Adaptive Adjustment Method of Power Transformer State Parameters

Abstract

The present application provides a method for adaptively adjusting the state parameters of a power transformer, which relates to the technical field of power systems, including: extracting a component service life set from a power transformer maintenance log; retrieving the operating state fluctuation amplitude of similar power transformers; obtaining an operating control parameter, and predicting the operating state through an operating state evaluation model; constructing a first operating state fluctuation interval; when the first operating state fluctuation interval intersects with the operating state warning interval, adjusting the operating control parameter, generating a control parameter adjustment strategy, adjusting the operating state according to the control parameter adjustment strategy, and issuing an early warning. This application can solve the technical problem in the prior art that the evaluation of the state parameters is not accurate due to the failure to accurately consider the impact of the different service conditions of each component on the fluctuation of the operating state, thereby improving the accuracy and adaptability of the adjustment of the state parameters of the power transformer.

Description

Technical Field

[0001] The present application relates to the technical field of power systems, and in particular to a method for adaptively adjusting state parameters of power transformers. Background Art

[0002] As a vital device in the power system, the operating status of the power transformer is directly related to the stability and safety of the power system. By real-time monitoring of the various status parameters of the transformer and combining historical operating data with fault records, the status parameters can be adjusted in real time to ensure that the transformer operates in the best working state, thereby reducing the probability of failure, extending the service life of the equipment, and improving the safety and reliability of the power system. Existing power transformer status parameter adjustment methods usually construct an operating status assessment model based on transformers of the same model for analysis. However, during the service process, each component of the transformer may experience varying degrees of aging and loss due to factors such as the working environment, load fluctuations, and operating time. Existing methods fail to fully consider the specific impact of these factors on operating status fluctuations, resulting in a lack of accuracy in the operating status assessment results, and may lead to incorrect state parameter adjustments, resulting in greater uncertainty when adjusting the state parameters.

[0003] In summary, the prior art has a technical problem in that the impact of different service conditions of various components on the fluctuation of the operating state is not accurately considered, resulting in inaccurate evaluation of state parameters, thereby affecting the flexibility of adjusting the state parameters of the power transformer. Summary of the Invention

[0004] The purpose of this application is to provide a method for adaptively adjusting the state parameters of a power transformer, so as to solve the technical problem in the prior art that the evaluation of the state parameters is not accurate enough due to the failure to accurately consider the impact of the different service conditions of each component on the fluctuation of the operating state, thereby affecting the flexibility of the adjustment of the state parameters of the power transformer.

[0005] In view of the above problems, the present application provides a method for adaptively adjusting the state parameters of a power transformer, wherein the method for adaptively adjusting the state parameters of a power transformer includes: extracting a component service life set from a power transformer maintenance log; retrieving the operating state fluctuation amplitude of similar power transformers based on the component service life set as a constraint; obtaining operating control parameters, predicting the operating state through an operating state evaluation model, and obtaining a benchmark operating state; taking the benchmark operating state as the center of the interval, and combining the operating state fluctuation amplitude, constructing a first operating state fluctuation interval; when the first operating state fluctuation interval has an intersection with the operating state warning interval, adjusting the operating control parameters, generating a control parameter adjustment strategy, adjusting the operating state according to the control parameter adjustment strategy, and issuing a warning.

[0006] The technical solution provided in this application has at least the following technical effects or advantages:

[0007] The invention relates to a method for extracting a component service life set from a power transformer maintenance log; retrieving the operating state fluctuation amplitude of similar power transformers based on the component service life set as a constraint; obtaining an operating control parameter, performing an operating state prediction using an operating state evaluation model, and obtaining a baseline operating state; constructing a first operating state fluctuation interval with the baseline operating state as the interval center and in combination with the operating state fluctuation amplitude; adjusting the operating control parameter when the first operating state fluctuation interval intersects with an operating state warning interval, generating a control parameter adjustment strategy, adjusting the operating state according to the control parameter adjustment strategy, and issuing an early warning. In other words, the invention relates to extracting a component service life set from a power transformer maintenance log, retrieving the operating state fluctuation amplitude of similar transformers, and performing an operating state prediction using an operating state evaluation model based on the obtained operating control parameter; constructing a first operating state fluctuation interval based on the baseline operating state and in combination with the operating state fluctuation amplitude; and if the first operating state fluctuation interval intersects with an operating state warning interval, adjusting the operating control parameter according to the prompt of the warning interval and performing an operating state prediction again, thereby obtaining an adjusted control parameter. This enhances the accuracy of state parameter evaluation and improves the accuracy and adaptability of power transformer state parameter adjustment.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, which can be implemented in accordance with the contents of the description, and to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically listed below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in this application or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely exemplary, and a person of ordinary skill in the art can obtain other drawings based on the provided drawings without creative work.

[0010] Figure 1 This is a flow chart of the method for adaptively adjusting the state parameters of a power transformer according to the present application;

[0011] Figure 2 This is a flow chart of obtaining the operating state fluctuation amplitude in the power transformer state parameter adaptive adjustment method of this application. DETAILED DESCRIPTION

[0012] The present application solves the technical problem in the prior art of inaccurate evaluation of state parameters due to the inaccurate consideration of the impact of different service conditions of various components on the fluctuation of operating status, thereby affecting the flexibility of adjusting the state parameters of the power transformer, by providing a method for adaptively adjusting the state parameters of the power transformer. By extracting the service life set of each component from the maintenance log of the power transformer, retrieving the operating state fluctuation amplitude of similar transformers, and performing operating state prediction based on the obtained operating control parameters through an operating state evaluation model, a first operating state fluctuation interval is constructed based on the baseline operating state and the operating state fluctuation amplitude; if the first operating state fluctuation interval intersects with the operating state warning interval, the operating control parameters are adjusted according to the prompt of the warning interval, and the operating state prediction is performed again to obtain the adjusted control parameters, thereby enhancing the accuracy of the state parameter evaluation and improving the accuracy and adaptability of the power transformer state parameter adjustment.

[0013] Below, the technical solutions in this application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited to the example embodiments described herein. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. It should also be noted that, for the convenience of description, only the parts related to this application, rather than all of them, are shown in the accompanying drawings.

[0014] For examples, please see the attached Figure 1 The present application provides a method for adaptively adjusting state parameters of a power transformer, wherein the method for adaptively adjusting state parameters of a power transformer specifically comprises the following steps:

[0015] S100: Extract component service life sets from power transformer maintenance logs.

[0016] Specifically, a power transformer maintenance log is a document that records the maintenance, inspection, upkeep, and troubleshooting of a power transformer during operation. It details each repair, failure, and inspection of each component, including the date, details, and component replacements. Components refer to the individual parts that make up a power transformer, such as windings, oil tanks, cooling systems, and insulation materials. Service life refers to the time it takes for the performance of individual components of a power transformer to degrade to the specified end-of-life standard after a certain period of actual use. Service life is affected by a variety of factors, such as load, ambient temperature, and operating hours.

[0017] By scanning, reading, and parsing the maintenance logs of power transformers, records containing the maintenance and service history of each component are collected, and the service life information of each component is extracted from the maintenance logs. For example, the usage time and service history of components are obtained from the power transformer maintenance logs based on keywords (such as replacement date and usage duration). The service life information of all components in the transformer is aggregated into a set, namely the component service life set. For example, suppose the service history information of each component of a power transformer is recorded in the maintenance log, and the service life of the winding is 5 years, the cooling system is 7 years, and the insulating oil is 6 years. This information is extracted from the text using a log data extraction tool and stored as a component service life collection: the winding service start time is 2020-01-01, the service end time is 2025-01-01, and the service duration is 5 years; the cooling system service start time is 2019-01-01, the service end time is 2026-01-01, and the service duration is 7 years; the insulating oil service start time is 2020-01-01, the service end time is 2026-01-01, and the service duration is 6 years. By understanding the service life of each component, the evaluation model can more accurately predict the operating status fluctuations of the transformer and avoid ignoring the impact of service history on the equipment's operating status.

[0018] S200: Retrieving operating state fluctuation amplitudes of similar power transformers based on the component service life set as a constraint.

[0019] Specifically, the database is searched for power transformers of the same type, and the set of component service life records and operating status fluctuation amplitude records for each similar power transformer are determined. Using the component service life set as a constraint, similar transformers are searched for in a step-by-step manner. Combining component service life and operating status fluctuation records, the operating status fluctuation amplitude is accurately assessed. Using the component service life set as a constraint, similar power transformers and their operating status fluctuation records are searched step-by-step. A centralized value evaluation of the multi-level operating status fluctuation records is performed to obtain the operating status fluctuation amplitude.

[0020] Specifically, using the component service life set as the first-level constraint, a search is performed on similar power transformers to obtain a first-level set of similar transformers. The first-level transformers have a first set of operating state fluctuation record amplitudes and a first-level set of transformer component service life records. The first-level transformer component service life records set corresponding to the first-level transformers is then used as a second-level constraint. A second-level set of similar transformers is retrieved from the first-level transformers. The second-level transformers have a second set of operating state fluctuation record amplitudes and a second-level set of transformer component service life records. This process is repeated until N-level transformers are retrieved, each with an N-th set of operating state fluctuation record amplitudes, where 5 ≥ N ≥ 2, and N is an integer.

[0021] A centralized value evaluation is performed on the operating status fluctuation record amplitude sets of each similar transformer at each level. Starting from the first operating status fluctuation record amplitude set, the operating status fluctuation record amplitude sets of the next layer are grouped in sequence until the Nth layer. For each layer of data, multiple groups of operating status fluctuation record amplitudes in the previous layer are used as the grouping basis. Starting from the Nth layer, a centralized value evaluation is performed on multiple groups of operating status fluctuation record amplitudes, and multiple centralized amplitude values ​​are calculated. The centralized value evaluation results of the Nth layer are backtracked and updated to the corresponding groups in the N-1th layer to generate an updated N-1th operating status fluctuation record amplitude set. The centralized value evaluation of the updated N-1th operating status fluctuation record amplitude set is continued, and the update is backtracked to the previous layer, iterating layer by layer until the centralized value evaluation of the first layer is completed. After iteration, the centralized value evaluation result of the first operating status fluctuation record amplitude set is obtained, which is called the operating status fluctuation amplitude. The multi-level retrieval method can effectively improve the accuracy and reliability of the data and avoid errors caused by single constraints.

[0022] S300: Obtaining operation control parameters, performing operation state prediction using an operation state evaluation model, and obtaining a reference operation state.

[0023] Furthermore, the present application S300 includes:

[0024] S310: Collecting the operation log of the power transformer that meets the preset model and the service life of each component is less than the service life threshold, wherein the power transformer operation log includes operation control parameter recording data and operation status monitoring data; S320: Using the operation status monitoring data as supervision and the operation control parameter recording data as input to train a random forest model to obtain the operation status evaluation model; S330: Inputting the operation control parameters into the operation status evaluation model and outputting the benchmark operation status.

[0025] Specifically, the preset model refers to the power transformer model that meets specific technical specifications or parameter requirements, which is used to ensure the uniformity and adaptability of data analysis. The service life threshold refers to the restriction condition used for screening, which represents the upper limit of the service life of the component. For example, if the service life threshold is 10 years, only transformers with a service life of less than 10 years are selected. The power transformer operation logs that meet the preset model and whose components have a service life of less than the service life threshold are screened out from the transformer database. The power transformer operation log records the data during the operation of the transformer, including operation control parameter record data (such as voltage and current set values) and operation status monitoring data (such as actual voltage, current, temperature, etc.).

[0026] Ensure that the operating control parameters and operating status data are aligned in time. Remove outliers and missing values, and process the time series data of the operation log into feature vectors. Use operating status monitoring data (such as temperature and vibration) as supervisory labels (target values) and operating control parameter recording data (such as load current and input voltage) as model input features. Use the random forest algorithm to train the operating status assessment model, construct 100 decision trees with a maximum tree depth of 10, and split the dataset into training and test sets (e.g., 80% training and 20% testing). By training the operating status assessment model with the training set, the random forest model learns the relationship between input data (operating control parameters) and output data (operating status monitoring data), enabling it to predict the operating status under new operating control parameters.

[0027] The model is verified on the test set, and the output is compared with the predicted value and the true value until the operation status evaluation model reaches the convergence condition, such as the accuracy reaches 98%. The trained operation status evaluation model is applied to the new transformer operation data to predict its operation status and evaluate the operation performance.

[0028] The real-time transformer operation monitoring system acquires operating control parameters, including key parameters set during power transformer operation, such as load current, input voltage, power factor, and switching mode. The extracted operating control parameters are organized into the format required by the model, including data normalization or standardization to ensure that the input data aligns with the feature scale used during model training. The operating control parameters are then fed into the trained operating status assessment model to predict the operating status, including indicators such as winding temperature, vibration amplitude, and oil level. The baseline operating status is defined as the theoretically predicted state of the device under ideal operating control parameters—that is, the ideal operating state that the power transformer should achieve under normal operating conditions. The model's predicted output is analyzed to determine the range or threshold for the normal operating status. The predicted operating status is compared with historical normal operating data to verify the accuracy of the prediction. If the predicted results are within the normal operating range, they are considered the baseline operating status. By acquiring the operating control parameters and using the operating status assessment model for prediction, the baseline operating status of the power transformer can be understood in real time, helping to promptly identify deviations between the actual operating status and the baseline operating status, allowing appropriate adjustments to ensure safe and stable operation of the power transformer.

[0029] S400: Constructing a first operating state fluctuation interval with the reference operating state as the interval center and in combination with the operating state fluctuation amplitude.

[0030] Specifically, the baseline operating state is used as the center of the interval, and combined with the operating state fluctuation amplitude, a fluctuation interval is constructed to represent the acceptable fluctuation range of the transformer's state parameters during normal operation. The range formed by adding or subtracting the fluctuation amplitude from the baseline operating state is used to define the normal operating boundary. The fluctuation interval is calculated using the following formula: Interval = [Baseline Value − Fluctuation Amplitude, Baseline Value + Fluctuation Amplitude]. For example, the baseline operating state is: winding temperature = 80°C, vibration amplitude = 0.010g; winding temperature fluctuation amplitude: ±5°C; vibration amplitude fluctuation amplitude: ±0.002g; winding temperature fluctuation range: [75°C, 85°C], and vibration amplitude fluctuation range: [0.008g, 0.012g]. Using the baseline operating state and fluctuation amplitude, the acceptable operating state range is clearly defined, defining the normal operating boundary of the equipment.

[0031] S500: When the first operating state fluctuation interval and the operating state warning interval have an intersection, adjust the operating control parameter, generate a control parameter adjustment strategy, adjust the operating state according to the control parameter adjustment strategy, and issue a warning.

[0032] Specifically, a pre-defined operating status warning interval, defined as a threshold range for equipment operational safety, identifies status intervals that may cause failures or abnormalities. A determination is made as to whether the first operating status fluctuation interval overlaps with the operating status warning interval. If so, the operational control parameters need to be adjusted. The intersection refers to the portion of the two intervals that overlaps, indicating that the operating status fluctuation interval falls within the warning range.

[0033] If there is an intersection, it means that some operating conditions have entered the warning interval and the operating control parameters need to be adjusted to restore them to a safe range. If there is no intersection, the equipment is operating normally and no adjustment is required. Adjust the operating control parameters based on the specific nature of the warning interval. Obtain the operating control parameter constraint interval, that is, determine the safe range of the operating control parameters. The operating control parameter constraint interval is used to limit the upper and lower bounds of the operating control parameter value range to ensure that parameter adjustments do not exceed the safety or design limits of the equipment. Within the operating control parameter constraint interval, a random algorithm is used to generate new control parameter values ​​for exploring different operating conditions. The updated operating control parameters are used as input to the operating condition evaluation model to predict the corresponding operating condition. The model prediction results are used as the new baseline operating condition, that is, the updated baseline operating condition. Once an intersection between the fluctuation interval and the warning interval is detected, the operating control parameters are adjusted to move the transformer's operating condition out of the warning interval and ensure that the equipment operates within the safe range. When the operating status fluctuation interval intersects the warning interval, an immediate warning is issued to alert personnel of potential risks in the current equipment status. The warning includes which transformer components are currently in a dangerous state, which parameters are outside the safe range, the risk level, and the generated control parameter adjustment strategy. Warning information is directly displayed on a visual panel and transmitted to the transformer's control center, triggering appropriate automated operations, including control parameter adjustment strategies. Warning information is not static. After the control parameters are adjusted, the transformer's operating status continues to be monitored, and the warning interval or adjustment strategy is adjusted based on real-time data.

[0034] If the transformer intersects with the operating status warning interval and is at the upper limit of the warning interval or has partially exceeded the warning interval, it means that it is in a relatively serious fault. At this time, it only needs to issue an alarm to remind the staff to deal with it as soon as possible. There is no need to generate a control parameter adjustment strategy. Therefore, the fault at this time cannot be adjusted and the staff needs to repair or replace it.

[0035] Based on the intersection of the first operating state fluctuation interval and the warning interval, a control parameter adjustment strategy is generated. This strategy is an adjustment plan formulated based on the specific conditions of the transformer (such as load, temperature, vibration, etc.). Specifically, with the updated baseline operating state as the center of the interval, the upper and lower bounds are calculated in combination with the operating state fluctuation amplitude to obtain the second operating state fluctuation interval, which is used to measure whether the adjusted operating state is within the safe range. Determine whether the constructed second operating state fluctuation interval and the operating state warning interval overlap, that is, whether there is an intersection. If there is an intersection, it means that the current operating state is within the warning range or close to the warning range, indicating that the operating state still has risks and it is necessary to continue adjusting the operating control parameters until the operating state fluctuation interval and the operating state warning interval have no intersection.

[0036] When the second operating state fluctuation interval and the operating state warning interval do not intersect, adjustments to the operating control parameters cease. Based on the updated operating control parameters, a control parameter adjustment strategy is generated, and specific parameter adjustments are executed to adjust the transformer's operating state and ensure its operating state returns to a safe range. By dynamically adjusting the operating control parameters, the operating state is maintained within a safe range. When the operating state approaches or enters the warning interval, rapid adjustments are made to prevent failure or further deterioration, ensuring that the transformer's operating state remains within a safe and effective range while avoiding unnecessary warnings and potential failure risks.

[0037] Further, as attached Figure 2 As shown, this application S200 includes:

[0038] S210: Using the component service life set as the first-level constraint, retrieve the first-level similar transformers, wherein the first-level similar transformers have a first operating state fluctuation record amplitude set and a first-level transformer component service record life set; S220: Using the first-level transformer component service record life set as the second-level constraint, retrieve the second-level similar transformers, wherein the second-level similar transformers have a second operating state fluctuation record amplitude set and a second-level transformer component service record life set; S230: Up to using the N-1 level transformer component service record life set as the N-level constraint, retrieve N-level similar transformers, wherein the N-level similar transformers have the N-th operating state fluctuation record amplitude set, 5≥N≥2, N is an integer; S240: Perform centralized value evaluation on the first operating state fluctuation record amplitude set, the second operating state fluctuation record amplitude set, and the N-th operating state fluctuation record amplitude set to obtain the operating state fluctuation amplitude.

[0039] Specifically, transformers of the same model are obtained, and the component service life set is used as the first-level constraint. A preliminary screening is performed on these transformers of the same model. That is, using the component service life set as a benchmark, the deviation in component service life from all transformers of the same model to be analyzed is calculated to obtain a deviation set. By setting a deviation distance threshold, transformers with large deviations are screened out, referred to as first-level similar transformers. Simultaneously, the component service life records for these screened transformers are added to the first-level transformer component service life record set and the first-level operating state fluctuation record amplitude set.

[0040] Using the first-level transformer component service life records set for the first-level similar transformers as the second-level constraint, repeat the above steps for a second-level screening to obtain second-level similar transformers. Similarly, the second-level similar transformers also have a second operating state fluctuation record amplitude set and a second-level transformer component service life record set. This process is repeated until N-level similar transformers are obtained, and the corresponding N-th operating state fluctuation record amplitude set is extracted. Where 5 ≥ N ≥ 2, and N is an integer. When searching for similar power transformers, one can start with the first-level similar transformers and gradually refine them to the second, third, fourth, and finally fifth levels. This process can be considered a step-by-step refinement process. Each level of search further refines the transformer similarity criteria based on the previous level, thereby finding a more accurate group of similar transformers. N is upper bounded at 5 because, in practical applications, a certain range of transformer component service life sets can meet most analysis and optimization requirements. Further refinement (i.e., N > 5) may result in the required number of similar transformers becoming too scarce, making effective data analysis impossible. Furthermore, the complexity of data retrieval and analysis increases, leading to higher costs. N is greater than or equal to 2 because if the sample size is too small, the evaluation results may be inaccurate or even lose statistical significance.

[0041] A centralized value evaluation is performed on the first, second, and Nth sets of operating state fluctuation amplitude records. Central values ​​are calculated for multiple data sets (e.g., multiple sets of operating state fluctuation amplitude values) to obtain representative central values ​​(e.g., mean, median). This is typically done by calculating the corresponding mean, median, and standard deviation. The weights of the mean, median, and standard deviation entered by the user are then obtained through the user-side interface, and the corresponding centralized value is calculated using a weighted average. This process is then backfilled and evaluated layer by layer until the centralized value of the first set of operating state fluctuation amplitude records is updated and calculated. After the first-level centralized value evaluation is completed, the result is defined as the operating state fluctuation amplitude, which characterizes the overall operating state fluctuation characteristics. By using a hierarchical constraint approach to screen similar transformer sets, the service life differences and operating state characteristics of the transformers can be comprehensively considered, thereby improving the accuracy of the condition assessment. This hierarchical constraint approach avoids the potential for bias accumulation caused by a single constraint, ensuring the scientific nature of the screening results.

[0042] Furthermore, the present application further comprises the following steps:

[0043] S241: Group the second operating state fluctuation record amplitude set according to the first operating state fluctuation record amplitude set to obtain multiple groups of second operating state fluctuation record amplitudes; S242: Group the third operating state fluctuation record amplitude set according to the multiple groups of second operating state fluctuation record amplitudes to obtain multiple groups of third operating state fluctuation record amplitudes; S243: Until the Nth operating state fluctuation record amplitude set is grouped according to multiple groups of N-1th operating state fluctuation record amplitudes to obtain multiple groups of Nth operating state fluctuation record amplitudes, wherein any one operating state fluctuation record amplitude of the multiple groups of N-1th operating state fluctuation record amplitudes One-to-one correspondence between a group of operating state fluctuation record amplitudes and the multiple groups of Nth operating state fluctuation record amplitudes; S244: traverse the multiple groups of Nth operating state fluctuation record amplitudes to perform centralized value evaluation, obtain multiple Nth operating state fluctuation centralized amplitudes, add the multiple Nth operating state fluctuation centralized amplitudes to the corresponding multiple groups of N-1th operating state fluctuation record amplitudes, and obtain updated multiple groups of N-1th operating state fluctuation record amplitudes; S245: perform iterative centralized value evaluation based on the updated multiple groups of N-1th operating state fluctuation record amplitudes until the centralized value evaluation is performed on the updated first operating state fluctuation record amplitude set to obtain the operating state fluctuation amplitude.

[0044] Specifically, starting from the first operating state fluctuation record amplitude set, the next layer of operating state fluctuation record amplitude sets are grouped in sequence. For example, based on the fluctuation records of the first layer, the second layer of data is divided into multiple groups, and each group of data corresponds to a certain operating state fluctuation record amplitude of the previous layer. Repeat the above grouping steps until the Nth layer. For each layer of data, multiple groups of operating state fluctuation record amplitudes of the previous layer are used as the basis for grouping. For example, the grouping results of the second layer are used to divide the third layer of data, and the grouping results of the third layer are used to divide the fourth layer of data, and so on. Specifically, first, the second operating state fluctuation record amplitude set is grouped according to the first operating state fluctuation record amplitude set to obtain multiple groups of second operating state fluctuation record amplitudes. Then, the third operating state fluctuation record amplitude set is grouped according to these grouped second operating state fluctuation record amplitudes, and so on, until the Nth operating state fluctuation record amplitude set is grouped. The grouping of each level is based on the grouping results of the previous level, thereby ensuring the consistency and comparability of the data.

[0045] Any one of the multiple groups of N-1th operating status fluctuation record amplitudes corresponds one-to-one to one of the groups of operating status fluctuation record amplitudes in the next layer, that is, each group of data in the Nth layer is a subset of a certain group of data in the N-1th layer. Traverse the multiple groups of Nth operating status fluctuation record amplitudes to perform centralized value evaluation, traverse each group of the multiple groups of Nth operating status fluctuation record amplitudes, calculate the centralized value of the group of data, and obtain multiple Nth operating status fluctuation centralized amplitudes. Centralized value evaluation refers to performing statistical analysis on a group of data and extracting values ​​that represent the centralized trend of the group of data. Add multiple Nth operating status fluctuation centralized amplitudes to the corresponding multiple groups of N-1th operating status fluctuation record amplitudes, that is, backfill the centralized amplitude obtained from the Nth layer evaluation to the corresponding N-1th layer grouping, and update the N-1th layer data.

[0046] This process is repeated continuously, performing a centralized value evaluation on multiple sets of N-1 operating status fluctuation record amplitudes, until the centralized value evaluation is performed on the updated first operating status fluctuation record amplitude set, resulting in the operating status fluctuation amplitude value. Iterative centralized value evaluation involves recursively performing centralized value evaluation on the updated data layer by layer until all data at all levels are processed into a single centralized value. The above steps are continuously repeated, backfilling and evaluating upwards layer by layer, until the centralized value of the first operating status fluctuation record amplitude set is updated and calculated. After the first layer of centralized value evaluation is completed, the result is defined as the operating status fluctuation amplitude value, which is used to characterize the fluctuation characteristics of the overall operating status.

[0047] Through layer-by-layer grouping and centralized value evaluation, structured management of complex data relationships is achieved, which facilitates hierarchical processing of operating status fluctuation data. The centralized value evaluation and update process effectively reduces the noise of operating status fluctuation data and improves the accuracy and reliability of the final evaluation results.

[0048] Furthermore, the present application further comprises the following steps:

[0049] S211: configuring a weight set for a component type set through the user end; S212: obtaining transformers of the same model to be analyzed, wherein the transformers to be analyzed have a service record life set of components to be analyzed and a set of operating state fluctuation record amplitudes to be analyzed; S213: calculating the deviation between the service record life set of the components to be analyzed and the component service life set, and obtaining a component service life deviation modulus set; S214: weighting the component service life deviation modulus set according to the weight set, and obtaining a deviation distance; S215: when the deviation distance is greater than the distance threshold, it is deemed that the first-level constraint is met, and the transformer to be analyzed is added to the first-level similar transformer, the service record life set of the components to be analyzed is added to the service record life set of the first-level transformer components, and the operating state fluctuation record amplitude set to be analyzed is added to the first operating state fluctuation record amplitude set.

[0050] Specifically, on the user side, a weight value is pre-configured for each component type based on the importance of the work of different components in the power transformer and their impact on the operating status. This constitutes a weight set, including the weight of each component, which is used to represent the importance or influence of the component on the overall performance. Transformers of the same model as the power transformer to be analyzed are retrieved from the power transformer management system. The transformer to be analyzed has a service life record set of the components to be analyzed and a set of operating status fluctuation records to be analyzed. The service life record set of the components to be analyzed contains the service history information of each component in the transformer to be analyzed, including the start time, end time, and service duration of each component. The operating status fluctuation record amplitude set to be analyzed records the operating status fluctuation of the transformer to be analyzed within a certain time period, including the fluctuation amplitude of status parameters (such as temperature, pressure, and oil and gas concentration) within a certain period. This reflects the fluctuation of the transformer's operating stability. The larger the fluctuation amplitude, the more likely it is that there is an abnormality or failure risk.

[0051] Calculate the deviation between the service life set of power transformer components and the service life record set of the components being analyzed. By calculating the service life deviation of each component in the set, we obtain the component service life deviation modulus set. The deviation modulus value is the deviation (difference) between the service life of the component being analyzed and the service life of other similar components. By calculating these deviations, we can quantify the differences in the service conditions of individual transformer components.

[0052] The component life deviation moduli are weighted according to the component type weight set. Specifically, each component's weight is multiplied by its life deviation, and the results are summed across all components to obtain the deviation distance, which combines the service life deviations and weights of each component. For example, assuming the winding weight is 0.6 and the cooling system weight is 0.4, the transformer to be analyzed has a winding service life of 5 years and a cooling system service life of 9 years, and the power transformer has a winding service life of 6 years and a cooling system service life of 7 years, the deviation distance is |7-9| × 0.4 + |6-5| × 0.6 = 1.4.

[0053] A distance threshold is pre-set, and by comparing the calculated deviation distance with the set distance threshold, it is determined whether the transformer to be analyzed meets the conditions for entering the first-level constraint. If the deviation distance is greater than the distance threshold, it means that there is a significant difference between the component life of the transformer to be analyzed and the power transformer, meeting the first-level constraint and requiring further attention and analysis. It is added to the first-level set of similar transformers, and its corresponding component service record life set to be analyzed is added to the first-level transformer component service record life set, and the operating status fluctuation record amplitude set to be analyzed is added to the first operating status fluctuation record amplitude set. Through deviation calculation and weighted evaluation, transformers of the same model with significant differences in component life as the transformer to be analyzed are accurately screened out, effectively identifying potential abnormal transformers and ensuring that they can enter a more in-depth inspection or maintenance process in a timely manner to avoid failures caused by ignoring subtle differences.

[0054] Furthermore, the present application S500 includes:

[0055] S510: Obtain the constraint interval of the operating control parameter; S520: Randomly update the operating control parameter according to the constraint interval of the operating control parameter to obtain the updated operating control parameter; S530: Perform operating state prediction on the updated operating control parameter through the operating state evaluation model to obtain the updated benchmark operating state; S540: With the updated benchmark operating state as the center of the interval, combined with the operating state fluctuation amplitude, construct a second operating state fluctuation interval; S550: When the second operating state fluctuation interval does not have an intersection with the operating state warning interval, determine the control parameter adjustment strategy.

[0056] Specifically, operating control parameter constraints are extracted from equipment operating specifications or historical data. These constraints represent the reasonable range of values ​​for operating control parameters (such as load voltage, current, and power factor), typically determined based on historical equipment operating data or standard specifications. Within these constraints, new parameter values ​​are generated using random sampling methods, effectively updating the operating control parameters. Random updates involve generating new operating control parameter values ​​within the constraints using a random algorithm to increase parameter diversity and explore possible optimization solutions. A random number generator is used to generate random values ​​within the constraints, and the voltage setpoint or temperature threshold is adjusted to ensure they meet the constraints. If the parameters do not meet the constraints (e.g., due to floating-point precision issues or invalid values), they are regenerated. The generated random values ​​are used as new operating control parameters, and the parameters are updated to obtain updated operating control parameters.

[0057] The updated operating control parameters are fed into the trained operating state assessment model for operating state prediction, yielding an updated baseline operating state. The updated baseline operating state is the new operating state calculated by the operating state assessment model, serving as the new reference baseline. For example, consider the operating control parameter constraint range, assuming the load current is [50A, 100A]. A random number generator is used to generate a load current value, such as 78A. This randomly generated parameter is fed into the operating state assessment model for operating state prediction.

[0058] Taking the updated benchmark operating state as the center of the interval, the upper and lower bounds are calculated in combination with the operating state fluctuation amplitude to obtain the second operating state fluctuation interval, which is used to measure whether the adjusted operating state is within a safe range. Determine whether the constructed second operating state fluctuation interval and the operating state warning interval have interval overlap, that is, whether there is an intersection. If there is an intersection, it means that the current operating state is within the warning range or close to the warning range, indicating that there is still a risk in the operating state, and it is necessary to continue adjusting the operating control parameters until the operating state fluctuation interval and the operating state warning interval have no intersection. When the second operating state fluctuation interval and the operating state warning interval have no intersection, stop adjusting the operating control parameters. According to the updated operating control parameters at this time, adjust the actual operating state of the transformer. By dynamically adjusting the operating control parameters, the operating state is kept within a safe range. When it is detected that the operating state is approaching or entering the warning range, it is quickly adjusted to prevent failure or further deterioration, ensuring that the operating state of the transformer remains within a safe and effective range, while avoiding unnecessary warnings and potential failure risks.

[0059] By randomly updating and exploring different parameter configurations, the equipment operating status is optimized. After detecting the warning status, the operating parameters are quickly responded and adjusted to reduce the risk of failure. By adjusting the operating control parameters, the equipment operating status is returned to a safe range to avoid further deterioration.

[0060] Furthermore, the present application further comprises the following steps:

[0061] S560: When the number of updates meets the preset number of times, the operation control parameters that do not intersect with the operation status warning interval are still not obtained, and a particle distribution space is constructed according to the operation control parameter constraint interval; S570: The historical updated operation control parameters are distributed in the particle distribution space to obtain a transformer history control particle swarm; S580: A fitness function is constructed, wherein the fitness function is used to evaluate the normalized sum of the intersection amplitudes of various control attributes; S590: According to the fitness function, a particle swarm optimization is performed on the transformer history control particle swarm to obtain the updated benchmark operation status.

[0062] Specifically, the preset number of times refers to the maximum number of attempts allowed when adjusting the operation control parameters. If the number of updates exceeds the preset number of times and the operation control parameters with non-intersection in the operation status warning interval are still not obtained, it means that the random adjustment method can no longer further optimize the operation status, and the next optimization process is entered. Construct the particle distribution space based on the constraint interval of the operation control parameters. List all operation control parameters and their value ranges, and each parameter is represented by a dimension. Define the spatial dimension, that is, the number of parameters. The range of each dimension is determined by the minimum and maximum values ​​of the corresponding parameters. Randomly generate a certain number of particles in the distribution space, and each particle corresponds to a set of parameter values. Each particle needs to have an initial velocity in space to control the movement of the particle during the optimization process. Set boundary conditions to ensure that the particles do not exceed the boundaries when moving in the distribution space.

[0063] All historically updated operating control parameters are distributed into a particle distribution space to form a particle swarm. Each particle represents a historical control parameter combination. All historically updated control parameter combinations are mapped into a particle swarm through the particle distribution space. The particle swarm consists of multiple particles, each corresponding to a historical control parameter combination.

[0064] A fitness function is constructed to evaluate the normalized sum of the intersection amplitudes of each control attribute. This is done by performing a weighted sum of the normalized intersection amplitudes of each control attribute, based on the weight of each operational control attribute. In optimization algorithms, the fitness function is used to evaluate the quality of each solution (particle). In particle swarm optimization (PSO), the fitness function calculates the target value corresponding to the particle position, reflecting the quality of the solution. The fitness function calculates the weighted sum of the normalized intersection amplitudes of the control attributes. The goal is to find a control parameter that disjoints the operational state from the warning interval.

[0065] The fitness function evaluates whether the position (i.e., the set of control parameters) of each particle in the particle swarm meets the target. Specifically, the control parameters corresponding to each particle meet the preset operating state requirements, particularly avoiding intersections with the operating state warning interval. Based on the fitness function, the normalized value of the intersection amplitude of the control attributes (such as current, voltage, and temperature) of the transformer's historical control particle swarm is calculated. The particle's fitness is evaluated based on this normalized value and the weight set. The particle's speed and position are continuously adjusted using the particle swarm optimization algorithm to find the solution (i.e., the optimal control parameter combination) that optimizes the fitness function. Particles update their speed and position based on their current position, their historical best position, and their global best position. With each iteration, each particle in the swarm searches for a direction that improves its fitness. The final result of the particle swarm optimization algorithm yields the optimal control parameter combination, which is then used to derive the updated baseline operating state. After multiple iterations, the particle swarm optimization algorithm finds an optimal particle, representing the most suitable operating control parameters. The control parameter combination corresponding to this particle is used as input for prediction by the operating state assessment model, resulting in an updated baseline operating state. Through particle swarm optimization, we can find a control parameter combination that is more suitable for the current operating environment, thereby improving the efficiency and stability of transformer operation.

[0066] Furthermore, the present application further comprises the following steps:

[0067] S561: Obtaining an operation control attribute weight set through a user terminal; S562: The fitness function is used to perform weighted sum calculation on the normalized intersection amplitude values ​​of the various control attributes according to the operation control attribute weight set.

[0068] Specifically, the user end refers to the process of obtaining relevant control parameters or weight information through user input or system operation. The user end can be a graphical interface, allowing operators to enter operational control attribute weights through sliders, input boxes, and other methods. The operational control attribute weight set is the importance weight assigned to each operational control attribute (such as current, voltage, and temperature), reflecting the priority of each attribute in the operational status assessment. The user end provides an interface that allows users to enter the importance weights of each operational control attribute based on experience or needs.

[0069] Determine the intersection magnitude of each control attribute. This intersection magnitude refers to the overlap in the interaction ranges of each control attribute during operation, reflecting the degree of joint fluctuation between multiple attributes under certain conditions. Using operational status monitoring data, calculate the intersection magnitude of each attribute under the same conditions. Normalize the intersection magnitudes of each control attribute, mapping them to a fixed range (typically 0, 1) for unified calculation. Normalization avoids the effects of inconsistent dimensionality across different attributes.

[0070] The normalized value is multiplied by the weight set and added together to obtain the final result. In other words, the fitness value is equal to the normalized value multiplied by the corresponding weight and summed. The result of the fitness function is used to guide the particle swarm optimization and help select the optimal combination of operation control parameters. By inputting the weight set on the user side, the importance of each control attribute can be dynamically adjusted according to different devices, environments or user needs. All control attributes are comprehensively considered to avoid the excessive influence of a single attribute on the result, thereby achieving more accurate operation status prediction. The result of the fitness function provides a clear direction for particle swarm optimization, making the optimization process more targeted and reducing unnecessary calculations.

[0071] In summary, the method for adaptively adjusting the state parameters of a power transformer provided in this application has the following technical effects:

[0072] The method extracts a component service life set from a power transformer maintenance log; retrieves the operating state fluctuation amplitude of similar power transformers using the component service life set as a constraint; obtains operating control parameters, predicts the operating state through an operating state evaluation model, and obtains a baseline operating state; constructs a first operating state fluctuation interval with the baseline operating state as the interval center and in combination with the operating state fluctuation amplitude; when the first operating state fluctuation interval intersects with the operating state warning interval, adjusts the operating control parameters, generates a control parameter adjustment strategy, adjusts the operating state according to the control parameter adjustment strategy, and issues a warning. That is to say, by extracting the service life set of each component from the maintenance log of the power transformer, retrieving the operating state fluctuation amplitude of the similar transformer, and based on the obtained operating control parameters, the operating state is predicted through the operating state evaluation model, and on the basis of the baseline operating state, combined with the operating state fluctuation amplitude, a first operating state fluctuation interval is constructed; if the first operating state fluctuation interval intersects with the operating state warning interval, the operating control parameters are adjusted according to the prompt of the warning interval, and the operating state prediction is performed again to obtain an updated baseline operating state; based on the updated baseline operating state, a second operating state fluctuation interval is constructed, and the control parameters are adjusted according to the intersection of different fluctuation intervals, thereby enhancing the accuracy of the state parameter evaluation, thereby improving the accuracy and adaptability of the power transformer state parameter adjustment.

[0073] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

[0074] Obviously, those skilled in the art may make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application is intended to include these modifications and variations.

Claims

1. A method for adaptively adjusting state parameters of a power transformer, characterized in that: include: Extract component service life sets from power transformer maintenance logs; Retrieving the operating state fluctuation amplitudes of similar power transformers based on the component service life set as a constraint; Obtain operation control parameters, predict the operation status through the operation status evaluation model, and obtain the benchmark operation status; Taking the reference operating state as the interval center and combining the operating state fluctuation amplitude, a first operating state fluctuation interval is constructed; When the first operating state fluctuation interval and the operating state warning interval have an intersection, adjusting the operating control parameter, generating a control parameter adjustment strategy, adjusting the operating state according to the control parameter adjustment strategy, and issuing a warning; Taking the component service life set as a constraint, the operating state fluctuation amplitude of similar power transformers is retrieved, including: Using the component service life set as a first-level constraint, searching for first-level similar transformers, wherein the first-level similar transformers have a first operating state fluctuation record amplitude set and a first-level transformer component service record life set; Using the first-level transformer component service record life set as a second-level constraint, searching for second-level similar transformers, wherein the second-level similar transformers have a second operating state fluctuation record amplitude set and a second-level transformer component service record life set; Until the service record life set of the N-1 level transformer components is used as the N-level constraint, N-level similar transformers are retrieved, wherein the N-level similar transformers have the N-th operating state fluctuation record amplitude set, 5≥N≥2, and N is an integer; Centralized value evaluation is performed on the first operating state fluctuation record amplitude set, the second operating state fluctuation record amplitude set, and up to the Nth operating state fluctuation record amplitude set to obtain the operating state fluctuation amplitude value.

2. The method for adaptively adjusting the state parameters of a power transformer according to claim 1, wherein: Performing centralized value evaluation on the first operating state fluctuation record amplitude set, the second operating state fluctuation record amplitude set, and up to the Nth operating state fluctuation record amplitude set to obtain the operating state fluctuation amplitude value includes: Grouping the second operating state fluctuation record amplitude set according to the first operating state fluctuation record amplitude set to obtain multiple groups of second operating state fluctuation record amplitudes; Grouping the third operating state fluctuation record amplitude set according to the multiple groups of second operating state fluctuation record amplitudes to obtain multiple groups of third operating state fluctuation record amplitudes; until the Nth operating state fluctuation record amplitude set is grouped according to the multiple groups of N-1th operating state fluctuation record amplitudes to obtain multiple groups of Nth operating state fluctuation record amplitudes, wherein any one operating state fluctuation record amplitude of the multiple groups of N-1th operating state fluctuation record amplitudes corresponds one-to-one to a group of operating state fluctuation record amplitudes of the multiple groups of Nth operating state fluctuation record amplitudes; Traversing the multiple groups of Nth running state fluctuation record amplitudes to perform concentration value evaluation, obtaining multiple Nth running state fluctuation concentration amplitudes, adding the multiple Nth running state fluctuation concentration amplitudes to the corresponding multiple groups of N-1th running state fluctuation record amplitudes, and obtaining updated multiple groups of N-1th running state fluctuation record amplitudes; An iterative centralized value evaluation is performed based on the updated multiple groups of N-1th running state fluctuation record amplitudes until a centralized value evaluation is performed on the updated first running state fluctuation record amplitude set to obtain the running state fluctuation amplitude.

3. The method for adaptively adjusting the state parameters of a power transformer according to claim 1, wherein: Using the component service life set as the first-level constraint, search for first-level similar transformers, including: Configuring a weight set for a component type set by the user end; Obtain a transformer to be analyzed of the same model, wherein the transformer to be analyzed has a service life record set of components to be analyzed and a amplitude set of operating state fluctuation records to be analyzed; Calculating the deviation between the service record life set of the component to be analyzed and the component service life set to obtain a component service life deviation modulus set; weighting the component service life deviation modulus set according to the weight set to obtain a deviation distance; When the deviation distance is greater than the distance threshold, it is considered that the first-level constraint is met, and the transformer to be analyzed is added to the first-level similar transformers, the service record life set of the component to be analyzed is added to the service record life set of the first-level transformer component, and the operating state fluctuation record amplitude set to be analyzed is added to the first operating state fluctuation record amplitude set.

4. The method for adaptively adjusting state parameters of a power transformer according to claim 1, wherein: Obtain operation control parameters, predict the operation status through the operation status assessment model, and obtain the baseline operation status, including: Collecting operation logs of power transformers that meet preset models and have components with service life less than a service life threshold, wherein the power transformer operation logs include operation control parameter record data and operation status monitoring data; Using the operation status monitoring data as supervision and the operation control parameter record data as input to train a random forest model to obtain the operation status evaluation model; The operation control parameters are input into the operation state evaluation model, and the reference operation state is output.

5. The method for adaptively adjusting state parameters of a power transformer according to claim 1, wherein: Adjusting the operation control parameters and generating a control parameter adjustment strategy includes: Get the constraint range of operation control parameters; Randomly updating the operation control parameter according to the operation control parameter constraint interval to obtain an updated operation control parameter; Performing an operating state prediction on the updated operating control parameters using an operating state evaluation model to obtain an updated reference operating state; Taking the updated reference operating state as the interval center and combining the operating state fluctuation amplitude, constructing a second operating state fluctuation interval; When the second operating state fluctuation interval and the operating state warning interval do not have an intersection, the control parameter adjustment strategy is determined according to the updated operating control parameter.

6. The method for adaptively adjusting the state parameters of a power transformer according to claim 5, wherein: Performing an operation state prediction on the updated operation control parameter by using an operation state evaluation model to obtain an updated reference operation state, and then further comprising: When the number of updates meets the preset number, and the operation control parameter that does not intersect with the operation status warning interval is still not obtained, the particle distribution space is constructed according to the operation control parameter constraint interval; Distributing the historical updated operation control parameters in the particle distribution space to obtain a transformer historical control particle swarm; Constructing a fitness function, wherein the fitness function is used to evaluate the normalized sum of the intersection amplitudes of the various control attributes; According to the fitness function, particle swarm optimization is performed on the transformer history control particle swarm to obtain the updated benchmark operating state.

7. The method for adaptively adjusting state parameters of a power transformer according to claim 6, wherein: Constructing the fitness function also includes: Obtain the weight set of operation control attributes through the user terminal; The fitness function is used to perform weighted sum calculation on the normalized intersection amplitude values ​​of the various control attributes according to the operation control attribute weight set.

Images