Transformer fault diagnosis method and system

Abstract

The present invention relates to the technical field of transformer fault diagnosis, and more specifically to a transformer fault diagnosis method and system. The method comprises the following steps: S1: acquiring raw data of dissolved gas in the transformer's on-load tapchanger oil; S2: normalizing and quantizing the raw data to obtain corrected gas concentration data; S3: separating steady-state background gas components from the corrected gas concentration data to obtain net gas production data reflecting the net gas production behavior of the transformer's on-load tapchanger; S4: identifying characteristic gas production events associated with the transformer's on-load tapchanger switching operation time in the net gas production data based on the transformer's on-load tapchanger operation records, and extracting gas production information of the characteristic gas production events; S5: constructing a characteristic gas production event sequence from multiple characteristic gas production events in chronological order; and S6: determining the internal fault state of the transformer's on-load tapchanger. This improves diagnostic accuracy.

Description

Technical Field

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

[0002] In the power system, the on-load tap changer of the oil-immersed transformer is an important device for regulating voltage, and its operating status has a key impact on the safe and stable operation of the transformer.

[0003] Existing on-load tap-changer fault diagnosis methods based on dissolved gas analysis typically rely on periodic oil sampling data collected from fixed sampling locations. However, due to the limitations of the sampling locations and the uneven flow of oil within the oil chamber, localized gas production characteristics often have difficulty promptly and fully disseminating to the sampling area. As a result, the collected gas concentration data fails to fully reflect the equipment's actual gas production status, reducing the accuracy of the diagnostic results. Summary of the Invention

[0004] The purpose of the present invention is to address the above-mentioned deficiencies and provide a transformer fault diagnosis method and system.

[0005] The present invention adopts the following technical solutions:

[0006] The present application provides a transformer fault diagnosis method, which includes the following steps:

[0007] S1: Obtain the raw data of dissolved gas in transformer on-load tap-changer oil;

[0008] S2: Normalize and quantify the distortion in the original data caused by gas escape, oil replenishment operation and oil temperature fluctuation to obtain the corrected gas concentration data;

[0009] 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 transformer on-load tap-changer;

[0010] S4: Combined with the operation record of the transformer on-load tap-changer, identify characteristic gas generation events related to the switching operation time of the transformer on-load tap-changer in the net gas generation data, and extract gas generation information of the characteristic gas generation events;

[0011] S5: construct a characteristic gas production event sequence by chronologically sequencing multiple characteristic gas production events;

[0012] S6: Based on the behavioral pattern of the characteristic gas generation event sequence and the preset known fault event sequence pattern library, the internal fault state of the transformer on-load tap changer is determined.

[0013] Through the above scheme, the accuracy and reliability of transformer on-load tap-changer fault diagnosis are improved, and problems such as original data distortion and background interference are overcome.

[0014] Optionally, the present application further proposes that step S6 specifically includes the following steps:

[0015] Identifying characteristic gas generation events associated with a specific transformer on-load tap-changer operation sequence in a characteristic gas generation event sequence;

[0016] For characteristic gas generation events associated with the operation sequence, analyzing the variation of characteristic gas components indicating different potential defect types contained therein, and obtaining variation data of each characteristic gas component;

[0017] Based on the change data of each characteristic gas component, the type of operation sequence, and the expected gas change response characteristics of each single fault mode under the corresponding operation stimulus in the preset known fault event sequence pattern library, it is judged whether there are gas production contributions from multiple defect sources in the characteristic gas production event, and the judgment result of gas production contributions from multiple defect sources is obtained;

[0018] Based on the judgment results, the existence and nature of each independent defect source in the concurrent fault are distinguished, and the distinction results of each independent defect source are obtained. Combined with the distinction results and the overall pattern of the characteristic gas production event sequence, a comparison analysis with the preset fault event pattern library is completed to determine the internal fault state of the transformer on-load tap-changer.

[0019] Through the above solution, independent defect sources in concurrent faults can be further identified, thereby improving the precision of diagnosis.

[0020] Optionally, the present application further proposes that the step of determining whether there are gas production contributions from multiple defect sources in a characteristic gas production event includes:

[0021] Based on the variation data of each characteristic gas component, characterization information indicating the nonlinear coupling effect between different characteristic gas components is analyzed to obtain coupling effect characterization information;

[0022] Combined with the coupling effect characterization information, the type of operation sequence, and the expected gas change response characteristics of each single fault mode under the corresponding operation stimulus in the known fault event sequence pattern library, the change data of each characteristic gas component is processed to separate or estimate the adjusted gas change contributed by different potential defect sources after considering the nonlinear coupling effect, and obtain the adjusted gas change of each defect source;

[0023] According to the adjusted gas change amount of each defect source, it is judged whether there are gas production contributions of multiple defect sources in the characteristic gas production event, and a judgment result of the gas production contributions of multiple defect sources is obtained.

[0024] Through the above scheme, the nonlinear coupling effect between gas components is taken into account, making the judgment of the gas production contribution of different defect sources more accurate.

[0025] Optionally, the present application further proposes that the step of obtaining coupling effect characterization information includes:

[0026] Obtaining current service information of the transformer on-load tap-changer associated with the characteristic gas generation event, where the current service information reflects the service duration or historical accumulated operating conditions of the transformer on-load tap-changer;

[0027] Based on the current service information, a nonlinear coupling effect reference characteristic applicable to the current state of the transformer on-load tap changer is determined from a preset data set, wherein the data set includes nonlinear coupling effect reference characteristics corresponding to different service stages or different historical accumulated operating conditions;

[0028] Based on the variation data of characteristic gas components and comparing them with the reference characteristics of nonlinear coupling effects, characterization information indicating the nonlinear coupling effects between different characteristic gas components is analyzed and obtained. The characterization information is used to reflect the current state of the transformer on-load tap changer.

[0029] Through the above scheme, service information is introduced to determine the reference characteristics of nonlinear coupling effects, making the coupling effect analysis more in line with the actual situation.

[0030] Optionally, the present application further proposes that the steps of analyzing and obtaining characterization information indicating the nonlinear coupling effect between different characteristic gas components based on the variation data of the characteristic gas components and comparing the variation data with the nonlinear coupling effect reference characteristics include:

[0031] Express the variation data of each characteristic gas component as the actual gas behavior pattern;

[0032] The expected gas behavior pattern contained in the nonlinear coupling effect reference characteristics is used as the reference pattern;

[0033] Based on the actual gas behavior pattern and the reference pattern, the quantitative difference between the two is calculated;

[0034] Based on the quantitative differences and the preset discrimination logic related to the nonlinear coupling effect, and combined with the actual gas behavior pattern and the reference pattern, the coupling effect characterization information indicating the nonlinear coupling effect between different characteristic gas components and reflecting the current state of the transformer on-load tap-changer is analyzed and obtained.

[0035] Through the above scheme, the actual and reference gas behavior patterns are analyzed by quantifying the differences, and the nonlinear coupling effect is characterized more accurately.

[0036] Optionally, the present application further proposes that the step of expressing the variation data of each characteristic gas component as an actual gas behavior pattern includes:

[0037] Obtaining the variation data of each characteristic gas component;

[0038] Extracting statistical characteristic parameters within a predetermined time window from the variation data to obtain a set of statistical characteristic parameters;

[0039] The statistical characteristic parameter set is used as the actual gas behavior model.

[0040] Through the above scheme, the actual gas behavior pattern is expressed by using a set of statistical characteristic parameters, which simplifies the analysis process.

[0041] Optionally, the present application further proposes that the steps of analyzing and obtaining coupling effect characterization information include:

[0042] Compare the quantitative differences to pre-set thresholds or rules in discriminant logic;

[0043] Based on the comparison results and the characteristics of the actual gas behavior pattern and the reference pattern, and in combination with the discrimination logic, one or more nonlinear coupling effect types are determined from a preset set of nonlinear coupling effect types;

[0044] The determined nonlinear coupling effect type is used as coupling effect characterization information that can reflect the current state of the transformer on-load tap changer.

[0045] Through the above scheme, the coupling effect type is determined based on quantitative differences and discriminant logic, which improves the accuracy of coupling effect characterization.

[0046] Optionally, the present application further proposes that the step of analyzing the type of nonlinear coupling effect includes:

[0047] Analyze the quantitative differences and characteristics of the actual gas behavior pattern and the reference pattern, identify the characteristics that indicate the reaction or solubility change of a specific gas, and obtain the relevant characteristics;

[0048] Compare and analyze the relevant features with a preset single gas reaction or solubility change pattern feature library to obtain a comparison analysis result;

[0049] Based on the comparative analysis results, determine whether the difference between the actual gas behavior pattern and the reference pattern can be explained by a combination of one or more preset single reaction patterns;

[0050] Based on the judgment results, one or more nonlinear coupling effect types that can reflect the actual complex coupling situation are determined.

[0051] Through the above scheme, the intrinsic mechanism of the coupling effect is deeply analyzed by comparing the single reaction pattern feature library.

[0052] Optionally, the present application further proposes that the step of obtaining the adjusted gas variation of each defect source includes:

[0053] Obtaining the type of transformer on-load tap-changer operation sequence associated with the characteristic gas generation event;

[0054] Based on the type of operation sequence, the expected change characteristics of the local environmental conditions related to the nonlinear coupling effect in the transformer oil chamber are determined, and the expected change characteristics of the local environmental conditions are obtained;

[0055] According to the expected change characteristics of the local environmental conditions, the parameters representing the nonlinear coupling effect are modified to obtain the adjusted nonlinear coupling effect representation parameters;

[0056] Based on the adjusted nonlinear coupling effect characterization parameters, and combined with the coupling effect characterization information, the type of transformer on-load tap-changer operation sequence, and the expected gas change response characteristics of each single fault mode under the corresponding operation excitation in the known fault event sequence pattern library, the change data of each characteristic gas component are processed to separate or estimate the adjusted gas change contributed by different potential defect sources after considering the nonlinear coupling effect, and obtain the adjusted gas change of each defect source.

[0057] Through the above scheme, the influence of the operation sequence on the local environment is taken into account, the characterization parameters of the nonlinear coupling effect are modified, and the accuracy of the gas production estimation of the defect source is improved.

[0058] Optionally, the present application further proposes a transformer fault diagnosis system, which is applied to the above-mentioned transformer fault diagnosis method, and the system includes:

[0059] The data processing module obtains the raw data of dissolved gas in the transformer on-load tap-changer oil, normalizes and quantifies the distortion caused by gas escape, oil replenishment operation and oil temperature fluctuation in the raw data, and obtains the corrected gas concentration data;

[0060] The 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 transformer on-load tap-changer;

[0061] The characteristic event recognition module combines the operation records of the transformer on-load tap changer to identify characteristic gas production events related to the switching operation time of the transformer on-load tap changer in the net gas production data and extract the gas production information of the characteristic gas production events;

[0062] An event sequence construction module constructs a characteristic gas production event sequence by chronologically combining multiple characteristic gas production events;

[0063] The fault status analysis module compares and analyzes the behavior pattern of the characteristic gas production event sequence with the preset known fault event sequence pattern library to determine the internal fault status of the transformer on-load tap changer.

[0064] Through the above solution, a system for implementing the above diagnostic method is provided, which is convenient for practical application.

[0065] To further understand the features and technical contents of the present invention, please refer to the following detailed description and drawings of the present invention. However, the drawings provided are only for reference and illustration and are not intended to limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 is a flow chart of the method of the present invention;

[0067] Figure 2 It is a schematic diagram of the overall structure of the present invention. DETAILED DESCRIPTION

[0068] The following is an explanation of the embodiments of the present invention through specific embodiments. Those skilled in the art can understand the advantages and effects of the present invention from the contents disclosed in this specification. The present invention can be implemented or applied through other different specific embodiments, and the 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. In addition, the drawings of the present invention are only for simple schematic illustrations and are not depicted according to actual dimensions. It is stated in advance. The following embodiments will further explain the relevant technical contents of the present invention in detail, but the disclosed contents are not intended to limit the scope of protection of the present invention.

[0069] This embodiment provides a transformer fault diagnosis method and system, combined with Figure 1 and Figure 2 shown.

[0070] refer to Figure 1 , a transformer fault diagnosis method, the method comprising the following steps:

[0071] S1: Obtain the raw data of dissolved gas in transformer on-load tap-changer oil;

[0072] S2: Normalize and quantify the distortion in the original data caused by gas escape, oil replenishment operation and oil temperature fluctuation to obtain the corrected gas concentration data;

[0073] 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 transformer on-load tap-changer;

[0074] S4: Combined with the operation record of the transformer on-load tap-changer, identify characteristic gas generation events related to the switching operation time of the transformer on-load tap-changer in the net gas generation data, and extract gas generation information of the characteristic gas generation events;

[0075] S5: construct a characteristic gas production event sequence by chronologically sequencing multiple characteristic gas production events;

[0076] S6: Based on the behavioral pattern of the characteristic gas generation event sequence and the preset known fault event sequence pattern library, the internal fault state of the transformer on-load tap changer is determined.

[0077] Among them, normalization and quantitative correction of the distortion in the original data caused by gas escape, oil replenishment operation, and oil temperature fluctuation refers to processing the initially acquired gas concentration data to eliminate or reduce data deviations caused by non-fault factors. This can be achieved by using correction models based on historical data, compensation algorithms based on physical principles, or dynamic adjustments combined with real-time monitoring data. For example, the influence of gas solubility can be compensated according to the oil temperature change curve, or the gas concentration can be diluted and corrected according to the oil replenishment record. The main purpose is to obtain gas concentration data that is closer to the actual gas production situation and provide a reliable basis for subsequent analysis.

[0078] Separating steady-state background gas components from the corrected gas concentration data refers to identifying and removing those long-standing or slowly changing gas components that are not related to the on-load tap-changer switching operation. This can be achieved using time series analysis methods, statistical filtering techniques, 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 subtraction. This is mainly to highlight the net gas production caused by the instantaneous or short-term behavior of the on-load tap-changer, focusing on gas production events related to the operation.

[0079] Identifying 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 in 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 by using technologies such as setting a time window, threshold detection or change rate analysis. For example, a fixed time period can be set before and after each switching operation, and gas concentration peaks or steep changes can be searched within this time period. The main purpose is to associate gas production behavior with specific equipment operations and distinguish between normal operation gas production and abnormal fault gas production.

[0080] Extracting gas production information of characteristic gas production events refers to obtaining key data used to describe the event from the identified characteristic gas production events, which may include parameters such as gas type, concentration change, change rate, duration, etc. Its main purpose is to quantify the characteristics of each operation-related gas production event and provide data input for subsequent model construction and comparative analysis.

[0081] Constructing a characteristic gas production event sequence by chronologically sequencing multiple characteristic gas production events means arranging all characteristic gas production events identified within a period of time in the order of their occurrence to form an ordered data set. This is mainly to reflect the dynamic evolution of the gas production behavior of the on-load tap-changer during continuous operation and multiple operations, and to reveal potential fault development trends or patterns.

[0082] Based on the comparative analysis of the behavioral patterns of characteristic gas production event sequences with a preset library of known fault event sequence patterns, the internal fault state of the transformer on-load tap-changer is determined. This means comparing the overall characteristics, change patterns, or specific combinations exhibited by the constructed characteristic gas production event sequence with pre-established gas production event sequence patterns representing different known fault types (such as poor contact, partial discharge, etc.). This can be achieved using techniques such as pattern recognition algorithms, machine learning models, or rule matching logic. For example, the similarity or distance between sequences can be calculated, or preset fault rules can be triggered based on specific event combinations in the sequence. Its main purpose is to diagnose whether the on-load tap-changer has a specific type of internal fault by matching the gas production behavior pattern.

[0083] This application solution normalizes and quantifies the original data of dissolved gas in the transformer on-load tap-changer oil, eliminates interference from factors such as gas escape, oil replenishment, and oil temperature fluctuations, and obtains more realistic gas concentration data; further separates the steady-state background gas and extracts the net gas production data related to the tap-changer operation; combines operation records to identify characteristic gas production events and construct a time series, and by comparing with a known fault event pattern library, determines the internal fault state, thereby achieving hierarchical, progressive, accurate and reliable fault diagnosis.

[0084] In some preferred embodiments, raw data on dissolved gas in transformer on-load tapchanger oil can be obtained through an online monitoring system or by periodically sampling oil for offline gas chromatography analysis. When correcting the raw data, a multivariate regression model can be established that incorporates parameters such as oil temperature, oil replenishment volume, and ambient temperature, and the measured gas concentration can be corrected based on these parameters. To isolate the steady-state background gas components, a moving average or exponential smoothing method can be used to calculate the long-term trend of gas concentration as the background value, which is then subtracted from the corrected data. To identify characteristic gas generation events, a time window, such as 30 minutes, can be set around each on-load tapchanger switching operation. Within this window, the net gas generation data is tested for gas concentration increments or rates of change exceeding a preset threshold. If present, the event is marked as a characteristic gas generation event, and the time of occurrence, the gas type involved, and the concentration change are recorded. To construct a sequence of characteristic gas generation events, all characteristic gas generation events identified over a period of time (e.g., a month or a year) can be arranged in chronological order to form an ordered list. When conducting comparative analysis, machine learning algorithms such as support vector machines or neural networks can be used to take the statistical characteristics of the characteristic gas production event sequence (such as event frequency, probability of occurrence of different gas combinations, distribution of concentration change amplitude, etc.) as input, and compare them with trained classification models representing different fault modes (such as slight contact wear, local overheating, initial discharge, etc.) to output the most matching fault type judgment result.

[0085] Through multiple data corrections and background gas separation, this solution improves the authenticity and pertinence of gas data; combines operation records to construct a gas production event sequence and accurately capture the evolution of faults; and achieves precise judgment of fault status based on sequence pattern comparison, effectively overcoming the limitations of traditional methods that rely on isolated data and improving the accuracy and reliability of diagnosis.

[0086] This application further proposes that step S6 specifically includes the following steps:

[0087] Identifying characteristic gas generation events associated with a specific transformer on-load tap-changer operation sequence in a characteristic gas generation event sequence;

[0088] For characteristic gas generation events associated with the operation sequence, analyzing the variation of characteristic gas components indicating different potential defect types contained therein, and obtaining variation data of each characteristic gas component;

[0089] Based on the change data of each characteristic gas component, the type of operation sequence, and the expected gas change response characteristics of each single fault mode under the corresponding operation stimulus in the preset known fault event sequence pattern library, it is judged whether there are gas production contributions from multiple defect sources in the characteristic gas production event, and the judgment result of gas production contributions from multiple defect sources is obtained;

[0090] Based on the judgment results, the existence and nature of each independent defect source in the concurrent fault are distinguished, and the distinction results of each independent defect source are obtained. Combined with the distinction results and the overall pattern of the characteristic gas production event sequence, a comparison analysis with the preset fault event pattern library is completed to determine the internal fault state of the transformer on-load tap-changer.

[0091] Among them, a specific transformer on-load tapchanger operation sequence refers to a series of switching actions performed by the transformer on-load tapchanger within a specific time period, such as upshifting, downshifting or continuous switching, which can be characterized by operation record data; characteristic gas components indicating different potential defect types refer to gases dissolved in transformer oil that can reflect abnormal internal conditions of the tapchanger (such as overheating, discharge), such as hydrogen, methane, ethane, ethylene, acetylene, etc., which can be detected by gas chromatography analysis and other methods; a preset known fault event sequence pattern library refers to a pre-established data set containing characteristic gas production behavior patterns corresponding to different known fault types (such as slight overheating, partial discharge, severe overheating and discharge concurrently) under different operational stimuli, which can be constructed using historical data analysis, experimental simulation or expert experience; the expected gas change response of each single fault mode under the corresponding operational stimulus The response feature refers to the change law or pattern of the characteristic gas component of a specific single fault type recorded in the pattern library when it undergoes a specific operation sequence. It can be represented by a curve, vector or statistical parameter set; the gas production contribution of multiple defect sources refers to the characteristic gas quantity or its change trend independently generated by two or more different potential defect sources (for example, the simultaneous existence of overheating and discharge) in the same characteristic gas production event, which can be judged by decomposition or attribution analysis of the gas component change; the existence and nature of each independent defect source in a concurrent fault refers to further determining which types of defect sources (such as overheating, discharge, poor contact, etc.) exist at the same time, after judging the existence of gas production contributions from multiple defect sources, and the severity or characteristics of each of these defect sources. It can be distinguished by methods based on gas component ratio, change rate or pattern matching.

[0092] Based on the construction of characteristic gas production event sequences, this solution focuses on gas production events associated with specific operation sequences, deeply analyzing the changes in each characteristic gas component. Combining the type of operation sequence and a preset pattern library, it determines whether multiple defect sources exist and further distinguishes different defect types and properties. By combining local component analysis with overall sequence pattern comparison, the accuracy and reliability of concurrent fault identification and diagnosis are improved.

[0093] In some preferred embodiments, the present application is implemented as follows. First, a sequence of characteristic gas production events is obtained after preprocessing, background gas separation, characteristic event identification, and sequence construction. Assume that the sequence contains a characteristic gas production event associated with the most recent transformer on-load tapchanger upshift operation. For this characteristic gas production event associated with the upshift operation, the characteristic gas components contained therein, such as hydrogen, methane, ethylene, and acetylene, are analyzed, and their changes before and after the operation are calculated to obtain change data for each characteristic gas component. For example, the changes in ethylene and acetylene are detected to exceed the normal range. Then, based on this gas change data and the fact that the operation was an upshift operation, a pre-set library of known fault event sequence patterns is consulted. The pattern library may record expected response characteristics: a single overheating fault primarily causes an increase in ethylene during an upshift operation, while a single discharge fault primarily causes an increase in acetylene during an upshift operation. By comparing the actual detected significant increases in both ethylene and acetylene with the expected responses in the pattern library, it is determined that multiple fault sources contribute to the gas production in this characteristic gas production event, such as the simultaneous presence of overheating and discharge. Based on this judgment, the presence and nature of each independent defect source in the concurrent fault are further distinguished. For example, based on the relative proportions and absolute values ​​of the changes in the amounts of ethylene and acetylene, combined with the gas characteristics of overheating and discharge of varying severity in the pattern library, it is possible to distinguish between two independent defect sources: moderate overheating and mild partial discharge. Finally, the "moderate overheating and mild partial discharge" distinction result is combined with the overall behavioral pattern of the entire sequence of characteristic gas production events (which may include multiple operations and corresponding gas production events) for a comprehensive analysis. This is then compared with the corresponding concurrent fault patterns in the pattern library to ultimately determine that the internal fault state of the transformer's on-load tap changer is "moderate overheating with mild partial discharge."

[0094] By analyzing the changes in characteristic gas components of operation-related gas production events and combining them with overall pattern comparison, this solution can identify and distinguish concurrent faults from multiple defect sources, effectively overcoming the limitations of relying solely on overall pattern comparison and improving the accuracy of fault diagnosis under complex working conditions.

[0095] This application further proposes that the steps of determining whether there are gas generation contributions from multiple defect sources in a characteristic gas generation event include:

[0096] Based on the variation data of each characteristic gas component, characterization information indicating the nonlinear coupling effect between different characteristic gas components is analyzed to obtain coupling effect characterization information;

[0097] Combined with the coupling effect characterization information, the type of operation sequence, and the expected gas change response characteristics of each single fault mode under the corresponding operation stimulus in the known fault event sequence pattern library, the change data of each characteristic gas component is processed to separate or estimate the adjusted gas change contributed by different potential defect sources after considering the nonlinear coupling effect, and obtain the adjusted gas change of each defect source;

[0098] According to the adjusted gas change amount of each defect source, it is judged whether there are gas production contributions of multiple defect sources in the characteristic gas production event, and a judgment result of the gas production contributions of multiple defect sources is obtained.

[0099] Among them, the coupling effect characterization information refers to the characterization information indicating the nonlinear coupling effect between different characteristic gas components. Specifically, it may refer to parameters or models that reflect the influence of environmental factors such as gas reaction rate, solubility change, adsorption and desorption behavior. Its purpose is to quantify the complex interactions 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, which is separated or estimated from the total gas change amount through data processing methods after considering the nonlinear coupling effect between gas components. Its purpose is to eliminate the interference of the coupling effect and obtain the gas production contribution of each defect source that is closer to the actual situation.

[0100] This solution analyzes the nonlinear coupling effects between characteristic gas components, combines the type of operation sequence and known fault modes, corrects and decouples the original gas data, and accurately estimates the actual gas contribution of each defect source, thereby improving the reliability and accuracy of concurrent fault identification and diagnosis.

[0101] In some preferred embodiments, specifically, when analyzing information indicating nonlinear coupling effects between different characteristic gas components, a machine learning model based on historical data can be established. This model's inputs include the variation data for each characteristic gas component, the type of operation sequence, and environmental parameters such as oil temperature and oil pressure. Its output is a coupling coefficient or coupling function that reflects the degree of mutual influence between the gas components. For example, a neural network model can be trained to predict the deviation between the actual variation of a particular gas under given operating and environmental conditions and the theoretical variation without coupling effects, with this deviation pattern used as information representative of the coupling effect. When processing the variation data for each characteristic gas component to obtain the adjusted gas variation for each defect source, a model-based inversion approach can be employed. For example, a joint model can be established that incorporates gas generation models for multiple potential defect sources and a model of coupling effects between gas components. Then, using an optimization algorithm, with the measured gas variation data as the target, the adjusted gas variation independently contributed by each defect source can be inverted to best explain the measured data. For example, we can assume there are two potential defect sources, A and B, which generate gases G1 and G2, respectively. Furthermore, there is a mutually reinforcing coupling effect between G1 and G2. The measured changes in G1 and G2 are the result of the combined effects of the independent gas production from A and B and the coupling effect. By establishing a mathematical model to describe this process and solving the model parameters using measured data, we can estimate the adjusted gas changes contributed by A and B.

[0102] By introducing nonlinear coupling effect analysis and correcting gas variation data, this scheme can accurately separate the gas production contributions of different defect sources, effectively overcome the interference caused by gas interactions, and improve the reliability and accuracy of concurrent fault diagnosis under complex working conditions.

[0103] This application further proposes that the steps of obtaining coupling effect characterization information include:

[0104] Obtaining current service information of the transformer on-load tap-changer associated with the characteristic gas generation event, where the current service information reflects the service duration or historical accumulated operating conditions of the transformer on-load tap-changer;

[0105] Based on the current service information, a nonlinear coupling effect reference characteristic applicable to the current state of the transformer on-load tap changer is determined from a preset data set, wherein the data set includes nonlinear coupling effect reference characteristics corresponding to different service stages or different historical accumulated operating conditions;

[0106] Based on the variation data of characteristic gas components and comparing them with the reference characteristics of nonlinear coupling effects, characterization information indicating the nonlinear coupling effects between different characteristic gas components is analyzed and obtained. The characterization information is used to reflect the current state of the transformer on-load tap changer.

[0107] Among them, the current service information refers to the information reflecting the historical operating status of the transformer on-load tap-changer, which can be obtained by recording the operating time, recording the number of operations, recording the cumulative switching load, etc., and its purpose is to quantify the aging degree or wear status of the transformer on-load tap-changer; the preset data set refers to a data set that stores the characteristic laws of the nonlinear coupling effects of different transformer on-load tap-changers under different service conditions, which can be established through laboratory simulation, statistical analysis of historical operating data, expert experience, etc., and its purpose is to provide a reference benchmark for the coupling effect analysis of the current transformer on-load tap-changer; the nonlinear coupling effect reference characteristic refers to the expected law of mutual influence or correlation between different characteristic gas components under the service state of a specific transformer on-load tap-changer, which can be expressed as a proportional relationship between the changes in gas components, a correlation model, or a specific gas The purpose of the combined gas production mode is to provide a coupling effect benchmark that matches the current state of the transformer on-load tap-changer; the control processing refers to the process of comparing and analyzing the actual measured change data of the characteristic gas components with the reference characteristics of the nonlinear coupling effect. It can adopt methods such as calculation difference, pattern matching, and statistical test. Its purpose is to identify the degree and pattern of deviation between the actual gas behavior and the expected reference characteristics; the characterization information indicating the nonlinear coupling effect between different characteristic gas components refers to the information obtained through the control processing that can quantify or describe the degree and nature of the nonlinear interaction between the gas components. It can be expressed as a quantitative difference value, deviation pattern type, coupling effect strength index, or a specific coupling effect type identifier. Its purpose is to provide a quantitative or qualitative description of the gas coupling effect under the current state of the transformer on-load tap-changer.

[0108] This scheme obtains the current service information and selects the matching nonlinear coupling effect reference characteristics, compares the actual gas change with the reference benchmark, and analyzes the coupling effect characterization information that incorporates the influence of the service status, thereby more accurately separating the gas production contribution of each defect source and improving the reliability of multi-defect source identification.

[0109] In some preferred embodiments, obtaining current service information of the transformer on-load tap changer associated with a characteristic gas production event can be accomplished by reading the total operating hours or total switching times recorded in the transformer operation and maintenance record system. Based on the obtained current service information, such as the total operating hours, a nonlinear coupling effect reference characteristic applicable to the operating hour range can be searched or calculated from a preset data set. This data set can be a database storing expected proportional relationships or correlation models between gas component changes corresponding to different operating time periods. Based on characteristic gas component change data, such as the measured acetylene and ethylene changes, this data is compared with a determined nonlinear coupling effect reference characteristic. This comparison can be performed by comparing the actual measured acetylene to ethylene change ratio with the expected ratio for that service stage in the reference characteristic and calculating the difference between the two. Alternatively, the comparison can be performed by inputting the actual gas change data into a model provided by the reference characteristic and calculating the degree of deviation between the actual value and the model's predicted value. Based on the calculated differences or deviations, characterization information indicating the nonlinear coupling effect between different characteristic gas components is analyzed and obtained. This characterization information can be a quantitative indicator or a qualitative description, which is used to reflect the gas coupling effect characteristics under the current state of the transformer on-load tap changer.

[0110] The present application further proposes that the steps of analyzing and obtaining characterization information indicating the nonlinear coupling effect between different characteristic gas components by comparing the variation data of the characteristic gas components with the nonlinear coupling effect reference characteristics include:

[0111] Express the variation data of each characteristic gas component as the actual gas behavior pattern;

[0112] The expected gas behavior pattern contained in the nonlinear coupling effect reference characteristics is used as the reference pattern;

[0113] Based on the actual gas behavior pattern and the reference pattern, the quantitative difference between the two is calculated;

[0114] Based on the quantitative differences and the preset discrimination logic related to the nonlinear coupling effect, and combined with the actual gas behavior pattern and the reference pattern, the coupling effect characterization information indicating the nonlinear coupling effect between different characteristic gas components and reflecting the current state of the transformer on-load tap-changer is analyzed and obtained.

[0115] The actual gas behavior pattern refers to an expression obtained by abstracting and refining the variation data of each characteristic gas component, which can reflect the overall characteristics and interrelationships of gas component changes. It can be implemented using a set of statistical characteristic parameters, time-series variation curve characteristics, or feature vectors extracted based on pattern recognition algorithms. The reference pattern refers to an expression of the gas behavior pattern that may be caused by various nonlinear coupling effects under specific operating conditions, contained in the nonlinear coupling effect reference characteristics. It can be implemented using a pattern description method that matches the actual gas behavior pattern. The quantitative difference refers to a numerical representation of the degree of deviation between the actual gas behavior pattern and the reference pattern. It can be calculated using Euclidean distance, cosine similarity, or other pattern similarity metrics. The preset nonlinear coupling effect-related discrimination logic refers to a set of rules or models used to determine the type or degree of nonlinear coupling effect based on the quantitative difference and the characteristics of the actual gas behavior pattern and the reference pattern. It can be implemented using a rule base based on expert experience, a decision tree model, or a machine learning classifier. The coupling effect characterization information refers to information used to indicate the nonlinear coupling effect between different characteristic gas components and can reflect the current state of the transformer on-load tap changer. It can be implemented by using the type identification of the nonlinear coupling effect, the coupling strength parameter, or the coupling effect influencing factor.

[0116] In some preferred embodiments, the variation data of each characteristic gas component is expressed as an actual gas behavior pattern. The variation data of each characteristic gas component within a predetermined time window can be obtained, and statistical characteristic parameters such as mean value, standard deviation, maximum value, minimum value, rate of change, etc. are extracted from these variation data. These statistical characteristic parameters are then combined into a set to serve as the actual gas behavior pattern. Using the expected gas behavior pattern contained in the nonlinear coupling effect reference characteristics as a reference pattern, a library of typical gas variation patterns for different nonlinear coupling effect types under specific operational stimuli can be pre-established. The patterns in these pattern libraries can be represented in the form of the same statistical characteristic parameter sets as the actual gas behavior patterns. Based on the actual gas behavior pattern and the reference pattern, the quantitative difference between the two is calculated, and the Euclidean distance between the statistical characteristic parameter set of the actual gas behavior pattern and the statistical characteristic parameter set of each typical pattern in the reference pattern can be calculated. Based on the quantitative differences and the preset discrimination logic related to the nonlinear coupling effect, and combined with the actual gas behavior pattern and the reference pattern, the coupling effect characterization information is analyzed and obtained. A distance threshold can be set. If the distance between the actual pattern and a reference pattern is less than the threshold, it is considered that the corresponding nonlinear coupling effect type exists; alternatively, a classification model can be constructed, which inputs the actual gas behavior pattern and the quantitative differences and outputs the type or intensity of the nonlinear coupling effect.

[0117] This technical solution enables more accurate analysis and acquisition of information characterizing the nonlinear coupling effects between different characteristic gas components. This information can more effectively reflect the current state of the transformer's on-load tap-changer, providing more reliable input for subsequent fault diagnosis steps, thereby improving the accuracy of transformer fault diagnosis.

[0118] The present application further proposes that the steps of expressing the variation data of each characteristic gas component as an actual gas behavior pattern include:

[0119] Obtaining the variation data of each characteristic gas component;

[0120] Extracting statistical characteristic parameters within a predetermined time window from the variation data to obtain a set of statistical characteristic parameters;

[0121] The statistical characteristic parameter set is used as the actual gas behavior model.

[0122] Among them, the variation data refers to the concentration change value of the characteristic gas components indicating different potential defect types within a specific time period. It can be obtained by calculating the difference in gas concentrations at adjacent time points or within a specific time interval. Its purpose is to reflect the dynamic process of gas production or consumption; the predetermined time window refers to a certain time interval used to analyze the variation data. It can be set by a fixed time period or a time range associated with a specific event (such as tap changer operation). Its purpose is to define the scope of data analysis in order to extract features with time locality; the statistical characteristic parameter refers to a quantitative indicator used to describe the distribution characteristics or change trends of the variation data within a predetermined time window. It can be set by the mean, variance, maximum value, etc. , minimum, median, quantile, skewness, kurtosis, etc., with the purpose of compressing and abstracting the original data and extracting key information; the statistical feature parameter set refers to a combination of multiple statistical feature parameters extracted from the variation data, which can be stored in data structures such as vectors, lists or structures. Its purpose is to aggregate multi-dimensional data features and comprehensively describe the gas behavior; the actual gas behavior pattern refers to a quantitative representation obtained by extracting statistical features from the variation data, which is used to characterize the gas behavior characteristics of the transformer on-load tap-changer under specific working conditions. It can be embodied in a statistical feature parameter set, with the purpose of converting the original and complex time series data into structured and comparable pattern information.

[0123] This scheme extracts the statistical characteristic parameters of gas changes within a predetermined time window, integrates them into a standardized multi-dimensional statistical feature set, and defines it as the actual gas behavior pattern, thereby achieving dimensionality reduction and structured expression of the original data, providing a reliable basis for subsequent pattern comparison and coupling effect analysis.

[0124] In some preferred embodiments, the present application is implemented as follows: First, concentration change data for characteristic gas components (such as H2, CH4, C2H4, and C2H2) associated with a specific operation of a transformer on-load tap changer (e.g., a single switching action) is obtained over a period of time before and after the operation. This data can be obtained from an online monitoring system or periodic sampling and analysis records, and the concentration change is calculated. Then, a predetermined time window is set, such as a fixed duration after the operation is completed, or the period from the start of the operation until the gas concentration change stabilizes. Within this predetermined time window, statistical characteristic parameters are calculated for the concentration change data for each characteristic gas component, such as the mean, standard deviation, and maximum value of each gas concentration within the time window. These calculated mean, standard deviation, and maximum values ​​are aggregated to form a set of statistical characteristic parameters, such as a vector containing the mean, standard deviation, and maximum value of each gas. Finally, this vector containing the statistical characteristic parameters for the multiple gases is used as the actual gas behavior pattern corresponding to the operation.

[0125] Through the above technical solution, the original gas variation data, which may contain noise and redundant information and has different dimensional scales, is converted into a structured, representative and comparable set of statistical feature parameters, thereby effectively extracting the key characteristics of gas behavior, providing standardized input for subsequent pattern recognition and comparative analysis, and improving the accuracy and reliability of analyzing the dissolved gas behavior in transformer on-load tap-changer oil.

[0126] This application further proposes that the steps of analyzing and obtaining coupling effect characterization information include:

[0127] Compare the quantitative differences to pre-set thresholds or rules in discriminant logic;

[0128] Based on the comparison results and the characteristics of the actual gas behavior pattern and the reference pattern, and in combination with the discrimination logic, one or more nonlinear coupling effect types are determined from a preset set of nonlinear coupling effect types;

[0129] The determined nonlinear coupling effect type is used as coupling effect characterization information that can reflect the current state of the transformer on-load tap changer.

[0130] Among them, the quantitative difference refers to the degree of difference between the actual gas behavior pattern and the reference pattern, which can be quantified by distance measurement, similarity index or statistical test results, and its purpose is to provide a basis for preliminary judgment of the existence or strength of nonlinear coupling effect; the discrimination logic refers to the set of rules, models or algorithms used to analyze the quantitative difference, actual gas behavior pattern, reference pattern characteristics and determine the type of nonlinear coupling effect, which can include a rule 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 inferences and judgments based on the input information to derive the type of nonlinear coupling effect; the preset threshold or A rule refers to a specific numerical limit or judgment condition used in discriminant logic to compare quantitative differences. It can be a critical value or logical judgment statement determined through historical data analysis, simulation, or expert knowledge. Its purpose is to convert quantitative differences into a judgment result of whether a specific condition is met; an actual gas behavior pattern refers to a dynamic or static characteristic representation based on characteristic gas component change data that reflects the changes in gas over time or operation. It can include the change curve of gas concentration over time, change rate, peak value, integrated area or statistical distribution characteristics. Its purpose is to capture the gas production or consumption characteristics under the actual operating state of the transformer on-load tap changer; reference A pattern is a pattern of gas behavior expected to occur under specific operational stimuli, determined based on the service information of the transformer's on-load tap-changer, reflecting normal or known single fault modes. It can be a typical gas variation pattern established through historical operating data, laboratory test results, or theoretical models. Its purpose is to provide a benchmark for comparison with actual gas behavior patterns. A preset set of nonlinear coupling effect types refers to a predefined classification of specific patterns of interactions between different gas components that may exist in the transformer's on-load tap-changer oil compartment. These patterns may include promotion or inhibition of specific gas reactions, abnormal solubility changes, and enhanced or weakened adsorption and desorption effects. Its purpose is to provide a candidate set for identifying nonlinear coupling effects. A nonlinear coupling effect type refers to a specific pattern determined from a preset set that can describe the interaction between different gas components in the actual gas behavior pattern. Its purpose is to reveal the underlying reasons why actual gas behavior deviates from the reference pattern. Coupling effect characterization information refers to information used to reflect the nonlinear interactions between different characteristic gas components in the current state of the transformer's on-load tap-changer. The determined nonlinear coupling effect type can be used as characterization information. Its purpose is to provide a basis for subsequent separation or estimation of the gas production contributions of different defect sources.

[0131] This solution accurately identifies and determines the type of nonlinear coupling effect by comparing the quantitative differences and detailed characteristics of the actual gas behavior pattern and the reference pattern with the preset discrimination logic, thereby improving the accuracy of the coupling effect characterization and providing a reliable basis for subsequent defect source isolation and fault diagnosis.

[0132] In some preferred embodiments, analyzing and obtaining coupling effect characterization information can be specifically implemented as follows: First, a quantitative difference is obtained between the actual gas behavior pattern and the reference pattern, for example, by calculating the Euclidean distance between the two patterns on key statistical features. This quantitative difference is then compared with a preset threshold in the discrimination logic, for example, to 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, the specific characteristics of the actual gas behavior pattern and the reference pattern are further analyzed, for example, analyzing the shape, peak occurrence time, and rate of change of each gas component variation curve. Combined with the preset discrimination logic, which can be a decision tree-based model, classification is performed within a preset set of nonlinear coupling effect types based on the comparison results of the quantitative differences and the extracted features. For example, if the quantitative difference is large and the actual pattern shows abnormal increases in both hydrogen and acetylene, the discrimination logic may determine the presence of a "compound coupling effect caused by high-temperature discharge" type. Ultimately, the determined "compound coupling effect caused by high-temperature discharge" type is used as coupling effect characterization information that reflects the current state of the transformer on-load tap changer.

[0133] Through the above-mentioned technical solution, the present application can more accurately determine the type of nonlinear coupling effect between dissolved gases in transformer on-load tap-changer oil. By comprehensively utilizing quantitative differences, the characteristics of actual gas behavior patterns and reference patterns, and pre-set judgment logic, it overcomes the shortcomings of single-metric judgment and more comprehensively captures the complexity of gas behavior, thereby identifying one or more nonlinear coupling effect types that best match the actual situation from a set of pre-set types. Using the determined type as coupling effect characterization information provides a more precise basis for subsequent analysis and diagnosis, improving the accuracy and reliability of transformer on-load tap-changer fault diagnosis.

[0134] This application further proposes that the steps for analyzing the type of nonlinear coupling effect include:

[0135] Analyze the quantitative differences and characteristics of the actual gas behavior pattern and the reference pattern, identify the characteristics that indicate the reaction or solubility change of a specific gas, and obtain the relevant characteristics;

[0136] Compare and analyze the relevant features with a preset single gas reaction or solubility change pattern feature library to obtain a comparison analysis result;

[0137] Based on the comparative analysis results, determine whether the difference between the actual gas behavior pattern and the reference pattern can be explained by a combination of one or more preset single reaction patterns;

[0138] Based on the judgment results, one or more nonlinear coupling effect types that can reflect the actual complex coupling situation are determined.

[0139] Among them, identifying features indicating specific gas reactions or solubility changes refers to extracting pattern details 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) from actual gas behavior patterns. This can be achieved by analyzing the proportional relationship between gas components, the rate of change of gas concentration with time or number of operations, the correlation between gas behavior and temperature or pressure, etc. Its purpose is to reveal the specific physical and chemical mechanisms behind gas behavior; a preset single gas reaction or solubility change pattern feature library refers to a feature set containing known, typical single gas reaction patterns (for example, gas component characteristics generated by a specific fault type) or solubility change patterns (for example, the effect of temperature change on the solubility of a specific gas), which can be constructed based on historical data, experimental studies or theoretical models, with the purpose of providing a basis for comparing and explaining actual complex gas behaviors; comparative analysis refers to comparing the identified relevant features with single gas reaction or solubility change patterns. Comparison is performed with known features in the feature library, which can be achieved through pattern matching algorithms, feature vector similarity calculation, rule reasoning, etc., with the aim of evaluating the degree of 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 a combination of one or more preset single reaction patterns refers to 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 nonlinearly) one or more single reactions or solubility change patterns in the library. This can be achieved through establishing a superposition model, applying expert system rules or machine learning models, with the aim of identifying whether the complex coupling effect is the result of the combined action of multiple known single effects. The type of nonlinear 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 produced by the interaction or superposition of multiple factors in the actual gas behavior, with the aim of more accurately characterizing the current state of the transformer on-load tap-changer.

[0140] This solution identifies the detailed characteristics of gas behavior and compares them with a single pattern library to determine whether the differences are caused by a combination of multiple single patterns, thereby more accurately determining the type of complex nonlinear coupling effects and improving the depth and reliability of coupling effect analysis.

[0141] In some preferred embodiments, specifically, assuming that the actual gas behavior pattern displays data on the variation of characteristic gas components such as hydrogen, methane, ethylene, and acetylene, this variation data is first analyzed to identify features such as abnormal acetylene / ethylene ratios, abnormal hydrogen / methane ratios, or nonlinear increases in the concentration of certain gases with the number of operations. These features indicate possible reactions such as partial discharge, overheating, or oil degradation, thereby generating relevant features. These identified relevant features are then compared with a pre-defined library of single gas reaction or solubility change pattern features. This library may include features such as "arcing pattern features" (high acetylene and hydrogen), "local overheating pattern features" (high methane and ethylene), and "oil-paper aging pattern features" (high CO and CO2). The comparison results indicate a certain degree of similarity between the actual features and both the "arcing pattern features" and the "local overheating pattern features," resulting in a comparison analysis result. Next, based on the comparative analysis results, a determination is made as to whether the difference between the actual gas behavior pattern and the reference pattern (e.g., the expected pattern under normal operation) can be explained by a combination of two pre-defined single reaction modes: "arcing" and "localized overheating." For example, by analyzing whether the gas component ratios meet expectations for the superposition of these two modes, a determination is made as to whether the difference between the actual gas behavior pattern and the reference pattern can be explained by a combination of one or more pre-defined single reaction modes. Finally, based on the determination results, a determination is made as to whether the nonlinear coupling effect type may include a "combined coupling effect of arcing and localized overheating," and one or more nonlinear coupling effect types are identified that can reflect the actual complex coupling situation.

[0142] Through the above technical solution, by analyzing the quantitative differences and the characteristics of the actual gas behavior pattern and the reference pattern, the characteristics indicating a specific gas reaction or solubility change are identified, and the relevant characteristics are compared and analyzed with a preset single gas reaction or solubility change pattern feature library. According to the comparison and analysis results, it is judged whether the difference between the actual gas behavior pattern and the reference pattern can be explained by a combination of one or more preset single reaction patterns. Based on the judgment result, one or more nonlinear coupling effect types that can reflect the actual complex coupling situation are determined, so that the nonlinear coupling effect type can be analyzed more accurately, thereby improving the accuracy and reliability of the coupling effect analysis.

[0143] The present application further proposes a method comprising:

[0144] Obtaining the type of transformer on-load tap-changer operation sequence associated with the characteristic gas generation event;

[0145] Based on the type of operation sequence, the expected change characteristics of the local environmental conditions related to the nonlinear coupling effect in the transformer oil chamber are determined, and the expected change characteristics of the local environmental conditions are obtained;

[0146] According to the expected change characteristics of the local environmental conditions, the parameters representing the nonlinear coupling effect are modified to obtain the adjusted nonlinear coupling effect representation parameters;

[0147] Based on the adjusted nonlinear coupling effect characterization parameters, and combined with the coupling effect characterization information, the type of transformer on-load tap-changer operation sequence, and the expected gas change response characteristics of each single fault mode under the corresponding operation excitation in the known fault event sequence pattern library, the change data of each characteristic gas component are processed to separate or estimate the adjusted gas change contributed by different potential defect sources after considering the nonlinear coupling effect, and obtain the adjusted gas change of each defect source.

[0148] Among them, obtaining the type of transformer on-load tapchanger operation sequence associated with the characteristic gas production event refers to identifying the specific operation behavior pattern that causes the gas production event, such as multiple consecutive switching, switching at specific intervals, etc.; the expected change characteristics of local environmental conditions refer to predicting the possible change patterns of environmental factors such as temperature, pressure, and oil flow state in the transformer oil chamber based on the operation sequence type, which can be determined by using a preset physical model or historical operation data analysis; the adjusted nonlinear coupling effect characterization parameters refer to adjusting the model parameters or correction factors that describe nonlinear coupling behaviors such as gas dissolution, diffusion, and reaction according to the expected change characteristics of local environmental conditions, which can be obtained by using a correction function or lookup table based on environmental change characteristics; each single fault mode in the known fault event sequence pattern library is adjusted in the corresponding operation The expected gas change response characteristics under excitation refer to pre-established reference patterns that describe the changes in the types, proportions, and quantities of characteristic gases generated by different single fault types under specific operational excitations. These patterns can be constructed using experimental data, simulation models, or historical fault data statistics. Processing the change data of each characteristic gas component to separate or estimate the adjusted gas changes contributed by different potential defect sources after considering the nonlinear coupling effect refers to the use of an algorithm or model to comprehensively utilize the adjusted nonlinear coupling effect characterization parameters, coupling effect characterization information, the type of operation sequence, and the expected response characteristics of the known fault mode to decompose the total characteristic gas change into parts contributed by different potential defect sources. This can be achieved using a linear / nonlinear decomposition algorithm, an optimization algorithm, or a rule-based reasoning system.

[0149] This solution more accurately reflects gas behavior by determining the expected changes in the local environment based on the type of operation sequence and correcting the parameters characterizing the nonlinear coupling effect. It then separates the gas production contribution of each defect source and improves the reliability of multi-defect source judgment under complex operating conditions.

[0150] In some preferred embodiments, the present application is implemented as follows. For example, when a characteristic gas generation event is identified as being associated with a "three consecutive switching" operation sequence of a transformer on-load tap changer, the system first obtains the type of the operation sequence. Based on the "three consecutive switching" operation sequence type, the system determines the expected changes in the local environmental conditions within the transformer oil chamber, such as a predicted increase in local temperature and increased oil flow disturbance. Based on these expected changes in local environmental conditions, the system modifies parameters representing the nonlinear coupling effect. For example, the solubility parameter of the gas in oil can be adjusted based on the temperature increase, and the gas diffusion coefficient can be adjusted based on the intensity of the oil flow disturbance, thereby obtaining adjusted parameters representing the nonlinear coupling effect. Subsequently, based on these adjusted nonlinear coupling effect parameters, combined with coupling effect characterization information (e.g., indicating the presence of a coupling effect type of thermal decomposition and arc discharge), the "three consecutive switching" operation sequence type, and the "expected gas generation characteristics of minor overheating under continuous switching" and "expected gas generation characteristics of micro-discharge under continuous switching" stored in a known fault event sequence pattern library, the system processes the change data of each characteristic gas component of the characteristic gas generation event. This process can employ a gas decomposition model based on an optimization algorithm that attempts to find a set of defect sources (e.g., slight overheating and microdischarges) and their respective gas contributions such that, when calculated based on the adjusted nonlinear coupling effect characterization parameters, the sum of these contributions most closely matches the actual measured gas component variation data. Through this process, the system separates or estimates the adjusted gas variation contributions from slight overheating and microdischarges, respectively, to obtain the adjusted gas variation for each defect source.

[0151] Through the above technical solution, this application can more accurately adjust the gas variation of each defect source based on the type of transformer on-load tap-changer operation sequence. By considering the impact of the operation sequence on local environmental conditions and correcting the parameters characterizing the nonlinear coupling effect, the processing of the variation of each characteristic gas component is more realistic, thereby more accurately separating or estimating the gas variation contributed by different potential defect sources. This improves the accuracy of determining whether multiple defect sources contribute to gas production in complex gas production events, thereby improving the reliability of transformer fault diagnosis.

[0152] refer to Figure 2 , the present application further proposes a transformer fault diagnosis system, which is applied to a transformer fault diagnosis method, and the system includes:

[0153] The data processing module obtains the raw data of dissolved gas in the transformer on-load tap-changer oil, normalizes and quantifies the distortion caused by gas escape, oil replenishment operation and oil temperature fluctuation in the raw data, and obtains the corrected gas concentration data;

[0154] The 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 transformer on-load tap-changer;

[0155] The characteristic event recognition module combines the operation records of the transformer on-load tap changer to identify characteristic gas production events related to the switching operation time of the transformer on-load tap changer in the net gas production data and extract the gas production information of the characteristic gas production events;

[0156] An event sequence construction module constructs a characteristic gas production event sequence by chronologically combining multiple characteristic gas production events;

[0157] The fault status analysis module compares and analyzes the behavior pattern of the characteristic gas production event sequence with the preset known fault event sequence pattern library to determine the internal fault status of the transformer on-load tap changer.

[0158] 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 background components unrelated to the fault from the preprocessed gas data, which can be implemented based on statistical models, signal filtering algorithms or machine learning models; the characteristic event recognition module refers to the functional unit responsible for locating and extracting gas production characteristics related to specific events in the gas data based on 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 reasoning system, an expert system, a machine learning classifier or a pattern recognition algorithm.

[0159] This solution organically connects all aspects of the diagnostic process through a system that includes modules such as data processing, background gas separation, characteristic event identification, event sequence construction, and fault status analysis. This forms a complete data processing and analysis chain, achieving automated and reliable diagnosis of transformer on-load tap-changer faults.

[0160] In some preferred embodiments, the present application is implemented as follows: The transformer fault diagnosis system can be implemented as a standalone hardware device containing a main control unit, such as an industrial-grade embedded computer. The data processing module can be implemented by a software program on the embedded computer. This program, through a communication interface with a gas sensor (e.g., an online dissolved gas monitoring device), acquires raw data on dissolved gas in the transformer's on-load tap-changer oil and performs 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, such as a statistical analysis or trend prediction method based on historical data, to the corrected gas concentration data to extract net gas production data. The characteristic event identification module, also implemented by a software program, receives transformer on-load tap-changer operation records from a transformer monitoring system (e.g., a SCADA system) and searches the net gas production data for gas production peaks or changes associated with the time of the operation. This program 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 generation events and their related information on an internal storage medium (such as a solid-state drive or database) within the device, sorting them by timestamp to construct a characteristic gas generation event sequence. The fault state analysis module is implemented by diagnostic software running on an embedded computer. This software loads a preset library of known fault event sequence patterns, which can be stored internally on the device or accessed via a network. The software then compares and analyzes the behavioral patterns of the constructed characteristic gas generation event sequence with those in the library, for example using a pattern matching algorithm or classification model. Ultimately, the module outputs a diagnosis of the internal fault state of the transformer's on-load tap changer. The entire system connects the various functional modules via an internal bus or network to enable data flow and processing.

[0161] The above-mentioned technical solution provides a system capable of practically implementing transformer fault diagnosis methods. This system effectively eliminates various distortions in raw gas data through the data processing module, improving data reliability. The background gas separation module accurately extracts net gas production data reflecting the transformer's on-load tapchanger's actual gas production behavior, enhancing the visualization of fault characteristics. The characteristic event recognition module, combined with operation records, precisely locates key gas production events related to switching operations, improving the targeted nature of diagnosis. The event sequence construction module organizes temporal information about fault evolution, providing a foundation for comprehensive analysis. The fault state analysis module intelligently determines the internal fault state of the transformer's on-load tapchanger by comparing event sequence patterns with a pattern library. This systematic implementation enables the automated and efficient execution of complex diagnostic methods, improving diagnostic accuracy and reliability, and effectively detecting and determining internal fault states of the transformer's on-load tapchanger.

[0162] The contents disclosed above are only preferred feasible embodiments of the present invention and do not limit the scope of protection of the present invention. Therefore, all equivalent technical changes made using the contents of the present invention description and drawings are included in the scope of protection of the present invention. In addition, the elements therein can be updated as technology develops.

Claims

1. A transformer fault diagnosis method, characterized in that: The method comprises the following steps: S1: Obtain the raw data of dissolved gas in transformer on-load tap-changer oil; S2: Normalize and quantify the distortion in the original data caused by gas escape, oil replenishment operation and oil temperature fluctuation 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 transformer on-load tap-changer; S4: Combined with the operation record of the transformer on-load tap-changer, identify characteristic gas generation events related to the switching operation time of the transformer on-load tap-changer in the net gas generation data, and extract gas generation information of the characteristic gas generation events; S5: construct a characteristic gas production event sequence by chronologically analyzing multiple characteristic gas production events; S6: Based on the behavioral pattern of the characteristic gas generation event sequence and the preset known fault event sequence pattern library, the internal fault state of the transformer on-load tap changer is determined.

2. A transformer fault diagnosis method according to claim 1, characterized in that: Step S6 specifically includes the following steps: Identifying characteristic gas generation events associated with a specific transformer on-load tap-changer operation sequence in a characteristic gas generation event sequence; For characteristic gas generation events associated with the operation sequence, analyzing the variation of characteristic gas components indicating different potential defect types contained therein, and obtaining variation data of each characteristic gas component; Based on the change data of each characteristic gas component, the type of operation sequence, and the expected gas change response characteristics of each single fault mode under the corresponding operation stimulus in the preset known fault event sequence pattern library, it is judged whether there are gas production contributions from multiple defect sources in the characteristic gas production event, and the judgment result of gas production contributions from multiple defect sources is obtained; Based on the judgment results, the existence and nature of each independent defect source in the concurrent fault are distinguished, and the distinction results of each independent defect source are obtained. Combined with the distinction results and the overall pattern of the characteristic gas production event sequence, a comparison analysis with the preset fault event pattern library is completed to determine the internal fault state of the transformer on-load tap-changer.

3. A transformer fault diagnosis method according to claim 2, characterized in that: The steps for determining whether there are gas generation contributions from multiple defect sources in a characteristic gas generation event include: Based on the variation data of each characteristic gas component, characterization information indicating the nonlinear coupling effect between different characteristic gas components is analyzed to obtain coupling effect characterization information; Combined with the coupling effect characterization information, the type of operation sequence, and the expected gas change response characteristics of each single fault mode under the corresponding operation stimulus in the known fault event sequence pattern library, the change data of each characteristic gas component is processed to separate or estimate the adjusted gas change contributed by different potential defect sources after considering the nonlinear coupling effect, and obtain the adjusted gas change of each defect source; According to the adjusted gas change amount of each defect source, it is judged whether there are gas production contributions of multiple defect sources in the characteristic gas production event, and a judgment result of the gas production contributions of multiple defect sources is obtained.

4. A transformer fault diagnosis method according to claim 3, characterized in that: The steps to obtain coupling effect characterization information include: Obtaining current service information of the transformer on-load tap-changer associated with the characteristic gas generation event, where the current service information reflects the service time or historical accumulated operating conditions of the transformer on-load tap-changer; Based on the current service information, a nonlinear coupling effect reference characteristic applicable to the current state of the transformer on-load tap changer is determined from a preset data set, wherein the data set includes nonlinear coupling effect reference characteristics corresponding to different service stages or different historical accumulated operating conditions; Based on the variation data of characteristic gas components and comparing them with the reference characteristics of nonlinear coupling effects, characterization information indicating the nonlinear coupling effects between different characteristic gas components is analyzed and obtained. The characterization information is used to reflect the current state of the transformer on-load tap changer.

5. A transformer fault diagnosis method according to claim 4, characterized in that: The steps of analyzing and obtaining characterization information indicating the nonlinear coupling effect between different characteristic gas components based on the variation data of the characteristic gas components and comparing the variation data with the nonlinear coupling effect reference characteristics include: Express the variation data of each characteristic gas component as the actual gas behavior pattern; The expected gas behavior pattern contained in the nonlinear coupling effect reference characteristics is used as the reference pattern; Based on the actual gas behavior pattern and the reference pattern, the quantitative difference between the two is calculated; Based on the quantitative differences and the preset discrimination logic related to the nonlinear coupling effect, and combined with the actual gas behavior pattern and the reference pattern, the coupling effect characterization information indicating the nonlinear coupling effect between different characteristic gas components and reflecting the current state of the transformer on-load tap-changer is analyzed and obtained.

6. A transformer fault diagnosis method according to claim 5, characterized in that: The steps of expressing the variation data of each characteristic gas component as an actual gas behavior pattern include: Obtaining the variation data of each characteristic gas component; Extracting statistical characteristic parameters within a predetermined time window from the variation data to obtain a set of statistical characteristic parameters; The statistical characteristic parameter set is used as the actual gas behavior model.

7. A transformer fault diagnosis method according to claim 6, characterized in that: The steps to analyze and obtain coupling effect characterization information include: Compare the quantitative differences to pre-set thresholds or rules in discriminant logic; Based on the comparison results and the characteristics of the actual gas behavior pattern and the reference pattern, and in combination with the discrimination logic, one or more nonlinear coupling effect types are determined from a preset set of nonlinear coupling effect types; The determined nonlinear coupling effect type is used as coupling effect characterization information that can reflect the current state of the transformer on-load tap changer.

8. A transformer fault diagnosis method according to claim 7, characterized in that: The steps to analyze nonlinear coupling effect types include: Analyze the quantitative differences and characteristics of actual gas behavior patterns and reference patterns, identify features that indicate specific gas reactions or solubility changes, and obtain relevant features; Compare and analyze the relevant features with a preset single gas reaction or solubility change pattern feature library to obtain a comparison analysis result; Based on the comparative analysis results, determine whether the difference between the actual gas behavior pattern and the reference pattern can be explained by a combination of one or more preset single reaction patterns; Based on the judgment results, one or more nonlinear coupling effect types that can reflect the actual complex coupling situation are determined.

9. A transformer fault diagnosis method according to claim 3, characterized in that: The steps of obtaining the adjusted gas variation of each defect source include: Obtaining the type of transformer on-load tap-changer operation sequence associated with the characteristic gas generation event; Based on the type of operation sequence, the expected change characteristics of the local environmental conditions related to the nonlinear coupling effect in the transformer oil chamber are determined, and the expected change characteristics of the local environmental conditions are obtained; According to the expected change characteristics of the local environmental conditions, the parameters representing the nonlinear coupling effect are modified to obtain the adjusted nonlinear coupling effect representation parameters; Based on the adjusted nonlinear coupling effect characterization parameters, and combined with the coupling effect characterization information, the type of transformer on-load tap-changer operation sequence, and the expected gas change response characteristics of each single fault mode under the corresponding operation excitation in the known fault event sequence pattern library, the change data of each characteristic gas component are processed to separate or estimate the adjusted gas change contributed by different potential defect sources after considering the nonlinear coupling effect, and obtain the adjusted gas change of each defect source.

10. A transformer fault diagnosis system, applied to the transformer fault diagnosis method according to claim 1, characterized in that: The system includes: The data processing module obtains the raw data of dissolved gas in the transformer on-load tap-changer oil, normalizes and quantifies the distortion caused by gas escape, oil replenishment operation and oil temperature fluctuation in the raw data, and obtains the corrected gas concentration data; The 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 transformer on-load tap-changer; The characteristic event recognition module combines the operation records of the transformer on-load tap changer to identify characteristic gas production events related to the switching operation time of the transformer on-load tap changer in the net gas production data and extract the gas production information of the characteristic gas production events; An event sequence construction module constructs a characteristic gas production event sequence by chronologically combining multiple characteristic gas production events; The fault status analysis module compares and analyzes the behavior pattern of the characteristic gas production event sequence with the preset known fault event sequence pattern library to determine the internal fault status of the transformer on-load tap changer.

Images