A Method for Predicting the Remaining Life of Transformer Bushings Considering Moisture Content and Accelerated Aging
By constructing a moisture content data simulation model and aging dynamic model, the problem of dynamic changes in moisture in the aging of transformer casing is solved, and more accurate residual life prediction is achieved, which improves the operating reliability and status monitoring of power grid equipment.
Patent Information
- Application Number
- CN202310812787.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-07-04
- Publication Date
- 2025-07-29
- Estimated Expiration
- 2043-07-04
AI Technical Summary
The prior art cannot effectively consider the dynamic changes in moisture content during the aging of transformer casing insulation, resulting in inaccurate life prediction, affecting the safety and reliability of power grid equipment operation.
By constructing a moisture content data simulation model, the moisture accelerated aging factor coefficient is obtained, and combined with the aging dynamic model, a transformation casing residual life prediction method is established to consider the impact of dynamic changes in moisture content on insulation aging.
It improves the accuracy and speed of the remaining life prediction of transformer casing, reduces dependence on full life data, enhances the sensitivity to humidity changes, and improves the real-time and accuracy of grid equipment status monitoring.
Smart Images
Figure CN116861780B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of transformer bushing life prediction, and in particular, to a method for predicting the remaining life of a transformer bushing considering accelerated aging due to water content. Background Art
[0002] The power grid structure in our country is complex and large in scale, and the safe and stable operation of the power grid has become increasingly important. Since there is less full-life data of power equipment, the accuracy of life prediction using traditional methods is not high, so a new life prediction model for transformer bushings is required. The oil-paper insulation of transformer bushings is always affected by environmental humidity during operation, which in turn causes changes in the water content of the transformer bushings. However, for the fluctuations of moisture during the insulation aging of transformer bushings, the moisture-related parameters in traditional aging models cannot use these data to sensitively reflect the impact of this situation on the equipment life. These two situations greatly limit the prediction of the insulation life of transformer bushings, and as a result, it is impossible to accurately grasp the equipment operation status in real time, and it is easy to misestimate equipment failures.
[0003] The existing technologies can be mainly divided into four categories according to the above introduction, namely data-driven methods, virtual sample data methods, Arrhenius model, and traditional oil-paper insulation aging kinetics models. Data-driven methods require a large amount of data for model training. In industry, transformer bushings are usually replaced before failure, resulting in less full-life data, which will directly affect the accuracy of the prediction results of data-driven life prediction methods. The virtual sample data method can expand the size of the sample dataset, but the generated virtual samples will be severely affected by the real samples, and their contribution to model training is highly similar to that of the real samples, and it is impossible to effectively improve the generalization and accuracy of the life prediction model. The Arrhenius model does not consider the influence of moisture content, and it is still a challenging task to obtain moisture content information from the insulation of transformer bushings. The research on traditional oil-paper insulation aging kinetics models mainly focuses on the default constant initial moisture content, without considering the fluctuating characteristics of moisture content over time. Therefore, it is crucial to re-establish a model for predicting the life of oil-paper capacitor bushings, and it is necessary to consider how moisture changes affect oil-paper insulation; there is an urgent need to design a technical solution for predicting the remaining life of transformer bushings on the premise of considering accelerated aging due to water content, so as to improve the accuracy of estimating the remaining life of transformer bushings by humidity. Summary of the Invention
[0004] In order to solve the above problems, the purpose of the present invention is to provide.
[0005] In order to achieve the above technical purpose, the present application provides a method for predicting the remaining life of a transformer bushing considering accelerated aging due to water content, including the following steps:
[0006] By collecting the off-line data and on-line data of the transformer bushing, a water content data simulation model of the transformer bushing is constructed, wherein the on-line data represents the first physical characteristic data of the transformer bushing itself, and the off-line data represents the second physical characteristic data of the space where the transformer bushing is located;
[0007] Based on the water content data simulation model, by analyzing the accelerating effect of moisture on the insulation aging of the transformer bushing, the moisture accelerated aging factor coefficient is obtained;
[0008] Based on the aging kinetics model of the transformer bushing, by introducing the moisture accelerated aging factor coefficient, a prediction model is constructed to predict the remaining life of the transformer bushing.
[0009] Preferably, in the process of constructing the water content data simulation model, a common feature extractor and a specific domain feature extractor are set as the feature extractors of the water content data simulation model, wherein the common feature extractor is composed of a combination of three convolutional layers and a max pooling layer plus a batch normalization layer, and the specific domain feature extractor is composed of two convolutional layers, a global average pooling layer and a fully connected layer.
[0010] Preferably, in the process of constructing the water content data simulation model, the loss function of the water content data simulation model is expressed as:
[0011]
[0012] Preferably, in the process of obtaining the moisture accelerated aging factor coefficient, transformer bushings with different initial water contents are obtained by the moisture absorption method, and after the thermal aging test with the same test cycle, the degree of polymerization of the insulating paper is measured. The aging rate of each group of bushings is calculated by dividing the reciprocal of the degree of polymerization of the insulating paper after the test by the reciprocal of the degree of polymerization of the insulating paper of the brand new bushing, and the aging rate of the transformer bushing with a certain initial water content is used as the reference thermal aging rate. The ratio of the aging rate corresponding to different initial water contents to the set reference thermal aging rate is calculated to obtain the moisture accelerated aging factor coefficient.
[0013] Preferably, in the process of obtaining the aging kinetics model, the aging kinetics model is expressed as:
[0014]
[0015] θ=(0.09873mc%+0.746665mc%+4.27556)10 5
[0016] where θ is the water content fixed influence coefficient; mc% is the initial water content value of the transformer bushing; E a is the activation energy; R is the molar gas constant; T is the thermodynamic temperature; t l is the equipment usage time.
[0017] Preferably, during the process of constructing the prediction model, the prediction model is expressed as:
[0018]
[0019] where F(m) is the moisture acceleration coefficient, which changes dynamically with the moisture content m in the transformer bushing.
[0020] Preferably, during the process of obtaining the moisture content m, the moisture content m is expressed as:
[0021] Y(x) = f(x, b) + ε
[0022] where Y(x) is the dynamic response of the moisture content in the transformer bushing in the real space, f(x, b) is the simulation calculation result of the transformer bushing data simulation model, x is the input variable of the transformer bushing simulation model, b is the parameter of the simulation model, and ε is the additional error.
[0023] Preferably, a transformer bushing remaining life prediction system for implementing the transformer bushing remaining life prediction method includes
[0024] a data acquisition module for collecting offline data and online data of the transformer bushing;
[0025] a data processing module for constructing a moisture content data simulation model through offline data and online data, and obtaining a moisture acceleration aging factor coefficient by analyzing the accelerating effect of moisture on the insulation aging of the transformer bushing, where the online data represents the first physical characteristic data of the transformer bushing itself, and the offline data represents the second physical characteristic data of the space where the transformer bushing is located;
[0026] a life prediction module for predicting the remaining life of the transformer bushing based on the prediction model according to the moisture acceleration aging factor coefficient, where a prediction model is constructed based on the aging kinetics model of the transformer bushing by introducing the moisture acceleration aging factor coefficient, and it changes dynamically with the moisture content m in the transformer bushing.
[0027] Compared with the prior art, the present invention has the following technical effects:
[0028] (1) Compared with the data method, the required amount of transformer full-life data is small, and less data can be made to play a role, which can make the life prediction of the transformer bushing more accurate.
[0029] (2) Considering the influence of humidity on the insulation of the transformer bushing, the close relationship between the operating state and the remaining life of the transformer bushing is further highlighted.
[0030] (3) Compared with using traditional virtual sample methods, the generated simulation data covers a wider range of situations, overcomes the limitations of the real data distribution, and can improve the calculation speed and accuracy of the remaining life of transformer bushings. Description of the Drawings
[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.
[0032] Figure 1 is a flowchart of the method for predicting the remaining life of transformer bushings according to the present invention;
[0033] Figure 2 is a schematic structural diagram of the data simulation model of the transformer bushing according to the present invention. Detailed Embodiments
[0034] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some, rather than all, of the embodiments of the present application. Usually, the components of the embodiments of the present application described and illustrated in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the present application claimed, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.
[0035] As Figure 1-2 shown, in view of the problems of small sample size faced by the data-driven life prediction method and the limitation of data processing due to the assumption of the same distribution of source data in the virtual sample data method, the remaining life determination method proposed by the present invention will construct a simulation model of the transformer bushing, establish the relationship between the real device and the bushing simulation model through data, expand the types and quantities of the sample set, obtain a large number of balanced and accessible data sets, meet the requirements of comprehensive working condition coverage, reduce the influence of real sample data on the simulation data, and provide a large number of labeled, multi-working-condition, and low-cost training data for the life prediction model.
[0036] Aiming at the problem that the traditional oil-paper insulation aging kinetics model does not consider the fluctuation characteristics of water content over time, a life prediction model considering the accelerating effect of the water content of transformer bushings on insulation aging is re-established, the corresponding moisture acceleration coefficient is calculated, the accelerating effect of moisture on aging is obtained, and the traditional life prediction model of transformer bushings is expanded.
[0037] The implementation of the specific technical solution of the present invention mainly includes the following three parts:
[0038] 1. A method for predicting the life of a transformer considering the dynamic change of water content;
[0039] 2. Construction of a water content data simulation model for transformer bushings;
[0040] 3. A calculation method for the moisture accelerated aging factor coefficient of the remaining life of transformer bushings.
[0041] 1. A method for predicting the life of a transformer bushing considering the dynamic change of water content:
[0042] According to
[0043]
[0044] The traditional aging kinetics model of transformer bushings can be obtained as
[0045]
[0046] θ = (0.09873mc% + 0.746665mc% + 4.27556)10 5 (3)
[0047] In the formula, θ is the fixed influence coefficient of water content and can be calculated by formula (3); mc% is the initial water content value of the transformer bushing; E a is the activation energy; R is the molar gas constant, R = 8.315 J / (mol·K); T is the thermodynamic temperature; t l is the equipment usage time.
[0048] As the water content of the bushing changes, the degradation rate of the oil-paper insulation in the transformer bushing also changes, causing the aging rate of the transformer bushing to change. However, the above model does not consider the accelerating effect of the change in the water content in the transformer bushing on the remaining life of the transformer bushing. Therefore, the moisture acceleration coefficient η is introduced into the formula to obtain a life prediction method considering the accelerating effect of the dynamic change of water content on the remaining life of the transformer bushing, as shown in formula (4).
[0049]
[0050] Wherein, F(m) is the moisture acceleration coefficient, which dynamically changes with the water content m in the transformer bushing.
[0051] 2. Construction of the water content data simulation model for the transformer bushing:
[0052] (1) Establish a water content data simulation model for the transformer bushing equipment using sensor data.
[0053] By arranging temperature and humidity sensors, ultrasonic sensors, mechanical sensors, current sensors, voltage sensors, and other sensing devices on the transformer bushing, offline data such as environment and location and online data of the configured sensors are obtained, and a water content data simulation model for the transformer bushing equipment is established.
[0054] The water content data simulation model for the transformer bushing equipment can be expressed as
[0055] Y(x) = f(x, b) + ε
[0056] Wherein, Y(x) is the dynamic response of the water content in the transformer bushing in the real space, f(x, b) is the simulation calculation result of the transformer bushing data simulation model, x is the input variable of the transformer bushing water content data simulation model, b is the parameter of the simulation model, and ε is the additional error.
[0057] Use multi-source data such as historical monitoring data and online sensor data of the transformer bushing to train a water content numerical predictor for the transformer bushing through a neural network, find the mapping f, that is, f(x, b) ≈ Y(x), to achieve reliable prediction of the water content data of the transformer bushing.
[0058] As Figure 2 shown, the transformer bushing data simulation model extracts two types of features through a common feature extractor and a specific domain feature extractor (the common feature extractor obtains common features and the specific domain feature extractor obtains specific domain features).
[0059] The common feature extractor consists of a combination of three convolutional layers and max pooling layers plus a batch normalization layer.
[0060] The specific domain feature extractor consists of two convolutional layers, a global average pooling layer, and a fully connected layer.
[0061] The structure of the predictor is that after extracting features through a shared convolutional neural network, the feature representations of two samples are concatenated to obtain a combined feature representation. Then, this combined feature is input into the fully connected layer to determine whether the two samples are similar or calculate the distance between them.
[0062] The water content data simulation model for the transformer bushing equipment can be described as
[0063]
[0064] In the formula, is the learning feature of the historical data source domain of the transformer bushing in the i-th group, is the learning feature of the data target domain of the transformer bushing in the i-th group. F Di (g) is a domain-specific feature extractor network, F C (g) is a general feature extractor network, where is the source domain data, x T is the target domain data.
[0065] Then, after obtaining the source features, input them into the predictor to train the model, and map the domain-related representation of the j-th data sample corresponding to the i-th source domain to an estimated label:
[0066]
[0067] where, R i represents a domain-specific regression quantity graph.
[0068] To achieve the alignment of the decision boundary, it is necessary to directly reduce the difference in the output values generated by different domain-specific regression quantities. Set the network loss function as
[0069]
[0070] In the loss function, N is the total number of input sample data, n T is the number of source domain data, is the target domain data, and k, like i and j, is the independent variable in the summation; by changing the operating and environmental parameters in the constructed simulation model of the water content of the transformer bushing, the water content life cycle data required for life prediction can be obtained.
[0071] 3. Calculation method for the moisture accelerated aging factor coefficient of the remaining life of the transformer bushing:
[0072] The accelerating effect of moisture on the insulation aging of the transformer bushing is a key factor in predicting the remaining life of the transformer bushing. For the moisture accelerated aging factor coefficient F(m), its calculation formula is obtained through the following method:
[0073] Transformer bushings with different initial water contents are obtained by the moisture absorption method. After the thermal aging test with the same test cycle, measure the degree of polymerization of the insulating paper. Divide the reciprocal of the degree of polymerization of the insulating paper after the test by the reciprocal of the degree of polymerization of the insulating paper of the brand-new bushing to calculate the aging rate of each group of bushings respectively. The aging rate of a transformer bushing with a certain initial water content is used as the reference thermal aging rate, and calculate the ratio of the aging rate corresponding to different initial water contents to the set reference thermal aging rate to obtain the moisture accelerated aging factor η.
[0074] The relationship η = F(m) between the water content and the moisture accelerated aging factor is obtained by the data fitting method.
[0075] The present invention proposes a method for determining the remaining life of a transformer bushing. The traditional remaining life prediction model is improved in view of the accelerating effect of the actual water content of the transformer bushing on the aging process of the transformer bushing, aiming to improve the accuracy of the remaining life prediction considering the dynamic influence of the water content. At the same time, considering the problem that it is difficult to obtain the water content during the operation of the transformer bushing, the characteristic correlation between the existing water content data of the transformer bushing and various sensors is learned to realize the prediction of the water content of the transformer bushing. First, an estimation model for the water content data of the transformer bushing is built using offline data such as environment and location and online data such as configured sensors. Secondly, the relationship between the moisture accelerated aging factor coefficient and the water content of the transformer bushing is obtained through the aging test of the transformer bushing. Finally, the estimated water content data is obtained from the constructed simulation model of the water content data of the transformer bushing, and the moisture accelerated aging factor coefficient is calculated and substituted into the method for the remaining life of the transformer bushing considering the accelerated aging of the water content to solve and estimate the remaining life of the transformer bushing.
[0076] In addition to being used for determining the remaining life of a transformer bushing, the present invention is also applicable to other power equipment in the power grid that identifies the operating state and fault diagnosis by setting sensors. For example, when the switchgear in a substation is affected by moisture, it will also affect the internal insulation part. Since it also identifies the state and diagnoses faults based on the voltage and current data and power data inside the cabinet, the method proposed in this invention first constructs a device data simulation model based on the monitoring data, then predicts its life, and finally compares it with the full life data to determine whether the threshold setting is reasonable, so it is also effective for determining the remaining life of the switchgear in the substation.
[0077] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in one Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.
[0078] In the description of the present invention, it should be understood that the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, "a plurality of" means two or more, unless otherwise specifically defined.
[0079] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these changes and modifications.
Claims
1. A method for predicting the remaining life of a transformer bushing considering moisture content accelerated aging, characterized in that, Including the following steps: By collecting the off-line data and on-line data of the transformer bushing, a water content data simulation model of the transformer bushing is constructed, where the on-line data represents the first physical characteristic data of the transformer bushing itself, and the off-line data represents the second physical characteristic data of the space where the transformer bushing is located; Among them, the water content data simulation model of the transformer bushing is expressed as: Y(x) = f(x, b) + ε In the formula, Y(x) is the dynamic response of the water content of the transformer bushing in the real space, f(x, b) is the simulation calculation result of the transformer bushing data simulation model, x is the input variable of the transformer bushing water content data simulation model, b is the parameter of the simulation model, and ε is the additional error; Use the collected data to train the water content numerical predictor of the transformer bushing through a neural network, find the mapping f, that is, f(x, b) ≈ Y(x), to realize the reliable prediction of the water content data of the transformer bushing; The transformer bushing data simulation model extracts two types of features through a common feature extractor and a specific domain feature extractor. The common feature extractor consists of a combination of three convolutional layers and a max pooling layer plus a batch normalization layer. The specific domain feature extractor consists of two convolutional layers, a global average pooling layer, and a fully connected layer; After the predictor extracts features in the shared convolutional neural network, the feature representations of the two samples are concatenated to obtain a combined feature representation, and the combined feature is input into the fully connected layer to determine whether the two samples are similar or calculate the distance between them; Describe the water content data simulation model of the transformer bushing as: Wherein, is the learning feature of the historical data source domain of the transformer bushing in the i-th group, is the learning feature of the data target domain of the transformer bushing in the i-th group, F Di (g) is a feature extractor network for a specific domain, F C (g) is a general feature extractor network, where is the source domain data, x T is the target domain data; After obtaining the source features, input them into the predictor to train the model, and obtain the j-th data sample corresponding to the i-th source domain, mapping the domain-related representation to an estimated label: wherein, R i represents a regression quantity diagram in a specific field; Express the loss function of the water content data simulation model as: In the loss function, N is the total number of input sample data, and n T is the number of source domain data, is the target domain data, and k, like i and j, is the independent variable in the summation; by changing the operating and environmental parameters in the constructed simulation model of the water content data of the transformer bushing, the water content life cycle data required for life prediction can be obtained; Based on the water content data simulation model of the transformer bushing, by analyzing the accelerating effect of moisture on the insulation aging of the transformer bushing, obtain the moisture accelerated aging factor coefficient, where In the process of obtaining the moisture accelerated aging factor coefficient, transformer bushings with different initial water contents are obtained by the moisture absorption method. After the thermal aging test with the same test cycle, measure the degree of polymerization of the insulating paper. Divide the reciprocal of the degree of polymerization of the insulating paper after the test by the reciprocal of the degree of polymerization of the insulating paper of the brand new bushing to calculate the aging rate of each group of bushings respectively, and use the aging rate of the transformer bushing with a certain initial water content as the reference thermal aging rate. Calculate the ratio of the aging rate corresponding to different initial water contents to the set reference thermal aging rate to obtain the moisture accelerated aging factor coefficient η, and obtain the relationship between the water content and the moisture accelerated aging factor η = F(m) through the data fitting method; Based on the aging kinetics model of the transformer bushing, by introducing the moisture accelerated aging factor coefficient, construct a prediction model to predict the remaining life of the transformer bushing, according to Obtain the aging kinetics model of the traditional transformer bushing: θ = (0.09873mc% + 0.746665mc% + 4.27556)10 5 where θ is the water content fixed influence coefficient; mc% is the initial water content value of the transformer bushing; E a is the activation energy; R is the molar gas constant; T is the thermodynamic temperature; t l is the equipment usage time; Introduce the moisture acceleration coefficient η into the above formula to obtain a life prediction method considering the accelerating effect of the dynamic change of water content on the remaining life of the transformer bushing, which is: Where F(m) is the moisture acceleration coefficient, which varies dynamically with the water content m in the transformer bushing.