Transformer oil detection method and device, computer device, storage medium and computer program product
By collecting data from multiple oil sampling points of transformer oil and processing photoacoustic signals, combined with a trained model, transformer oil detection is performed, solving the problem of low accuracy in traditional manual detection and achieving higher detection accuracy.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SOUTHERN POWER GRID SENSING TECHNOLOGY (GUANGDONG) CO LTD
- Filing Date
- 2024-12-14
- Publication Date
- 2026-07-24
AI Technical Summary
Traditional transformer oil testing methods rely on manual testing, which involves subjective factors and results in low testing accuracy.
Transformer oil was collected from multiple oil inlets of the transformer, and after separation and processing, gas photoacoustic signals were obtained to determine the acetylene content. Feature vectors were extracted, and the trained model was used for prediction probability screening to obtain target detection results.
This improves the accuracy of transformer oil testing, avoids subjective errors caused by manual testing, and ensures the objectivity and accuracy of the test results.
Smart Images
Figure CN119804340B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of power grid technology, and in particular to a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for testing transformer oil. Background Technology
[0002] In power systems, testing the transformer oil in transformers is crucial for ensuring the safe operation of transformers.
[0003] In traditional technology, transformer oil testing is generally done manually; however, this manual testing method is subject to subjective factors and prone to errors, resulting in low accuracy in transformer oil testing. Summary of the Invention
[0004] Therefore, it is necessary to provide a transformer oil testing method, apparatus, computer equipment, computer-readable storage medium, and computer program product that can improve the accuracy of transformer oil testing in response to the above-mentioned technical problems.
[0005] Firstly, this application provides a method for testing transformer oil, including:
[0006] The transformer oil to be tested is separated to obtain the corresponding gas to be tested; the transformer oil is collected through multiple oil sampling ports of the transformer to be tested.
[0007] Acquire the photoacoustic signal corresponding to the gas to be detected;
[0008] Based on the photoacoustic signal, the current acetylene content value corresponding to the gas to be detected is determined;
[0009] Extract the feature vector of the current acetylene content value;
[0010] The feature vectors are input into multiple trained transformer oil detection models to obtain the target prediction probability of the transformer oil under various preset detection results.
[0011] From the various preset detection results, the preset detection result with the highest target prediction probability is selected as the target detection result corresponding to the transformer oil.
[0012] In one embodiment, before separating the transformer oil to be tested to obtain the gas to be tested corresponding to the transformer oil, the method further includes:
[0013] Obtain the structural information and oil flow path information of the transformer under test;
[0014] Extract the key structural information from the structural information, and extract the key oil flow path information from the oil flow path information;
[0015] Based on the key structural information and the key oil flow path information, multiple oil intake ports of the transformer to be tested are determined.
[0016] In one embodiment, extracting the feature vector of the current acetylene content value includes:
[0017] Obtain the target data type corresponding to the current acetylene content value;
[0018] Based on the target data type, query the correspondence between the data type and the feature extraction model to obtain the target feature extraction model corresponding to the current acetylene content value;
[0019] The historical acetylene content value corresponding to the gas to be detected is obtained, and the current acetylene content value is used as the main data, while the historical acetylene content value is used as the auxiliary data. These values are then input into the target feature extraction model to obtain the feature vector.
[0020] In one embodiment, the step of inputting the feature vector into multiple trained transformer oil detection models to obtain the target prediction probability of the transformer oil under various preset detection results includes:
[0021] The feature vectors are respectively input into the multiple trained transformer oil detection models to obtain the predicted probability of the transformer oil under various preset detection results output by each trained transformer oil detection model.
[0022] The predicted probability of the transformer oil output by each trained transformer oil detection model under various preset detection results is verified to obtain the verification result.
[0023] If the verification result meets the preset verification result, the predicted probability of the transformer oil output by each trained transformer oil detection model under the various preset detection results is weighted and summed to obtain the target predicted probability.
[0024] In one embodiment, after selecting the preset detection result with the highest target prediction probability from the various preset detection results as the target detection result corresponding to the transformer oil, the method further includes:
[0025] If the target detection result indicates that the transformer oil is abnormal, obtain the identification information corresponding to the transformer to be tested;
[0026] Based on the target detection results and the identification information, an early warning message corresponding to the transformer to be detected is generated;
[0027] The warning information is sent to the target terminal associated with the transformer to be tested.
[0028] In one embodiment, each trained transformer oil detection model is obtained by training it in the following manner:
[0029] The sample transformer oil is separated to obtain the corresponding sample gas; the sample transformer oil is collected from multiple oil sampling ports of the sample transformer.
[0030] Obtain the sample photoacoustic signal corresponding to the sample gas;
[0031] Based on the photoacoustic signal of the sample, the current acetylene content value corresponding to the sample gas is determined;
[0032] Extract the sample feature vector of the acetylene content value of the current sample;
[0033] The sample feature vectors are input into each transformer oil detection model to be trained, and the predicted detection results corresponding to the sample transformer oil are output by each transformer oil detection model to be trained.
[0034] Obtain the actual detection results corresponding to the sample transformer oil, and based on the difference between the predicted detection results and the actual detection results output by each transformer oil detection model to be trained, iteratively train each transformer oil detection model to be trained to obtain each trained transformer oil detection model.
[0035] Secondly, this application also provides a transformer oil testing device, comprising:
[0036] The gas acquisition module is used to separate the transformer oil to be tested to obtain the gas to be tested corresponding to the transformer oil; the transformer oil is collected through multiple oil sampling ports of the transformer to be tested;
[0037] The signal acquisition module is used to acquire the photoacoustic signal corresponding to the gas to be detected;
[0038] The content determination module is used to determine the current acetylene content value corresponding to the gas to be detected based on the photoacoustic signal.
[0039] The feature extraction module is used to extract the feature vector of the current acetylene content value;
[0040] The probability determination module is used to input the feature vector into multiple trained transformer oil detection models respectively to obtain the target prediction probability of the transformer oil under various preset detection results;
[0041] The result determination module is used to select the preset detection result with the highest target prediction probability from the various preset detection results, and use it as the target detection result corresponding to the transformer oil.
[0042] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:
[0043] The transformer oil to be tested is separated to obtain the corresponding gas to be tested; the transformer oil is collected through multiple oil sampling ports of the transformer to be tested.
[0044] Acquire the photoacoustic signal corresponding to the gas to be detected;
[0045] Based on the photoacoustic signal, the current acetylene content value corresponding to the gas to be detected is determined;
[0046] Extract the feature vector of the current acetylene content value;
[0047] The feature vectors are input into multiple trained transformer oil detection models to obtain the target prediction probability of the transformer oil under various preset detection results.
[0048] From the various preset detection results, the preset detection result with the highest target prediction probability is selected as the target detection result corresponding to the transformer oil.
[0049] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:
[0050] The transformer oil to be tested is separated to obtain the corresponding gas to be tested; the transformer oil is collected through multiple oil sampling ports of the transformer to be tested.
[0051] Acquire the photoacoustic signal corresponding to the gas to be detected;
[0052] Based on the photoacoustic signal, the current acetylene content value corresponding to the gas to be detected is determined;
[0053] Extract the feature vector of the current acetylene content value;
[0054] The feature vectors are input into multiple trained transformer oil detection models to obtain the target prediction probability of the transformer oil under various preset detection results.
[0055] From the various preset detection results, the preset detection result with the highest target prediction probability is selected as the target detection result corresponding to the transformer oil.
[0056] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:
[0057] The transformer oil to be tested is separated to obtain the corresponding gas to be tested; the transformer oil is collected through multiple oil sampling ports of the transformer to be tested.
[0058] Acquire the photoacoustic signal corresponding to the gas to be detected;
[0059] Based on the photoacoustic signal, the current acetylene content value corresponding to the gas to be detected is determined;
[0060] Extract the feature vector of the current acetylene content value;
[0061] The feature vectors are input into multiple trained transformer oil detection models to obtain the target prediction probability of the transformer oil under various preset detection results.
[0062] From the various preset detection results, the preset detection result with the highest target prediction probability is selected as the target detection result corresponding to the transformer oil.
[0063] The aforementioned transformer oil detection method, apparatus, computer equipment, storage medium, and computer program product first collect data from multiple oil sampling ports of the transformer to be tested to obtain the transformer oil to be tested. The transformer oil is then separated to obtain the corresponding gas to be tested. Next, the photoacoustic signal corresponding to the gas to be tested is acquired. Then, based on the photoacoustic signal, the current acetylene content value corresponding to the gas to be tested is determined, and the feature vector of the current acetylene content value is extracted. Then, the feature vector is input into multiple trained transformer oil detection models to obtain the target prediction probability of the transformer oil under various preset detection results. Finally, from the various preset detection results, the preset detection result with the highest target prediction probability is selected as the target detection result corresponding to the transformer oil. In this way, by collecting samples from multiple oil sampling ports of the transformer during the testing process, a more comprehensive and representative sample of the transformer oil can be obtained. This allows the subsequent gas samples to more accurately reflect the current acetylene content. Through a series of processes such as feature extraction and probability screening, the target detection results corresponding to the transformer oil can be obtained more accurately, which helps to improve the accuracy of the determination of the target detection results of the transformer oil, thereby improving the accuracy of transformer oil testing. Moreover, the entire process does not require manual intervention, avoiding the subjective factors and errors that are prone to occur in manual testing, which lead to lower accuracy of transformer oil testing, further improving the accuracy of transformer oil testing. Attached Figure Description
[0064] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0065] Figure 1 This is a flowchart illustrating a transformer oil detection method in one embodiment;
[0066] Figure 2 This is a flowchart illustrating the transformer oil detection method in another embodiment;
[0067] Figure 3 This is a flowchart illustrating a photoacoustic spectroscopy detection method for multi-channel oil-acetylene parallel measurement in a transformer, as shown in one embodiment.
[0068] Figure 4 This is a structural block diagram of a transformer oil detection device in one embodiment;
[0069] Figure 5This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation
[0070] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0071] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.
[0072] In one exemplary embodiment, such as Figure 1 As shown, a method for detecting transformer oil is provided. This embodiment illustrates the application of this method to a server; it is understood that this method can also be applied to a terminal, or to a system including a terminal and a server, and is implemented through interaction between the terminal and the server. The terminal can be, but is not limited to, various personal computers, laptops, smartphones, and tablets; the server can be a standalone server or a server cluster composed of multiple servers. In this embodiment, the method includes the following steps:
[0073] Step S101: Separate the transformer oil to be tested to obtain the gas to be tested corresponding to the transformer oil; the transformer oil is collected through multiple oil sampling ports of the transformer to be tested.
[0074] Transformer oil refers to the oil sample in a transformer.
[0075] The gas to be tested refers to the gas obtained by separating the transformer oil to be tested.
[0076] Among them, the transformer to be tested refers to the transformer that needs to be tested.
[0077] The oil sampling port refers to the location in the transformer under test used to collect transformer oil.
[0078] For example, the server identifies multiple oil sampling ports of the transformer to be tested; then, the server collects data from the multiple oil sampling ports of the transformer to be tested to obtain the transformer oil to be tested; then, the server separates the gas and liquid in the transformer oil to be tested to obtain the gas to be tested corresponding to the transformer oil.
[0079] Step S102: Obtain the photoacoustic signal corresponding to the gas to be detected.
[0080] Among them, photoacoustic signal refers to the signal generated by the gas to be detected based on the photoacoustic effect.
[0081] For example, the server introduces the gas to be detected into the photoacoustic cell and determines the light source corresponding to the gas to be detected based on the absorption spectrum characteristics of the gas to be detected; then, the server illuminates the gas to be detected in the photoacoustic cell with the light source according to the preset light path to obtain the photoacoustic signal corresponding to the gas to be detected.
[0082] Step S103: Determine the current acetylene content value corresponding to the gas to be detected based on the photoacoustic signal.
[0083] The current acetylene content value refers to the acetylene content of the gas to be detected at the current time.
[0084] For example, the server queries the correspondence between the photoacoustic signal and the acetylene content value based on the photoacoustic signal to obtain the current acetylene content value corresponding to the gas to be detected.
[0085] Step S104: Extract the feature vector of the current acetylene content value.
[0086] Among them, the feature vector is used to represent the characterization vector of the current acetylene content value.
[0087] For example, the server inputs the current acetylene content value into the feature extraction model, and the feature extraction model performs feature extraction processing on the current acetylene content value to obtain the feature vector of the current acetylene content value.
[0088] Step S105: Input the feature vectors into multiple trained transformer oil detection models to obtain the target prediction probability of transformer oil under various preset detection results.
[0089] Among them, the transformer oil detection model refers to a network model that can obtain the target detection result of transformer oil by using the feature vector of the current acetylene content value of the gas to be detected.
[0090] It should be noted that the various trained transformer oil detection models have different structures, such as RNN (Recurrent Neural Network) structure and CNN (Convolutional Neural Network) structure.
[0091] The preset test results refer to pre-defined test results, including whether the transformer oil is abnormal or not. It should be noted that the preset test results depend on the specific circumstances.
[0092] The target prediction probability is used to represent the overall likelihood that the preset detection result is correct.
[0093] For example, the server inputs the feature vectors into multiple trained transformer oil detection models to obtain the predicted probability of transformer oil under various preset detection results output by each trained transformer oil detection model; then, the server performs weighted summation on the predicted probabilities of transformer oil under various preset detection results output by each trained transformer oil detection model to obtain the target predicted probability of transformer oil under various preset detection results.
[0094] Step S106: Select the preset detection result with the highest target prediction probability from various preset detection results, and use it as the target detection result corresponding to transformer oil.
[0095] Among them, the target detection result refers to the preset detection result with the highest target prediction probability.
[0096] For example, the server selects the preset detection result with the highest target prediction probability from various preset detection results, and uses this preset detection result as the target detection result corresponding to the transformer oil.
[0097] In the above-mentioned transformer oil detection method, the transformer oil to be tested is first obtained by collecting samples from multiple oil sampling ports of the transformer to be tested. The transformer oil to be tested is then separated to obtain the gas to be tested corresponding to the transformer oil. The photoacoustic signal corresponding to the gas to be tested is then acquired. Next, based on the photoacoustic signal, the current acetylene content value corresponding to the gas to be tested is determined, and the feature vector of the current acetylene content value is extracted. Then, the feature vector is input into multiple trained transformer oil detection models to obtain the target prediction probability of the transformer oil under various preset detection results. Finally, the preset detection result with the highest target prediction probability is selected from the various preset detection results as the target detection result corresponding to the transformer oil. In this way, by collecting samples from multiple oil sampling ports of the transformer during the testing process, a more comprehensive and representative sample of the transformer oil can be obtained. This allows the subsequent gas samples to more accurately reflect the current acetylene content. Through a series of processes such as feature extraction and probability screening, the target detection results corresponding to the transformer oil can be obtained more accurately, which helps to improve the accuracy of the determination of the target detection results of the transformer oil, thereby improving the accuracy of transformer oil testing. Moreover, the entire process does not require manual intervention, avoiding the subjective factors and errors that are prone to occur in manual testing, which lead to lower accuracy of transformer oil testing, further improving the accuracy of transformer oil testing.
[0098] In an exemplary embodiment, step S101, before separating the transformer oil to be tested to obtain the gas to be tested corresponding to the transformer oil, specifically includes the following: obtaining the structural information and oil flow path information of the transformer to be tested; extracting key structural information from the structural information and key oil flow path information from the oil flow path information; and determining multiple oil sampling ports of the transformer to be tested based on the key structural information and key oil flow path information.
[0099] Among them, structural information is used to represent information related to the structure of the transformer under test.
[0100] Among them, the oil flow path information is used to represent the path information of the transformer oil in the transformer to be tested.
[0101] Among them, key structural information is used to represent structural information with an importance greater than the preset importance, such as the core and windings of the transformer to be tested.
[0102] Among them, the key oil flow path information is used to represent the oil flow path information with an importance greater than the preset importance, such as the path information of transformer oil in the main channel of the transformer to be tested.
[0103] For example, the server obtains the operating time and load information of multiple candidate transformers; then, the server selects candidate transformers whose operating time is greater than a preset operating time and whose load information is greater than a preset load information from the multiple candidate transformers, as the transformers to be tested; then, the server obtains the identification information of the transformers to be tested, and obtains the structural information and oil flow path information corresponding to the identification information, as the structural information and oil flow path information of the transformers to be tested; then, the server inputs the structural information and oil flow path information into the importance prediction model, and obtains the importance of each structural information and the importance of each oil flow path information according to the importance prediction model; then, the server selects structural information with an importance greater than a preset importance from each structural information, as key structural information, and selects oil flow path information with an importance greater than a preset importance from each oil flow path information, as key oil flow path information; finally, the server determines multiple oil sampling ports of the transformers to be tested based on the key structural information and key oil flow path information.
[0104] In this embodiment, by first obtaining the overall structural information and oil flow path information of the transformer to be tested, and then filtering out the key structural information and key oil flow path information, multiple oil sampling ports can be accurately determined, avoiding the problem that blind sampling will result in the samples not being able to fully reflect the real situation of the transformer.
[0105] In an exemplary embodiment, step S104 above, extracting the feature vector of the current acetylene content value, specifically includes the following: obtaining the target data type corresponding to the current acetylene content value; querying the correspondence between the data type and the feature extraction model according to the target data type to obtain the target feature extraction model corresponding to the current acetylene content value; obtaining the historical acetylene content value corresponding to the gas to be detected, and using the current acetylene content value as the main data and the historical acetylene content value as the auxiliary data, inputting them into the target feature extraction model to obtain the feature vector.
[0106] The target data type refers to the data type corresponding to the current acetylene content value, such as numerical data and range data.
[0107] The correspondence between data types and feature extraction models represents the relationship between them. For example, numerical data corresponds to a linear regression model, while interval data corresponds to a decision tree model.
[0108] The target feature extraction model refers to the feature extraction model corresponding to the current acetylene content value.
[0109] Among them, the historical acetylene content value refers to the acetylene content value of the gas to be tested within a historical time period (such as the past week, the past month, etc.).
[0110] Among them, the main data can refer to the data with a relatively large weight.
[0111] Auxiliary data can refer to data with relatively small weights.
[0112] For example, the server identifies the data type corresponding to the current acetylene content value according to preset rules, and uses it as the target data type. Then, the server queries the correspondence between the data type and the feature extraction model according to the target data type to obtain the feature extraction model corresponding to the current acetylene content value, which is used as the target feature extraction model. Next, the server retrieves the historical acetylene content values corresponding to the gas to be detected from the database. Then, the server uses the current acetylene content value as the primary data and the historical acetylene content value as the auxiliary data, and inputs them into the target feature extraction model to obtain the feature vector.
[0113] In this embodiment, by first determining the target data type corresponding to the current acetylene content value, and then selecting the target feature extraction model based on the correspondence between the data type and the feature extraction model, it is possible to ensure a high degree of fit between the selected model and the data type. Moreover, by using the current acetylene content value as the primary data and combining it with historical acetylene content values as auxiliary data, the target feature extraction model is input into the model, thereby achieving effective integration of data from different time periods, which is beneficial to improving the accuracy of feature vector determination.
[0114] In an exemplary embodiment, step S105, which involves inputting feature vectors into multiple trained transformer oil detection models to obtain target prediction probabilities of transformer oil under various preset detection results, specifically includes the following: inputting feature vectors into multiple trained transformer oil detection models to obtain the prediction probability of transformer oil output by each trained transformer oil detection model under various preset detection results; performing verification processing on the prediction probabilities of transformer oil output by each trained transformer oil detection model under various preset detection results to obtain verification results; and performing weighted summation processing on the prediction probabilities of transformer oil output by each trained transformer oil detection model under various preset detection results when the verification results meet the preset verification results to obtain the target prediction probability.
[0115] The prediction probability is used to represent the likelihood that each trained transformer oil detection model will correctly determine the preset detection result.
[0116] The verification result is used to indicate whether the difference between the predicted probabilities of transformer oil output by each trained transformer oil detection model under various preset detection results is greater than the preset difference.
[0117] The preset verification result indicates that the difference between the predicted probabilities of the transformer oil output by each trained transformer oil detection model under various preset detection results is not greater than the preset difference. For example, if the difference between the predicted probabilities of the output transformer oil under various preset detection results is less than or equal to 0.8, such as (0.7, 0.3) and (0.2, 0.8) respectively. It should be noted that if the predicted probabilities of the output transformer oil under various preset detection results are (0.01, 0.99) and (0.95, 0.05), that is, the difference between the two predicted probabilities is greater than 0.8, it indicates that the difference between the predicted probabilities of the transformer oil output by the trained transformer oil detection model under various preset detection results is greater than the preset difference.
[0118] For example, the server inputs feature vectors into multiple trained transformer oil detection models to obtain the predicted probability of transformer oil output by each trained transformer oil detection model under various preset detection results. Then, the server subtracts the predicted probabilities of transformer oil output by each trained transformer oil detection model under various preset detection results to obtain the prediction probability difference value corresponding to each trained transformer oil detection model. Next, the server verifies these prediction probability differences to obtain verification results. Then, if the verification results meet the preset verification results, the server performs a weighted summation of the predicted probabilities of transformer oil output by each trained transformer oil detection model under various preset detection results to obtain the target prediction probability.
[0119] In this embodiment, by verifying the predicted probabilities output by each model, unreasonable results that may be caused by model anomalies can be filtered out, thereby ensuring that the data used in subsequent calculations are relatively reliable, further optimizing the prediction results, so that the obtained target predicted probabilities can more accurately reflect the actual state of transformer oil.
[0120] In an exemplary embodiment, step S106, after selecting the preset detection result with the highest target prediction probability from various preset detection results as the target detection result corresponding to the transformer oil, specifically includes the following: when the target detection result indicates that the transformer oil is abnormal, obtaining the identification information corresponding to the transformer to be tested; generating early warning information corresponding to the transformer to be tested based on the target detection result and the identification information; and sending the early warning information to the target terminal associated with the transformer to be tested.
[0121] The identification information refers to the information used to uniquely distinguish the transformer, such as the transformer number.
[0122] Among them, the early warning information is used to indicate alarm information indicating that the transformer under test has an abnormality.
[0123] The target terminal refers to the terminal associated with the transformer to be tested.
[0124] For example, when the target detection result indicates that there is an abnormality in the transformer oil, the server retrieves the identification information corresponding to the transformer to be tested from the database; then, it retrieves a preset warning information template, and combines the preset warning information template, the target detection result, and the identification information to obtain the warning information corresponding to the transformer to be tested; then, the server determines the terminal associated with the transformer to be tested as the target terminal; finally, the server sends the warning information to the target terminal.
[0125] In this embodiment, early warning information is generated based on the target detection results and identification information, so that the abnormal situation of the transformer can be communicated to the personnel who need to know in a timely manner, so that relevant personnel can prepare corresponding measures in advance, and thus respond to possible transformer failures more efficiently, which helps to improve the timeliness of transformer failure handling.
[0126] In an exemplary embodiment, the transformer oil detection method provided in this application further includes a training step for each trained transformer oil detection model, specifically including the following: separating the sample transformer oil to obtain the sample gas corresponding to the sample transformer oil; collecting the sample transformer oil through multiple oil sampling ports of the sample transformer; acquiring the sample photoacoustic signal corresponding to the sample gas; determining the current sample acetylene content value corresponding to the sample gas based on the sample photoacoustic signal; extracting the sample feature vector of the current sample acetylene content value; inputting the sample feature vector into each transformer oil detection model to be trained to obtain the predicted detection result corresponding to the sample transformer oil output by each transformer oil detection model to be trained; obtaining the actual detection result corresponding to the sample transformer oil, and iteratively training each transformer oil detection model to be trained based on the difference between the predicted detection result and the actual detection result corresponding to the sample transformer oil output by each transformer oil detection model to be trained, to obtain each trained transformer oil detection model.
[0127] Among them, the sample transformer oil refers to the transformer oil used to train each transformer oil detection model to be trained.
[0128] The sample gas refers to the gas obtained by separating the sample transformer oil.
[0129] Among them, the sample transformer refers to the transformer in which the sample transformer oil is located.
[0130] Among them, the sample photoacoustic signal refers to the signal generated by the sample gas based on the photoacoustic effect.
[0131] The current acetylene content value refers to the acetylene content of the sample gas at the current time.
[0132] Among them, the sample feature vector refers to the feature vector of the current sample acetylene content value.
[0133] Each transformer oil detection model to be trained has a different structure, such as RNN structure, CNN structure, etc.
[0134] The predicted detection result refers to the predicted value of the detection result corresponding to the sample transformer oil output by each transformer oil detection model to be trained.
[0135] The actual test result refers to the actual value of the test result corresponding to the sample transformer oil.
[0136] For example, in response to a model training instruction for each transformer oil detection model to be trained, the server collects sample transformer oil from multiple oil sampling ports of the sample transformer; then, the server separates the sample transformer oil to obtain the corresponding sample gas; next, the server acquires the sample photoacoustic signal corresponding to the sample gas; then, the server determines the current sample acetylene content value corresponding to the sample gas based on the sample photoacoustic signal; then, the server extracts the sample feature vector of the current sample acetylene content value; finally, the server inputs the sample feature vector into each transformer oil detection model to be trained, obtaining the output sample of each transformer oil detection model to be trained. The server then obtains the predicted detection result for the sample transformer oil; next, it acquires the actual detection result for the sample transformer oil and calculates the loss value based on the difference between the predicted and actual detection results output by each transformer oil detection model to be trained; then, the server adjusts the model parameters of each transformer oil detection model to be trained based on the loss value; finally, the server retrains each transformer oil detection model with adjusted model parameters until the loss value obtained by each trained transformer oil detection model is less than the loss value threshold, at which point training stops, and each trained transformer oil detection model is taken as a completed transformer oil detection model.
[0137] In this embodiment, by pre-training the transformer oil detection model, it is convenient to predict the target detection result of the transformer oil after obtaining the current acetylene content value of the gas to be detected corresponding to the transformer oil in practical applications. Moreover, the transformer oil detection model receives new data in each iteration, and performs internal model improvement and optimization, which makes it easier to make predictions more effectively and improves the prediction accuracy of the transformer oil detection model.
[0138] In one exemplary embodiment, such as Figure 2 As shown, another method for testing transformer oil is provided. Taking the application of this method to a server as an example, the method includes the following steps:
[0139] Step S201: Obtain the structural information and oil flow path information of the transformer to be tested; extract the key structural information from the structural information and the key oil flow path information from the oil flow path information; determine multiple oil sampling ports of the transformer to be tested based on the key structural information and the key oil flow path information.
[0140] Step S202: Separate the transformer oil to be tested to obtain the gas to be tested corresponding to the transformer oil; the transformer oil is collected through multiple oil sampling ports of the transformer to be tested.
[0141] Step S203: Obtain the photoacoustic signal corresponding to the gas to be detected.
[0142] Step S204: Determine the current acetylene content value corresponding to the gas to be detected based on the photoacoustic signal.
[0143] Step S205: Obtain the target data type corresponding to the current acetylene content value; based on the target data type, query the correspondence between the data type and the feature extraction model to obtain the target feature extraction model corresponding to the current acetylene content value.
[0144] Step S206: Obtain the historical acetylene content value corresponding to the gas to be detected, and use the current acetylene content value as the main data and the historical acetylene content value as the auxiliary data, input them into the target feature extraction model to obtain the feature vector.
[0145] Step S207: Input the feature vectors into multiple trained transformer oil detection models respectively to obtain the predicted probability of transformer oil under various preset detection results output by each trained transformer oil detection model.
[0146] Step S208: Verify the predicted probability of transformer oil output by each trained transformer oil detection model under various preset detection results to obtain the verification result.
[0147] Step S209: If the verification result meets the preset verification result, the predicted probability of transformer oil output by each trained transformer oil detection model under various preset detection results is weighted and summed to obtain the target predicted probability.
[0148] Step S210: Select the preset detection result with the highest target prediction probability from various preset detection results, and use it as the target detection result corresponding to transformer oil.
[0149] In the aforementioned transformer oil testing method, by collecting samples from multiple oil sampling ports of the transformer under test, a more comprehensive and representative sample of the transformer oil can be obtained. This allows the subsequently obtained gas sample to more accurately reflect the current acetylene content. Through a series of processes such as feature extraction and probability screening, the target detection result corresponding to the transformer oil can be obtained more accurately, thus improving the accuracy of the target detection result and consequently increasing the overall accuracy of transformer oil testing. Furthermore, the entire process requires no manual intervention, avoiding the subjective factors and errors inherent in manual testing methods that can lead to lower accuracy in transformer oil testing, further enhancing the overall accuracy of transformer oil testing.
[0150] In an exemplary embodiment, to more clearly illustrate the transformer oil detection method provided in this application, the following detailed description uses a specific embodiment. In one embodiment, as... Figure 3 As shown, this application also provides a multi-channel parallel measurement photoacoustic spectroscopy detection method for transformer oil acetylene. In the process of detecting transformer oil, the method first collects data from multiple oil sampling ports of the transformer to be tested, obtains the transformer oil to be tested, and then separates the transformer oil to obtain the corresponding gas to be tested. Next, the photoacoustic signal corresponding to the gas to be tested is acquired. Then, based on the photoacoustic signal, the current acetylene content value corresponding to the gas to be tested is determined, and the feature vector of the current acetylene content value is extracted. Then, the feature vector is input into multiple trained transformer oil detection models to obtain the target prediction probability of the transformer oil under various preset detection results. Finally, from the various preset detection results, the preset detection result with the highest target prediction probability is selected as the target detection result corresponding to the transformer oil. Specifically, it includes the following:
[0151] The multi-channel parallel measurement architecture involved in this embodiment includes a multi-channel sampling system, an oil and gas separation unit, a photoacoustic spectroscopy detection unit, a field control and data acquisition and processing unit, and a station control server.
[0152] 1. Multi-channel sampling system: The multi-channel sampling system is the core component of this embodiment, simultaneously collecting oil samples from multiple oil sampling ports of the transformer. This design significantly improves the efficiency and coverage of the detection. Compared with traditional single-channel sampling, multi-channel sampling can more comprehensively reflect the distribution of dissolved gases in transformer oil.
[0153] The device includes at least three independent oil sampling ports, each connected to an oil sampling unit for collecting oil samples from different locations on the transformer. This allows the device to acquire a large amount of detection data in a short time, significantly improving the quality of the dissolved gas volume fraction-time series, and enabling accurate and rapid prediction of the dissolved gas equilibrium volume fraction using degassing theory models. Furthermore, the multi-channel sampling system can simulate the process of localized gas generation and slow dissolution and diffusion in real large transformers, thereby improving the applicability and accuracy of monitoring.
[0154] Each sampling port is designed to operate independently to ensure the representativeness of oil samples collected from different locations. The location of the sampling ports should be carefully selected based on the transformer's structure and oil flow path to maximize detection accuracy. Each sampling port should be connected to an oil sampling unit capable of precisely controlling the sampling volume and speed. These units should also be equipped with anti-contamination measures to ensure that the oil samples are not contaminated during collection. To ensure data comparability, all sampling units should be able to operate synchronously. This may require a central control system to coordinate the actions of each sampling unit.
[0155] 2. Oil-Gas Separation Unit: Each oil sample collection unit is connected to an oil-gas separation unit to separate the gas and liquid in the collected oil sample for subsequent spectroscopic analysis. The design of the oil-gas separation unit ensures the purity of the gas, thereby improving the accuracy of the detection.
[0156] The oil-gas separation unit is designed to efficiently separate gas and liquid while minimizing gas loss.
[0157] 3. Photoacoustic Spectroscopy Detection Unit: Each oil-gas separation unit is connected to a photoacoustic spectroscopy detection unit, which uses laser photoacoustic spectroscopy technology to detect the acetylene content in the separated gas. It has advantages such as high target selectivity, short reaction time, and non-destructive treatment of the sample, making it suitable for online monitoring. To distinguish the photoacoustic signals of different gases, the detection unit should have high spectral resolution, using a high-quality spectrometer and advanced signal processing technology. It is equipped with powerful data processing capabilities, enabling real-time analysis of detection data and identification of the presence and concentration of acetylene.
[0158] 4. On-site Control and Data Acquisition Processing Unit: Each photoacoustic spectroscopy detection unit is connected to an on-site control and data acquisition processing unit, used to control the detection process, acquire detection data, and perform preliminary processing. The design of these units ensures real-time data processing and analysis, providing support for subsequent decision-making.
[0159] The field control unit features an intuitive user interface, facilitating operator monitoring and control of the testing process. The data acquisition and processing unit can acquire data from each photoacoustic spectroscopy detection unit in real time, and perform preliminary data processing and storage. The field control and data acquisition and processing units have a high-speed and stable communication interface to transmit data to the station control server.
[0160] 5. Station Control Server: Data from all field control and data acquisition and processing units is transmitted to the station control server via fiber optic cable. The station control server is responsible for the final processing, storage, and remote monitoring of the data. The design of the station control server enables remote monitoring and data analysis, improving monitoring flexibility and response speed.
[0161] The station control server has sufficient storage space to store historical data for trend analysis and fault diagnosis, and can handle large amounts of data from multiple field control and data acquisition and processing units. The station control server supports remote access for remote monitoring.
[0162] like Figure 3 As shown, the device includes three independent oil sampling ports (1), each connected to an oil sample collection unit (2), which in turn connects to an oil-gas separation unit (3), and the oil-gas separation unit connects to a photoacoustic spectroscopy detection unit (4). Each photoacoustic spectroscopy detection unit connects to a field control and data acquisition and processing unit (5), and all field control and data acquisition and processing units connect to a station control server (7). The station control server (7) is responsible for receiving data from each field control and data acquisition and processing unit (5), performing centralized processing and analysis, and feeding back the results to maintenance personnel so that maintenance measures can be taken in a timely manner. This multi-channel parallel measurement architecture enables the device to achieve rapid, accurate, and real-time monitoring of the acetylene content in transformer oil, effectively improving the operational safety and maintenance efficiency of the transformer. It not only improves the efficiency of monitoring but also helps to detect potential transformer faults in a timely manner, thereby taking preventive measures to avoid possible accidents and losses.
[0163] In the above embodiments, by collecting data from multiple oil sampling ports of the transformer to be tested during the transformer oil testing process, a more comprehensive and representative transformer oil sample can be obtained. This allows the subsequently obtained gas sample to more accurately reflect the current acetylene content. Through a series of processes such as feature extraction and probability screening, the target detection result corresponding to the transformer oil can be obtained more accurately, which helps to improve the accuracy of the determination of the target detection result corresponding to the transformer oil, thereby improving the detection accuracy of the transformer oil. Moreover, the entire process does not require manual intervention, avoiding the subjective factors and errors that are prone to occur in manual testing, which lead to low detection accuracy of transformer oil, further improving the detection accuracy of transformer oil.
[0164] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.
[0165] Based on the same inventive concept, this application also provides a transformer oil testing device for implementing the transformer oil testing method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the transformer oil testing device provided below can be found in the limitations of the transformer oil testing method described above, and will not be repeated here.
[0166] In one exemplary embodiment, such as Figure 4 As shown, a transformer oil detection device is provided, comprising: a gas acquisition module 401, a signal acquisition module 402, a content determination module 403, a feature extraction module 404, a probability determination module 405, and a result determination module 406, wherein:
[0167] The gas acquisition module 401 is used to separate the transformer oil to be tested and obtain the gas to be tested corresponding to the transformer oil; the transformer oil is collected through multiple oil sampling ports of the transformer to be tested.
[0168] The signal acquisition module 402 is used to acquire the photoacoustic signal corresponding to the gas to be detected.
[0169] The content determination module 403 is used to determine the current acetylene content value of the gas to be detected based on the photoacoustic signal.
[0170] The feature extraction module 404 is used to extract the feature vector of the current acetylene content value.
[0171] The probability determination module 405 is used to input the feature vectors into multiple trained transformer oil detection models to obtain the target prediction probability of transformer oil under various preset detection results.
[0172] The result determination module 406 is used to select the preset detection result with the highest target prediction probability from various preset detection results, and use it as the target detection result corresponding to transformer oil.
[0173] In an exemplary embodiment, the transformer oil detection device further includes an oil sampling port determination module, used to acquire the structural information and oil flow path information of the transformer to be tested; extract key structural information from the structural information and key oil flow path information from the oil flow path information; and determine multiple oil sampling ports of the transformer to be tested based on the key structural information and key oil flow path information.
[0174] In an exemplary embodiment, the feature extraction module 404 is further configured to obtain the target data type corresponding to the current acetylene content value; query the correspondence between the data type and the feature extraction model according to the target data type to obtain the target feature extraction model corresponding to the current acetylene content value; obtain the historical acetylene content value corresponding to the gas to be detected, and input the current acetylene content value as the main data and the historical acetylene content value as the auxiliary data into the target feature extraction model to obtain the feature vector.
[0175] In an exemplary embodiment, the probability determination module 405 is further configured to input feature vectors into multiple trained transformer oil detection models respectively, to obtain the predicted probability of transformer oil output by each trained transformer oil detection model under various preset detection results; to perform verification processing on the predicted probability of transformer oil output by each trained transformer oil detection model under various preset detection results, to obtain verification results; and, if the verification results meet the preset verification results, to perform weighted summation processing on the predicted probability of transformer oil output by each trained transformer oil detection model under various preset detection results, to obtain the target predicted probability.
[0176] In an exemplary embodiment, the transformer oil detection device further includes an information sending module, which is used to obtain the identification information corresponding to the transformer to be tested when the target detection result indicates that there is an abnormality in the transformer oil; generate warning information corresponding to the transformer to be tested based on the target detection result and the identification information; and send the warning information to the target terminal associated with the transformer to be tested.
[0177] In an exemplary embodiment, the transformer oil detection device further includes a model training module for separating the sample transformer oil to obtain the sample gas corresponding to the sample transformer oil; the sample transformer oil is collected by sampling multiple oil ports of the sample transformer; the sample photoacoustic signal corresponding to the sample gas is acquired; the current sample acetylene content value corresponding to the sample gas is determined based on the sample photoacoustic signal; the sample feature vector of the current sample acetylene content value is extracted; the sample feature vector is input to each transformer oil detection model to be trained to obtain the predicted detection result corresponding to the sample transformer oil output by each transformer oil detection model to be trained; the actual detection result corresponding to the sample transformer oil is acquired, and the transformer oil detection model to be trained is iteratively trained based on the difference between the predicted detection result and the actual detection result output by each transformer oil detection model to be trained to obtain each trained transformer oil detection model.
[0178] Each module in the aforementioned transformer oil testing device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.
[0179] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 5 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores data such as current acetylene content and target prediction probabilities. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a transformer oil detection method.
[0180] Those skilled in the art will understand that Figure 5The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0181] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.
[0182] In one exemplary embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above-described method embodiments.
[0183] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.
[0184] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.
[0185] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0186] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.
Claims
1. A method for detecting transformer oil, characterized in that, The method includes: Obtain the structural information and oil flow path information of the transformer under test; input the structural information and oil flow path information into an importance prediction model to obtain the first importance of each structural information and the second importance of each oil flow path information; from the structural information, select structural information with the first importance greater than a preset importance as key structural information, and from the oil flow path information, select oil flow path information with the second importance greater than the preset importance as key oil flow path information; determine multiple oil sampling ports of the transformer under test based on the key structural information and the key oil flow path information. The transformer oil to be tested is separated to obtain the corresponding gas to be tested; the transformer oil is obtained by collecting samples from the multiple oil sampling ports. Acquire the photoacoustic signal corresponding to the gas to be detected; Based on the photoacoustic signal, the correspondence between the photoacoustic signal and the acetylene content value is queried to obtain the current acetylene content value corresponding to the gas to be detected; Obtain the target data type corresponding to the current acetylene content value; based on the target data type, query the correspondence between the data type and the feature extraction model to obtain the target feature extraction model corresponding to the current acetylene content value; obtain the historical acetylene content value corresponding to the gas to be detected, and use the current acetylene content value as the main data and the historical acetylene content value as the auxiliary data, input them into the target feature extraction model to obtain the feature vector of the current acetylene content value; The feature vector is input into multiple trained transformer oil detection models to obtain the predicted probability of the transformer oil under various preset detection results output by each trained transformer oil detection model. The predicted probabilities of the transformer oil under various preset detection results output by each trained transformer oil detection model are verified to obtain a verification result. The verification result indicates whether the difference between the predicted probabilities of the transformer oil under various preset detection results output by each trained transformer oil detection model is greater than a preset difference. If the verification result satisfies the preset verification result, the predicted probabilities of the transformer oil under various preset detection results output by each trained transformer oil detection model are weighted and summed to obtain the target predicted probability. The preset verification result indicates that the difference between the predicted probabilities of the transformer oil under various preset detection results output by each trained transformer oil detection model is not greater than the preset difference. From the various preset detection results, the preset detection result with the highest target prediction probability is selected as the target detection result corresponding to the transformer oil.
2. The method according to claim 1, characterized in that, The photoacoustic signal refers to the signal generated by the gas to be detected based on the photoacoustic effect.
3. The method according to claim 1, characterized in that, The current acetylene content value refers to the acetylene content value of the gas to be detected at the current time.
4. The method according to claim 1, characterized in that, After selecting the target detection result with the highest prediction probability from the various preset detection results as the target detection result corresponding to the transformer oil, the method further includes: If the target detection result indicates that the transformer oil is abnormal, obtain the identification information corresponding to the transformer to be tested; Based on the target detection results and the identification information, an early warning message corresponding to the transformer to be detected is generated; The warning information is sent to the target terminal associated with the transformer to be tested.
5. The method according to any one of claims 1 to 4, characterized in that, Each successfully trained transformer oil detection model is obtained through the following method: The sample transformer oil is separated to obtain the corresponding sample gas; the sample transformer oil is collected from multiple oil sampling ports of the sample transformer. Obtain the sample photoacoustic signal corresponding to the sample gas; Based on the photoacoustic signal of the sample, the current acetylene content value corresponding to the sample gas is determined; Extract the sample feature vector of the acetylene content value of the current sample; The sample feature vectors are input into each transformer oil detection model to be trained, and the predicted detection results corresponding to the sample transformer oil are output by each transformer oil detection model to be trained. Obtain the actual detection results corresponding to the sample transformer oil, and based on the difference between the predicted detection results and the actual detection results output by each transformer oil detection model to be trained, iteratively train each transformer oil detection model to be trained to obtain each trained transformer oil detection model.
6. A transformer oil testing device, characterized in that, The device includes: An oil port determination module is used to acquire structural information and oil flow path information of the transformer under test; input the structural information and oil flow path information into an importance prediction model to obtain a first importance of each structural information and a second importance of each oil flow path; from the structural information, select structural information with a first importance greater than a preset importance as key structural information, and from the oil flow path information, select oil flow path information with a second importance greater than the preset importance as key oil flow path information; based on the key structural information and the key oil flow path information, determine multiple oil ports of the transformer under test. A gas acquisition module is used to separate the transformer oil to be tested to obtain the gas to be tested corresponding to the transformer oil; the transformer oil is obtained by collecting data from the multiple oil sampling ports; The signal acquisition module is used to acquire the photoacoustic signal corresponding to the gas to be detected; The content determination module is used to query the correspondence between the photoacoustic signal and the acetylene content value based on the photoacoustic signal, and obtain the current acetylene content value corresponding to the gas to be detected; The feature extraction module is used to obtain the target data type corresponding to the current acetylene content value; according to the target data type, query the correspondence between the data type and the feature extraction model to obtain the target feature extraction model corresponding to the current acetylene content value; obtain the historical acetylene content value corresponding to the gas to be detected, and use the current acetylene content value as the main data and the historical acetylene content value as the auxiliary data, input them into the target feature extraction model to obtain the feature vector of the current acetylene content value; A probability determination module is used to input the feature vector into the plurality of trained transformer oil detection models respectively, to obtain the predicted probability of the transformer oil output by each trained transformer oil detection model under various preset detection results; to perform a verification process on the predicted probability of the transformer oil output by each trained transformer oil detection model under various preset detection results, to obtain a verification result; the verification result is used to indicate whether the difference between the predicted probabilities of the transformer oil output by each trained transformer oil detection model under various preset detection results is greater than a preset difference; if the verification result meets the preset verification result, a weighted summation process is performed on the predicted probabilities of the transformer oil output by each trained transformer oil detection model under various preset detection results to obtain the target predicted probability; the preset verification result is used to indicate that the difference between the predicted probabilities of the transformer oil output by each trained transformer oil detection model under various preset detection results is not greater than the preset difference. The result determination module is used to select the preset detection result with the highest target prediction probability from the various preset detection results, and use it as the target detection result corresponding to the transformer oil.
7. The apparatus according to claim 6, characterized in that, The device further includes an information sending module, used to acquire the identification information corresponding to the transformer under test when the target detection result indicates that the transformer oil is abnormal; generate warning information corresponding to the transformer under test based on the target detection result and the identification information; and send the warning information to a target terminal associated with the transformer under test.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.