Transformer oil chromatography fault diagnosis processing method and device based on large language model

Abstract

The invention discloses a transformer oil chromatography fault diagnosis processing method and device based on a large language model. The method comprises the steps that oil chromatography data, obtained through real-time detection, of a target transformer are obtained; if the oil chromatogram data meets the oil chromatogram dynamic threshold value condition, determining that the target transformer is suspected to be abnormal, then collecting the off-line oil chromatogram of the target transformer, and carrying out fault study and judgment based on the off-line oil chromatogram threshold value; if the analysis result of the fault study and judgment is abnormal, performing fault diagnosis on the target transformer based on the oil chromatography data of the target transformer and a preset transformer fault diagnosis model to obtain a diagnosis result; and sending the fault type in the diagnosis result and the oil chromatography data to a preset transformer oil chromatography analysis and diagnosis question-answer model to obtain a fault processing scheme corresponding to the target transformer output by the transformer oil chromatography analysis and diagnosis question-answer model. According to the invention, the accuracy of transformer fault diagnosis and the effect of transformer fault processing can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of transformers, and in particular, to a method and device for diagnosing and processing transformer oil chromatogram faults based on a large language model. Background Art

[0002] As a core device of the power system, the operating state of a transformer is directly related to the safety and stability of the entire power system. During the operation of a transformer, due to excessive internal environmental temperature or partial discharge of the device, some trace gases are generated and dissolved in the insulating oil. Generally, the higher the degree of device failure, the higher the content of the dissolved gases in the oil. For this reason, the traditional chromatographic analysis technology came into being. This technology regularly collects samples of insulating oil and uses a gas chromatograph to analyze the types and contents of dissolved gases. According to the gas analysis results, problems such as overheating and discharge inside the transformer are diagnosed.

[0003] In order to timely grasp the operating state of the transformer and identify potential faults as early as possible, the monitoring technology is gradually shifting from traditional regular offline oil chromatogram analysis to real-time online oil chromatogram monitoring. Online monitoring of transformer oil chromatogram can detect in real time and upload the content of dissolved gases in the insulating oil at regular intervals, enabling early detection of abnormal gas changes during the offline detection interval. This is of great significance for the early identification, warning, and long-term continuous monitoring of potential faults, and helps to improve the safety and reliability of transformer operation. The current oil chromatogram analysis technology for online monitoring data still mainly based on static threshold analysis, that is, analysis based on fixed thresholds, which is difficult to cope with the complex changes of transformer oil chromatogram under different operating environments, resulting in insufficient accuracy of diagnostic results. Therefore, how to provide a more accurate transformer fault diagnosis method is an urgent problem to be solved in the prior art. Summary of the Invention

[0004] In order to solve at least one of the technical problems in the above background art, the present invention proposes a method and device for diagnosing and processing transformer oil chromatogram faults based on a large language model.

[0005] To achieve the above object, according to one aspect of the present invention, there is provided a method for diagnosing and processing transformer oil chromatogram faults based on a large language model, the method comprising:

[0006] Obtaining the oil chromatogram data of the target transformer obtained by real-time detection;

[0007] If the oil chromatogram data of the target transformer meets the oil chromatogram dynamic threshold condition, it is determined that the target transformer is suspected of having an abnormality, and then the offline oil chromatogram of the target transformer is collected and a fault judgment is made based on the offline oil chromatogram threshold, wherein the oil chromatogram dynamic threshold condition is determined according to the oil chromatogram data of multiple transformers obtained by real-time detection;

[0008] If the analysis result of the fault judgment is abnormal, then based on the oil chromatogram data of the target transformer and a preset transformer fault diagnosis model, perform fault diagnosis on the target transformer to obtain a diagnosis result, where the transformer fault diagnosis model is obtained by training a preset machine learning model with training samples, and the training samples are oil chromatogram data marked with diagnosis results for model training;

[0009] Send the fault type in the diagnosis result and the oil chromatogram data of the target transformer to a preset transformer oil chromatogram analysis and diagnosis Q&A model to obtain a fault handling solution corresponding to the target transformer output by the transformer oil chromatogram analysis and diagnosis Q&A model, where the transformer oil chromatogram analysis and diagnosis Q&A model is obtained by training a preset large language model with a knowledge base constructed from oil chromatogram analysis and diagnosis data and transformer condition-based maintenance report data.

[0010] Optionally, the method for transformer oil chromatogram fault diagnosis and processing based on a large language model further includes:

[0011] Obtain the oil chromatogram data of multiple transformers obtained by real-time detection;

[0012] According to the maximum and minimum values of each item in the oil chromatogram data of the multiple transformers, determine the dynamic threshold of each item in the oil chromatogram data, and then summarize the dynamic thresholds of each item in the oil chromatogram data to obtain the oil chromatogram dynamic threshold condition.

[0013] Optionally, the determining the dynamic threshold of each item in the oil chromatogram data according to the maximum and minimum values of each item in the oil chromatogram data of the multiple transformers includes:

[0014] Use the following formula to determine the dynamic threshold of each item in the oil chromatogram data:

[0015] T=λ(x max -x min )+x min

[0016] where T is the dynamic threshold of the target item in the oil chromatogram data, x max is the maximum value of the target item in the oil chromatogram data of the multiple transformers, x min is the minimum value of the target item in the oil chromatogram data of the multiple transformers, and λ is a coefficient.

[0017] Optionally, the method for transformer oil chromatogram fault diagnosis and processing based on a large language model further includes:

[0018] Obtain a training sample set, where the training sample set contains multiple training samples, and the training samples are oil chromatogram data marked with diagnosis results for model training;

[0019] Train the XGBoost model according to the training sample set to obtain the transformer fault diagnosis model.

[0020] Optionally, the transformer oil chromatogram fault diagnosis processing method based on the large language model further includes:

[0021] Obtain the data of the materials collected with keywords of oil chromatogram analysis, oil chromatogram faults, and oil chromatogram fault diagnosis, as well as the data of the transformer condition maintenance report, and construct a knowledge base based on these data;

[0022] Train the preset large language model according to the knowledge base to obtain the transformer oil chromatogram analysis diagnosis Q&A model.

[0023] To achieve the above object, according to another aspect of the present invention, there is provided a transformer oil chromatogram fault diagnosis processing device based on the large language model, and the device includes:

[0024] An oil chromatogram data acquisition unit, configured to acquire the oil chromatogram data of the target transformer obtained by real-time detection;

[0025] An anomaly detection unit, configured to determine that the target transformer is suspected of having an anomaly if the oil chromatogram data of the target transformer meets the oil chromatogram dynamic threshold condition, and then collect the offline oil chromatogram of the target transformer and conduct fault research and judgment based on the offline oil chromatogram threshold, where the oil chromatogram dynamic threshold condition is determined according to the oil chromatogram data of multiple transformers obtained by real-time detection;

[0026] A fault diagnosis unit, configured to conduct fault diagnosis on the target transformer based on the oil chromatogram data of the target transformer and the preset transformer fault diagnosis model to obtain a diagnosis result if the analysis result of the fault research and judgment is an anomaly, where the transformer fault diagnosis model is obtained by training a preset machine learning model with training samples, and the training samples are the oil chromatogram data marked with diagnosis results for model training;

[0027] A fault handling scheme determination unit, configured to send the fault type in the diagnosis result and the oil chromatogram data of the target transformer to the preset transformer oil chromatogram analysis diagnosis Q&A model to obtain the fault handling scheme corresponding to the target transformer output by the transformer oil chromatogram analysis diagnosis Q&A model, where the transformer oil chromatogram analysis diagnosis Q&A model is obtained by training a preset large language model with a knowledge base constructed from oil chromatogram analysis diagnosis data and transformer condition maintenance report data.

[0028] Optionally, the transformer oil chromatogram fault diagnosis processing device based on the large language model further includes:

[0029] A multi-data summarization unit for obtaining oil chromatographic data of multiple transformers obtained by real-time detection;

[0030] An oil chromatographic dynamic threshold condition determination unit for determining the dynamic threshold of each item in the oil chromatographic data according to the maximum and minimum values of each item in the oil chromatographic data of the multiple transformers, and then summarizing the dynamic thresholds of each item in the oil chromatographic data to obtain the oil chromatographic dynamic threshold condition.

[0031] Optionally, the oil chromatographic dynamic threshold condition determination unit includes:

[0032] A calculation module for determining the dynamic threshold of each item in the oil chromatographic data by using the following formula:

[0033] T = λ(x max - x min ) + x min

[0034] where T is the dynamic threshold of the target item in the oil chromatographic data, x max is the maximum value of the target item in the oil chromatographic data of the multiple transformers, x min is the minimum value of the target item in the oil chromatographic data of the multiple transformers, and λ is a coefficient.

[0035] Optionally, the transformer oil chromatographic fault diagnosis processing device based on a large language model further includes:

[0036] A training sample set acquisition unit for acquiring a training sample set, where the training sample set contains multiple training samples, and the training samples are oil chromatographic data marked with diagnosis results for model training;

[0037] A transformer fault diagnosis model generation unit for training an XGBoost model according to the training sample set to obtain the transformer fault diagnosis model.

[0038] Optionally, the transformer oil chromatographic fault diagnosis processing device based on a large language model further includes:

[0039] A knowledge base construction unit for acquiring data on oil chromatographic analysis, oil chromatographic faults, and oil chromatographic fault diagnosis collected with keywords, as well as transformer condition maintenance report data, and constructing a knowledge base based on these data;

[0040] A transformer oil chromatographic analysis diagnosis Q&A model generation unit for training a preset large language model according to the knowledge base to obtain the transformer oil chromatographic analysis diagnosis Q&A model.

[0041] To achieve the above object, according to another aspect of the present invention, there is also provided a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned transformer oil chromatogram fault diagnosis processing method based on the large language model are implemented.

[0042] To achieve the above object, according to another aspect of the present invention, there is also provided a computer-readable storage medium, on which a computer program / instructions are stored. When the computer program / instructions are executed by a processor, the steps of the above-mentioned transformer oil chromatogram fault diagnosis processing method based on the large language model are implemented.

[0043] To achieve the above object, according to another aspect of the present invention, there is also provided a computer program product, including a computer program / instructions. When the computer program / instructions are executed by a processor, the steps of the above-mentioned transformer oil chromatogram fault diagnosis processing method based on the large language model are implemented.

[0044] The beneficial effects of the present invention are as follows:

[0045] In the embodiments of the present invention, by setting dynamic oil chromatogram threshold conditions and training a transformer fault diagnosis model through a machine learning model, the beneficial effect of accurately diagnosing transformer faults is achieved. In addition, the present invention also trains a transformer oil chromatogram analysis and diagnosis Q&A model through a large language model, which can quickly provide accurate fault handling solutions. Therefore, the present invention also achieves the beneficial effect of improving the effect of transformer fault handling. Description of the Drawings

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. In the drawings:

[0047] Figure 1 is the first flow chart of the transformer oil chromatogram fault diagnosis processing method based on the large language model in the embodiments of the present invention;

[0048] Figure 2 is the second flow chart of the transformer oil chromatogram fault diagnosis processing method based on the large language model in the embodiments of the present invention;

[0049] Figure 3 is the third flow chart of the transformer oil chromatogram fault diagnosis processing method based on the large language model in the embodiments of the present invention;

[0050] Figure 4It is the fourth flowchart of the method for diagnosing and processing transformer oil chromatogram faults based on large language models in the embodiments of the present invention;

[0051] Figure 5 It is the first structural block diagram of the device for diagnosing and processing transformer oil chromatogram faults based on large language models in the embodiments of the present invention;

[0052] Figure 6 It is the second structural block diagram of the device for diagnosing and processing transformer oil chromatogram faults based on large language models in the embodiments of the present invention;

[0053] Figure 7 It is the third structural block diagram of the device for diagnosing and processing transformer oil chromatogram faults based on large language models in the embodiments of the present invention;

[0054] Figure 8 It is the fourth structural block diagram of the device for diagnosing and processing transformer oil chromatogram faults based on large language models in the embodiments of the present invention;

[0055] Figure 9 It is a schematic diagram of a computer device in the embodiments of the present invention. Detailed implementation manners

[0056] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0057] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0058] It should be noted that the terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above accompanying drawings are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0059] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0060] Figure 1 is the first flowchart of the transformer oil chromatogram fault diagnosis and processing method based on the large language model in the embodiment of the present invention. As Figure 1 shown, in an embodiment of the present invention, the transformer oil chromatogram fault diagnosis and processing method based on the large language model of the present invention includes steps S101 to S104.

[0061] Step S101, obtain the oil chromatogram data of the target transformer obtained by real-time detection.

[0062] The oil chromatogram analysis of transformers is usually used to detect the dissolved gases in the insulating oil of transformers to evaluate the operating status and potential faults of transformers. These oil chromatogram data usually include the following types of information: Gas components: mainly analyze the gases dissolved in transformer oil, including: hydrogen (H2), methane (CH4), ethylene (C2H4), ethane (C2H6), propylene (C3H6), oxygen (O2), carbon dioxide (CO2), carbon monoxide (CO), other gases (such as nitrogen, etc.); Gas concentration: the specific concentration value of each gas, usually expressed in ppm (parts per million); Gas ratio: the ratio relationship between different gases, which can help judge the type of fault. For example, the change in the ratio of certain specific gases may indicate a specific fault mode (such as thermal fault, discharge fault or overload).

[0063] Step S102, if the oil chromatogram data of the target transformer meets the oil chromatogram dynamic threshold condition, it is determined that the target transformer is suspected of having an abnormality, and then the offline oil chromatogram of the target transformer is collected and fault judgment is carried out based on the offline oil chromatogram threshold, where the oil chromatogram dynamic threshold condition is determined according to the oil chromatogram data of multiple transformers obtained by real-time detection.

[0064] Existing oil chromatogram analysis technologies are based on fixed thresholds for analysis, which are difficult to cope with the complex changes in transformer oil chromatograms under different operating environments, and may result in false alarms or missed alarms. The present invention proposes a threshold condition that can be dynamically updated, improving the accuracy of oil chromatogram analysis.

[0065] In the present invention, if it is judged that the target transformer is suspected of having an abnormality through the oil chromatogram dynamic threshold condition, the offline oil chromatogram of the target transformer is collected, and fault judgment is carried out based on the offline oil chromatogram and the offline oil chromatogram threshold to judge whether an abnormality really occurs. If the offline oil chromatogram meets the offline oil chromatogram threshold, the analysis result is that an abnormality has occurred, otherwise, the analysis result is that no abnormality has occurred.

[0066] Offline oil chromatography refers to the offline analysis of transformer insulating oil, using techniques such as gas chromatography to detect and analyze the components and concentrations of dissolved gases in the oil. Offline oil chromatography usually requires sampling from the transformer and analyzing it in a laboratory. The offline oil chromatography threshold refers to a specific reference value or limit value used to evaluate the gas component concentrations in the transformer insulating oil under offline conditions.

[0067] Step S103, if the analysis result of the fault judgment is abnormal, then based on the oil chromatography data of the target transformer and a preset transformer fault diagnosis model, perform fault diagnosis on the target transformer to obtain a diagnosis result, where the transformer fault diagnosis model is obtained by training a preset machine learning model with training samples, and the training samples are oil chromatography data marked with diagnosis results for model training.

[0068] In an embodiment of the present invention, the diagnosis result includes specific fault types. In an embodiment of the present invention, the fault types may include: discharge fault, overheating fault, arc fault, insulation aging, short - circuit fault, and other potential faults, etc.

[0069] Step S104, send the fault types in the diagnosis result and the oil chromatography data of the target transformer to a preset transformer oil chromatography analysis and diagnosis Q&A model to obtain a fault handling solution corresponding to the target transformer output by the transformer oil chromatography analysis and diagnosis Q&A model, where the transformer oil chromatography analysis and diagnosis Q&A model is obtained by training a preset large - language model with a knowledge base constructed from oil chromatography analysis and diagnosis data and transformer condition - based maintenance report data.

[0070] In the present invention, the transformer oil chromatography analysis and diagnosis Q&A model is obtained by training a preset large - language model with a knowledge base constructed from oil chromatography analysis and diagnosis data and transformer condition - based maintenance report data.

[0071] The formulation of existing fault handling solutions overly relies on experts' experience and lacks intelligence. The present invention proposes a transformer oil chromatography analysis and diagnosis Q&A model based on a large - language model. This model can independently analyze and judge to obtain a fault handling solution, reduce the dependence on experts' experience, provide decision - making suggestions for maintenance personnel with different experience levels, and reduce the risk of human errors.

[0072] It can be seen that the present invention realizes the beneficial effect of accurately diagnosing transformer faults by setting dynamic oil chromatogram threshold conditions and training a transformer fault diagnosis model through a machine learning model. In addition, the present invention also trains a transformer oil chromatogram analysis and diagnosis Q&A model through a large language model, which can quickly provide accurate fault handling solutions. Therefore, the present invention also realizes the beneficial effect of improving the effect of transformer fault handling.

[0073] Figure 2 It is the second flowchart of the transformer oil chromatogram fault diagnosis and processing method based on a large language model in an embodiment of the present invention. As Figure 2 shown, in an embodiment of the present invention, the transformer oil chromatogram fault diagnosis and processing method based on a large language model of the present invention further includes step S201 and step S202.

[0074] Step S201, obtaining the oil chromatogram data of multiple transformers obtained by real-time detection.

[0075] Step S202, determining the dynamic threshold of each item in the oil chromatogram data according to the maximum and minimum values of each item in the oil chromatogram data of the multiple transformers, and then summarizing the dynamic thresholds of each item in the oil chromatogram data to obtain the oil chromatogram dynamic threshold condition.

[0076] The present invention dynamically determines the oil chromatogram threshold condition based on the oil chromatogram data of multiple transformers obtained by real-time detection. Compared with the current analysis method based on a fixed threshold, it can better cope with the complex changes of transformer oil chromatograms under different operating environments and improve the accuracy of oil chromatogram analysis.

[0077] In an embodiment of the present invention, the multiple transformers are specifically multiple transformers of the same type and with similar working environments.

[0078] In an embodiment of the present invention, the step S202 of determining the dynamic threshold of each item in the oil chromatogram data according to the maximum and minimum values of each item in the oil chromatogram data of the multiple transformers includes:

[0079] Using the following formula to determine the dynamic threshold of each item in the oil chromatogram data:

[0080] T = λ(x max - x min ) + x min

[0081] where, T is the dynamic threshold of the target item in the oil chromatogram data, x max is the maximum value of the target item in the oil chromatogram data of the multiple transformers, x min is the minimum value of the target item in the oil chromatogram data of the multiple transformers, and λ is a coefficient.

[0082] In one embodiment of the present invention, λ can specifically be the α - quantile of the beita4 distribution.

[0083] Figure 3 It is the third flowchart of the method for diagnosing and processing transformer oil chromatogram faults based on a large - language model in an embodiment of the present invention. As Figure 3 shown, in one embodiment of the present invention, the method for diagnosing and processing transformer oil chromatogram faults based on a large - language model of the present invention further includes step S301 and step S302.

[0084] Step S301: Obtain a training sample set, where the training sample set contains multiple training samples, and the training samples are oil chromatogram data marked with diagnosis results for model training.

[0085] In the present invention, in order to establish a training sample set, the present invention first obtains a large amount of historical oil chromatogram data, and then pre - processes these large amounts of historical oil chromatogram data to obtain a large amount of oil chromatogram data for model training.

[0086] In one embodiment of the present invention, the steps of pre - processing may specifically include:

[0087] Step 1.1: Process missing data using the adjacent - value filling method, and replace the missing value with the average value of the data at the adjacent positions before and after.

[0088] Step 1.2: Delete duplicate values.

[0089] Step 1.3: Identify abnormal data using a manually set range and treat it as a missing value for processing.

[0090] Step 1.4: To improve the stability of model training, use the standardization method to compress the data to make it close to a normal distribution and eliminate the influence of dimensions. The specific conversion formula is as follows:

[0091]

[0092]

[0093]

[0094] where, x * is the normalized data, x is the actual value of the historical data, μ is the mean of the historical data, and σ is the standard deviation of the historical data.

[0095] Step S302: Train the XGBoost model according to the training sample set to obtain the transformer fault diagnosis model.

[0096] In one embodiment of the present invention, the steps of training the XGBoost model specifically include:

[0097] Step 2.1: Randomly sample from the training sample set using the bootstrap sampling method to generate multiple different training subsets;

[0098] Step 2.2: For each training subset, construct a decision tree. During the splitting process of each node, randomly select a subset of features instead of using all features;

[0099] Step 2.3: The decision tree usually grows to the maximum depth until a certain stopping condition is met (such as reaching the minimum number of samples, reaching the maximum depth, or the node purity meets the requirements);

[0100] Step 2.4: Repeat steps 2.2 to 2.3 until the required number of decision trees is generated;

[0101] Step 2.5: Adopt a "voting" mechanism to select the category with the highest frequency as the final prediction result;

[0102] Step 2.6: Evaluate the model using the F1-score metric. The specific formula is as follows:

[0103]

[0104]

[0105]

[0106] Where TP is the true positive, FP is the false positive, and FN is the false negative.

[0107] Figure 4 is the fourth flowchart of the transformer oil chromatogram fault diagnosis and processing method based on the large language model in the embodiment of the present invention. As Figure 4 shown, in one embodiment of the present invention, the transformer oil chromatogram fault diagnosis and processing method based on the large language model of the present invention further includes step S401 and step S402.

[0108] Step S401: Obtain the data of materials collected with keywords such as oil chromatogram analysis, oil chromatogram fault, and oil chromatogram fault diagnosis, as well as the data of the transformer condition maintenance report, and construct a knowledge base based on these data.

[0109] In one embodiment of the present invention, the present invention collects data of materials such as literature, standards, and books with keywords such as "oil chromatogram analysis", "oil chromatogram fault", and "oil chromatogram fault diagnosis". At the same time, collect the transformer condition maintenance report. These data of materials are used to construct a local knowledge base to provide professional knowledge support for the large model to answer.

[0110] Step S402: Train a pre - set large - language model according to the knowledge base to obtain the transformer oil chromatogram analysis and diagnosis Q&A model.

[0111] In an embodiment of the present invention, the large - language model of the present invention can specifically adopt various open - source models such as BERT, RoBERTa, XLNet, ALBERT, and LLaMA.

[0112] To improve the accuracy and professionalism of answers, the present invention adopts a question - prompting method to guide the transformer oil chromatogram analysis and diagnosis Q&A model to better perform fine - grained answer extraction and processing, enabling the transformer oil chromatogram analysis and diagnosis Q&A model to more effectively process user queries and provide more accurate and comprehensive answers. In addition, the present invention constructs guiding prompts and integrates them into the transformer oil chromatogram analysis and diagnosis Q&A model.

[0113] In an embodiment of the present invention, the training steps of the large - language model can specifically include:

[0114] Data collection: Collect technical literature, fault cases, diagnostic rules, maintenance reports, etc. covering key fields such as oil chromatogram analysis, oil chromatogram faults, and oil chromatogram fault diagnosis. Transformer condition - based maintenance report data is particularly important and should include content such as fault types, oil chromatogram data, and historical maintenance records.

[0115] Data cleaning and pre - processing: Clean the redundancy and noise in the data and standardize the data. For example, uniformly format fault types, diagnostic results, oil chromatogram data, etc. to form structured data (such as JSON or CSV format) for subsequent processing.

[0116] Knowledge base structure design: Divide the knowledge base into multiple modules, such as "fault types", "oil chromatogram data", "diagnostic rules", "fault handling solutions", etc., and establish a structured or semi - structured data model.

[0117] Knowledge extraction and storage: Use natural language processing (NLP) tools to extract key diagnostic rules, fault patterns, and corresponding handling measures from text data. It can be stored as a graph database or a NoSQL - based database for quick query and cross - table association.

[0118] Data annotation: Annotate key data such as diagnostic processes, fault types, and fault handling solutions. Especially in oil chromatogram data, annotate the abnormal values and threshold ranges of gas component contents so that the model can effectively identify fault characteristics.

[0119] Feature extraction: Extract oil chromatogram analysis features (such as gas components, threshold judgment, etc.) and fault features (such as fault types, typical manifestations, etc.) and convert them into input features available for the model.

[0120] Model Selection: Based on the collected data and diagnostic Q&A requirements, BERT or similar large language models can be selected. For Q&A requirements in specific industries, open-source pre-trained models can be selected and then fine-tuned.

[0121] Training Data Preparation: Use the Q&A data in the knowledge base to generate training samples, combine different oil chromatogram data with diagnostic results to form a large number of Q&A pairs. Ensure that the training set includes various typical fault scenarios and treatment solutions to improve the generalization ability of the model.

[0122] Model Training: Use the knowledge base data to fine-tune and train the large language model so that it can generate accurate fault diagnosis and treatment plan outputs. Consider using transfer learning techniques to fine-tune based on existing pre-trained language models (such as BERT or similar models) to reduce the consumption of computing resources and training time.

[0123] Model Inference: Input the preset oil chromatogram data and fault types into the Q&A model, and let the model generate corresponding diagnosis and treatment plans. The model can be integrated into the diagnostic system through the API interface to automatically output diagnostic results.

[0124] Result Verification and Iteration: Verify the accuracy of the model output through actual cases, and collect feedback data for retraining. Gradually optimize the model parameters and knowledge base data to improve the accuracy and practicality of the Q&A model.

[0125] As can be seen from the above embodiments, the transformer oil chromatogram fault diagnosis and treatment method based on the large language model of the present invention has at least achieved the following beneficial effects:

[0126] 1. In the embodiment of the present invention, by setting dynamic oil chromatogram threshold conditions and training a transformer fault diagnosis model through a machine learning model, the beneficial effect of accurately diagnosing transformer faults is achieved;

[0127] 2. The present invention also trains a transformer oil chromatogram analysis and diagnosis Q&A model through a large language model, and this model can quickly provide accurate fault treatment plans. Therefore, the present invention also achieves the beneficial effect of improving the effect of transformer fault treatment.

[0128] It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0129] Based on the same inventive concept, an embodiment of the present invention further provides a transformer oil chromatogram fault diagnosis and processing device based on a large language model, which can be used to implement the transformer oil chromatogram fault diagnosis and processing method based on the large language model described in the above embodiments, as described in the following embodiments. Since the principle of the transformer oil chromatogram fault diagnosis and processing device based on the large language model for solving problems is similar to that of the transformer oil chromatogram fault diagnosis and processing method based on the large language model, the embodiments of the transformer oil chromatogram fault diagnosis and processing device based on the large language model can refer to the embodiments of the transformer oil chromatogram fault diagnosis and processing method based on the large language model, and the repeated parts will not be described again. As used hereinafter, the term "unit" or "module" may be a combination of software and / or hardware that can achieve a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0130] Figure 5 is the first structural block diagram of the transformer oil chromatogram fault diagnosis and processing device based on the large language model in the embodiment of the present invention, as Figure 5 shown. In an embodiment of the present invention, the transformer oil chromatogram fault diagnosis and processing device based on the large language model of the present invention includes:

[0131] An oil chromatogram data acquisition unit 1, configured to acquire the oil chromatogram data of a target transformer obtained by real-time detection;

[0132] An anomaly detection unit 2, configured to determine that the target transformer is suspected of having an anomaly if the oil chromatogram data of the target transformer meets the oil chromatogram dynamic threshold condition, and then collect the offline oil chromatogram of the target transformer and perform fault judgment based on the offline oil chromatogram threshold, where the oil chromatogram dynamic threshold condition is determined based on the oil chromatogram data of multiple transformers obtained by real-time detection;

[0133] A fault diagnosis unit 3, configured to perform fault diagnosis on the target transformer based on the oil chromatogram data of the target transformer and a preset transformer fault diagnosis model to obtain a diagnosis result if the analysis result of the fault judgment is an anomaly, where the transformer fault diagnosis model is obtained by training a preset machine learning model with training samples, and the training samples are oil chromatogram data marked with diagnosis results for model training;

[0134] A fault handling solution determination unit 4, configured to send the fault type in the diagnosis result and the oil chromatogram data of the target transformer to a preset transformer oil chromatogram analysis and diagnosis Q&A model, and obtain the fault handling solution corresponding to the target transformer output by the transformer oil chromatogram analysis and diagnosis Q&A model, where the transformer oil chromatogram analysis and diagnosis Q&A model is obtained by training a preset large language model using a knowledge base constructed from oil chromatogram analysis and diagnosis data and transformer condition-based maintenance report data.

[0135] Figure 6 It is the second structural block diagram of the transformer oil chromatogram fault diagnosis and processing device based on a large language model in an embodiment of the present invention, as Figure 6 shown. In an embodiment of the present invention, the transformer oil chromatogram fault diagnosis and processing device based on a large language model of the present invention further includes:

[0136] A multi-data summary unit 5, configured to obtain the oil chromatogram data of multiple transformers obtained by real-time detection;

[0137] An oil chromatogram dynamic threshold condition determination unit 6, configured to determine the dynamic threshold of each item in the oil chromatogram data according to the maximum and minimum values of each item in the oil chromatogram data of the multiple transformers, and then summarize the dynamic thresholds of each item in the oil chromatogram data to obtain the oil chromatogram dynamic threshold condition.

[0138] In an embodiment of the present invention, the oil chromatogram dynamic threshold condition determination unit 6 includes:

[0139] A calculation module, configured to determine the dynamic threshold of each item in the oil chromatogram data using the following formula:

[0140] T = λ(x max - x min ) + x min

[0141] where T is the dynamic threshold of the target item in the oil chromatogram data, x max is the maximum value of the target item in the oil chromatogram data of the multiple transformers, x min is the minimum value of the target item in the oil chromatogram data of the multiple transformers, and λ is a coefficient.

[0142] Figure 7 It is the third structural block diagram of the transformer oil chromatogram fault diagnosis and processing device based on a large language model in an embodiment of the present invention, as Figure 7 shown. In an embodiment of the present invention, the transformer oil chromatogram fault diagnosis and processing device based on a large language model of the present invention further includes:

[0143] A training sample set acquisition unit 7, configured to acquire a training sample set, where the training sample set includes a plurality of training samples, and the training samples are oil chromatogram data marked with diagnostic results for model training;

[0144] A transformer fault diagnosis model generation unit 8, configured to train an XGBoost model according to the training sample set to obtain the transformer fault diagnosis model.

[0145] Figure 8 It is the fourth structural block diagram of the transformer oil chromatogram fault diagnosis processing device based on the large language model in the embodiments of the present invention. As Figure 8 shown, in an embodiment of the present invention, the transformer oil chromatogram fault diagnosis processing device based on the large language model of the present invention further includes:

[0146] A knowledge base construction unit 9, configured to acquire data of materials collected with keywords of oil chromatogram analysis, oil chromatogram faults, and oil chromatogram fault diagnosis, as well as transformer condition maintenance report data, and construct a knowledge base based on these data;

[0147] A transformer oil chromatogram analysis diagnosis Q&A model generation unit 10, configured to train a preset large language model according to the knowledge base to obtain the transformer oil chromatogram analysis diagnosis Q&A model.

[0148] To achieve the above object, according to another aspect of the present application, a computer device is further provided. As Figure 9 shown, the computer device includes a memory, a processor, a communication interface, and a communication bus. A computer program that can run on the processor is stored on the memory. When the processor executes the computer program, the steps in the method of the above embodiment are implemented.

[0149] The processor may be a central processing unit (CPU). The processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. chips, or a combination of the above various types of chips.

[0150] The memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer-executable programs, and units, such as the corresponding program units in the above method embodiments of the present invention. The processor executes various functional applications and work data processing of the processor by running the non-transitory software programs, instructions, and modules stored in the memory, that is, the method in the above method embodiments is implemented.

[0151] The memory may include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created by the processor and the like. In addition, the memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory may optionally include a memory remotely provided relative to the processor, and these remote memories can be connected to the processor through a network. Examples of the above network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0152] The one or more units are stored in the memory and, when executed by the processor, execute the method in the above embodiments.

[0153] Specific details of the above computer device can be understood by referring to the corresponding relevant descriptions and effects in the above embodiments, and will not be elaborated here.

[0154] To achieve the above object, according to another aspect of the present application, there is also provided a computer-readable storage medium storing a computer program, and when the computer program is executed in a computer processor, the steps in the above transformer oil chromatogram fault diagnosis and processing method based on a large language model are implemented. Those skilled in the art can understand that to implement all or part of the processes in the above method embodiments, it can be completed by instructing relevant hardware through a computer program. The program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the above method embodiments. Among them, the storage medium can be a magnetic disk, an optical disc, a read-only memory (ROM), a random access memory (RAM), a flash memory, a hard disk drive (HDD), or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above types of memories.

[0155] To achieve the above object, according to another aspect of the present application, there is also provided a computer program product, including a computer program / instructions, which when executed by a processor, implement the steps of the above-mentioned method for diagnosing transformer oil chromatogram faults based on a large language model.

[0156] Obviously, those skilled in the art should understand that the above-mentioned modules or steps of the present invention can be implemented by a general-purpose computing device. They can be concentrated on a single computing device or distributed on a network composed of multiple computing devices. Optionally, they can be implemented by program codes executable by the computing device. Thus, they can be stored in a storage device and executed by the computing device, or they can be separately fabricated into individual integrated circuit modules, or multiple modules or steps among them can be fabricated into a single integrated circuit module to be implemented. In this way, the present invention is not limited to any specific combination of hardware and software.

[0157] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A transformer oil chromatographic fault diagnosis and processing method based on a large language model, characterized in that: include: Acquire oil chromatogram data of the target transformer obtained through real-time detection; If the oil chromatogram data of the target transformer meets the oil chromatogram dynamic threshold condition, it is determined that the target transformer is suspected to be abnormal, and then the offline oil chromatogram of the target transformer is collected and the fault is diagnosed based on the offline oil chromatogram threshold, wherein the oil chromatogram dynamic threshold condition is determined based on the oil chromatogram data of multiple transformers detected in real time; If the analysis result of the fault diagnosis is abnormal, the target transformer is diagnosed for fault based on the oil chromatogram data of the target transformer and a preset transformer fault diagnosis model to obtain a diagnosis result, wherein the transformer fault diagnosis model is obtained by training a preset machine learning model using training samples, and the training samples are oil chromatogram data for model training with the diagnosis results marked; The fault type in the diagnosis result and the oil chromatography data of the target transformer are sent to a preset transformer oil chromatography analysis diagnosis question and answer model to obtain a fault handling solution corresponding to the target transformer output by the transformer oil chromatography analysis diagnosis question and answer model, wherein the transformer oil chromatography analysis diagnosis question and answer model is obtained by training a preset large language model using a knowledge base constructed by oil chromatography analysis diagnosis data and transformer status inspection report data.

2. The transformer oil chromatographic fault diagnosis and processing method based on a large language model according to claim 1 is characterized in that: Also includes: Acquire oil chromatogram data of multiple transformers detected in real time; According to the maximum value and the minimum value of each item in the oil chromatogram data of the multiple transformers, the dynamic threshold of each item in the oil chromatogram data is determined, and then the dynamic threshold of each item in the oil chromatogram data is summarized to obtain the oil chromatogram dynamic threshold condition.

3. The transformer oil chromatographic fault diagnosis and processing method based on a large language model according to claim 2 is characterized in that: Determining the dynamic threshold of each item in the oil chromatogram data according to the maximum value and the minimum value of each item in the oil chromatogram data of the plurality of transformers includes: The dynamic threshold value for each item in the oil chromatography data is determined using the following formula: T=λ(x max -x min )+x min Where T is the dynamic threshold of the target item in the oil chromatography data, x max is the maximum value of the target item in the oil chromatogram data of the multiple transformers, x min is the minimum value of the target item in the oil chromatogram data of the multiple transformers, and λ is a coefficient.

4. The transformer oil chromatographic fault diagnosis and processing method based on a large language model according to claim 1 is characterized in that: Also includes: Acquire a training sample set, wherein the training sample set includes a plurality of training samples, and the training samples are oil chromatogram data for model training with diagnostic results marked; The XGBoost model is trained according to the training sample set to obtain the transformer fault diagnosis model.

5. The transformer oil chromatographic fault diagnosis and processing method based on a large language model according to claim 1 is characterized in that: Also includes: Obtain data collected with oil chromatography analysis, oil chromatography failure and oil chromatography failure diagnosis as keywords, as well as transformer condition maintenance report data, and build a knowledge base based on these data; The preset large language model is trained according to the knowledge base to obtain the transformer oil chromatographic analysis diagnosis question-answering model.

6. A transformer oil chromatographic fault diagnosis and processing device based on a large language model, characterized in that: include: An oil chromatogram data acquisition unit, used to acquire oil chromatogram data of a target transformer detected in real time; an abnormality detection unit, for determining that the target transformer is suspected to be abnormal if the oil chromatogram data of the target transformer meets the oil chromatogram dynamic threshold condition, and then collecting the offline oil chromatogram of the target transformer and performing fault analysis based on the offline oil chromatogram threshold, wherein the oil chromatogram dynamic threshold condition is determined based on the oil chromatogram data of multiple transformers detected in real time; A fault diagnosis unit, for performing fault diagnosis on the target transformer based on the oil chromatogram data of the target transformer and a preset transformer fault diagnosis model to obtain a diagnosis result if the analysis result of the fault analysis is abnormal, wherein the transformer fault diagnosis model is obtained by training a preset machine learning model using training samples, and the training samples are oil chromatogram data for model training with the diagnosis results marked; A fault handling solution determination unit is used to send the fault type in the diagnosis result and the oil chromatography data of the target transformer to a preset transformer oil chromatography analysis diagnosis question and answer model to obtain a fault handling solution corresponding to the target transformer output by the transformer oil chromatography analysis diagnosis question and answer model, wherein the transformer oil chromatography analysis diagnosis question and answer model is obtained by training a preset large language model using a knowledge base constructed by oil chromatography analysis diagnosis data and transformer status inspection and maintenance report data.

7. The transformer oil chromatographic fault diagnosis and processing device based on a large language model according to claim 6 is characterized in that: Also includes: A multi-data aggregation unit, used to obtain oil chromatogram data of multiple transformers detected in real time; The oil chromatogram dynamic threshold condition determination unit is used to determine the dynamic threshold of each item in the oil chromatogram data according to the maximum value and the minimum value of each item in the oil chromatogram data of the multiple transformers, and then summarize the dynamic threshold of each item in the oil chromatogram data to obtain the oil chromatogram dynamic threshold condition.

8. The transformer oil chromatographic fault diagnosis and processing device based on a large language model according to claim 7 is characterized in that: The oil chromatography dynamic threshold condition determination unit comprises: A calculation module is used to determine the dynamic threshold value of each item in the oil chromatography data using the following formula: T=λ(x max -x min )+x min Where T is the dynamic threshold of the target item in the oil chromatography data, x max is the maximum value of the target item in the oil chromatogram data of the multiple transformers, x min is the minimum value of the target item in the oil chromatogram data of the multiple transformers, and λ is a coefficient.

9. The transformer oil chromatographic fault diagnosis and processing device based on a large language model according to claim 6 is characterized in that: Also includes: A training sample set acquisition unit, used for acquiring a training sample set, wherein the training sample set includes a plurality of training samples, and the training samples are oil chromatogram data for model training with diagnostic results marked; The transformer fault diagnosis model generation unit is used to train the XGBoost model according to the training sample set to obtain the transformer fault diagnosis model.

10. The transformer oil chromatographic fault diagnosis and processing device based on a large language model according to claim 6, characterized in that: Also includes: A knowledge base construction unit is used to obtain information data collected with oil chromatography analysis, oil chromatography failure and oil chromatography failure diagnosis as keywords, as well as transformer status maintenance report data, and to construct a knowledge base based on these data; The transformer oil chromatographic analysis diagnosis question-answering model generating unit is used to train a preset large language model according to the knowledge base to obtain the transformer oil chromatographic analysis diagnosis question-answering model.

11. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

12. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

13. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images