Transformer fault diagnosis method and system

Abstract

The invention relates to the technical field of transformer fault diagnosis, in particular to a transformer fault diagnosis method and system, and the method comprises the following steps: S1, obtaining the original data of dissolved gas in transformer on-load tap-changer oil; s2, performing normalization and quantitative correction on the original data to obtain corrected gas concentration data; s3, separating a steady-state background gas component from the corrected gas concentration data to obtain net gas production data reflecting the net gas production behavior of the on-load tap-changer of the transformer; s4, identifying a characteristic gas production event related to the switching operation time of the on-load tap-changer of the transformer in the net gas production data in combination with the operation record of the on-load tap-changer of the transformer, and extracting gas production information of the characteristic gas production event; s5, constructing a characteristic gas production event sequence by using the plurality of characteristic gas production events according to a time sequence; and S6, judging the internal fault state of the on-load tap-changer of the transformer. And the diagnosis accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer fault diagnosis, and particularly to a transformer fault diagnosis method and system. Background Art

[0002] In the power system, the on-load tap-changer of an oil-immersed transformer, as an important device for voltage regulation, has a crucial impact on the safe and stable operation of the transformer.

[0003] In the prior art, the on-load tap-changer fault diagnosis method based on dissolved gas analysis in oil usually relies on the oil sample data obtained regularly from fixed sampling positions. However, due to the limitations of the sampling positions and the non-uniformity of the oil flow inside the oil chamber, the local gas production characteristics are often difficult to diffuse to the sampling area in a timely and sufficient manner, resulting in the gas concentration data collected being difficult to comprehensively reflect the true gas production state of the equipment and reducing the accuracy of the diagnosis results. Summary of the Invention

[0004] The object of the present invention is to propose a transformer fault diagnosis method and system for the above-mentioned existing deficiencies.

[0005] The present invention adopts the following technical solutions: The present application provides a transformer fault diagnosis method, which includes the following steps: S1: Obtain the original data of the dissolved gases in the oil of the on-load tap-changer of the transformer; S2: Perform normalization and quantization correction on the distortion caused by gas escape, oil replenishment operation, and oil temperature fluctuation in the original data to obtain the corrected gas concentration data; S3: Separate the steady-state background gas components from the corrected gas concentration data to obtain the net gas production data reflecting the net gas production behavior of the on-load tap-changer of the transformer; S4: Combine the operation records of the on-load tap-changer of the transformer, identify the characteristic gas production events related to the switching operation time of the on-load tap-changer of the transformer in the net gas production data, and extract the gas production information of the characteristic gas production events; S5: Construct a characteristic gas production event sequence for multiple characteristic gas production events in chronological order; S6: Based on the behavior pattern of the characteristic gas production event sequence, compare and analyze it with the preset known fault event sequence pattern library to judge the internal fault state of the on-load tap-changer of the transformer.

[0006] Through the above solution, the accuracy and reliability of the on-load tap-changer fault diagnosis of the transformer are improved, and problems such as original data distortion and background interference are overcome.

[0007] Optionally, the present application also proposes that step S Identify the characteristic gas generation events associated with the operation sequence of a specific on-load tap-changer of a transformer in the gas generation event sequence of identification features; For the characteristic gas generation events associated with the operation sequence, analyze the change amounts of the characteristic gas components indicating different potential defect types included therein, and obtain the change amount data of each characteristic gas component; Based on the change amount data of each characteristic gas component, the type of the operation sequence, and the expected gas change response characteristics of each single fault mode in the preset known fault event sequence pattern library under the corresponding operation excitation, determine whether there is a gas generation contribution from multiple defect sources in the characteristic gas generation event, and obtain the determination result of the gas generation contribution from multiple defect sources; Based on the determination result, distinguish the existence and nature of each independent defect source in the concurrent fault, obtain the distinction result of each independent defect source, and combine the distinction result with the overall pattern of the characteristic gas generation event sequence to complete the comparison and analysis with the preset fault event pattern library, so as to determine the internal fault state of the on-load tap-changer of the transformer.

[0008] Through the above solution, the independent defect sources in the concurrent fault can be further identified, and the diagnostic fineness is improved.

[0009] Optionally, the present application also proposes that the steps of determining whether there is a gas generation contribution from multiple defect sources in the characteristic gas generation event include: Based on the change amount data of each characteristic gas component, analyze the characterization information indicating the non-linear coupling effect between different characteristic gas components, and obtain the coupling effect characterization information; Combine the coupling effect characterization information, the type of the operation sequence, and the expected gas change response characteristics of each single fault mode in the known fault event sequence pattern library under the corresponding operation excitation, and process the change amount data of each characteristic gas component to separate or estimate the adjusted gas change amounts respectively contributed by different potential defect sources after considering the non-linear coupling effect, and obtain the adjusted gas change amounts of each defect source; According to the adjusted gas change amounts of each defect source, determine whether there is a gas generation contribution from multiple defect sources in the characteristic gas generation event, and obtain the determination result of the gas generation contribution from multiple defect sources.

[0010] Through the above solution, the non-linear coupling effect between gas components is considered, making the determination of the gas generation contribution from different defect sources more accurate.

[0011] Optionally, the present application also proposes that the steps of obtaining the coupling effect characterization information include: Obtain the current service information of the on-load tap-changer of the transformer associated with the characteristic gas generation event, where the current service information reflects the service duration or historical cumulative operation conditions of the on-load tap-changer of the transformer; According to the current service information, determine the reference characteristics of the non-linear coupling effect applicable to the current state of the on-load tap-changer of the transformer from a preset data set, where the data set includes the reference characteristics of the non-linear coupling effect corresponding to different service stages or different historical cumulative operating conditions; Based on the change amount data of the characteristic gas components, and compare it with the reference characteristics of the non-linear coupling effect to analyze and obtain the characterization information indicating the non-linear coupling effect between different characteristic gas components, and the characterization information is used to reflect the current state of the on-load tap-changer of the transformer.

[0012] Through the above solution, introducing the service information to determine the reference characteristics of the non-linear coupling effect makes the coupling effect analysis more in line with the actual state.

[0013] Optionally, the present application also proposes that the steps of analyzing and obtaining the characterization information indicating the non-linear coupling effect between different characteristic gas components based on the change amount data of the characteristic gas components and comparing it with the reference characteristics of the non-linear coupling effect include: Express the change amount data of each characteristic gas component as an actual gas behavior pattern; Take the expected gas behavior pattern included in the reference characteristics of the non-linear coupling effect as the reference pattern; Based on the actual gas behavior pattern and the reference pattern, calculate the quantitative difference between the two; According to the quantitative difference and the preset discrimination logic related to the non-linear coupling effect, and combined with the actual gas behavior pattern and the reference pattern, analyze and obtain the coupling effect characterization information indicating the non-linear coupling effect between different characteristic gas components and capable of reflecting the current state of the on-load tap-changer of the transformer.

[0014] Through the above solution, analyzing the actual and reference gas behavior patterns through the quantitative difference can more accurately characterize the non-linear coupling effect.

[0015] Optionally, the present application also proposes that the steps of expressing the change amount data of each characteristic gas component as an actual gas behavior pattern include: Obtain the change amount data of each characteristic gas component; Extract the statistical characteristic parameters within a predetermined time window from the change amount data to obtain a set of statistical characteristic parameters; Take the set of statistical characteristic parameters as the actual gas behavior pattern.

[0016] Through the above solution, using the set of statistical characteristic parameters to express the actual gas behavior pattern simplifies the analysis process.

[0017] Optionally, the present application also proposes that the steps of analyzing and obtaining the coupling effect characterization information include: Compare the quantified differences with preset thresholds or rules in the discrimination logic; Based on the comparison results, as well as the characteristics of the actual gas behavior pattern and the reference pattern, and in combination with the discrimination logic, determine one or more types of non - linear coupling effects from the preset set of non - linear coupling effect types; Use the determined non - linear coupling effect type as the coupling effect characterization information that can reflect the current state of the on - load tap - changer of the transformer.

[0018] Through the above - mentioned solution, determining the coupling effect type based on the quantified differences and the discrimination logic improves the accuracy of the coupling effect characterization.

[0019] Optionally, the present application also proposes that the steps for analyzing the non - linear coupling effect type include: Analyze the quantified differences and the characteristics of the actual gas behavior pattern and the reference pattern, identify the characteristics indicating specific gas reactions or solubility changes, and obtain relevant characteristics; Compare and analyze the relevant characteristics with the preset characteristic library of single - gas reaction or solubility change patterns to obtain the comparison and analysis results; According to the comparison and analysis results, judge whether the differences between the actual gas behavior pattern and the reference pattern can be explained by the combination of one or more preset single - reaction patterns; Based on the judgment results, determine one or more non - linear coupling effect types that can reflect the actual complex coupling situation.

[0020] Through the above - mentioned solution, by comparing with the single - reaction pattern characteristic library, deeply analyze the internal mechanism of the coupling effect.

[0021] Optionally, the present application also proposes that the steps for obtaining the adjusted gas change amounts of each defect source include: Obtain the type of the on - load tap - changer operation sequence associated with the characteristic gas - generating event; Based on the type of the operation sequence, determine the expected change characteristics of the local environmental conditions related to the non - linear coupling effect in the transformer oil chamber to obtain the expected change characteristics of the local environmental conditions; According to the expected change characteristics of the local environmental conditions, correct the parameters characterizing the non - linear coupling effect to obtain the adjusted non - linear coupling effect characterization parameters; Based on the adjusted non - linear coupling effect characterization parameters, and in combination with the coupling effect characterization information, the type of the on - load tap - changer operation sequence, and the expected gas change response characteristics of each single - fault mode in the known fault event sequence pattern library under the corresponding operation excitation, process the change amount data of each characteristic gas component to separate or estimate the adjusted gas change amounts respectively contributed by different potential defect sources considering the non - linear coupling effect, and obtain the adjusted gas change amounts of each defect source.

[0022] Through the above solution, considering the influence of the operation sequence on the local environment, the characterization parameters of the non-linear coupling effect are corrected, and the accuracy of estimating the gas production of the defect source is improved.

[0023] Optionally, the present application also provides a transformer fault diagnosis system applied to the above-mentioned transformer fault diagnosis method. The system includes: A data processing module, which acquires the original data of the dissolved gases in the oil of the on-load tap-changer of the transformer, and performs normalization and quantization correction on the distortion caused by gas escape, oil replenishment operation, and oil temperature fluctuation in the original data to obtain the corrected gas concentration data; A background gas separation module, which separates the steady-state background gas components from the corrected gas concentration data to generate net gas production data reflecting the net gas production behavior of the on-load tap-changer of the transformer; A characteristic event recognition module, which combines the operation records of the on-load tap-changer of the transformer to identify characteristic gas production events related to the switching operation time of the on-load tap-changer of the transformer in the net gas production data, and extracts the gas production information of the characteristic gas production events; An event sequence construction module, which constructs a characteristic gas production event sequence by arranging multiple characteristic gas production events in chronological order; A fault state analysis module, which compares and analyzes the behavior pattern of the characteristic gas production event sequence with a preset known fault event sequence pattern library to judge the internal fault state of the on-load tap-changer of the transformer.

[0024] Through the above solution, a system for implementing the above diagnosis method is provided, which is convenient for practical application.

[0025] To enable a further understanding of the features and technical content of the present invention, please refer to the following detailed description of the present invention and the accompanying drawings. However, the provided drawings are only for reference and illustration, and are not used to limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 is a flowchart of the method of the present invention; Figure 2 is a schematic diagram of the overall structure of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] The following are specific embodiments to illustrate the implementation manners of the present invention. Those skilled in the art can understand the advantages and effects of the present invention from the content disclosed in this specification. The present invention can be implemented or applied through other different specific embodiments, and various details in this specification can also be modified and changed based on different viewpoints and applications without departing from the spirit of the present invention. Additionally, the drawings of the present invention are only for simple schematic illustration and are not drawn according to actual dimensions. The following embodiments will further detail the related technical content of the present invention, but the disclosed content is not used to limit the protection scope of the present invention.

[0028] This embodiment provides a transformer fault diagnosis method and system, as shown in combination with Figure 1 and Figure 2 shown.

[0029] Referring to Figure 1 , a transformer fault diagnosis method, the method includes the following steps: S1: Obtain the original data of the dissolved gases in the oil of the on-load tap-changer of the transformer; S2: Normalize and quantify the distortion caused by gas escape, oil replenishment operation, and oil temperature fluctuation in the original data to obtain the corrected gas concentration data; S3: Separate the steady-state background gas components from the corrected gas concentration data to obtain the net gas production data reflecting the net gas production behavior of the on-load tap-changer of the transformer; S4: Combine the operation records of the on-load tap-changer of the transformer, identify the characteristic gas production events related to the switching operation time of the on-load tap-changer of the transformer in the net gas production data, and extract the gas production information of the characteristic gas production events; S5: Construct a characteristic gas production event sequence for multiple characteristic gas production events in chronological order; S6: Based on the behavior pattern of the characteristic gas production event sequence, compare and analyze it with the preset known fault event sequence pattern library to judge the internal fault state of the on-load tap-changer of the transformer.

[0030] Among them, normalizing and quantifying the distortion caused by gas escape, oil replenishment operation, and oil temperature fluctuation in the original data means processing the initially obtained gas concentration data to eliminate or reduce the data deviation caused by non-fault factors. It can be achieved by using a correction model established based on historical data, a compensation algorithm based on physical principles, or dynamically adjusting in combination with real-time monitoring data. For example, the influence of oil temperature change curve on gas solubility can be compensated, or the gas concentration can be diluted and corrected according to the oil replenishment record. Its main purpose is to obtain gas concentration data closer to the real gas production situation and provide a reliable basis for subsequent analysis.

[0031] Separating the steady-state background gas components from the calibrated gas concentration data means identifying and removing those gas components that are long-term or slowly changing and unrelated to the on-load tap-changer switching operations. This can be achieved using techniques such as time series analysis, statistical filtering, or prediction models based on historical trends. For example, the long-term average value or trend line of the gas concentration can be calculated as the background component for deduction, mainly to highlight the net gas production caused by the instantaneous or short-term behavior of the on-load tap-changer and focus on the gas production events related to the operation.

[0032] Identifying the characteristic gas production events related to the switching operation time of the transformer on-load tap-changer means locating and marking the significant gas concentration changes that occur within a specific time window before and after the on-load tap-changer performs the switching operation in the net gas production data. This can be achieved using techniques such as setting a time window, threshold detection, or rate-of-change analysis. For example, a fixed time period can be set before and after the occurrence time point of each switching operation, and the gas concentration peak or steep change can be searched within this time period. The main purpose is to associate the gas production behavior with specific equipment operations and distinguish normal operation gas production from abnormal fault gas production.

[0033] Extracting the gas production information of the characteristic gas production events means obtaining the key data used to describe the event from the identified characteristic gas production events, which can include parameters such as gas type, concentration change amount, change rate, and duration. The main purpose is to quantify the characteristics of each operation-related gas production event and provide data input for subsequent pattern construction and comparison analysis.

[0034] Constructing a sequence of characteristic gas production events in chronological order for multiple characteristic gas production events means arranging all the identified characteristic gas production events within a period of time in the order of their occurrence time to form an ordered data set. The main purpose is to reflect the dynamic evolution process of the gas production behavior of the on-load tap-changer during continuous operation and multiple operations, and reveal potential fault development trends or patterns.

[0035] Comparing and analyzing the behavior pattern based on the sequence of characteristic gas production events with a pre-set pattern library of known fault event sequences to judge the internal fault state of the transformer on-load tap-changer means comparing the overall characteristics, change rules, or specific combinations shown by the constructed sequence of characteristic gas production events with the gas production event sequences representing different known fault types (such as poor contact, partial discharge, etc.) established in advance. This can be achieved using techniques such as pattern recognition algorithms, machine learning models, or rule matching logics. For example, the similarity or distance between sequences can be calculated, or the pre-set fault rules can be triggered according to specific event combinations in the sequence. The main purpose is to diagnose whether there is a specific type of internal fault in the on-load tap-changer by matching the gas production behavior pattern.

[0036] The solution of this application normalizes and quantifies the original data of dissolved gases in the oil of the on-load tap-changer of the transformer, eliminates the interference of factors such as gas escape, oil replenishment, and oil temperature fluctuation, and obtains more real gas concentration data; further separates the steady-state background gas, extracts the net gas production data related to the tap-changer operation; combines the operation records to identify characteristic gas production events and constructs a time series, and judges the internal fault state by comparing with the known fault event pattern library, realizing hierarchical and progressive, accurate and reliable fault diagnosis.

[0037] In some preferred embodiments, obtaining the original data of dissolved gases in the oil of the on-load tap-changer of the transformer can be achieved through an on-line monitoring system or by taking oil samples regularly for off-line gas chromatography analysis. When correcting the original data, a multivariate regression model including parameters such as oil temperature, oil replenishment amount, and ambient temperature can be established, and the measured gas concentration can be corrected according to these parameters. When separating the steady-state background gas components, the moving average method or the exponential smoothing method can be used to calculate the long-term trend of the gas concentration as the background, and this background value can be subtracted from the corrected data. When identifying characteristic gas production events, a time window of, for example, 30 minutes can be set before and after the time point of each on-load tap-changer switching operation, and whether there is an increment or change rate of gas concentration exceeding the preset threshold is detected in the net gas production data within this window. If it exists, it is marked as a characteristic gas production event, and the time when the event occurs, the types of gases involved, and the change amount of their concentrations are recorded. When constructing the sequence of characteristic gas production events, all the characteristic gas production events identified within a period of time (such as one month or one year) can be arranged in chronological order to form an ordered list. When performing the comparison analysis, machine learning algorithms such as support vector machines or neural networks can be used, and the statistical characteristics of the sequence of characteristic gas production events (such as event frequency, the probability of different gas combinations appearing, the distribution of concentration change ranges, etc.) are used as inputs, and compared with the trained classification models representing different fault modes (such as slight contact wear, local overheating, initial discharge, etc.), and the judgment result of the most matching fault type is output.

[0038] Through multiple data corrections and background gas separation, this solution improves the authenticity and pertinence of gas data; constructs a gas production event sequence in combination with operation records to accurately capture the fault evolution; realizes accurate judgment of the fault state based on sequence pattern comparison, effectively overcomes the limitations of traditional methods relying on isolated data, and improves the accuracy and reliability of diagnosis.

[0039] This application further proposes that step S6 specifically includes the following steps: Identify the characteristic gas production events associated with the operation sequence of a specific on-load tap-changer of the transformer in the sequence of characteristic gas production events; For a characteristic gas generation event associated with an operation sequence, analyze the change amounts of characteristic gas components that indicate different potential defect types included therein, and obtain the change amount data of each characteristic gas component; Based on the change amount data of each characteristic gas component, the type of the operation sequence, and the expected gas change response characteristics of each single fault mode in the preset known fault event sequence pattern library under the corresponding operation excitation, determine whether there is a gas generation contribution from multiple defect sources in the characteristic gas generation event, and obtain the judgment result of the gas generation contribution from multiple defect sources; Based on the judgment result, distinguish the existence and nature of each independent defect source in the concurrent fault, obtain the distinguishing result of each independent defect source, and combine the distinguishing result with the overall pattern of the characteristic gas generation event sequence to complete the comparison and analysis with the preset fault event pattern library, so as to judge the internal fault state of the on-load tap-changer of the transformer.

[0040] Among them, a specific on-load tap-changer operation sequence of a transformer refers to a series of switching actions performed by the on-load tap-changer of the transformer within a specific time period, such as raising the tap, lowering the tap, or continuous switching, etc., which can be characterized by operation record data; the characteristic gas components indicating different potential defect types refer to the gases dissolved in the transformer oil that can reflect the abnormal state inside the tap-changer (such as overheating, discharge), such as hydrogen, methane, ethane, ethylene, acetylene, etc., which can be detected by methods such as gas chromatography analysis; the preset known fault event sequence pattern library refers to a data set established in advance that contains the characteristic gas generation behavior patterns corresponding to different known fault types (such as mild overheating, partial discharge, concurrent severe overheating and discharge, etc.) under different operation excitations, which can be constructed by historical data analysis, experimental simulation, or expert experience; the expected gas change response characteristics of each single fault mode under the corresponding operation excitation refer to the change rules or patterns that the characteristic gas components are expected to generate when a certain specific single fault type undergoes a certain specific operation sequence, which are recorded in the pattern library and can be represented by curves, vectors, or statistical parameter sets; the gas generation contribution from multiple defect sources refers to the amount of characteristic gas or its change trend independently generated by two or more different potential defect sources (for example, overheating and discharge existing simultaneously) in the same characteristic gas generation event, which can be judged by the decomposition or attribution analysis of the change amount of gas components; the existence and nature of each independent defect source in the concurrent fault refer to, after judging the existence of the gas generation contribution from multiple defect sources, further determining which specific types of defect sources (such as overheating, discharge, poor contact, etc.) exist simultaneously, and the severity or characteristics of these defect sources respectively, which can be distinguished by methods based on gas component ratio, change rate, or pattern matching.

[0041] Based on the construction of the characteristic gas generation event sequence, this solution focuses on the gas generation events related to specific operation sequences, deeply analyzes the change amounts of each characteristic gas component, and combines the type of the operation sequence and the preset pattern library to determine whether there are multiple defect sources, and further distinguishes different defect types and natures. By combining local component analysis with overall sequence pattern comparison, the accuracy and reliability of concurrent fault identification and diagnosis are improved.

[0042] In some preferred embodiments, the present application is specifically implemented as follows. First, obtain the characteristic gas generation event sequence after preprocessing, background gas separation, and characteristic event recognition and sequence construction. Assume that the sequence contains a characteristic gas generation event associated with the most recent on-load tap-changer upshift operation of the transformer. For the characteristic gas generation event related to the upshift operation, analyze the characteristic gas components contained therein, such as hydrogen, methane, ethylene, acetylene, etc., and calculate their change amounts before and after the operation to obtain the change amount data of each characteristic gas component. For example, it is detected that the change amounts of ethylene and acetylene both exceed the normal range. Then, based on these gas change amount data, combine the type of this operation being an upshift operation, and consult the preset known fault event sequence pattern library. The pattern library may record the expected response characteristics that a single overheating fault mainly causes an increase in ethylene under an upshift operation, while a single discharge fault mainly causes an increase in acetylene under an upshift operation. By comparing the situation where both ethylene and acetylene are significantly increased detected actually with the expected response in the pattern library, judge that there is a gas generation contribution from multiple defect sources in this characteristic gas generation event, such as there may be overheating and discharge simultaneously. Based on this judgment result, further distinguish the existence and nature of each independent defect source in the concurrent fault. For example, according to the relative ratio and absolute value of the change amounts of ethylene and acetylene, and combining the gas characteristics of different severity levels of overheating and discharge in the pattern library, it can be distinguished that there may be two independent defect sources: medium overheating and slight partial discharge. Finally, comprehensively analyze the discrimination result of "there is medium overheating and slight partial discharge" with the overall behavior pattern of the entire characteristic gas generation event sequence (which may include multiple operations and corresponding gas generation events), and compare it with the corresponding concurrent fault pattern in the pattern library to finally judge the internal fault state of the on-load tap-changer of the transformer as "medium overheating accompanied by slight partial discharge".

[0043] By analyzing the change of the characteristic gas components of the operation-related gas generation events and combining the overall pattern comparison, this solution can identify and distinguish the concurrent faults of multiple defect sources, effectively overcome the limitations of only relying on overall pattern comparison, and improve the accuracy of fault diagnosis under complex working conditions.

[0044] The steps for the present application to further determine whether there is a gas generation contribution from multiple defect sources in the characteristic gas generation event include: Based on the change amount data of each characteristic gas component, analyze the characterization information indicating the non-linear coupling effect between different characteristic gas components to obtain the coupling effect characterization information; Combine the coupling effect characterization information, the type of operation sequence, and the expected gas change response characteristics of each single fault mode in the known fault event sequence pattern library under the corresponding operation excitation, and process the change amount data of each characteristic gas component to separate or estimate the adjusted gas change amounts respectively contributed by different potential defect sources considering the non-linear coupling effect, so as to obtain the adjusted gas change amounts of each defect source; According to the adjusted gas change amounts of each defect source, judge whether there is a gas production contribution from multiple defect sources in the characteristic gas production event to obtain the judgment result of the gas production contribution from multiple defect sources.

[0045] Among them, the coupling effect characterization information refers to the characterization information indicating the non-linear coupling effect between different characteristic gas components, which can specifically refer to parameters or models reflecting the influence of environmental factors on gas reaction rate, solubility change, adsorption and desorption behavior, etc., and its purpose is to quantify the complex interaction between gas components; among them, the adjusted gas change amount of each defect source refers to the gas change amount actually contributed by each independent potential defect source separated or estimated from the total gas change amount through data processing methods after considering the non-linear coupling effect between gas components, and its purpose is to eliminate the interference of the coupling effect and obtain the gas production contribution of each defect source closer to the actual situation.

[0046] This solution analyzes the non-linear coupling effect between characteristic gas components, combines the type of operation sequence and known fault modes, corrects and decouples the original gas data, accurately estimates the actual gas contribution of each defect source, thereby improving the reliability and accuracy of concurrent fault identification and diagnosis.

[0047] In some preferred embodiments, specifically, when analyzing the characterization information indicating the non-linear coupling effect between different characteristic gas components, a machine learning model based on historical data can be established. The inputs of this model are the change data of each characteristic gas component, the type of operation sequence, environmental parameters such as oil temperature and oil pressure, and the output is the coupling coefficient or coupling function reflecting the degree of mutual influence between gas components. For example, a neural network model can be trained to predict the deviation between the actual change of a certain gas and the theoretical change without coupling effect under given operation and environmental conditions, and this deviation pattern can be used as the characterization information of the coupling effect. When processing the change data of each characteristic gas component to obtain the adjusted gas change of each defect source, a method based on model inversion can be adopted. For example, a joint model including multiple potential defect source gas generation models and a coupling effect model between gas components is established, and then an optimization algorithm is used to invert, with the measured gas change data as the target, the adjusted gas change independently contributed by each defect source that can best explain these measured data. For example, it can be assumed that there are two potential defect sources A and B, which generate gases G1 and G2 respectively, and there is a mutually promoting coupling effect between G1 and G2. Then the measured changes of G1 and G2 are the result of the independent gas generation of A and B and the combined effect of the coupling effect. By establishing a mathematical model to describe this process and using the measured data to solve the model parameters, the adjusted gas changes contributed by A and B respectively can be estimated.

[0048] By introducing the non-linear coupling effect analysis and correcting the gas change data, this solution can accurately separate the gas generation contributions of different defect sources, effectively overcome the interference caused by gas interaction, and improve the reliability and accuracy of concurrent fault diagnosis under complex working conditions.

[0049] The present application further proposes that the steps to obtain the characterization information of the coupling effect include: Obtain the current service information of the on-load tap-changer of the transformer associated with the characteristic gas generation event, and the current service information reflects the service duration or historical cumulative operation conditions of the on-load tap-changer of the transformer; According to the current service information, determine the non-linear coupling effect reference characteristics applicable to the current state of the on-load tap-changer of the transformer from a preset data set, and the data set contains non-linear coupling effect reference characteristics corresponding to different service stages or different historical cumulative operation conditions; Based on the change data of the characteristic gas components, and comparing it with the non-linear coupling effect reference characteristics for control processing, to analyze and obtain the characterization information indicating the non-linear coupling effect between different characteristic gas components, and the characterization information is used to reflect the current state of the on-load tap-changer of the transformer.

[0050] Among them, the current service information refers to the information reflecting the historical operating state of the on-load tap-changer of the transformer, which can be obtained by recording the operating time, recording the number of operations, recording the cumulative switched load, etc. The purpose is to quantify the aging degree or wear state of the on-load tap-changer of the transformer; the preset data set refers to the data set storing the characteristic laws of the non-linear coupling effect of different on-load tap-changers of transformers in different service states, which can be established by means of laboratory simulation, statistical analysis of historical operation data, expert experience, etc. The purpose is to provide a reference benchmark for the coupling effect analysis of the current on-load tap-changer of the transformer; the reference characteristics of the non-linear coupling effect refer to the expected laws of the mutual influence or correlation between different characteristic gas components under a specific service state of the on-load tap-changer of the transformer, which can be manifested as the proportional relationship between the change amounts of gas components, the correlation model, or the gas production mode of a specific gas combination. The purpose is to provide a coupling effect benchmark matching the current state of the on-load tap-changer of the transformer; the comparison process refers to the process of comparing and analyzing the change amount data of the actually measured characteristic gas components with the reference characteristics of the non-linear coupling effect, which can adopt methods such as calculating the difference, pattern matching, statistical test, etc. The purpose is to identify the deviation degree and pattern between the actual gas behavior and the expected reference characteristics; the characterization information indicating the non-linear coupling effect between different characteristic gas components refers to the information obtained through the comparison process and capable of quantifying or describing the degree and nature of the non-linear interaction between gas components, which can be manifested as a quantified difference value, a deviation pattern type, a coupling effect strength index, or a specific coupling effect type identifier. The purpose is to provide a quantitative or qualitative description reflecting the gas coupling effect of the current on-load tap-changer of the transformer under the current state.

[0051] This solution obtains the current service information and selects the matching reference characteristics of the non-linear coupling effect, compares the actual gas change amount with the reference benchmark, and analyzes the characterization information of the coupling effect incorporating the influence of the service state, so as to more accurately separate the gas production contributions of each defect source and improve the reliability of multi-defect source identification.

[0052] In some preferred embodiments, obtaining the current service information of the on-load tap-changer of the transformer associated with the characteristic gas generation event can be achieved by reading the total operating hours or the total number of switching operations recorded in the transformer operation and maintenance record system. According to the obtained current service information, such as the total operating hours, the reference characteristics of the non-linear coupling effect applicable to the range of the operating hours can be found or calculated from a preset data set. This data set can be a database storing the expected proportional relationships or correlation models between the gas component change amounts corresponding to different operating time periods. Based on the change amount data of the characteristic gas components, such as the change amounts of acetylene and ethylene measured, and comparing it with the determined reference characteristics of the non-linear coupling effect. The comparison process can be to compare the ratio of the actually measured change amounts of acetylene and ethylene with the expected ratio in the reference characteristics at this service stage, and calculate the difference between the two. Or, the comparison process can be to input the actual gas change amount data into the model provided by the reference characteristics and calculate the deviation degree between the actual value and the model predicted value. According to the calculated difference or deviation degree, analyze and obtain the characterization information indicating the non-linear coupling effect between different characteristic gas components. This characterization information can be a quantitative index or a qualitative description, used to reflect the gas coupling effect characteristics of the on-load tap-changer of the transformer in the current state.

[0053] The steps of this application further propose, based on the change amount data of the characteristic gas components, and comparing it with the reference characteristics of the non-linear coupling effect to analyze and obtain the characterization information indicating the non-linear coupling effect between different characteristic gas components, include: Express the change amount data of each characteristic gas component as an actual gas behavior pattern; Use the expected gas behavior pattern included in the reference characteristics of the non-linear coupling effect as the reference pattern; Based on the actual gas behavior pattern and the reference pattern, calculate the quantitative difference between the two; According to the quantitative difference and the preset discrimination logic related to the non-linear coupling effect, and combining the actual gas behavior pattern and the reference pattern, analyze and obtain the coupling effect characterization information indicating the non-linear coupling effect between different characteristic gas components and capable of reflecting the current state of the on-load tap-changer of the transformer.

[0054] Among them, the actual gas behavior pattern refers to the expression form obtained by abstracting and refining the change amount data of each characteristic gas component, which can reflect the overall characteristics and mutual relationships of the gas component changes. It can be implemented by using a set of statistical characteristic parameters, the characteristics of the time series change curve, or the feature vectors extracted based on the pattern recognition algorithm. The reference pattern refers to the expression form included in the reference characteristics of the nonlinear coupling effect, which represents the gas behavior pattern that may be caused by various nonlinear coupling effects under specific working conditions. It can be implemented by using a pattern description method that matches the actual gas behavior pattern. The quantization difference refers to the numerical representation of the deviation degree between the actual gas behavior pattern and the reference pattern. It can be calculated by using the Euclidean distance, cosine similarity, or other pattern similarity measurement methods. The preset discrimination logic related to the nonlinear coupling effect refers to the set of rules or models used to judge the type or degree of the nonlinear coupling effect according to the quantization difference and the characteristics of the actual gas behavior pattern and the reference pattern. It can be implemented by using a rule base based on expert experience, a decision tree model, or a machine learning classifier. The coupling effect characterization information refers to the information used to indicate the nonlinear coupling effect between different characteristic gas components and can reflect the current state of the on-load tap-changer of the transformer. It can be implemented by using the type identifier of the nonlinear coupling effect, the coupling strength parameter, or the coupling effect influence factor.

[0055] In some preferred embodiments, expressing the change amount data of each characteristic gas component as the actual gas behavior pattern, the change amount data of each characteristic gas component within a predetermined time window can be obtained, and statistical characteristic parameters such as the average value, standard deviation, maximum value, minimum value, change rate, etc. can be extracted from these change amount data. These statistical characteristic parameters are formed into a set as the actual gas behavior pattern. Taking the expected gas behavior pattern included in the reference characteristics of the nonlinear coupling effect as the reference pattern, a typical gas change pattern library of different nonlinear coupling effect types under specific operation excitations can be established in advance. The patterns in these pattern libraries can be represented in the same form of a set of statistical characteristic parameters as the actual gas behavior pattern. Based on the actual gas behavior pattern and the reference pattern, calculating the quantization difference between the two, the Euclidean distance between the set of statistical characteristic parameters of the actual gas behavior pattern and the set of statistical characteristic parameters of each typical pattern in the reference pattern can be calculated. According to the quantization difference and the preset discrimination logic related to the nonlinear coupling effect, and combining the actual gas behavior pattern and the reference pattern, the coupling effect characterization information can be analyzed and obtained. A distance threshold can be set. If the distance between the actual pattern and a certain reference pattern is less than the threshold, it is considered that there is a corresponding type of nonlinear coupling effect; or, a classification model can be constructed, and the actual gas behavior pattern and the quantization difference are input, and the type or intensity of the nonlinear coupling effect is output.

[0056] Through the above technical solutions, it is possible to more accurately analyze and obtain the characterization information indicating the non-linear coupling effect between different characteristic gas components. This characterization information can more effectively reflect the current state of the on-load tap-changer of the transformer, provide a more reliable input for subsequent fault diagnosis steps, and thus improve the accuracy of transformer fault diagnosis.

[0057] The steps of expressing the change amount data of each characteristic gas component as an actual gas behavior pattern further proposed in this application include: Obtain the change amount data of each characteristic gas component; Extract the statistical characteristic parameters within a predetermined time window from the change amount data to obtain a set of statistical characteristic parameters; Use the set of statistical characteristic parameters as the actual gas behavior pattern.

[0058] Among them, the change amount data refers to the concentration change values of characteristic gas components indicating different potential defect types within a specific time period, which can be obtained by calculating the gas concentration difference between adjacent time points or at specific time intervals, and its purpose is to reflect the dynamic process of gas production or consumption; the predetermined time window refers to a definite time interval used to analyze the change amount data, which can be set as a time period with a fixed duration or a time range associated with a specific event (such as tap-changer operation), and its purpose is to define the data analysis scope to extract features with time locality; the statistical characteristic parameters refer to the quantitative indicators used to describe the distribution characteristics or change trends of the change amount data within the predetermined time window, which can be expressed by mean, variance, maximum value, minimum value, median, quantile, skewness, kurtosis, etc., and its purpose is to compress and abstract the original data and extract key information; the set of statistical characteristic parameters refers to the combination of multiple statistical characteristic parameters extracted from the change amount data, which can be stored in data structures such as vectors, lists or structures, and its purpose is to gather multi-dimensional data features and comprehensively describe the gas behavior; the actual gas behavior pattern refers to the quantitative representation obtained by extracting the statistical characteristics of the change amount data and used to characterize the gas behavior characteristics of the on-load tap-changer of the transformer under specific working conditions, which can be reflected by the set of statistical characteristic parameters, and its purpose is to transform the original and complex time series data into structured and comparable pattern information.

[0059] This solution extracts the statistical characteristic parameters of the gas change amount within a predetermined time window, integrates them into a standardized multi-dimensional statistical characteristic set, and defines it as the actual gas behavior pattern, realizing the dimensionality reduction and structured expression of the original data, and providing a reliable basis for subsequent pattern comparison and coupling effect analysis.

[0060] In some preferred embodiments, the present application is specifically implemented as follows: First, obtain the concentration change data of each characteristic gas component (such as H2, CH4, C2H4, C2H2, etc.) associated with a specific operation (such as a single switching operation) of the on-load tap-changer of the transformer within a period of time before and after the operation. These data can be obtained from an on-line monitoring system or regular sampling and analysis records, and calculate their change amounts. Then, set a predetermined time window, for example, set it to a fixed duration after the operation is completed, or set it to the time period from the start of the operation to when the gas concentration change tends to be stable. Within this predetermined time window, calculate the statistical characteristic parameters for the change amount data of each characteristic gas component, such as calculating the average value, standard deviation, and maximum value of the change amounts of each gas within this time window. Pool these calculated parameters such as the average value, standard deviation, and maximum value to form a set of statistical characteristic parameters. For example, a vector can be constructed, which contains the average value, standard deviation, and maximum value of each gas. Finally, use this vector containing the statistical characteristic parameters of multiple gases as the actual gas behavior pattern corresponding to this operation.

[0061] Through the above technical solution, the original gas change amount data, which may contain noise, redundant information, and have different dimensional scales, is transformed into a structured, representative, and comparable set of statistical characteristic parameters, thereby effectively extracting the key characteristics of the gas behavior, providing a standardized input for subsequent pattern recognition and comparison analysis, and improving the accuracy and reliability of analyzing the behavior of dissolved gases in the oil of the on-load tap-changer of the transformer.

[0062] The steps for the present application to further analyze and obtain the characterization information of the coupling effect include: Compare the quantization difference with a preset threshold or rule in the discrimination logic; Based on the comparison result, the characteristics of the actual gas behavior pattern and the reference pattern, and in combination with the discrimination logic, determine one or more non-linear coupling effect types from the preset set of non-linear coupling effect types; Use the determined non-linear coupling effect type as the coupling effect characterization information that can reflect the current state of the on-load tap-changer of the transformer.

[0063] Among them, the quantification difference refers to the degree of difference between the actual gas behavior pattern and the reference pattern, which can be quantified using distance metrics, similarity indices, or statistical test results, and its purpose is to provide a basis for preliminarily judging the existence, strength, or weakness of the nonlinear coupling effect; the discrimination logic refers to a set of rules, models, or algorithms used to analyze the quantification difference, the actual gas behavior pattern, and the characteristics of the reference pattern and determine the type of nonlinear coupling effect, which can include a rule set set based on expert experience, a decision tree model, a support vector machine model, or a neural network model, and its purpose is to make an inference and judgment based on the input information to obtain the type of nonlinear coupling effect; the preset threshold or rule refers to a specific numerical boundary or judgment condition used in the discrimination logic to compare the quantification difference, which can be a critical value or a logical judgment statement determined through historical data analysis, simulation, or expert knowledge, and its purpose is to convert the quantification difference into a judgment result of whether a specific condition is met; the actual gas behavior pattern refers to a dynamic or static feature representation based on the data of the change amount of characteristic gas components and reflecting the gas's change over time or operation, which can include the change curve of gas concentration over time, the change rate, the peak value, the integral area, or the statistical distribution characteristics, and its purpose is to capture the gas production or consumption characteristics of the on-load tap-changer of the transformer under the actual operating state; the reference pattern refers to a pattern determined according to the service information of the on-load tap-changer of the transformer, expected to appear under specific operation excitations, and reflecting the gas behavior under normal or known single fault modes, which can be a typical gas change pattern established through historical operation data, laboratory test results, or theoretical models, and its purpose is to provide a benchmark for comparison with the actual gas behavior pattern; the preset set of nonlinear coupling effect types refers to a predefined specific pattern classification of the mutual influence between different gas components that may exist in the oil chamber of the on-load tap-changer of the transformer, which can include types such as promotion or inhibition of specific gas reactions, abnormal changes in solubility, and enhancement or weakening of adsorption and desorption effects, and its purpose is to provide a candidate set for the identification of the nonlinear coupling effect; the type of nonlinear coupling effect refers to a specific pattern determined from the preset set and capable of describing the interaction between different gas components in the actual gas behavior pattern, and its purpose is to reveal the deep reason for the deviation of the actual gas behavior from the reference pattern; the coupling effect characterization information refers to the information used to reflect the nonlinear interaction between different characteristic gas components in the current state of the on-load tap-changer of the transformer, which can use the determined type of nonlinear coupling effect as the characterization information, and its purpose is to provide a basis for subsequent separation or estimation of the gas production contributions of different defect sources.

[0064] This solution compares the quantification difference and detailed characteristics between the actual gas behavior pattern and the reference pattern with the preset discrimination logic to accurately identify and determine the type of nonlinear coupling effect, improving the accuracy of coupling effect characterization and providing a reliable basis for subsequent defect source separation and fault diagnosis.

[0065] In some preferred embodiments, the analysis and acquisition of the coupling effect characterization information may be specifically implemented as follows: First, obtain the quantitative difference between the actual gas behavior pattern and the reference pattern. For example, calculate the Euclidean distance between the two in terms of key statistical features. Compare this quantitative difference with a preset threshold in the discrimination logic. For example, determine whether the Euclidean distance exceeds a preset critical value. Based on the comparison result, for example, if the Euclidean distance exceeds the critical value, indicating a significant difference, further analyze the specific features of the actual gas behavior pattern and the reference pattern. For example, analyze the shape of the change curve of each gas component, the time of peak appearance, the change rate, etc. Combine the preset discrimination logic, which can be a decision tree-based model, and classify and judge in the preset set of non-linear coupling effect types according to the comparison result of the quantitative difference and the extracted features. For example, if the quantitative difference is large and the actual pattern shows that both hydrogen and acetylene increase abnormally at the same time, the discrimination logic may determine that there is a type of "complex coupling effect caused by high-temperature discharge". Finally, use the determined type of "complex coupling effect caused by high-temperature discharge" as the coupling effect characterization information that can reflect the current state of the on-load tap-changer of the transformer.

[0066] Through the above technical solutions, the present application can more accurately determine the non-linear coupling effect types among the dissolved gases in the on-load tap-changer of the transformer. By comprehensively using the quantitative difference, the features of the actual gas behavior pattern and the reference pattern, and the preset discrimination logic, the deficiencies of single-index judgment can be overcome, and the complexity of gas behavior can be captured more comprehensively, so as to identify one or more non-linear coupling effect types that most conform to the actual situation from the preset type set. Using the determined type as the coupling effect characterization information provides a more accurate basis for subsequent analysis and diagnosis, and improves the accuracy and reliability of the fault diagnosis of the on-load tap-changer of the transformer.

[0067] The present application further proposes that the steps for analyzing the non-linear coupling effect types include: Analyze the quantitative difference and the features of the actual gas behavior pattern and the reference pattern, identify the features indicating specific gas reactions or solubility changes, and obtain the relevant features; Compare and analyze the relevant features with the preset feature library of single gas reaction or solubility change patterns to obtain the comparison and analysis results; According to the comparison and analysis results, judge whether the difference between the actual gas behavior pattern and the reference pattern can be explained by the combination of one or more preset single reaction patterns; Based on the judgment result, determine one or more non-linear coupling effect types that can reflect the actual complex coupling situation.

[0068] Among them, identifying the features indicating specific gas reactions or solubility changes means extracting the pattern details from the actual gas behavior patterns that can point to specific chemical reaction processes (for example, the generation or consumption of a certain gas) or physical solubility changes (for example, affected by temperature or pressure). This can be achieved by analyzing the proportional relationship between gas components, the change rate of gas concentration over time or the number of operations, the correlation between gas behavior and temperature or pressure, etc. The purpose is to reveal the specific physical and chemical mechanisms behind the gas behavior; the preset single gas reaction or solubility change pattern feature library refers to a set of features containing known and typical single gas reaction patterns (for example, the gas component characteristics generated by specific fault types) or solubility change patterns (for example, the effect of temperature change on the solubility of a specific gas). It can be constructed based on historical data, experimental research or theoretical models. The purpose is to provide a basis for comparing and explaining the actual complex gas behavior; the comparison and analysis means comparing the identified relevant features with the known features in the single gas reaction or solubility change pattern feature library. This can be achieved by means of pattern matching algorithms, feature vector similarity calculations, rule reasoning, etc. The purpose is to evaluate the similarity between the actual gas behavior and the known single pattern; judging whether the difference between the actual gas behavior pattern and the reference pattern can be explained by the combination of one or more preset single reaction patterns means evaluating whether the deviation between the observed actual gas behavior pattern and the expected reference pattern can be reproduced or explained by superimposing (linearly or non-linearly) one or more single reaction or solubility change patterns in the library. This can be achieved by establishing a superimposed model, applying expert system rules or machine learning models, etc. The purpose is to identify whether the complex coupling effect is the result of the combined action of multiple known single effects; the type of non-linear coupling effect that can reflect the actual complex coupling situation refers to the type determined through the above analysis that can describe the composite effect generated by the interaction or superposition of multiple factors in the actual gas behavior. The purpose is to more accurately characterize the current state of the on-load tap-changer of the transformer.

[0069] This solution identifies the detailed features of gas behavior, compares them with the single pattern library, and judges whether the difference is caused by the combination of multiple single patterns, so as to more accurately determine the type of complex non-linear coupling effect and improve the depth and reliability of coupling effect analysis.

[0070] In some preferred embodiments, specifically, assume that the actual gas behavior pattern shows the change amount data of characteristic gas components such as hydrogen, methane, ethylene, acetylene, etc. First, analyze these change amount data to identify characteristics such as abnormal acetylene / ethylene ratio, abnormal hydrogen / methane ratio, or non-linear growth of a certain gas concentration with the number of operations. These characteristics indicate specific reactions such as possible partial discharge, overheating, or oil deterioration, and obtain relevant characteristics. Then, compare the identified relevant characteristics with a preset feature library of single gas reaction or solubility change patterns. This library can include "arc discharge mode characteristics" (high acetylene, hydrogen), "local overheating mode characteristics" (high methane, ethylene), "oil-paper aging mode characteristics" (high CO, CO2), etc. The comparison result shows that the actual characteristics have certain similarities with both "arc discharge mode characteristics" and "local overheating mode characteristics", and obtain the comparison analysis result. Next, according to the comparison analysis result, judge whether the difference between the actual gas behavior pattern and the reference pattern (for example, the expected pattern under normal operation) can be explained by the combination of the two preset single reaction patterns of "arc discharge" and "local overheating". For example, by analyzing whether the gas component ratio conforms to the expectation of the superposition of these two patterns, judge whether the difference between the actual gas behavior pattern and the reference pattern can be explained by the combination of one or more preset single reaction patterns. Finally, based on the judgment result, determine that the type of non-linear coupling effect may include "compound coupling effect of arc discharge and local overheating", and determine one or more types of non-linear coupling effects that can reflect the actual complex coupling situation.

[0071] Through the above technical solutions, by analyzing the quantitative differences and the characteristics of the actual gas behavior pattern and the reference pattern, identifying the characteristics indicating specific gas reactions or solubility changes, and comparing and analyzing the relevant characteristics with the preset feature library of single gas reaction or solubility change patterns, according to the comparison analysis result, judge whether the difference between the actual gas behavior pattern and the reference pattern can be explained by the combination of one or more preset single reaction patterns, and based on the judgment result, determine one or more types of non-linear coupling effects that can reflect the actual complex coupling situation, so as to be able to more accurately analyze the type of non-linear coupling effect and improve the accuracy and reliability of the coupling effect analysis.

[0072] The present application further proposes a method, which includes: Obtain the type of the on-load tap-changer operation sequence associated with the characteristic gas generation event; Based on the type of the operation sequence, determine the expected change characteristics generated by it on the local environmental conditions related to the non-linear coupling effect in the transformer oil chamber, and obtain the expected change characteristics of the local environmental conditions; Modify the parameters characterizing the non - linear coupling effect according to the expected change characteristics of the local environmental conditions to obtain the adjusted parameters for characterizing the non - linear coupling effect; Based on the adjusted parameters for characterizing the non - linear coupling effect, and in combination with the characterization information of the coupling effect, the type of the on - load tap - changer operation sequence of the transformer, and the expected gas change response characteristics of each single fault mode in the known fault event sequence pattern library under the corresponding operation excitation, process the change amount data of each characteristic gas component to separate or estimate the adjusted gas change amounts respectively contributed by different potential defect sources after considering the non - linear coupling effect, so as to obtain the adjusted gas change amounts of each defect source.

[0073] Among them, obtaining the type of the on - load tap - changer operation sequence of the transformer associated with the characteristic gas - generating event refers to identifying the specific operation behavior pattern that causes the gas - generating event, such as continuous multiple switching, switching at a specific interval, etc.; the expected change characteristics of the local environmental conditions refer to predicting the possible change patterns of environmental factors such as the temperature, pressure, and oil flow state in the transformer oil chamber based on the type of the operation sequence, which can be determined by using a preset physical model or historical operation data analysis; the adjusted parameters for characterizing the non - linear coupling effect refer to adjusting the model parameters or correction factors describing the non - linear coupling behaviors such as gas dissolution, diffusion, and reaction according to the expected change characteristics of the local environmental conditions, which can be obtained by using a correction function or look - up table based on the environmental change characteristics; the expected gas change response characteristics of each single fault mode in the known fault event sequence pattern library under the corresponding operation excitation refer to the pre - established reference patterns describing the types, proportions, and quantity changes of the characteristic gases generated by different single fault types under specific operation excitations, which can be constructed by using experimental data, simulation models, or historical fault data statistics; processing the change amount data of each characteristic gas component to separate or estimate the adjusted gas change amounts respectively contributed by different potential defect sources after considering the non - linear coupling effect means using an algorithm or model to comprehensively utilize the adjusted parameters for characterizing the non - linear coupling effect, the characterization information of the coupling effect, the type of the operation sequence, and the expected response characteristics of the known fault modes to decompose the total characteristic gas change amount into parts respectively contributed by different potential defect sources, which can be realized by using linear / non - linear decomposition algorithms, optimization algorithms, or rule - based inference systems.

[0074] This solution determines the expected changes in the local environment according to the type of the operation sequence and modifies the parameters characterizing the non - linear coupling effect, which can more accurately reflect the gas behavior, and then separate the gas - generating contributions of each defect source, thereby improving the reliability of multi - defect source judgment under complex operation conditions.

[0075] In some preferred embodiments, the present application is specifically implemented as follows. For example, when it is recognized that a characteristic gas generation event is associated with the "continuous three - time switching" operation sequence of the on - load tap - changer of a transformer, the system first obtains the type of this operation sequence. Based on the operation sequence type of "continuous three - time switching", the system determines the expected change characteristics of the local environmental conditions in the transformer oil chamber, such as predicting that the local temperature will rise and the oil flow disturbance will increase. According to these expected change characteristics of the local environmental conditions, the system corrects the parameters characterizing the non - linear coupling effect. For example, the solubility parameter of the gas in the oil can be adjusted according to the temperature rise amplitude, and the gas diffusion coefficient can be adjusted according to the intensity of the oil flow disturbance, so as to obtain the adjusted parameters characterizing the non - linear coupling effect. Subsequently, based on these adjusted parameters characterizing the non - linear coupling effect, combined with the coupling effect characterization information (for example, indicating the type of coupling effect of thermal decomposition and arc discharge), the operation sequence type of "continuous three - time switching", and the "expected gas generation characteristics of slight overheating under continuous switching" and "expected gas generation characteristics of micro - discharge under continuous switching" stored in the known fault event sequence pattern library, the system processes the change amount data of each characteristic gas component of this characteristic gas generation event. This processing can adopt a gas decomposition model based on an optimization algorithm, which attempts to find a set of defect sources (such as slight overheating and micro - discharge) and their respective gas contribution amounts, such that when these contribution amounts are calculated based on the adjusted parameters characterizing the non - linear coupling effect, their sum is closest to the actually measured change amount data of each characteristic gas component. Through this processing process, the system separates or estimates the adjusted gas change amounts contributed by slight overheating and micro - discharge respectively, and obtains the adjusted gas change amounts of each defect source.

[0076] Through the above - mentioned technical solution, the present application can more accurately adjust the gas change amounts of each defect source by combining the type of the on - load tap - changer operation sequence of the transformer. By considering the influence of the operation sequence on the local environmental conditions and correcting the parameters characterizing the non - linear coupling effect, the processing of the change amount data of each characteristic gas component is made closer to the actual situation, so that the gas change amounts contributed by different potential defect sources can be more accurately separated or estimated. This improves the accuracy of judging whether there are gas generation contributions from multiple defect sources in complex gas generation events, and further enhances the reliability of transformer fault diagnosis.

[0077] Reference Figure 2 , the present application further proposes a transformer fault diagnosis system, which is applied to a transformer fault diagnosis method. The system includes: A data processing module, which acquires the original data of the dissolved gases in the oil of the on - load tap - changer of the transformer, and performs normalization and quantization correction on the distortion caused by gas escape, oil replenishment operation, and oil temperature fluctuation in the original data to obtain the corrected gas concentration data; A background gas separation module separates the steady-state background gas components from the corrected gas concentration data to generate net gas production data reflecting the net gas production behavior of the on-load tap-changer of the transformer; A characteristic event recognition module combines the operation records of the on-load tap-changer of the transformer, identifies characteristic gas production events related to the switching operation time of the on-load tap-changer of the transformer in the net gas production data, and extracts the gas production information of the characteristic gas production events; An event sequence construction module constructs a characteristic gas production event sequence by arranging multiple characteristic gas production events in chronological order; A fault state analysis module compares and analyzes the behavior pattern of the characteristic gas production event sequence with a preset known fault event sequence pattern library to judge the internal fault state of the on-load tap-changer of the transformer.

[0078] Among them, the data processing module refers to the functional unit responsible for receiving the original gas data and performing preprocessing, which can be implemented by a dedicated hardware circuit, a software program executed by a general-purpose processor, or a combination of software and hardware; the background gas separation module refers to the functional unit responsible for distinguishing the background components unrelated to the fault from the preprocessed gas data, which can be implemented by a statistical model, a signal filtering algorithm, or a machine learning model; the characteristic event recognition module refers to the functional unit responsible for locating and extracting the gas production characteristics related to specific events in the gas data according to the external operation information, which can be implemented by a time series analysis algorithm, a pattern matching algorithm, or a rule engine; the event sequence construction module refers to the functional unit responsible for organizing the identified multiple characteristic events according to the time relationship, which can be implemented by a database management system, a data structure organization algorithm, or an event stream processing engine; the fault state analysis module refers to the functional unit responsible for comparing the constructed event sequence with the known pattern and drawing a diagnostic conclusion, which can be implemented by a rule inference system, an expert system, a machine learning classifier, or a pattern recognition algorithm.

[0079] This solution realizes the automatic and reliable diagnosis of the on-load tap-changer fault of the transformer by a system including modules such as data processing, background gas separation, characteristic event recognition, event sequence construction, and fault state analysis, organically connecting all links of the diagnosis process to form a complete data processing and analysis chain.

[0080] In some preferred embodiments, the present application is specifically implemented as follows: The transformer fault diagnosis system can be implemented as an independent hardware device, which contains a main control unit inside, such as an industrial-grade embedded computer. The data processing module can be implemented by a software program on the embedded computer. This program is connected through a communication interface with a gas sensor (such as an on-line dissolved gas monitoring device) to obtain the original data of the dissolved gases in the oil of the on-load tap-changer of the transformer, and execute data cleaning, normalization, and quantization correction algorithms. The background gas separation module can also be implemented by a software program on the embedded computer. This program applies a background separation algorithm to the corrected gas concentration data, such as statistical analysis or trend prediction methods based on historical data, to extract the net gas production data. The characteristic event recognition module is also implemented by a software program. This program receives the operation records of the on-load tap-changer of the transformer from a transformer monitoring system (such as a SCADA system), and searches for gas production peaks or changes related to the time points near the operation in the net gas production data, identifies characteristic gas production events, and extracts information such as their gas production volume and duration. The event sequence construction module can store the identified characteristic gas production events and their related information in the storage medium inside the device (such as a solid-state drive or a database), and sort them according to the time stamp to construct a characteristic gas production event sequence. The fault state analysis module is implemented by the diagnostic software running on the embedded computer. This software loads a preset pattern library of known fault event sequences, which can be stored inside the device or accessed through the network. Then, it compares and analyzes the behavior pattern of the constructed characteristic gas production event sequence with the patterns in the pattern library, such as using a pattern matching algorithm or a classification model, and finally outputs the diagnosis result of the internal fault state of the on-load tap-changer of the transformer. The entire system connects each functional module through an internal bus or network to achieve the flow and processing of data.

[0081] Through the above technical solutions, a system capable of actually executing the transformer fault diagnosis method is provided. The system effectively eliminates various distortions in the original gas data through the data processing module, improving the reliability of the data; accurately extracts the net gas production data reflecting the true gas production behavior of the on-load tap-changer of the transformer through the background gas separation module, enhancing the manifestation of fault characteristics; accurately locates the key gas production events related to the switching operation by combining the operation records through the characteristic event recognition module, improving the pertinence of the diagnosis; organizes the time-series information of the fault evolution through the event sequence construction module, providing a basis for comprehensive analysis; and realizes the intelligent judgment of the internal fault state of the on-load tap-changer of the transformer based on the comparison of the event sequence pattern with the pattern library through the fault state analysis module. This systematic implementation method enables the complex diagnosis method to be executed automatically and efficiently, improving the accuracy and reliability of the diagnosis, and can effectively detect and judge the internal fault state of the on-load tap-changer of the transformer.

[0082] The content disclosed above is only a preferred and feasible embodiment of the present invention, and does not limit the protection scope of the present invention. Therefore, all equivalent technical changes made by using the content of the specification and drawings of the present invention are included in the protection scope of the present invention. In addition, with the development of technology, the elements therein can be updated.

Claims

1. A transformer fault diagnosis method, characterized in that, The method includes the following steps: S1: Obtain the original data of the dissolved gases in the on-load tap-changer oil of the transformer; S2: Perform normalization and quantization correction on the distortion caused by gas escape, oil replenishment operation, and oil temperature fluctuation in the original data to obtain the corrected gas concentration data; S3: Separate the steady-state background gas components from the corrected gas concentration data to obtain the net gas production data reflecting the net gas production behavior of the on-load tap-changer of the transformer; S4: Combine the operation records of the on-load tap-changer of the transformer, identify the characteristic gas production events related to the switching operation time of the on-load tap-changer of the transformer in the net gas production data, and extract the gas production information of the characteristic gas production events; S5: Construct a characteristic gas production event sequence for multiple characteristic gas production events in chronological order; S6: Compare and analyze the behavior pattern of the characteristic gas production event sequence with the preset known fault event sequence pattern library to judge the internal fault state of the on-load tap-changer of the transformer.

2. The method for diagnosing transformer faults according to claim 1, characterized in that, Step S6 specifically includes the following steps: Identify the characteristic gas production events associated with a specific on-load tap-changer operation sequence in the characteristic gas production event sequence; For the characteristic gas production events associated with the operation sequence, analyze the change amounts of the characteristic gas components indicating different potential defect types to obtain the change amount data of each characteristic gas component; Based on the change amount data of each characteristic gas component, the type of the operation sequence, and the expected gas change response characteristics of each single fault mode in the preset known fault event sequence pattern library under the corresponding operation excitation, judge whether there is a gas production contribution from multiple defect sources in the characteristic gas production events to obtain the judgment result of the gas production contribution from multiple defect sources; Based on the judgment result, distinguish the existence and nature of each independent defect source in the concurrent faults to obtain the distinguishing result of each independent defect source, and combine the distinguishing result with the overall pattern of the characteristic gas production event sequence to complete the comparison and analysis with the preset fault event pattern library to judge the internal fault state of the on-load tap-changer of the transformer.

3. The method for diagnosing transformer faults according to claim 2, characterized in that, The steps for judging whether there is a gas production contribution from multiple defect sources in the characteristic gas production events include: Based on the change amount data of each characteristic gas component, analyze the characterization information indicating the non-linear coupling effect between different characteristic gas components to obtain the coupling effect characterization information; Combine the coupling effect characterization information, the type of the operation sequence, and the expected gas change response characteristics of each single fault mode in the known fault event sequence pattern library under the corresponding operation excitation to process the change amount data of each characteristic gas component to separate or estimate the adjusted gas change amounts respectively contributed by different potential defect sources considering the non-linear coupling effect to obtain the adjusted gas change amounts of each defect source; Based on the adjusted gas change amounts of each defect source, judge whether there is a gas production contribution from multiple defect sources in the characteristic gas production events to obtain the judgment result of the gas production contribution from multiple defect sources.

4. The method for diagnosing transformer faults according to claim 3, characterized in that, The steps for obtaining the coupling effect characterization information include: Obtain the current service information of the on-load tap-changer of the transformer associated with the characteristic gas production events, where the current service information reflects the service duration or historical cumulative operation conditions of the on-load tap-changer of the transformer; Determine the reference characteristics of the non-linear coupling effect applicable to the current state of the on-load tap-changer of the transformer from a preset data set according to the current service information, where the data set includes the reference characteristics of the non-linear coupling effect corresponding to different service stages or different historical cumulative operating conditions; Based on the change amount data of the characteristic gas components, and comparing it with the reference characteristics of the non-linear coupling effect for control processing, so as to analyze and obtain the characterization information indicating the non-linear coupling effect between different characteristic gas components, and the characterization information is used to reflect the current state of the on-load tap-changer of the transformer.

5. A transformer fault diagnosis method according to claim 4, characterized in that, The steps of analyzing and obtaining the characterization information indicating the non-linear coupling effect between different characteristic gas components based on the change amount data of the characteristic gas components and comparing it with the reference characteristics of the non-linear coupling effect for control processing include: Express the change amount data of each characteristic gas component as an actual gas behavior pattern; Take the expected gas behavior pattern included in the reference characteristics of the non-linear coupling effect as the reference pattern; Based on the actual gas behavior pattern and the reference pattern, calculate the quantitative difference between the two; According to the quantitative difference and the preset discrimination logic related to the non-linear coupling effect, and combining the actual gas behavior pattern and the reference pattern, analyze and obtain the coupling effect characterization information indicating the non-linear coupling effect between different characteristic gas components and capable of reflecting the current state of the on-load tap-changer of the transformer.

6. The method for diagnosing transformer faults according to claim 5, wherein The steps of expressing the change amount data of each characteristic gas component as an actual gas behavior pattern include: Obtain the change amount data of each characteristic gas component; Extract the statistical characteristic parameters within a predetermined time window from the change amount data to obtain a set of statistical characteristic parameters; Take the set of statistical characteristic parameters as the actual gas behavior pattern.

7. The method for diagnosing transformer faults according to claim 6, wherein The steps of analyzing and obtaining the coupling effect characterization information include: Compare the quantitative difference with the preset threshold or rule in the discrimination logic; Based on the comparison result, the characteristics of the actual gas behavior pattern and the reference pattern, and combining the discrimination logic, determine one or more non-linear coupling effect types from the preset set of non-linear coupling effect types; Take the determined non-linear coupling effect type as the coupling effect characterization information capable of reflecting the current state of the on-load tap-changer of the transformer.

8. A transformer fault diagnosis method according to claim 7, characterized in that, The steps of analyzing the non-linear coupling effect type include: Analyze the quantitative difference and the characteristics of the actual gas behavior pattern and the reference pattern, identify the characteristics indicating specific gas reactions or solubility changes, and obtain relevant characteristics; Compare and analyze the relevant characteristics with the preset single gas reaction or solubility change pattern feature library to obtain the comparison and analysis result; According to the comparison and analysis result, judge whether the difference between the actual gas behavior pattern and the reference pattern can be explained by the combination of one or more preset single reaction patterns; Based on the judgment result, determine one or more non-linear coupling effect types capable of reflecting the actual complex coupling situation.

9. The method for diagnosing transformer faults according to claim 3, wherein, The steps of obtaining the adjusted gas change amount of each defect source include: Obtain the type of the operation sequence of the on-load tap-changer of the transformer associated with the characteristic gas generation event; Based on the type of the operation sequence, determine the expected change characteristics of the local environmental conditions related to the nonlinear coupling effect in the transformer oil chamber, and obtain the expected change characteristics of the local environmental conditions; According to the expected change characteristics of the local environmental conditions, correct the parameters characterizing the nonlinear coupling effect to obtain the adjusted parameters for characterizing the nonlinear coupling effect; Based on the adjusted parameters for characterizing the nonlinear coupling effect, and in combination with the characterization information of the coupling effect, the type of the on-load tap-changer operation sequence of the transformer, and the expected gas change response characteristics of each single fault mode in the known fault event sequence pattern library under the corresponding operation excitation, process the change amount data of each characteristic gas component to separate or estimate the adjusted gas change amounts respectively contributed by different potential defect sources after considering the nonlinear coupling effect, and obtain the adjusted gas change amounts of each defect source.

10. A transformer fault diagnosis system is applied to a transformer fault diagnosis method described in claim 1, characterized in that, The system includes: A data processing module, which acquires the original data of the dissolved gases in the oil of the on-load tap-changer of the transformer, and performs normalization and quantization correction on the distortion caused by gas escape, oil replenishment operation, and oil temperature fluctuation in the original data to obtain the corrected gas concentration data; A background gas separation module, which separates the steady-state background gas components from the corrected gas concentration data to generate net gas production data reflecting the net gas production behavior of the on-load tap-changer of the transformer; A characteristic event recognition module, which combines the operation records of the on-load tap-changer of the transformer to identify the characteristic gas production events related to the switching operation time of the on-load tap-changer of the transformer in the net gas production data, and extracts the gas production information of the characteristic gas production events; An event sequence construction module, which constructs a characteristic gas production event sequence by arranging multiple characteristic gas production events in chronological order; A fault state analysis module, which performs a comparison and analysis based on the behavior pattern of the characteristic gas production event sequence and the preset known fault event sequence pattern library to judge the internal fault state of the on-load tap-changer of the transformer.

Images