missing key changes. By quantifying the rate ratio and verifying continuous points, the accuracy of identification can be greatly improved, and the aging process can be better understood, problems can be detected in time and measures can be taken, thereby extending the service life of the transformer and improving the reliability and safety of the transformer during operation.

[0066] 503. Based on the confirmed inflection point information, the breakdown voltage change curve is segmented to obtain multiple curve segments;

[0067] 504. The pre-constructed natural aging model and enhanced aging model are used to fit multiple curve segments to obtain fitting information, and the aging information is confirmed based on the fitting information.

[0068] In this embodiment, based on the confirmed inflection point information, the breakdown voltage change curve is meticulously divided into multiple curve segments, and then fitted separately. By using the curve segment fitting method, the accuracy of aging analysis is significantly improved, thereby capturing the aging characteristics of each stage more precisely. This not only effectively distinguishes the specific mechanisms of different aging stages, but also provides more detailed data support for in-depth research on the aging process of materials or equipment, making the aging analysis work more comprehensive and accurate.

[0069] Furthermore, in this embodiment of the invention, the step of using a pre-constructed natural aging model and an enhanced aging model to fit multiple curve segments to obtain fitting information, and then confirming aging information based on the fitting information, includes:

[0070] 601. A natural aging model is pre-constructed based on the first-order kinetic equations, and an enhanced aging model is pre-constructed based on the zero-order kinetic equations;

[0071] In this embodiment, the natural aging model is constructed based on the first-order kinetic equation, which is specifically expressed as follows:

[0072] ;

[0073] in, For aging rate, k 1 represents the first rate coefficient. t for U AI The corresponding working time inside the transformer; integrating the equations of the first-order dynamic model, we can obtain:

[0074] ;

[0075] in, U New The initial breakdown voltage of the new insulating oil as defined above is equivalent to when time U AI From the above formula, we can see that ln U AI and t The relationship is linear; this integral equation theoretically describes U AI The mathematical relationship between t and t in the current situation; U AIThe actual value of t is fitted using the least squares method according to the above integral equation, and the coefficients of determination are obtained, which are denoted as the first goodness of fit. R 1 2 , to describe the degree of agreement between the actual value and the first-order dynamic model; where 0 ≤ R 1 2 ≤1, and can be solved at the same time k The value of 1.

[0076] When insulating oil is contaminated with metal particles, insoluble substances, water, or other substances that easily increase conductivity due to various complex circumstances, the insulating oil will age rapidly over time. In fact, the aging rate of the insulating oil may be unrelated to its inherent properties but rather to the continuous infiltration of external conductive substances. In this embodiment, an enhanced aging model is used to describe the aforementioned situation. This enhanced aging model is constructed based on a 0th-order kinetic equation, which is specifically expressed as follows:

[0077] ;

[0078] in, For aging rate, k 0 represents the second rate coefficient. t for U AI The corresponding operating time inside the transformer; integrating the equations of the 0th-order dynamic model, we can obtain:

[0079] ;

[0080] in, U New The initial breakdown voltage of the new insulating oil as defined above is equivalent to when time U AI From the above formula, we can see that ln U AI and t The relationship is linear; this integral equation theoretically describes U AI The mathematical relationship between t and t in the current situation; U AI The actual value of t is fitted using the least squares method according to the above integral equation, and the coefficients of determination are obtained, which are denoted as the second goodness of fit. R 0 2 , to describe the degree of agreement between the actual values ​​and the 0th-order dynamic model; where 0 ≤ R 0 2 ≤1, and can be solved at the same time k The value of 0.

[0081] 602. The pre-constructed natural aging model is used to fit multiple curve segments to obtain the first goodness of fit and the first rate coefficient corresponding to each curve segment.

[0082] 603. The pre-constructed enhanced aging model is used to fit multiple curve segments to obtain the second goodness of fit and the second rate coefficient corresponding to each curve segment.

[0083] 604. Compare the first and second goodness-of-fit scores, and determine the aging type of each curve segment based on the comparison results;

[0084] In this embodiment, comparison R 1 2 and R 0 2 ,like R 1 2 > R 0 2 And 0.8≤ R 1 2 If the value is ≤1, then the curve segment is determined to be a natural aging process, such as the initial aging of a well-sealed transformer; if... R 0 2 > R 1 2 And 0.8≤ R 0 2 If ≤1, the curve segment is determined to be an accelerated aging process, such as aging caused by leaks in the housing; if 0.4≤ R 0 2 ≈ R 1 2 If the value is less than 0.8, the curve segment is determined to be of mixed aging type, and the mixed aging rate is calculated. k H :

[0085] ;

[0086] If none of the above three conditions are met, it is considered an unknown aging type, and an alarm command needs to be generated to remind relevant personnel to promptly investigate potential problems with the transformer itself.

[0087] In this embodiment, the natural and enhanced models each correspond to different dynamic models. These two models have different focuses and applicable ranges when dealing with aging problems. By using the goodness-of-fit index, the degree of fit between the model and the actual aging process can be effectively quantified, thereby making the judgment of aging type more scientific and reliable. The higher the goodness-of-fit, the more accurate the model's description of the aging mechanism; conversely, adjustments and optimizations are needed. In addition, clear judgment conditions provide quantifiable standards and repeatable operating procedures for the identification of aging types, which not only avoids errors caused by subjective judgment but also greatly improves the efficiency of fault location.

[0088] 605. Assess the aging status based on the changing trends of the first and second rate coefficients of each curve segment;

[0089] In this embodiment, the rate coefficient of each curve segment is analyzed ( k 1. k 0 or k H The aging status is determined by the changes in the rate coefficient, with the following specific rules: if the rate coefficient shows a gradually increasing trend, the aging status is determined to be gradually accelerating; if the rate coefficient shows a gradually decreasing trend, the aging status is determined to be gradually slowing down; if the rate coefficient changes irregularly, the aging status is determined to be complex and without obvious patterns; if there is only one curve and the curve has no inflection point, the aging status is determined to be stable. Rate coefficient and status analysis can achieve a quantitative description of the aging process and can directly reproduce the analysis process; by analyzing the dynamic changes of the rate coefficient, maintenance plans can be formulated in advance, thereby improving the pertinence and effectiveness of maintenance work.

[0090] 606. Integrate the aging type and the aging state to obtain aging information.

[0091] Furthermore, in this embodiment of the invention, generating an early warning report based on the judgment result and aging information includes:

[0092] 701. When the judgment result indicates that there is a risk of failure, integrate the judgment result and aging information to generate an alarm command;

[0093] In this embodiment, when a failure risk is determined, the failure progress and aging information, including aging type and aging status, are first integrated. Then, based on the integrated information, a standardized alarm command is generated. The generated alarm command includes the failure progress, the time interval of each curve segment and the corresponding aging type and aging status, as well as suggested measures. Finally, the generated alarm command is sent to the operation and maintenance system, which can be done through SMS, platform push, or other means.

[0094] 702. When the judgment result is no risk of failure, obtain the preset early warning template, fill in the judgment result and aging information into the obtained early warning template, and obtain an early warning report;

[0095] In this embodiment, a pre-set warning template is invoked, which includes modules such as "basic information", "detection results", "aging analysis" and "recommendations". Then, the judgment result (e.g., "no risk"), failure progress (e.g., 38.5%) and aging information (e.g., "natural aging, stable condition") are accurately filled into the corresponding positions of the warning template, and finally a PDF report is automatically generated.

[0096] In this embodiment, the template-based generation of early warning reports ensures the uniformity of the report format. At the same time, the content of alarm commands and early warning reports is standardized, clearly including key information such as risk level and recommended measures, enabling maintenance personnel to respond quickly to relevant situations and effectively reduce decision-making costs. In addition, the early warning reports and alarm commands cover equipment information and situation analysis, providing technical personnel with comprehensive reference data, which helps to formulate reasonable maintenance strategies and reduce overall maintenance costs.

[0097] The above describes the insulating oil aging failure early warning method in the embodiments of the present invention. The following describes the insulating oil aging failure early warning device in the embodiments of the present invention. Please refer to [link / reference]. Figure 2 One embodiment of the insulating oil aging failure early warning device of the present invention includes:

[0098] Module 801 is used to pre-build the breakdown voltage prediction model;

[0099] The acquisition module 802 is used to periodically acquire the absorption spectrum information of insulating oil and record the sampling time points corresponding to the absorption spectrum information. The absorption spectrum information is input into the pre-constructed breakdown voltage prediction model to obtain the breakdown voltage corresponding to each sampling time point.

[0100] The judgment module 803 is used to construct a breakdown voltage change curve based on the sampling time and its corresponding breakdown voltage. When the preset detection time point is reached, it determines whether there is a failure problem and confirms the aging information based on the constructed breakdown voltage change curve.

[0101] The early warning module 804 is used to generate early warning reports based on the judgment results and aging information.

[0102] Based on the same ideas as the methods in the above embodiments, the apparatus provided in this application can implement the methods in the above embodiments.

[0103] above Figure 2The insulating oil aging failure early warning device in this embodiment of the invention is described in detail from the perspective of modular functional entities. The insulating oil aging failure early warning device in this embodiment of the invention is described in detail below from the perspective of hardware processing.

[0104] Figure 3 This is a schematic diagram of the structure of an insulating oil aging failure early warning device 900 provided in an embodiment of the present invention. The insulating oil aging failure early warning device 900 can vary considerably due to different configurations or performance. It may include one or more central processing units (CPUs) 910 and memory 920, and one or more storage media 930 (e.g., one or more mass storage devices) storing application programs 933 or data 932. The memory 920 and storage media 930 can be temporary or persistent storage. The program stored in the storage media 930 may include one or more modules (not shown in the diagram), each module may include a series of instruction operations on the insulating oil aging failure early warning device 900. Furthermore, the processor 910 may be configured to communicate with the storage media 930 and execute a series of instruction operations in the storage media 930 on the insulating oil aging failure early warning device 900 to implement the steps of the insulating oil aging failure early warning method provided in the above-described method embodiments.

[0105] The insulating oil aging failure early warning device 900 may also include one or more power supplies 940, one or more wired or wireless network interfaces 950, one or more input / output interfaces 960, and / or one or more operating systems 931, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 3 The structure of the insulating oil aging failure early warning device shown does not constitute a limitation on the insulating oil aging failure early warning device. It may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0106] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of the insulating oil aging failure early warning method.

[0107] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0108] If the integrated unit is implemented as 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, in essence, 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. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0109] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An insulation oil aging failure early warning method, characterized in that, The method comprises the following steps: Pre-constructing a breakdown voltage prediction model, specifically, measuring the voltage information of a plurality of transformer oil samples throughout their life cycles to obtain sample voltage information, the sample voltage information including initial breakdown voltage and first breakdown voltage; measuring the absorption characteristics of the plurality of transformer oil samples in a set spectral region based on a preset wavelength selection rule to obtain sample absorption spectrum information; Taking the sample absorption spectrum information as features and the sample voltage information as labels, a sample data set is constructed, and the data set is divided into a training set and a test set based on a preset division ratio; Training and testing a machine learning model using the training set and the test set to minimize the error between the predicted breakdown voltage and the true value, thereby obtaining a breakdown voltage prediction model; Periodically collecting absorption spectrum information of the insulating oil and recording the sampling time points corresponding to the absorption spectrum information, and inputting the absorption spectrum information into the pre-constructed breakdown voltage prediction model to obtain the breakdown voltage corresponding to each sampling time point, specifically, periodically collecting the absorption spectrum information of the insulating oil in the set spectral region based on the preset wavelength selection rule, and recording the sampling time points corresponding to the absorption spectrum information; Respectively inputting the periodically collected absorption spectrum information into the pre-constructed breakdown voltage prediction model to obtain the breakdown voltage corresponding to the sampling time point; Based on the sampling time and the corresponding breakdown voltage, a breakdown voltage change curve is constructed, and when a preset detection time point is reached, it is determined whether there is a failure problem based on the constructed breakdown voltage change curve and the aging information is confirmed, specifically, taking the sampling time as the abscissa and the breakdown voltage as the ordinate, connecting the breakdown voltage corresponding to each sampling time point in turn to obtain a breakdown voltage change curve; when the preset detection time point is reached, the breakdown voltage corresponding to the sampling time closest to the detection time point is used to calculate the failure progress; based on the calculated failure progress, it is determined whether there is a failure problem, and based on the calculated failure progress and the constructed breakdown voltage change curve, the aging information is confirmed; Based on the judgment result and the aging information, a warning report is generated.

2. The insulation oil aging failure early warning method according to claim 1, characterized in that, The aging information is confirmed based on the calculated failure progress and the constructed breakdown voltage change curve, which comprises: Calculating the aging speed of each sampling time in the breakdown voltage change curve; Confirming inflection point information based on the calculated aging speed, the inflection point information including one or more accelerated inflection points and / or one or more decelerated inflection points; Segmenting the breakdown voltage change curve based on the confirmed inflection point information to obtain a plurality of segments of curves; Using a pre-constructed natural aging model and a pre-constructed accelerated aging model to respectively fit the plurality of segments of curves to obtain fitting information, and confirming the aging information based on the fitting information.

3. The insulation oil aging failure early warning method according to claim 2, characterized in that, The fitting information is obtained by using the pre-constructed natural aging model and the pre-constructed accelerated aging model to respectively fit the plurality of segments of curves, and the aging information is confirmed based on the fitting information, which comprises: Pre-constructing a natural aging model based on a first-order kinetic equation and a pre-constructed accelerated aging model based on a zero-order kinetic equation; The pre-constructed natural aging model is used for fitting the multiple segments of the curve respectively, and a first fitting goodness and a first rate coefficient corresponding to each segment of the curve are obtained; The pre-constructed accelerated aging model is used for fitting the multiple segments of the curve respectively, and a second fitting goodness and a second rate coefficient corresponding to each segment of the curve are obtained; The first fitting goodness and the second fitting goodness are compared, and the aging type of each segment of the curve is determined according to the comparison result; The aging trend is evaluated according to the change trend of the first rate coefficient and the second rate coefficient of each segment of the curve; The aging information is obtained by integrating the aging type and the aging trend.

4. The insulation oil aging failure early warning method according to claim 1, characterized in that, The pre-warning report is generated based on the judgment result and the aging information, including: When the judgment result is that there is a failure risk, the judgment result and the aging information are integrated to generate an alarm instruction; When the judgment result is that there is no failure risk, a pre-set pre-warning template is obtained, the judgment result and the aging information are filled into the obtained pre-warning template respectively, and a pre-warning report is obtained.

5. An insulation oil aging failure early warning device, characterized in that, It includes: The construction module is used for pre-constructing a breakdown voltage prediction model. Specifically, the voltage information of a plurality of transformer oil samples throughout the life cycle is measured to obtain sample voltage information, which includes an initial breakdown voltage and a first breakdown voltage. The absorption characteristics of the plurality of transformer oil samples in a set spectral region are measured based on a pre-set wavelength selection rule to obtain sample absorption spectrum information. The sample absorption spectrum information is used as a feature, and the sample voltage information is used as a label to construct a sample data set, and the data set is divided into a training set and a test set based on a pre-set division ratio. The training set and the test set are used to train and test the machine learning model to minimize the error between the predicted breakdown voltage and the true value, and the breakdown voltage prediction model is obtained. The acquisition module is used for periodically acquiring the absorption spectrum information of the insulating oil and recording the sampling time point corresponding to the absorption spectrum information, and inputting the absorption spectrum information into the pre-constructed breakdown voltage prediction model to obtain the breakdown voltage corresponding to each sampling time point. Specifically, based on the pre-set wavelength selection rule, the absorption spectrum information of the insulating oil in the set spectral region is periodically acquired, and the sampling time point corresponding to the absorption spectrum information is recorded. The periodically acquired absorption spectrum information is input into the pre-constructed breakdown voltage prediction model to obtain the breakdown voltage corresponding to the sampling time point. The judgment module is used for constructing a breakdown voltage change curve based on the sampling time and the corresponding breakdown voltage, and determining whether there is a failure problem and confirming the aging information when reaching a pre-set detection time point. Specifically, the sampling time is used as the horizontal coordinate, the breakdown voltage is used as the vertical coordinate, the breakdown voltage corresponding to each sampling time point is connected in sequence to obtain a breakdown voltage change curve graph; when the pre-set detection time point is reached, the breakdown voltage corresponding to the sampling time closest to the detection time point is used to calculate the failure progress; whether there is a failure problem is determined based on the calculated failure progress, and the aging information is confirmed based on the calculated failure progress and the constructed breakdown voltage change curve graph. The pre-warning module is used for generating a pre-warning report based on the judgment result and the aging information.

6. An insulation oil aging failure early warning device characterized by, The insulation oil aging failure early warning device comprises a memory and at least one processor, and the memory stores instructions; The at least one processor invokes the instructions in the memory, so that the insulation oil aging failure early warning device performs the steps of the insulation oil aging failure early warning method according to any one of claims 1-4.

7. A computer-readable storage medium having stored thereon instructions, the computer-readable storage medium comprising: The instructions are executed by the processor to implement the steps of the insulation oil aging failure early warning method according to any one of claims 1-4.

Images