Oil paper insulation aging state evaluation method, device, equipment and medium

Abstract

The application relates to an oil-paper insulation aging state evaluation method and device, equipment and medium, and relates to the technical field of transformer insulation aging evaluation. The method comprises the following steps: obtaining content data of full-cycle and stage characteristic quantities in an oil sample and normalizing the content data, determining an aging stage based on the stage characteristic quantities, extracting a standard fingerprint vector of the corresponding stage from a fingerprint database, calculating a comprehensive matching degree of a measured vector and a standard vector through multidimensional distance matching, and selecting a maximum matching degree corresponding to insulation paper polymerization degree as an evaluation result. The fingerprint database contains normalized data of full-cycle characteristic quantities and stage characteristic quantities screened through correlation under different aging degrees. The application solves the problems of low evaluation accuracy, poor generalization and insufficient stage adaptability caused by few characteristic quantities and ignoring stage products in the existing method by fusing full-cycle and stage characteristic quantities and combining a stage-by-stage fingerprint matching strategy, and improves the accuracy and refinement level of aging evaluation.

Description

Technical Field

[0001] This application relates to the field of transformer insulation aging assessment technology, specifically to a method, apparatus, equipment, and medium for assessing the aging status of oil-paper insulation. Background Technology

[0002] Power transformers are core transmission and transformation equipment in power systems, and the operational status of their oil-paper composite insulation systems directly affects the safe and stable operation of the power grid. In oil-paper insulation systems, the aging of the insulating paper is irreversible and is a core factor limiting the overall insulation life of the transformer. Statistical data shows that the failure rate of insulating paper is significantly higher than that of insulating oil; therefore, the aging state of the insulating paper is a key indicator determining the reliability of the oil-paper insulation system. Establishing a scientific and effective method for assessing the aging state of insulating paper is crucial for ensuring the safe operation of transformers and formulating operation and maintenance strategies.

[0003] Currently, chemical assessment is the mainstream online assessment method, which assesses the condition of insulating paper by detecting aging products such as furfural, carbon oxides, and methanol dissolved in the oil. This method has advantages such as no downtime, online monitoring, and results unaffected by on-site electromagnetic interference, and has been widely recognized and applied in the industry. However, existing chemical assessment methods mostly rely on single characteristic quantities for analysis, which has significant limitations in practical applications: First, furfural, as a traditional indicator, is generated in extremely low quantities in the early stages of aging, making it difficult to reflect the early aging state, and is easily affected by environmental factors and the catalytic decomposition of copper materials in the later stages of aging, becoming almost undetectable in heat-resistant paper insulation systems; second, CO and CO2 in the oil do not only originate from the degradation of insulating paper, but also from the oxidation of the insulating oil itself and air intrusion, leading to insufficient specificity in the assessment results; third, although methanol has attracted attention as a novel characteristic quantity, its assessment effect in the later stages of insulating paper aging is poor. Research shows that assessments based on single aging products have limited accuracy and are subject to many limitations in application scenarios.

[0004] To overcome the limitations of single indicators, scholars have recently conducted research on the assessment of the aging status of insulating paper based on the combination of multiple characteristic quantities, and have made some progress. However, the number of chemical characteristic quantities used in existing joint assessment methods is usually no more than five. In fact, dozens of chemical substances are generated during the aging process of oil-paper insulation. Assessment based on only a few characteristic quantities not only fails to fully utilize the complementary information between different aging products, but may also lead to information overlap between indicators, making the assessment results susceptible to external interference and lacking generalization ability.

[0005] Furthermore, existing screening criteria for chemical characteristics are relatively simplistic, typically requiring products to persist throughout the entire aging cycle and maintain a stable correlation with the degree of polymerization (DP) of the insulating paper. This screening criterion leads to the neglect of effective products that only appear or undergo significant changes at specific aging stages. While such stage-specific products cannot independently characterize the entire aging process, they often contain rich information on the evolution of aging stages, providing crucial support for accurate identification of aging stages and assessment of the state within each stage. Currently, there are no reports on the application of stage-specific chemical products in the aging assessment of oil-paper insulation, and a systematic classification system and utilization method are lacking, resulting in blind spots in aging assessment regarding stage division and refined diagnosis. Summary of the Invention

[0006] In view of this, this application provides a method, apparatus, equipment and medium for assessing the aging status of oil-paper insulation. The main purpose is to solve the technical problems of low assessment accuracy, poor generalization and insufficient stage adaptability caused by the existing transformer oil-paper insulation aging assessment methods relying on too few characteristic quantities and ignoring stage characteristic products.

[0007] In a first aspect, this application provides a method for assessing the aging condition of oil-paper insulation, including: The content data of various aging characteristics dissolved in the transformer oil sample to be tested are obtained, and the content data are normalized to obtain a standardized feature vector; wherein, the various aging characteristics include full-cycle characteristics and stage characteristics; the full-cycle characteristics are aging characteristic products continuously detected throughout the entire aging cycle of the oil-paper insulation; the stage characteristics are aging characteristic products detected at a specific aging stage of the oil-paper insulation. Based on the stage feature quantities in the standardized feature vector, the aging stage of the transformer oil sample to be tested is determined. The standard fingerprint vector corresponding to the aging stage is extracted from the pre-constructed fingerprint database, and the measured fingerprint vector of the transformer oil sample to be tested is matched with the standard fingerprint vector in a multi-dimensional distance to obtain the comprehensive matching degree between the transformer oil sample to be tested and the standard fingerprint vector; wherein, each standard fingerprint vector in the fingerprint database includes the normalized content data of the full-cycle feature quantity and the stage feature quantity after correlation screening under different aging degrees. The degree of polymerization of the insulating paper corresponding to the standard fingerprint vector with the highest comprehensive matching degree is selected as the evaluation result of the aging state of the insulating paper of the transformer oil sample to be tested.

[0008] Secondly, this application provides a device for assessing the aging status of oil-paper insulation, comprising: The acquisition module is used to acquire the content data of various aging characteristics dissolved in the transformer oil sample to be tested, and to normalize the content data to obtain a standardized feature vector; wherein, the various aging characteristics include full-cycle characteristics and stage characteristics; the full-cycle characteristics are aging characteristic products continuously detected throughout the entire aging cycle of the oil-paper insulation; the stage characteristics are aging characteristic products detected at a specific aging stage of the oil-paper insulation. The determination module is used to determine the aging stage of the transformer oil sample under test based on the stage feature quantity in the standardized feature vector; The matching module is used to extract standard fingerprint vectors corresponding to the aging stage from a pre-built fingerprint database, and perform multi-dimensional distance matching between the measured fingerprint vector of the transformer oil sample to be tested and the standard fingerprint vector to obtain the comprehensive matching degree between the transformer oil sample to be tested and the standard fingerprint vector; wherein, each standard fingerprint vector in the fingerprint database includes the normalized content data of the full-cycle feature quantity and the stage feature quantity after correlation screening under different aging degrees. The selection module is used to select the degree of polymerization of the insulating paper corresponding to the standard fingerprint vector with the highest comprehensive matching degree as the evaluation result of the aging state of the insulating paper of the transformer oil sample to be tested.

[0009] Thirdly, this application provides an electronic device, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the computer program to implement the oil-paper insulation aging state assessment method described in the first aspect.

[0010] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the oil-paper insulation aging state assessment method described in the first aspect.

[0011] By employing the above technical solution, this application provides a method, apparatus, equipment, and medium for assessing the aging state of oil-paper insulation. Compared with existing technologies, this application can obtain content data of multiple aging characteristic quantities dissolved in the transformer oil sample to be tested, and normalize the content data to obtain a standardized feature vector. The multiple aging characteristic quantities include full-cycle characteristic quantities and stage characteristic quantities. Full-cycle characteristic quantities are aging characteristic products continuously detected throughout the entire aging cycle of the oil-paper insulation; stage characteristic quantities are aging characteristic products detected at specific aging stages of the oil-paper insulation. Based on the stage characteristic quantities in the standardized feature vector, the aging state of oil-paper insulation is determined. The aging stage of the transformer oil sample under test is determined. Standard fingerprint vectors corresponding to the aging stage are extracted from a pre-constructed fingerprint database. The measured fingerprint vectors of the transformer oil sample under test are then matched with the standard fingerprint vectors in multiple dimensions to obtain the comprehensive matching degree between the transformer oil sample under test and the standard fingerprint vectors. Each standard fingerprint vector in the fingerprint database includes full-cycle feature quantities and normalized content data of stage feature quantities after correlation screening under different aging degrees. The degree of polymerization of the insulating paper corresponding to the standard fingerprint vector with the highest comprehensive matching degree is selected as the evaluation result of the aging state of the insulating paper of the transformer oil sample under test.

[0012] By adopting the above technical solution, this application obtains and normalizes the content data of various aging characteristics dissolved in the transformer oil sample under test, including full-cycle and stage characteristics. This expands the aging assessment feature dimension from a single or few indicators to the fusion of complementary information of full-cycle and stage characteristics, overcoming the problems of information loss and poor generalization caused by the insufficient number of features in existing methods. By determining the aging stage of the transformer oil sample under test based on the stage characteristics in the standardized feature vector, and extracting the standard fingerprint vector corresponding to the aging stage from the fingerprint database for matching, the non-uniform dynamic adaptation of the aging assessment strategy with the deterioration process of the insulation paper is realized, solving the technical defects of stage division blind spots and insufficient fine diagnosis capabilities caused by ignoring stage characteristic products. By constructing a fingerprint database together with the stage characteristics after correlation screening and the full-cycle characteristics, and using a multi-dimensional distance fusion matching algorithm to calculate the comprehensive matching degree to select the corresponding degree of polymerization of the insulation paper as the evaluation result, the objective weighting and accurate comparison of multi-source chemical fingerprint information are realized, eliminating the limitation of single distance measurement being easily affected by feature amplitude fluctuations. This method solves the technical problems of low evaluation accuracy, poor generalization and insufficient stage adaptability of existing transformer oil-paper insulation aging assessment methods, which rely on too few characteristic quantities and ignore stage characteristic products. By making full use of the complementary aging information of different chemical products in the whole cycle and local stages, it significantly improves the stage adaptability and prediction accuracy of oil-paper insulation aging assessment.

[0013] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0014] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A flowchart illustrating a method for assessing the aging state of oil-paper insulation provided in this application embodiment; Figure 2 A graph showing the degree of polymerization of insulating paper provided in this application embodiment; Figure 3 A graph showing the change in the content of alcohol compounds provided in an embodiment of this application; Figure 4 A graph showing the change in the content of aldehyde compounds provided in an embodiment of this application; Figure 5 A graph showing the change in the content of an acid compound is provided in an embodiment of this application. Figure 6 A staged feature quantity content change and stage division diagram provided for embodiments of this application; Figure 7 This is a schematic diagram of a device for assessing the aging status of oil-paper insulation, provided in an embodiment of this application. Detailed Implementation

[0017] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of this application, including various details to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0018] The following description, with reference to the accompanying drawings, describes the method, apparatus, equipment, and medium for assessing the aging status of oil-paper insulation according to embodiments of this application.

[0019] This application provides a method, apparatus, equipment, and medium for assessing the aging status of oil-paper insulation. The main purpose is to solve the technical problems of low assessment accuracy, poor generalization, and insufficient stage adaptability caused by existing transformer oil-paper insulation aging assessment methods, which rely on too few characteristic quantities and ignore stage characteristic products.

[0020] like Figure 1 As shown in the embodiments of this application, a method for assessing the aging state of oil-paper insulation is provided, which may specifically include: Step 101: Obtain the content data of various aging characteristics dissolved in the transformer oil sample to be tested, and normalize the content data to obtain a standardized feature vector; the various aging characteristics include full-cycle characteristics and stage characteristics.

[0021] This application focuses on the construction of potential characteristic quantities and their fingerprint system for assessing the aging of transformer oil-paper insulation, along with corresponding assessment methods. Through thermal aging experiments on oil-paper insulation, the types and content variations of chemical characteristic quantities in the oil are analyzed; full-cycle and stage-specific characteristic quantities that can be used for assessment are established. Based on the variation patterns of stage-specific characteristic quantities, a strategy for dividing different aging stages is proposed. An insulation paper aging chemical fingerprint database containing full-cycle and stage-specific characteristic quantities is constructed using multiple correlation coefficients. An aging state assessment strategy based on a fusion comparison algorithm is proposed. Finally, the accuracy of the constructed fingerprint database and assessment method is verified.

[0022] Specifically, to obtain test samples with different aging degrees, this paper describes the preparation of oil-paper insulation samples. Following the aforementioned thermal aging experiments, sampling and equilibration were performed. The DP value of the insulation paper was measured using the viscosity method according to the IEC 60450 standard. The change in the degree of polymerization of the insulation paper with thermal aging time is shown below. Figure 2 As shown.

[0023] Depend on Figure 2 It is evident that the DP value of insulating paper decreases monotonically with increasing aging time, exhibiting an overall characteristic of "rapid decline in the early stage and slow decay in the later stage": in the early stage of aging (0–10d), the DP value drops rapidly from about 1100 to about 700, and the weak bonds at the ends of the cellulose chains break violently under thermal stress; as the aging process progresses, the rate of DP decline gradually slows down, dropping below 200 by 50d. The DP range measured in this aging experiment covers the entire life cycle of transformer oil paper insulation, laying the foundation for the completeness of the fingerprint database.

[0024] In this embodiment of the application, through comprehensive detection and analysis of chemical products in oil during the thermal aging test of oil-paper insulation, from the 52 compounds detected, after eliminating interfering substances with no obvious change or too low content according to the product source, content level and its change law with the aging process, 18 chemical characteristic quantities with aging assessment value were finally screened out and divided into two categories: full-cycle characteristic quantities and stage characteristic quantities.

[0025] The full-cycle characteristic quantity refers to aging characteristic products that are continuously detected throughout the entire aging cycle of the oil-paper insulation and have a certain correlation with the degree of polymerization (DP) value of the insulation paper. In the embodiments of this application, the full-cycle characteristic quantity includes at least one of the following 11 compounds: Alcohols (4 types): methanol, ethanol, n-propanol (1-propanol), n-butanol (1-butanol); Aldehydes (3 types): formaldehyde, acetaldehyde, furfural; Acids (4 types): formic acid, acetic acid, propionic acid, and butyric acid.

[0026] The aforementioned full-cycle characteristic quantities can be continuously detected throughout the aging process.

[0027] in, Figure 3 The graph shows the changes in the content of alcohol compounds such as methanol, ethanol, 1-propanol, and 1-butanol. Figure 3 As shown, the concentrations of all four types of alcohols monotonically increase with aging time. Methanol production is the highest, reaching a maximum of 2.3 ppm, and it grows the fastest, making it the core characteristic quantity of alcohols. Although the production of the other alcohols is relatively low, they can still be used for evaluation at a specific aging stage.

[0028] Figure 4 The graph shows the changes in the content of aldehyde compounds in 2-furfural, formaldehyde, and acetaldehyde, such as... Figure 4 As shown, furfural has the highest production, and its content increases sharply in the middle and late stages of aging, reaching a maximum of about 3.3 ppm. Formaldehyde and acetaldehyde are more sensitive to the initial aging, but their overall growth rate is relatively slow.

[0029] Figure 5 The graph shows the changes in the content of acidic compounds such as formic acid, acetic acid, propionic acid, and butyric acid. Figure 5 As shown, the overall content of acids is low, with a maximum value not exceeding 0.6 ppm. Initial formation is slow, the rate of increase accelerates significantly in the middle stage, formic acid content maintains an upward trend in the later stage, while the rate of increase for other acids slows down. Overall, there are significant differences in the order of magnitude of content among different categories and characteristic products, requiring subsequent normalization processing.

[0030] Stage characteristic quantities refer to aging characteristic products detected only at specific aging stages of oil-paper insulation. In the embodiments of this application, stage characteristic quantities include the following seven compounds, and their content changes are as follows: Figure 6 As shown: 2-Methylfuran, furfuryl alcohol, methyl acetate, methyl formate, cyclohexanone, 2-pentanone, and 2-heptanone.

[0031] The aforementioned stage characteristic quantities appear gradually with increasing aging in the order of 2-methylfuran, furfuryl alcohol, methyl acetate and methyl formate, cyclohexanone, 2-pentanone, and 2-heptanone. Their combination and content levels can preliminarily reflect the evolution of oil-paper insulation from early to middle and late aging stages. Combined with... Figure 2 and Figure 6 Specifically: 2-Methylfuran: Detected in the DP value range of 696 to 519, its content first increases and then decreases, reaching a peak at around 540 DP value, making it more suitable for early aging stage location. Furfuryl alcohol: It begins to appear after the DP value drops to 519. Its content generally increases with the degree of aging, and then decreases slightly in the later stage. Methyl acetate: It is gradually generated when the DP value drops below 387, and the overall value continues to increase, showing a relatively stable negative correlation with the DP value; Methyl formate: It is gradually generated when the DP value drops below 387, showing a change characteristic of first increasing and then decreasing, making it more suitable as an indicator for identifying intermediate stages; Cyclohexanone: It begins to appear and continues to increase when the DP value reaches below 355, and can be used as an auxiliary assessment of the degree of aging in the middle and late stages; 2-Pentanone: It begins to be produced after the DP value reaches 309, and its content increases significantly in the later stages of aging; 2-Hepanotone: It begins to appear after the DP value reaches 290, and its content increases significantly in the later stages of aging.

[0032] In the embodiments of this application, the content data of various aging characteristic substances dissolved in the transformer oil sample to be tested can be obtained by gas chromatography-mass spectrometry (such as HS-GC-MS). Those skilled in the art will understand that other applicable chemical analysis instruments or methods can also be used to obtain the above content data, and this application does not make specific limitations in this regard.

[0033] To ensure the reliability of the test data, the content data of each aging characteristic quantity can be repeatedly tested multiple times (e.g., 3 times) and the average value is taken. At the same time, the standard deviation of the test data is calculated to evaluate the reproducibility of the data.

[0034] Because there are significant differences in the order of magnitude of the content of different categories and characteristic products (for example, the content of alcohols ranges from approximately 0 to 2.3 ppm, and the content of acids ranges from approximately 0 to 0.6 ppm), in order to eliminate the influence of order of magnitude on the subsequent calculation results, this embodiment of the application can use the maximum-minimum normalization method to standardize the original content data, linearly mapping the data to the [0,1] interval. The normalization formula is as follows: (1); In the formula, For the first The sampling point of the th sampling point Normalized content of the product, For the first The minimum content of this product throughout the entire aging test. For the first The maximum content of this product throughout the entire aging test. For the first The sampling point of the th sampling point The original measured content of the product.

[0035] After normalization, the content data of each aging characteristic quantity are mapped to the [0,1] interval, thus obtaining the standardized feature vector of the transformer oil sample to be tested.

[0036] Step 102: Determine the aging stage of the transformer oil sample under test based on the stage feature quantities in the standardized feature vector.

[0037] Since the characteristic products of different stages appear in the order of 2-methylfuran, furfuryl alcohol, methyl acetate and methyl formate, cyclohexanone, 2-pentanone and 2-heptanone, their combination and content levels can reflect the evolution of the aging process of oil paper insulation from the early to the middle and late stages. Therefore, the stage characteristic quantities can be used to preliminarily locate the aging stage of the oil sample to be tested, narrow the subsequent fingerprint matching range, and improve the stage adaptability and accuracy of the assessment.

[0038] For embodiments of this disclosure, determining the aging stage of the transformer oil sample under test based on the stage feature quantity in the standardized feature vector may specifically include: For each sampling point, based on the detection limit of the stage characteristic quantity, the detection status of each stage characteristic product in the current sampling point is determined. The detection status includes the detection status of the stage characteristic product being detected in the sampling point and the non-detection status of the stage characteristic product not being detected in the current sampling point. The component of the stage feature product in the detected state at the current sampling point is assigned the first value, and the component of the stage feature product in the undetected state at the current sampling point is assigned the second value. The assigned components are combined in a preset order to construct the product type state vector of the sampling point; wherein, the first value is different from the second value. Obtain the product type state vector of the current sampling point and the product type state vector of the previous sampling point that is adjacent to the current sampling point in time. Calculate the absolute value of the difference between the product type state vector of the current sampling point and the product type state vector of the previous sampling point to obtain the absolute value of the state change corresponding to the characteristic products of each stage. The absolute values ​​of all state changes are summed, and the sum is determined as the type change discriminant variable corresponding to the current sampling point. If the type change discriminant variable is greater than the preset type change threshold, it is determined that there is a change in the type of stage characteristic product between adjacent sampling points, and the time node corresponding to the type change discriminant variable is determined as the candidate stage boundary. The magnitude of the type change discriminant variable represents the overall degree of change in the detection state of stage characteristic products between adjacent sampling points. Using the candidate stage boundary as the dividing point, a first preset number of samples are selected before the candidate stage boundary as the front sample group, and a second preset number of samples are selected after the candidate stage boundary as the back sample group. Calculate the arithmetic mean of the normalized content of each type of stage characteristic product in the front sample group to obtain the front average normalized content; at the same time, calculate the arithmetic mean of the normalized content of each type of stage characteristic product in the back sample group to obtain the back average normalized content. For each type of stage characteristic product, the difference between the back-side average normalized content and the front-side average normalized content is calculated, and the difference is squared to obtain the content difference square value corresponding to each type of stage characteristic product. The total sum of squared differences in content of all stage characteristic products is obtained by summing the squared values ​​of the differences. The square root of the total sum of squared differences is then taken, and the result of the square root operation is determined as the overall content difference index value before and after the candidate stage boundary. The magnitude of the overall content difference index value represents the comprehensive deviation of the normalized content of all stage characteristic products before and after the candidate stage boundary. The aging stage of the transformer oil sample to be tested can be determined based on the candidate stage boundary and the overall content difference index value, which may include: If multiple adjacent time points are simultaneously identified as candidate stage boundaries, then the magnitude of the overall content difference index value corresponding to the multiple adjacent time points is compared. The candidate stage boundary with the largest overall content difference index value is selected as the aging stage boundary. The aging cycle is divided into intervals according to the aging stage boundaries to obtain the aging stage of the transformer oil sample to be tested. The aging stages, in chronological order, include stages one, two, three, four, five, and six. Each stage is separated by adjacent effective stage boundaries, and the combination of characteristic products of each stage satisfies the following progressive relationship: The first stage corresponds to an empty set of characteristic product combinations. The combination of stage-specific characteristic products corresponding to the second stage only includes the first type of stage-specific characteristic products; The combination of stage-specific characteristic products corresponding to the third stage includes the first stage-specific characteristic products, and adds the second stage-specific characteristic products. The combination of stage-specific characteristic products corresponding to the fourth stage includes the first and second stage-specific characteristic products, and adds a third stage-specific characteristic product. The fifth stage corresponds to a combination of stage-specific characteristic products that, in addition to the first, second, and third stage-specific characteristic products, also includes a fourth stage-specific characteristic product. The combination of stage-specific characteristic products corresponding to the sixth stage includes the first, second, third, and fourth stage characteristic products, and adds a fifth stage characteristic product. Furthermore, the normalized content of the fifth stage characteristic product changes at a rate greater than a preset rate of change threshold over time. Among them, the first type of stage feature product, the second type of stage feature product, the third type of stage feature product, the fourth type of stage feature product and the fifth type of stage feature product are all stage feature products corresponding to the stage feature quantities in the standardized feature vector, and each type of stage feature product is different from the others.

[0039] Specifically, it can be based on the first Detection limit of the product Its detection status is represented as: (2); In the formula, Indicates the first Characteristic products of each stage in the first Detection status of each sampling point Indicates the first Characteristic products of each stage in the first The detection status of each sampling point. Indicates the first Characteristic products of each stage in the first Undetected status at each sampling point Indicates that the characteristic product of the j-th stage is in the first stage. The actual measured concentration of oil samples from each sampling point. Indicates the first The detection limit of characteristic products at each stage.

[0040] Therefore, based on the detection status, the first... The product type state vector of each sampling point can be represented as: (3); In the formula, For the first Product type state vector of each sampling point For the first Characteristic products of each stage in the first The detection status of each sampling point.

[0041] To quantify the degree of change in the types of characteristic products between adjacent sampling points, a type change index is defined: (4); In the formula, Indicates the first The sampling point and the first The number of products whose state changes were detected between sampling points. Indicates the first Characteristic products of each stage in the first Detection status of each sampling point Indicates the first Characteristic products of each stage in the first The detection status of each sampling point, where m represents the total number of types of stage-specific characteristic products.

[0042] To further determine whether there has been a change in the types of characteristic products between adjacent sampling points, a discriminant variable for type change is defined. for: (5); In the formula, =1 indicates the first The sampling point and the first There are changes in the types of characteristic products at each sampling point, and the corresponding time point can be regarded as the boundary of the candidate stage. =0 indicates the first The sampling point and the first There was no change in species between the sampling points. Indicates the first The sampling point and the first The number of products whose state changes were detected between sampling points.

[0043] To reduce the impact of fluctuations in individual detection values ​​on the stage division results, candidate boundaries are used. Centered on a certain number of samples before and after it, calculate the first... The average normalized content of the product on both sides of the boundary: (6); In the formula, Candidate Boundary The previous chapter The average normalized content of the characteristic products of each stage, Candidate Boundary The next The average normalized content of the characteristic products of each stage, The number of samples selected before the candidate boundary. This refers to the number of samples selected after the candidate boundary. Let be the normalized content value of the j-th stage characteristic product at the r-th sampling point. This paper takes two adjacent samples before and after the boundary, i.e. .

[0044] The overall content difference index value of characteristic product content distribution before and after the candidate boundary is calculated using Euclidean distance: (7); In the formula, Candidate stage boundary The overall content difference index before and after, Candidate Boundary The previous chapter The average normalized content of the characteristic products of each stage, Candidate Boundary The next The average normalized content of each stage-specific characteristic product, where m is the total number of stage-specific characteristic products. The larger the value, the more significant the change in the distribution of multiple product contents before and after that time point.

[0045] When the time node appears If the value is 1, then this point is considered a candidate stage boundary node. If adjacent nodes appear simultaneously... If the value is 1, then compare adjacent nodes. ,choose Larger points serve as stage boundaries. It should be noted that in days 0-26, due to the limited variety of aging products, the first three aging stages (stage I-stage III) can be determined solely based on the production times of 2-methylfuran and furfuryl alcohol; there is no need to utilize [other methods]. and Stage division. As the aging process deepens, the types of characteristic products of each stage become more diverse, making it difficult to divide stages based solely on changes in the types of products. Using equations (1)-(7), the discriminant variables for the changes in the types of products at each time point after stage III are calculated. and content difference indicators The following table serves as the basis for dividing the stages: Table 1. Boundary Species Change Indicators and Content Difference Indicators

[0046] As shown in Table 1, for adjacent time points 26 and 28, Z 26 With Z 28 Both are equal to 1, because D 26 Greater than D 28 Therefore, 26 was chosen as the boundary node between Stage III and Stage IV. On day 32, Z... 32 Equal to 1 but Z 32 With Z 34 Since the value equals 0, 32 is chosen as the boundary node between Stage IV and Stage V. Z appears on days 38 and 40. 38 With Z 40 In the case where both are equal to 1, since D 40 Greater than D 38 Therefore, 40 was chosen as the boundary node between Stage V and Stage VI. The aging cycle was ultimately divided into 6 stages, as follows: Figure 6 As shown. From Figure 6As can be seen, the stages are not divided according to equal time intervals, but rather based on the evolution of chemical products. Stage I (0–8 days): No characteristic products of this stage were detected; the DP value decreased from 1115 to 696, indicating the insulation paper was still in a healthy state. Stage II (8–18 days): 2-methylfuran began to appear; the DP value further decreased from 696 to 519, indicating that cellulose degradation was gradually increasing. Stage III (18–26 days): furfuryl alcohol began to form and coexist with 2-methylfuran; the DP value decreased from 519 to 387, indicating that the insulation paper had entered a stage of continuous deterioration. Stage IV (26–32 days): methyl formate and methyl acetate appeared successively; the DP value decreased to 355, and the chemical product system became more abundant. Stage V (32–40 days): cyclohexanone began to be detected; the DP value decreased to 309, indicating that the insulation paper had entered the final stage of aging. VI (40–50 days) marks the exit from service. 2-Pentanone and 2-Heptanone appear and accumulate rapidly, further reducing the DP value from 309 to 186. The insulation paper reaches the end of its lifespan, and the transformer should be taken out of service. This stage division method is based on the evolutionary characteristics of chemical fingerprint information during the aging process, effectively reflecting the entire process of insulation paper transitioning from a healthy state to a severely aged state.

[0047] Step 103: Extract the standard fingerprint vector corresponding to the aging stage from the pre-constructed fingerprint database, and perform multi-dimensional distance matching between the measured fingerprint vector of the transformer oil sample to be tested and the standard fingerprint vector to obtain the comprehensive matching degree between the transformer oil sample to be tested and the standard fingerprint vector.

[0048] In this embodiment, a chemical fingerprint database for transformer oil-paper insulation is pre-constructed. Each standard fingerprint vector in the database includes normalized content data of full-cycle features and stage features after correlation screening at different aging degrees (i.e., different DP values). The construction process of the fingerprint database includes the following steps: Step 201: Obtain the full-cycle characteristic content data and stage characteristic content data of the oil-paper insulation aging sample.

[0049] In this embodiment, oil-paper insulation samples at different aging levels were obtained by conducting accelerated thermal aging experiments, and the content of aging products and the DP value of the insulation paper in each sample were measured. For example, Kunlun No. 45 transformer oil and Weidmann insulation paper with a thickness of 0.25 mm were selected as raw materials. The oil-paper materials were loaded into a sealed container at a mass ratio of 14:1, and accelerated thermal aging tests were conducted at a constant temperature of 130°C. Samples were taken at preset sampling intervals, and finally, samples of different aging states covering the entire life cycle of transformer oil-paper insulation were obtained.

[0050] Step 202: Calculate the comprehensive correlation coefficient between the characteristic quantities of each stage and the degree of polymerization (DP) of the insulating paper. The comprehensive correlation coefficient includes Kendall's rank correlation coefficient, Pearson's linear correlation coefficient and Spearman's rank correlation coefficient.

[0051] Some of the stage-specific features can only be used to identify and classify aging stages, while others can not only be used to identify and classify aging stages but also serve as auxiliary features in DP evaluation. Therefore, this application employs a joint correlation coefficient discrimination method to analyze the correlation between each stage-specific feature and the DP value.

[0052] Specifically, the following three correlation coefficients are calculated respectively: (8); In the formula, τ is the Kendall rank correlation coefficient, C is the number of pairs of variables with the same direction of rank change, D is the number of pairs of variables with inconsistent directions of rank change, and r is the Pearson linear correlation coefficient, used to characterize the degree of linear correlation between the concentration of aging products and the degree of polymerization (DP) of insulating paper. Let be the concentration of aging products in the i-th sample. Let be the DP value of the i-th sample. This is the arithmetic mean of the concentrations of aging products across all samples. Let ρ be the arithmetic mean of the degree of polymerization (DP) of all insulating papers, and ρ be the Spearman rank correlation coefficient. Let be the difference between the concentration of aging products and the rank corresponding to the degree of polymerization (DP) of the insulating paper in the i-th sample, and n be the total number of samples.

[0053] The comprehensive correlation coefficient γ is constructed using the equal-weighted average fusion method: (9); The closer γ is to 1, the stronger the correlation between the feature quantity and the DP value at this stage.

[0054] Step 203: Include the stage features with a comprehensive correlation coefficient greater than a preset threshold as evaluation features in the fingerprint database.

[0055] In this embodiment, the preset threshold can be set to 0.95, that is, the stage feature quantity with γ>0.95 is determined as the evaluation feature quantity. Through the above screening, the stage feature quantities that have a strong correlation with the DP value can be retained for subsequent DP value evaluation, such as cyclohexanone, 2-pentanone, and 2-heptanone, which have a good correspondence with the DP value change within their aging range.

[0056] Step 204: Construct a fingerprint database based on full-cycle features and evaluation features.

[0057] Due to the significant differences in the content of aging products in oil, the assessment is prone to large errors. To eliminate the influence of the order-of-magnitude differences between different aging characteristics on the assessment results, this paper uses the maximum-minimum normalization method to standardize the original concentration data of the full-cycle characteristics and the assessment characteristics, linearly mapping the data to the [0,1] interval. The normalization formula is as follows: (10); In the formula, These are the original measured values ​​of aging characteristic quantities. The minimum measured value of this characteristic quantity This is the maximum measured value of the characteristic quantity. These are the normalized standard values. After normalization, the DP value is used as an index, and the normalized content of each feature constitutes the standard fingerprint vector, thereby constructing the fingerprint database. The fingerprint database constructed after normalization is shown in Table 2.

[0058] Table 2 Feature Fingerprint Database

[0059] After the fingerprint database is built, the matching algorithm is crucial for predicting the aging state of oil-paper insulation. Currently, most fingerprint matching algorithms calculate Euclidean distance, but a single Euclidean distance can only characterize the differences in the vector space and is susceptible to the limitations of fluctuations in the amplitude of chemical characteristic quantities. To address this, this paper establishes a multi-distance-similarity fusion fingerprint matching model. This model integrates spatial distance, extreme value deviation, trend similarity, and statistical correlation. By using the entropy weighting method, it achieves the fusion of objective weighting and the matching algorithm, significantly improving the accuracy of fingerprint matching for oil-paper insulation aging.

[0060] In this embodiment of the disclosure, the measured fingerprint vector of the transformer oil sample to be tested is matched with the standard fingerprint vector in a multi-dimensional distance to obtain the comprehensive matching degree between the transformer oil sample to be tested and the standard fingerprint vector. Specifically, this may include: Step 301: Calculate the Euclidean distance, Manhattan distance, Chebyshev distance, and Mahalanobis distance between the measured fingerprint vector and the standard fingerprint vector, respectively.

[0061] Step 302: Map Euclidean distance, Manhattan distance, Chebyshev distance, and Mahalanobis distance to positive matching scores.

[0062] Step 303: Calculate the objective weights of each positive matching score based on the entropy weight method, which may specifically include: Construct an initial decision matrix containing the positive matching scores corresponding to Euclidean distance, Manhattan distance, Chebyshev distance and Mahalanobis distance; The initial decision matrix is ​​column-normalized to obtain the weight matrix; Calculate the information entropy of each matching metric based on the weight matrix; The objective weights corresponding to each matching metric are calculated based on the information entropy; the smaller the information entropy, the greater the corresponding objective weight.

[0063] Step 304: Calculate the overall matching degree by weighting and merging the positive matching scores according to the objective weights.

[0064] Specifically, let the fingerprint vector of the sample to be tested be... The standard fingerprint vector in the fingerprint database is , where n is the dimension of the aging feature. This paper uses four types of distance metrics and one type of similarity metric to jointly characterize the degree of matching: (1) Euclidean distance mainly reflects the straight-line distance between two vectors in the feature space, representing the overall difference: (11); In the formula, Let be the Euclidean distance between the sample to be tested and the standard fingerprint vector. This represents the normalized content value of the i-th aging characteristic in the sample to be tested. is the normalized content value of the i-th aging feature in the standard fingerprint vector, and n is the total number of dimensions of the aging features.

[0065] The smaller the Euclidean distance, the smaller the overall feature difference between the sample to be tested and the standard fingerprint, and the higher the matching degree.

[0066] (2) The Manhattan distance mainly represents the sum of the absolute deviations of each dimension, and is highly stable and insensitive to outliers: (12); In the formula, The Manhattan distance between the sample to be tested and the standard fingerprint vector. This represents the normalized content value of the i-th aging characteristic in the sample to be tested. is the normalized content value of the i-th aging feature in the standard fingerprint vector, and n is the total number of dimensions of the aging features.

[0067] The smaller the Manhattan distance, the smaller the overall deviation of each feature, and the better the matching effect.

[0068] (3) Chebyshev distance focuses only on the maximum single-dimensional deviation, which can sensitively identify aging anomalies: (13); In the formula, The Chebyshev distance between the sample to be tested and the standard fingerprint vector is given. This represents the normalized content value of the i-th aging characteristic in the sample to be tested. is the normalized content value of the i-th aging feature in the standard fingerprint vector, and n is the total number of dimensions of the aging features.

[0069] The smaller the maximum deviation, the less obvious the abnormal features of the sample, and the higher the matching reliability.

[0070] (4) Mahalanobis distance takes into account the correlation between features and eliminates the influence of dimensional coupling, making it suitable for joint matching of multiple chemical products: (14); In the formula, Let X be the Mahalanobis distance between the test sample and the standard fingerprint vector, X be the measured fingerprint vector of the test sample, and S be the covariance matrix of the aging features. This represents the standard fingerprint vector from the fingerprint database. The smaller the Mahalanobis distance, the smaller the difference after deducting feature coupling, and the more accurate the match.

[0071] To standardize evaluation criteria and achieve objective weighting, distance-based indicators are mapped to positive indicators as follows: (15); In the formula, This represents the positive matching score corresponding to the Euclidean distance. Let be the Euclidean distance between the sample to be tested and the standard fingerprint vector. The Manhattan distance between the sample to be tested and the standard fingerprint vector. The Chebyshev distance between the sample to be tested and the standard fingerprint vector is... The Mahalanobis distance between the sample to be tested and the standard fingerprint vector. The positive matching score corresponding to the Manhattan distance. This represents the positive matching score corresponding to Chebyshev. This represents the positive matching score corresponding to the Mahalanobis distance.

[0072] Let m be the number of samples to be tested. The initial decision matrix for the five metrics is constructed as follows: (16); In the formula, Let be the initial decision matrix. Each column in the matrix corresponds to a matching metric, and each row corresponds to a sample to be tested. To eliminate the difference between the units and numerical magnitudes, the decision matrix is ​​column-normalized to obtain the weight matrix P of each sample under the corresponding metric, calculated as follows: (17); In the formula, For the element in the i-th row and j-th column of the decision matrix, Let represent the proportion of the i-th sample under the j-th metric, and m represent the total number of types of stage-specific feature products. For the j-th matching metric, its information entropy is... The calculation formula is: (18); In the formula, This is an adjustment coefficient used to normalize the entropy value to the [0,1] interval. Let represent the proportion of the i-th sample under the j-th metric, and m represent the total number of types of stage-specific characteristic products. The smaller the information entropy, the higher the dispersion of the metric's results, the stronger its ability to distinguish different aging states, and the higher its decision-making value. Finally, the objective weight corresponding to the j-th matching metric... for: (19); In the formula, The objective weight corresponding to the j-th matching metric. Let be the information entropy corresponding to the j-th matching metric.

[0073] This yields the objective fusion weights for the four types of matching metrics. Finally, the four positive matching metrics are weighted and fused according to the weights obtained by the entropy weight method to obtain the comprehensive matching degree: (20); In the formula, The overall matching degree between the test sample and the standard fingerprint vector. To match the objective fusion weights corresponding to the metrics. This represents the positive matching score corresponding to the Euclidean distance. The positive matching score corresponding to the Manhattan distance. This represents the positive matching score corresponding to Chebyshev. This represents the positive matching score corresponding to the Mahalanobis distance.

[0074] The higher the overall matching degree, the more similar the aging fingerprint of the sample under test is to a certain standard state in the fingerprint database, and the more reliable the corresponding aging degree, aggregation degree, and service life determination results are. By comparing fingerprints in the fingerprint information database and selecting the one with the highest overall matching degree, the aging state of transformer oil-paper insulation can be accurately identified.

[0075] Step 104: Select the degree of polymerization of the insulating paper corresponding to the standard fingerprint vector with the highest comprehensive matching degree as the evaluation result of the aging state of the insulating paper of the transformer oil sample to be tested.

[0076] That is, by traversing the standard fingerprint vectors in the fingerprint database corresponding to the aging stage of the oil sample to be tested, calculating each comprehensive matching degree and selecting the one with the largest comprehensive matching degree, the DP value of the standard fingerprint vector corresponding to the largest comprehensive matching degree is the evaluation result of the aging state of the insulation paper of the transformer oil sample to be tested, and thus the aging degree, remaining life and whether the transformer needs to be taken out of service can be determined.

[0077] To verify the accuracy and effectiveness of the proposed oil-paper insulation fingerprint database and multi-dimensional fusion fingerprint matching algorithm, this section conducts verification experiments based on laboratory accelerated thermal aging test data to systematically test the accuracy and versatility of the fingerprint database and evaluation model. This verification experiment uses three different types of aged insulating oil-paper samples (No. 45 oil and non-aqueous insulating paper, No. 45 oil and aqueous insulating paper, and No. 25 oil and non-aqueous insulating paper) subjected to accelerated thermal aging at 130℃ as the research object. Independent test samples not involved in the fingerprint database construction were extracted from the full-cycle experimental data as the verification set. The samples cover 6 aging stages, totaling 12 oil samples, as shown in Table 3. Samples 1-6 are No. 45 oil and non-aqueous insulating paper samples, samples 7-9 are No. 45 oil and aqueous insulating paper samples, and samples 10-12 are No. 25 oil and non-aqueous insulating paper samples.

[0078] The verification process is as follows: First, the full-cycle aging characteristics and stage characteristics of the test oil sample are extracted; second, the maximum-minimum normalization method is used to map the original concentration data to the [0,1] interval to complete the data standardization process; then, the aging stage of the sample is located by the stage characteristics, and then the aging degree is evaluated by a multi-dimensional fusion matching algorithm. The predicted results are compared with the measured aging time and the actual value of the polymerization degree DP of the insulating paper to verify the effectiveness of the method.

[0079] Table 3. Experimental Prediction Results

[0080] Polymerization degree was predicted for six groups of oil-paper insulation samples at different aging stages. The results showed that the average relative error between the predicted and measured values ​​was only 4.76%, and the prediction error for each sample was controlled within 10%, indicating high overall prediction accuracy. This demonstrates that the aging fingerprint database and multi-dimensional matching algorithm constructed in this paper can accurately characterize the aging and deterioration degree of oil-paper insulation, providing reliable technical support for the condition assessment of transformer oil-paper insulation.

[0081] Based on the above Figure 1 The specific implementation of the method shown in this embodiment provides a device for assessing the aging state of oil-paper insulation, such as... Figure 7 As shown, the device includes: an acquisition module 31, a determination module 32, a matching module 33, and a selection module 34; The acquisition module 31 is used to acquire the content data of various aging characteristics dissolved in the transformer oil sample to be tested, and to normalize the content data to obtain a standardized feature vector; wherein, the various aging characteristics include full-cycle characteristics and stage characteristics; the full-cycle characteristics are aging characteristic products continuously detected throughout the full aging cycle of the oil-paper insulation; the stage characteristics are aging characteristic products detected at a specific aging stage of the oil-paper insulation. The determination module 32 is used to determine the aging stage of the transformer oil sample under test based on the stage feature quantity in the standardized feature vector; The matching module 33 is used to extract the standard fingerprint vector corresponding to the aging stage from the pre-built fingerprint database, and perform multi-dimensional distance matching between the measured fingerprint vector of the transformer oil sample to be tested and the standard fingerprint vector to obtain the comprehensive matching degree between the transformer oil sample to be tested and the standard fingerprint vector; wherein, each standard fingerprint vector in the fingerprint database includes the normalized content data of the full-cycle feature quantity and the stage feature quantity after correlation screening under different aging degrees. Module 34 is selected to select the degree of polymerization of the insulating paper corresponding to the standard fingerprint vector with the highest comprehensive matching degree as the evaluation result of the aging state of the insulating paper of the transformer oil sample to be tested.

[0082] In a specific application scenario, module 31 obtains the following full-cycle characteristic quantities: at least one of methanol, ethanol, n-propanol, n-butanol, formaldehyde, acetaldehyde, furfural, formic acid, acetic acid, propionic acid, and butyric acid; and the following stage characteristic quantities: 2-methylfuran, furfuryl alcohol, methyl acetate, methyl formate, cyclohexanone, 2-pentanone, and 2-heptanone.

[0083] In a specific application scenario, the determining module 32 can be used to determine the detection status of each stage feature product at the current sampling point based on the detection limit of the stage feature quantity. The detection status includes a detection status where the stage feature product is detected at the sampling point and an undetected status where the stage feature product is not detected at the current sampling point. The component of the stage feature product in the detected state at the current sampling point is assigned a first value, and the component of the stage feature product in the undetected state at the current sampling point is assigned a second value. The assigned components are combined according to a preset arrangement order to construct the product type status vector of the sampling point. Wherein, the first value is not... Similar to the second value; obtain the product type state vector of the current sampling point and the product type state vector of the previous sampling point that is temporally adjacent to the current sampling point; calculate the absolute value of the difference between the product type state vector of the current sampling point and the product type state vector of the previous sampling point to obtain the absolute value of the state change corresponding to the characteristic products of each stage; sum all the absolute values ​​of the state change, and determine the sum as the type change discriminant variable corresponding to the current sampling point; if the type change discriminant variable is greater than a preset type change threshold, it is determined that there is a stage characteristic product type change between adjacent sampling points, and the time node corresponding to the type change discriminant variable is determined as... Candidate stage boundary; wherein, the numerical value of the category change discriminant variable characterizes the overall change in the detection status of stage feature products between adjacent sampling points; taking the candidate stage boundary as the dividing point, a first preset number of samples are selected before the candidate stage boundary as the front sample group, and a second preset number of samples are selected after the candidate stage boundary as the back sample group; the arithmetic mean of the normalized content of each type of stage feature product in the front sample group is calculated to obtain the front average normalized content; at the same time, the arithmetic mean of the normalized content of each type of stage feature product in the back sample group is calculated to obtain the back average normalized content; for each type of stage feature product, the back average normalized content is calculated. The difference between the normalized content and the preceding average normalized content is calculated, and the difference is squared to obtain the content difference square value corresponding to each type of stage characteristic product. The content difference square values ​​corresponding to all stage characteristic products are summed to obtain the total difference square. The square root of the total difference square is then taken to determine the overall content difference index value before and after the candidate stage boundary. The magnitude of the overall content difference index value represents the comprehensive deviation of the normalized content of all stage characteristic products before and after the candidate stage boundary. Based on the candidate stage boundary and the overall content difference index value, the aging stage of the transformer oil sample to be tested is determined.

[0084] In a specific application scenario, the determining module 32 can be used to compare the magnitudes of the overall content difference index values ​​corresponding to the multiple adjacent time nodes if they are simultaneously determined as candidate stage boundaries; select the candidate stage boundary with the largest overall content difference index value as the aging stage boundary; divide the aging cycle into intervals according to the aging stage boundaries to obtain the aging stage of the transformer oil sample to be tested; wherein, the aging stages include, in chronological order, a first stage, a second stage, a third stage, a fourth stage, a fifth stage, and a sixth stage, each stage being separated by adjacent valid stage boundaries, and the stage characteristic product combinations of each stage satisfy the following progressive relationship: the stage characteristic product combination corresponding to the first stage is an empty set; the stage characteristic product combination corresponding to the second stage only contains the first type of stage characteristic product; the stage characteristic product combination corresponding to the third stage, in addition to containing the first type of stage characteristic product, adds the second type of stage characteristic product; the stage characteristic product combination corresponding to the fourth stage... The phased feature product combination, based on the first and second types of phased feature products, adds a third type of phased feature product; the phased feature product combination corresponding to the fifth stage, based on the first, second, and third types of phased feature products, adds a fourth type of phased feature product; the phased feature product combination corresponding to the sixth stage, based on the first, second, third, and fourth types of phased feature products, adds a fifth type of phased feature product, and the normalized content of the fifth type of phased feature product changes at a rate greater than a preset rate of change threshold; wherein, the first, second, third, fourth, and fifth types of phased feature products are all phased feature products corresponding to the phased feature quantities in the standardized feature vector, and each type of phased feature product is different from the others.

[0085] In specific application scenarios, the matching module 33 can be used to obtain the full-cycle feature quantity and stage feature quantity of the oil-paper insulation aging sample; calculate the comprehensive correlation coefficient between each stage feature quantity and the degree of polymerization of the insulation paper; wherein, the comprehensive correlation coefficient includes Kendall's rank correlation coefficient, Pearson's linear correlation coefficient and Spearman's rank correlation coefficient; use the stage feature quantity whose comprehensive correlation coefficient is greater than a preset threshold as the evaluation feature quantity; and construct the fingerprint database based on the full-cycle feature quantity and the evaluation feature quantity.

[0086] In specific application scenarios, the matching module 33 can be used to calculate the Euclidean distance, Manhattan distance, Chebyshev distance, and Mahalanobis distance between the measured fingerprint vector and the standard fingerprint vector, respectively; map the Euclidean distance, Manhattan distance, Chebyshev distance, and Mahalanobis distance to positive matching scores; calculate the objective weight of each positive matching score based on the entropy weight method; and perform weighted fusion calculation on the positive matching scores according to the objective weights to obtain the comprehensive matching degree.

[0087] In a specific application scenario, the matching module 33 can be used to construct an initial decision matrix containing the positive matching scores corresponding to the Euclidean distance, the Manhattan distance, the Chebyshev distance, and the Mahalanobis distance; perform column normalization on the initial decision matrix to obtain a weight matrix; calculate the information entropy of each matching metric based on the weight matrix; and calculate the objective weight corresponding to each matching metric based on the information entropy; wherein, the smaller the information entropy, the larger the corresponding objective weight.

[0088] It should be noted that other corresponding descriptions of the functional units involved in the oil-paper insulation aging condition assessment device provided in this embodiment can be found in [reference]. Figure 1 The corresponding description in [the document] will not be repeated here.

[0089] Based on the above, Figure 1 Accordingly, this embodiment also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method. Figure 1 The method shown.

[0090] Based on this understanding, the technical solution of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as CD-ROM, USB flash drive, mobile hard drive, etc.) and includes several instructions to cause a computer device (such as personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of this application.

[0091] Based on the above, Figure 1 The method shown, and Figure 7 To achieve the above objectives, the present application also provides an electronic device, comprising a storage medium and a processor; the storage medium for storing a computer program; and the processor for executing the computer program to implement the above-described virtual system embodiment. Figure 1 The method shown.

[0092] Optionally, the aforementioned physical devices may also include a user interface, a network interface, a camera, radio frequency (RF) circuitry, sensors, audio circuitry, a Wi-Fi module, etc. The user interface may include a display screen, input units such as a keyboard, etc., and optional user interfaces may also include USB interfaces, card reader interfaces, etc. The network interface may optionally include standard wired interfaces, wireless interfaces (such as Wi-Fi interfaces), etc.

[0093] Those skilled in the art will understand that the physical device structure provided in this embodiment does not constitute a limitation on the physical device, and may include more or fewer components, or combine certain components, or have different component arrangements.

[0094] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the aforementioned physical device, supporting the operation of the oil-paper insulation aging condition assessment program and other software and / or programs. The network communication module is used to enable communication between the various components within the storage medium, as well as communication with other hardware and software in the oil-paper insulation aging condition assessment physical device.

[0095] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platform, or it can be implemented by hardware.

[0096] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the term "comprising" or any other variations thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0097] The above are merely specific embodiments of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to these embodiments, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for assessing the aging condition of oil-paper insulation, characterized in that, include: The content data of various aging characteristics dissolved in the transformer oil sample to be tested are obtained, and the content data are normalized to obtain a standardized feature vector; wherein, the various aging characteristics include full-cycle characteristics and stage characteristics; the full-cycle characteristics are aging characteristic products continuously detected throughout the entire aging cycle of the oil-paper insulation; the stage characteristics are aging characteristic products detected at a specific aging stage of the oil-paper insulation. Based on the stage feature quantities in the standardized feature vector, the aging stage of the transformer oil sample to be tested is determined. The standard fingerprint vector corresponding to the aging stage is extracted from the pre-constructed fingerprint database, and the measured fingerprint vector of the transformer oil sample to be tested is matched with the standard fingerprint vector in a multi-dimensional distance to obtain the comprehensive matching degree between the transformer oil sample to be tested and the standard fingerprint vector; wherein, each standard fingerprint vector in the fingerprint database includes the normalized content data of the full-cycle feature quantity and the stage feature quantity after correlation screening under different aging degrees. The degree of polymerization of the insulating paper corresponding to the standard fingerprint vector with the highest comprehensive matching degree is selected as the evaluation result of the aging state of the insulating paper of the transformer oil sample to be tested.

2. The method for assessing the aging state of oil-paper insulation according to claim 1, characterized in that, The full-cycle characteristic quantities include at least one of methanol, ethanol, n-propanol, n-butanol, formaldehyde, acetaldehyde, furfural, formic acid, acetic acid, propionic acid, and butyric acid; the stage characteristic quantities include 2-methylfuran, furfuryl alcohol, methyl acetate, methyl formate, cyclohexanone, 2-pentanone, and 2-heptanone.

3. The method for assessing the aging state of oil-paper insulation according to claim 1, characterized in that, The determination of the aging stage of the transformer oil sample under test based on the stage feature quantities in the standardized feature vector includes: For each sampling point, based on the detection limit of the stage characteristic quantity, the detection status of each stage characteristic product in the current sampling point is determined. The detection status includes the detection status of the stage characteristic product being detected in the sampling point and the non-detection status of the stage characteristic product not being detected in the current sampling point. The component of the stage feature product in the detected state corresponding to the current sampling point is assigned a first value, and the component of the stage feature product in the undetected state corresponding to the current sampling point is assigned a second value. The assigned components are combined in a preset order to construct the product type state vector of the sampling point; wherein, the first value is different from the second value. Obtain the product type state vector of the current sampling point and the product type state vector of the previous sampling point that is temporally adjacent to the current sampling point. Calculate the absolute value of the difference between the product type state vector of the current sampling point and the product type state vector of the previous sampling point to obtain the absolute value of the state change corresponding to the characteristic products of each stage. The absolute values ​​of all the state changes are summed, and the sum is determined as the type change discriminant variable corresponding to the current sampling point. If the type change discriminant variable is greater than a preset type change threshold, it is determined that there is a change in the type of stage characteristic product between adjacent sampling points, and the time node corresponding to the type change discriminant variable is determined as the candidate stage boundary. The magnitude of the type change discriminant variable represents the overall degree of change in the detection state of stage characteristic products between adjacent sampling points. Using the candidate stage boundary as the dividing point, a first preset number of samples are selected before the candidate stage boundary as the front sample group, and a second preset number of samples are selected after the candidate stage boundary as the back sample group. The arithmetic mean of the normalized content of each type of stage characteristic product in the front sample group is calculated to obtain the front average normalized content; at the same time, the arithmetic mean of the normalized content of each type of stage characteristic product in the back sample group is calculated to obtain the back average normalized content. For each type of stage characteristic product, the difference between the rear average normalized content and the front average normalized content is calculated, and the difference is squared to obtain the content difference square value corresponding to each type of stage characteristic product. The total sum of squared differences in content corresponding to all stage characteristic products is obtained by summing the squared differences. The square root of the total sum of squared differences is then taken, and the result of the square root operation is determined as the overall content difference index value before and after the candidate stage boundary. The magnitude of the overall content difference index value represents the comprehensive deviation of the normalized content of all stage characteristic products before and after the candidate stage boundary. The aging stage of the transformer oil sample to be tested is determined based on the candidate stage boundary and the overall content difference index value.

4. The method for assessing the aging status of oil-paper insulation according to claim 3, characterized in that, The step of determining the aging stage of the transformer oil sample to be tested based on the candidate stage boundary and the overall content difference index value includes: If multiple adjacent time points are simultaneously identified as the candidate stage boundaries, then the magnitudes of the overall content difference index values ​​corresponding to the multiple adjacent time points are compared. The candidate stage boundary with the largest overall content difference index value is selected as the aging stage boundary. The aging cycle is divided into intervals according to the aging stage boundaries to obtain the aging stage of the transformer oil sample to be tested. The aging stages, arranged chronologically, include a first stage, a second stage, a third stage, a fourth stage, a fifth stage, and a sixth stage. Each stage is separated by adjacent effective stage boundaries, and the combination of characteristic products of each stage satisfies the following progressive relationship: The combination of stage-specific characteristic products corresponding to the first stage is an empty set; The combination of stage-specific characteristic products corresponding to the second stage only includes the first type of stage-specific characteristic products; The combination of stage feature products corresponding to the third stage includes the first type of stage feature products, and adds a second type of stage feature products. The combination of stage feature products corresponding to the fourth stage includes the first type of stage feature products and the second type of stage feature products, and adds a third type of stage feature products. The fifth stage corresponding to the combination of stage feature products includes the first type of stage feature products, the second type of stage feature products and the third type of stage feature products, and adds a fourth type of stage feature products. The sixth stage corresponding to the stage feature product combination includes the first type of stage feature product, the second type of stage feature product, the third type of stage feature product and the fourth type of stage feature product, and adds a fifth type of stage feature product, and the normalized content of the fifth type of stage feature product changes over time at a rate greater than a preset change rate threshold. Among them, the first type of stage feature product, the second type of stage feature product, the third type of stage feature product, the fourth type of stage feature product and the fifth type of stage feature product are all stage feature products corresponding to the stage feature quantities in the standardized feature vector, and each type of stage feature product is different from the others.

5. The method for assessing the aging state of oil-paper insulation according to claim 1, characterized in that, The process of constructing the fingerprint database includes: Obtain the full-cycle and stage-specific characteristics of oil-paper insulation aging samples; Calculate the comprehensive correlation coefficient between each stage characteristic quantity and the degree of polymerization of the insulating paper; wherein, the comprehensive correlation coefficient includes Kendall's rank correlation coefficient, Pearson's linear correlation coefficient and Spearman's rank correlation coefficient; The stage feature quantity whose comprehensive correlation coefficient is greater than a preset threshold is used as the evaluation feature quantity; The fingerprint database is constructed based on the full-cycle feature quantities and the evaluation feature quantities.

6. The method for assessing the aging status of oil-paper insulation according to claim 1, characterized in that, The step of performing multi-dimensional distance matching between the measured fingerprint vector of the transformer oil sample to be tested and the standard fingerprint vector to obtain the comprehensive matching degree between the transformer oil sample to be tested and the standard fingerprint vector includes: Calculate the Euclidean distance, Manhattan distance, Chebyshev distance, and Mahalanobis distance between the measured fingerprint vector and the standard fingerprint vector, respectively. The Euclidean distance, Manhattan distance, Chebyshev distance, and Mahalanobis distance are mapped to positive matching scores; The objective weights of each positive matching score are calculated based on the entropy weight method; The positive matching score is weighted and fused according to the objective weights to obtain the comprehensive matching degree.

7. The method for assessing the aging state of oil-paper insulation according to claim 6, characterized in that, The objective weights for calculating each positive matching score based on the entropy weight method include: Construct an initial decision matrix containing the positive matching scores corresponding to the Euclidean distance, the Manhattan distance, the Chebyshev distance, and the Mahalanobis distance; The initial decision matrix is ​​column normalized to obtain the weight matrix; Calculate the information entropy of each matching metric based on the weight matrix; The objective weights corresponding to each matching metric are calculated based on the information entropy; wherein, the smaller the information entropy, the larger the corresponding objective weight.

8. A device for assessing the aging status of oil-paper insulation, characterized in that, include: The acquisition module is used to acquire the content data of various aging characteristics dissolved in the transformer oil sample to be tested, and to normalize the content data to obtain a standardized feature vector; wherein, the various aging characteristics include full-cycle characteristics and stage characteristics; the full-cycle characteristics are aging characteristic products continuously detected throughout the entire aging cycle of the oil-paper insulation; the stage characteristics are aging characteristic products detected at a specific aging stage of the oil-paper insulation. The determination module is used to determine the aging stage of the transformer oil sample under test based on the stage feature quantity in the standardized feature vector; The matching module is used to extract standard fingerprint vectors corresponding to the aging stage from a pre-built fingerprint database, and perform multi-dimensional distance matching between the measured fingerprint vector of the transformer oil sample to be tested and the standard fingerprint vector to obtain the comprehensive matching degree between the transformer oil sample to be tested and the standard fingerprint vector; wherein, each standard fingerprint vector in the fingerprint database includes the normalized content data of the full-cycle feature quantity and the stage feature quantity after correlation screening under different aging degrees. The selection module is used to select the degree of polymerization of the insulating paper corresponding to the standard fingerprint vector with the highest comprehensive matching degree as the evaluation result of the aging state of the insulating paper of the transformer oil sample to be tested.

9. An electronic device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method for assessing the aging state of oil-paper insulation as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for assessing the aging state of oil-paper insulation as described in any one of claims 1 to 7.

Images