Predictive maintenance method and device of transformer, computer equipment, readable storage medium and program product

Abstract

The invention relates to a predictive maintenance method and device of a transformer, computer equipment, a readable storage medium and a program product. When the prediction condition is satisfied, according to the abnormal operation characteristics of the transformer and the preset association rule, the abnormal information such as the current time, the current abnormal type, the current abnormal level and the current operation state information corresponding to the transformer is determined; and determining target maintenance time and a maintenance strategy corresponding to the transformer according to the abnormal information through the degradation prediction model. And maintaining the transformer before the target maintenance time based on the maintenance strategy. Compared with a traditional mode of maintaining limited fault types through threshold limitation and the like, the method has the advantages that the current abnormal information is determined by combining the abnormal operation characteristics and the association rules, the degradation trend and the maintenance strategy of the transformer are predicted through the degradation prediction model, and the maintenance efficiency is improved based on the maintenance strategy. Maintenance of the transformer is completed within the target maintenance time before abnormal degradation, and timeliness of maintenance of the transformer is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power equipment maintenance, in particular to a transformer predictive maintenance method and device, computer equipment, computer readable storage medium and computer program product. BACKGROUND

[0002] With the development of smart grid and digital substation, as the key equipment of power transmission, the maintenance and management mode of transformer is transforming from the traditional preventive maintenance system combining periodic maintenance and after-service maintenance to the predictive maintenance based on the actual operation state of the equipment. At present, the method of fault diagnosis and maintenance of transformer depends on the threshold limit such as the attention value of dissolved gas analysis in oil, which can only identify limited fault types for maintenance, and is difficult to cope with diversified faults under complex operating environment, so that the timeliness of detecting the abnormality of transformer and maintenance under complex operating environment is reduced.

[0003] Therefore, the current transformer maintenance method has the defect of low timeliness. SUMMARY

[0004] Therefore, it is necessary to provide a transformer predictive maintenance method, device, computer equipment, computer readable storage medium and computer program product capable of improving timeliness in view of the above technical problems.

[0005] In a first aspect, the present application provides a transformer predictive maintenance method, comprising:

[0006] When it is detected that the prediction condition is met, an abnormal operation feature corresponding to the transformer is obtained; the abnormal operation feature represents an operation feature corresponding to an abnormal operation parameter;

[0007] According to the abnormal operation feature and a preset association rule, current abnormal information corresponding to the transformer is determined; the preset association rule includes an association relationship between each abnormal operation feature and abnormal information; the current abnormal information includes current time, current abnormal type, current abnormal level and current operation state information;

[0008] The current abnormal information is input into a degradation prediction model; the degradation prediction model is used to determine target maintenance time and a maintenance strategy of the transformer according to the current time, current abnormal type, current abnormal level and current operation state information in the abnormal information;

[0009] According to the maintenance strategy, the transformer is maintained before the target maintenance time.

[0010] In a second aspect, the present application further provides a transformer predictive maintenance device, comprising:

[0011] an acquisition module configured to acquire a corresponding abnormal operation feature of the transformer when it is detected that the prediction condition is met, the abnormal operation feature representing a corresponding operation feature when an operation parameter is abnormal;

[0012] a determination module configured to determine corresponding current abnormal information of the transformer according to the abnormal operation feature and a preset association rule, the preset association rule including an association relationship between each abnormal operation feature and abnormal information, and the current abnormal information including a current time, a current abnormal type, a current abnormal level, and current operation state information;

[0013] a prediction module configured to input the current abnormal information into a deterioration prediction model, the deterioration prediction model configured to determine a target maintenance time and a maintenance strategy of the transformer according to the current time, the current abnormal type, the current abnormal level, and the current operation state information in the abnormal information;

[0014] a maintenance module configured to perform maintenance on the transformer before the target maintenance time according to the maintenance strategy.

[0015] In a third aspect, the present application also provides a computer device including a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the method described above when executing the computer program.

[0016] In a fourth aspect, the present application also provides a computer-readable storage medium storing a computer program, the computer program being executed by a processor to implement the steps of the method described above.

[0017] In a fifth aspect, the present application also provides a computer program product including a computer program, the computer program being executed by a processor to implement the steps of the method described above.

[0018] The predictive maintenance method, device, computer device, computer-readable storage medium, and computer program product of the transformer described above determine the current time, the current abnormal type, the current abnormal level, and the current operation state information of the transformer according to the abnormal operation feature and the preset association rule when the prediction condition is met, and determine the target maintenance time and the maintenance strategy of the transformer according to the abnormal information by the deterioration prediction model. The maintenance on the transformer is performed before the target maintenance time based on the maintenance strategy. Compared with the traditional maintenance on limited fault types by threshold limit and the like, the current abnormal information is determined by combining the abnormal operation feature and the association rule, the deterioration trend and the maintenance strategy of the transformer are predicted by the deterioration prediction model, the maintenance on the transformer is completed before the target maintenance time when the deterioration is abnormal based on the maintenance strategy, and the timeliness of the maintenance on the transformer is improved. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the following will briefly introduce the drawings needed to be used in the description of the embodiments of the present application or the related art. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other related drawings can also be obtained without creative effort based on these drawings.

[0020] Figure 1 Flowchart of a predictive maintenance method for a transformer in an embodiment;

[0021] Figure 2 Flowchart of an anomaly prediction step in an embodiment;

[0022] Figure 3 Timing diagram of an anomaly deterioration maintenance step in an embodiment;

[0023] Figure 4 Flowchart of a maintenance step in an embodiment;

[0024] Figure 5 Flowchart of a running state determination step in an embodiment;

[0025] Figure 6 Flowchart of a predictive maintenance method for a transformer in another embodiment;

[0026] Figure 7 Block diagram of a predictive maintenance device for a transformer in an embodiment;

[0027] Figure 8 Internal structure diagram of a computer device in an embodiment. DETAILED DESCRIPTION

[0028] In order to make the purposes, technical solutions and advantages of the present application clearer, the following will further describe the present application in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0029] It should be noted that the terms "first", "second", etc. used in the present application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "include" and "have" and any variations thereof used in the present application are intended to cover non-exclusive inclusion. The term "multiple" used in the present application refers to two and more than two. The term "and / or" used in the present application refers to one of the solutions, or any combination of multiple solutions.

[0030] In related technologies, with the development of smart grids and digital substations, the maintenance and management model of transformers, as key equipment for power transmission, is shifting from a traditional preventative maintenance system combining periodic inspections and reactive repairs to a predictive maintenance approach based on the actual operating status of the equipment. However, many problems still urgently need to be addressed in transformer maintenance and management:

[0031] Incompleteness of fault diagnosis: Traditional fault diagnosis methods often rely on threshold limitations, such as the attention values ​​in dissolved gas analysis of oil, which can only identify a limited number of fault types and are difficult to cope with the diverse faults in complex operating environments. Furthermore, transformer operating data includes periodic pre-inspection test data and real-time monitoring data, which have significant differences in timeliness. However, these differences are not adequately addressed, leading to a one-sided diagnosis of potential faults, such as insufficient timeliness in capturing minor faults or early fault signals.

[0032] The limitations of condition assessment: Transformers have complex structures and intricate relationships between components and performance conditions. Assessments often rely on only certain characteristic quantities (such as assessing insulation condition solely through oil gas content), neglecting comprehensive operational information; or they use expert experience for weighting, failing to objectively reflect the correlation between characteristic indicators and faults. In particular, condition quantity scoring methods are heavily influenced by expert subjectivity, while partial characteristic quantity discrimination methods struggle to comprehensively describe the overall operating condition of the equipment, leading to discrepancies between assessment results and actual conditions.

[0033] The limitations of fault prediction: Faults progress from symptom onset to complete failure in a continuous degradation process with multiple ambiguous intermediate states. However, predictions often focus on the probability or timing of fault occurrence, failing to quantify the degree of degradation within the maintenance period. Furthermore, these methods only consider the impact of equipment age, neglecting differences in the equipment's current operational status and its resilience to faults (e.g., equipment in good health is more resistant to faults) as well as the inherent degradation characteristics of different faults (e.g., insulation aging and partial discharge deteriorate at different rates), leading to insufficient prediction accuracy. This results in frequent over-maintenance or under-maintenance in transformer maintenance, increasing maintenance costs and potentially impacting the stable operation of the power system due to unplanned outages.

[0034] Based on this, this application determines the current abnormal information by combining abnormal operation characteristics and association rules, predicts the deterioration trend and maintenance strategy of the transformer through a deterioration prediction model, and completes the maintenance of the transformer within the target maintenance time before abnormal deterioration based on the maintenance strategy, thereby improving the timeliness of transformer maintenance.

[0035] In one embodiment, such as Figure 1As shown, a predictive maintenance method for transformers is provided. This embodiment illustrates the application of this method to a terminal. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and implemented through interaction between the terminal and the server, including the following steps S202 to S208. Wherein:

[0036] Step S202: When the prediction conditions are met, the corresponding abnormal operation characteristics of the transformer are obtained; the above abnormal operation characteristics represent the operation characteristics corresponding to the abnormal operation parameters.

[0037] The terminal can be a computer device that manages the transformer, such as a management terminal in a power system. The prediction conditions can be conditions indicating that the transformer requires anomaly prediction and maintenance. For example, the terminal can set a prediction period as a prediction condition, and when the terminal detects that the prediction period has arrived, it performs anomaly prediction for the transformer.

[0038] The aforementioned prediction cycle can be determined based on the transformer's pre-set lifespan. Specifically, the terminal divides the transformer's entire lifespan into three stages: initial, middle, and late. In the initial stage, the transformer begins operation, and the operation of new equipment, as well as its connection and coordination with other related equipment, may be unstable, requiring frequent monitoring of its operating status; therefore, the prediction cycle interval is relatively short. In the middle stage, the transformer's operation has reached a stable state, and the prediction cycle interval is longer. In the late stage, the transformer's components and functions begin to age, and the frequency of failures continuously increases, requiring a continuous reduction in the prediction cycle interval.

[0039] In the process of anomaly prediction, the terminal can acquire the transformer's operating parameters and extract features from these parameters to obtain the corresponding operating characteristics of the transformer. The terminal can acquire multiple types of operating parameters for the transformer, resulting in multiple types of extracted operating characteristics. The terminal can then identify anomalies in each type of operating characteristic to obtain the corresponding abnormal operating characteristics of the transformer.

[0040] For example, when operating parameters are abnormal, their corresponding operating characteristics fluctuate. The terminal identifies the fluctuation information of the operating characteristics corresponding to the operating parameters, and identifies operating characteristics whose fluctuation information exceeds the change threshold corresponding to the type of operating parameter as abnormal operating characteristics. In other words, abnormal operating characteristics characterize the operating features corresponding to the abnormality of the operating parameters.

[0041] Step S204: Based on the above-mentioned abnormal operation characteristics and preset association rules, determine the current abnormal information corresponding to the above-mentioned transformer; the above-mentioned preset association rules include the association relationship between each abnormal operation characteristic and the abnormal information; the above-mentioned current abnormal information includes the current time, current abnormality type, current abnormality level and current operation status information.

[0042] The terminal can pre-set preset association rules. These rules include the association relationships between various abnormal operation features and abnormal information. The terminal can extract abnormal operation sample features and abnormal sample information from a sample library, and construct the association relationships between abnormal operation features and abnormal information by determining the correlation degree (support) between them. Since abnormal operation features include multiple types, the aforementioned association relationships can include multiple types. The terminal can match these abnormal operation features with the preset association rules, identify the association relationships corresponding to the abnormal operation features from multiple relationships, and obtain the abnormal information associated with the abnormal operation features based on these relationships. This yields the current abnormal information for the transformer, and from the current abnormal information, it obtains the current time, current abnormality type, current abnormality level, and current operating status information.

[0043] Among them, the current time indicates the time when the anomaly occurred, the current anomaly type indicates the type of anomaly that occurred in the transformer, the current anomaly level indicates the severity of the anomaly that occurred in the transformer at the current time, and the current operating status information indicates the current health of the transformer.

[0044] Step S206: Input the above-mentioned current anomaly information into the deterioration prediction model; the above-mentioned deterioration prediction model is used to determine the target maintenance time and maintenance strategy corresponding to the above-mentioned transformer based on the above-mentioned current time, current anomaly type, current anomaly level and current operating status information in the above-mentioned anomaly information.

[0045] The terminal can pre-train a degradation prediction model iteratively based on abnormal sample information, maintenance sample time, and maintenance sample strategy to obtain a trained degradation prediction model. This model can be used to predict abnormal degradation trends in transformers (such as the severity of anomalies over time). The terminal can input current anomaly information into the degradation prediction model, which can extract the current time, current anomaly type, current anomaly level, and current operating status information. Based on these information, the model predicts the degradation of the transformer's current anomaly, forecasting the change in the anomaly level of the current anomaly type over time. This yields the degradation trend of the anomaly level for the current anomaly type, allowing the terminal to determine the target maintenance time and strategy for the transformer based on this degradation trend.

[0046] The target maintenance time represents the time when the anomaly level is highest, and the predicted time when the aforementioned anomaly of the transformer reaches its most severe state. The terminal needs to complete the maintenance before this time. The aforementioned maintenance strategy can be determined based on the current anomaly type and the aforementioned degradation trend. For example, the terminal can determine a maintenance strategy that meets the degradation trend of the anomaly based on the current anomaly type and the current anomaly level.

[0047] Specifically, taking anomalies as an example, the terminal can accurately predict the evolution of potential fault states in a transformer during maintenance, ensuring the effective execution of predictive maintenance strategies. For example... Figure 2 As shown, Figure 2 This is a flowchart illustrating the anomaly prediction steps in one embodiment. The terminal can classify fault severity levels. For example, based on the impact of a fault on equipment stability, the terminal classifies transformer faults into three levels, clearly defining the degradation characteristics and maintenance strategies for each level. Level 1 faults represent the most severe faults, requiring immediate shutdown and maintenance upon occurrence; Level 2 faults represent moderate severity but affect transformer stability, requiring short-term maintenance; Level 3 faults represent minor severity with a long degradation period, suitable for rational maintenance. The fault severity level classification (Level 1, 2, and 3) is a predefined, relatively static classification system. Fluctuating anomaly characteristics act as triggers, detecting transformer anomalies. As a diagnostic basis, the terminal infers potential fault types through association rules and maps the fault types to their corresponding preset severity levels.

[0048] The terminal can also construct a degradation prediction model. For example, based on parameters such as the current operating status of the transformer, the severity of the fault, and the equipment's operating time, and combined with the transformer's fault degradation patterns, the terminal can establish a degradation trend prediction model. Here, the aforementioned operating status represents the equipment's operating status level, which is determined by the operating status score.

[0049] The operational status score, denoted as R, integrates the severity and weight of all types of faults and serves as a quantitative indicator of the overall health of the equipment. The status level l is a discrete classification value (l=1, 2, 3, 4, 5), corresponding to five preset levels: normal, attention, minor, abnormal, and severe. This level l represents the operational status referred to in the subsequent degradation prediction model.

[0050] The degradation prediction model can call the state level l parameter. When the terminal starts the prediction of a specific fault, the terminal directly obtains the current operating state level l of the transformer and inputs it into the degradation prediction model.

[0051] The degradation prediction model for transformer fault conditions can be expressed as: H z,l (t)~F z,l(t)=a z,l tb z,l Among them, the terminal definition function H z,l (t) is used to describe the change in state score over time for different levels of faults under a specific operating condition.

[0052] Among them, the above-mentioned status score refers to the score (abnormal level) of the comprehensive index of fault categories p. j .

[0053] function H z,l (t) describes the change in state score for a specific fault class (such as insulation fault) over time t. This state score is related to the overall fault class score p. j They are the same concept. It can be understood as: H z,l (t) represents the future time P. j Value prediction. Z represents the severity level of the fault (1, 2, 3); l represents the current overall operating status level of the transformer (1-5). F z,l (t) is H z,l The specific mathematical model or calculation formula for (t), H z,l The behavior of (t) is determined by model F. z,l (t) is used to describe it. z,l and b z,l : These are empirical constants that encapsulate the degradation rate and initial offset of a specific level of fault z under a specific device state l. z,l This represents the rate of degradation for a specific fault, b z,l It is an offset or baseline adjustment constant, whose function is to ensure that the model at time t=0 is aligned with the current state score p when the fault is identified. j Alignment.

[0054] The terminal sets the time of the first occurrence of the fault signal as t0 (with t0=0 as the starting reference point for the maintenance plan). Variables z=1,2,3 represent the fault's severity level, where smaller values ​​indicate a greater threat to the system's safe operation. Simultaneously, the equipment's operating status level (operating status information, l=1,2,…,5) is also characterized in a similar way; lower level values ​​indicate better equipment health. The time point when the transformer fault deteriorates to its most severe state (target maintenance time point) is TF. z,l (e.g., T23 or T24), a z,l and b z,lThese represent empirical constants for a transformer in a specific state at fault level z. Their values ​​can be determined based on statistical analysis of historical operating data and the experience of domain experts. During prediction, the terminal inputs the transformer's current fault degradation time (current time), empirical constants, fault type, fault hazard level (current anomaly level), and equipment operating status level (current operating status information) into the degradation prediction model. The degradation prediction model then quantitatively describes the degradation trend of the fault from its current state to the degradation stage.

[0055] For example, suppose a transformer being monitored has an operating status score R currently mapped to l=2 (belonging to the watchful state). The terminal, through association rule analysis, identifies a potential level 2 partial discharge fault (z=2). The terminal then calculates the current score (current anomaly level) p for this fault type (partial discharge). j =0.75. The terminal can determine the parameters of the degradation prediction model, for example, by statistically fitting empirical constants based on historical databases and expert experience: a 2,2 =0.02, b 2,2 =0.10. The terminal establishes a prediction equation and inputs the parameters into the degradation prediction model. t=0 corresponds to the moment when the fault is just identified and H(0) is 0.75. Based on this model, the terminal predicts a state score of 0.65 for the next 10 days (t=10 days).

[0056] Specifically, when determining the maintenance strategy, the terminal combines a fixed prediction cycle with a fault development state maintenance method based on fault degradation time, as follows: Figure 3 As shown, Figure 3 This is a timing diagram of the abnormal degradation maintenance steps in one embodiment.

[0057] The terminal can pre-set the prediction cycle. For example, the terminal divides the entire life cycle of a transformer into three stages: initial, middle, and late. In the initial stage, the transformer begins operation, and the operation of new equipment, as well as its connection and coordination with other related equipment, may be unstable, requiring frequent monitoring of its operating status; therefore, the prediction cycle interval is relatively short. In the middle stage, the transformer's operation has reached a stable state, and the prediction cycle interval is longer. In the late stage, the transformer's components and functions begin to age, and the frequency of failures continues to increase, requiring a continuous reduction in the prediction cycle interval.

[0058] The terminal can also perform fault development status maintenance based on the fault deterioration time (target maintenance time). For example, within the prediction period, when the terminal detects abnormal fluctuations in a real-time monitorable indicator (operating parameter) (such as the appearance of abnormal operating characteristics), the terminal performs an online assessment of the potential faults in the transformer and its current operating status (such as current operating status information). Based on this, it predicts the deterioration trend of potential faults during the maintenance phase, and then allocates maintenance resources and arranges maintenance work.

[0059] For example, at the (i-1)th prediction time, the terminal did not detect any abnormal fluctuations in the transformer's operating parameters. After time interval T1, the i-th prediction time begins. The time interval between this prediction and the (i+1)th prediction is T2. In this prediction, the terminal detects abnormal fluctuations in the operating parameters at time T21. At this point, the terminal can perform fault diagnosis (identify abnormal information), status assessment (determine the transformer's operating status information), and fault prediction (determine the target maintenance time and maintenance strategy) for the transformer. The terminal completes the fault analysis at time T22 and predicts that the fault will deteriorate to its most severe state at time T23. Therefore, the terminal can perform maintenance on the transformer within time T4 and complete the maintenance work before time T24 (the target maintenance time).

[0060] Step S208: According to the above maintenance strategy, perform maintenance on the transformer before the target maintenance time.

[0061] Once the terminal determines the maintenance strategy, it can perform maintenance on the transformer based on the strategy and complete the maintenance before the target maintenance time.

[0062] Specifically, such as Figure 4 As shown, Figure 4 This is a flowchart illustrating the maintenance steps in one embodiment, taking a fault as an example. The terminal assesses the transformer's health status after the (i-1)th prediction. When an abnormal fluctuation in operating characteristics is detected, the terminal determines that a fault signal has occurred and performs fault diagnosis, status assessment, and fault deterioration prediction. It calculates the time point T23 when the fault deteriorates to its most severe state, completes preventative maintenance work before time T23, and determines the completion time T24. The terminal then assesses the transformer's health status after maintenance. If time T23 is greater than time T24, it indicates that the target maintenance time was determined too early. The predicted time point when the fault deteriorates to its most severe state is earlier than or equal to the expected completion time of the maintenance work, meaning that according to the normal procedure, there is not enough time to repair before the fault occurs. The terminal can repeat the fault diagnosis and prediction steps. If time T23 is less than time T24, the predicted time point when the fault develops to its most severe state is later than the end time of the current maintenance work, meaning that the terminal has sufficient time to complete the repair before the fault becomes critical. The terminal can end the maintenance work within this prediction cycle after the maintenance.

[0063] In the aforementioned predictive maintenance method for transformers, when prediction conditions are met, abnormal information such as the current time, current anomaly type, current anomaly level, and current operating status of the transformer is determined based on its abnormal operating characteristics and preset association rules. A degradation prediction model then determines the target maintenance time and maintenance strategy for the transformer based on this abnormal information. Maintenance is then performed on the transformer before the target maintenance time based on the maintenance strategy. Compared to traditional methods that limit maintenance to a limited number of fault types using threshold restrictions, this application combines abnormal operating characteristics and association rules to determine current anomaly information, predicts the transformer's degradation trend and maintenance strategy using a degradation prediction model, and completes transformer maintenance before the target maintenance time before abnormal degradation occurs, thus improving the timeliness of transformer maintenance.

[0064] In one embodiment, the step of obtaining current running status information includes: determining the current abnormality type corresponding to the abnormal running feature and the current abnormality level corresponding to the current abnormality type according to the preset association rule; determining the second weight corresponding to the current abnormality level according to the first weight of the abnormal running feature and the number of the current abnormality types; and determining the current running status information according to the current abnormality level and the second weight.

[0065] In this embodiment, the terminal can monitor the transformer's operating status in real time, thereby determining the transformer's current operating status information. Specifically, the terminal can combine preset association relationships to determine the current anomaly type corresponding to the abnormal operating characteristics, i.e., the type of anomaly currently existing in the transformer, and can also determine the current anomaly level corresponding to the current anomaly type, i.e., the current severity of the current anomaly type.

[0066] The terminal can also obtain a first weight for abnormal operation characteristics. This first weight is determined based on the confidence level of the associations corresponding to the abnormal operation characteristics. The terminal can determine a second weight corresponding to the current abnormality level based on the first weight and the number of current abnormality types. Therefore, the terminal can determine the current operation status information based on the current abnormality level and the second weight. For example, the terminal can obtain the current operation status information based on the status score corresponding to the current abnormality level, combined with the second weight. Here, the confidence level represents the reliability of the association.

[0067] In one embodiment, the step of obtaining the first weight includes: obtaining the abnormal operation sample feature corresponding to the abnormal operation feature and the abnormal sample type corresponding to the abnormal operation sample feature; obtaining the first occurrence frequency of the abnormal sample type and the second occurrence frequency of the abnormal operation sample feature and the abnormal sample type occurring simultaneously; determining the confidence level of the target association corresponding to the abnormal operation feature based on the first occurrence frequency and the second occurrence frequency; and determining the first weight based on the confidence level.

[0068] In this embodiment, the terminal can determine the aforementioned first weight based on the confidence level of the association relationship. For example, the terminal can obtain abnormal operation sample features corresponding to abnormal operation features and abnormal sample types corresponding to the aforementioned abnormal operation sample features from the sample library. The sample library stores multiple abnormal operation sample features and their corresponding abnormal sample types. Each of these samples can be obtained by analyzing historical operation features and historical abnormal types. The terminal can obtain a first occurrence frequency of the aforementioned abnormal sample type and a second occurrence frequency of the aforementioned abnormal operation sample feature and the aforementioned abnormal sample type occurring simultaneously. The first occurrence frequency represents the number of times the aforementioned abnormal sample type occurs within a historical time period. The second occurrence frequency represents the number of times the abnormal operation sample feature and the abnormal sample type occur simultaneously within a historical time period.

[0069] The terminal can determine the confidence level of the target association corresponding to the above-mentioned abnormal operation characteristics based on the above-mentioned first occurrence frequency and the above-mentioned second occurrence frequency, and determine the above-mentioned first weight based on the above-mentioned confidence level.

[0070] Specifically, the terminal can determine the confidence level of association rules, the weight of abnormal operation features, the score of anomaly level, and the operational status information. For example, based on a classified sample set and normalized feature score data, the terminal mines meaningful association rules between underlying feature indicators (abnormal operation features) and comprehensive fault indicators (abnormal information) and calculates their confidence levels. Based on the obtained association rule confidence levels, the terminal further determines the weight coefficients (first weights) of the corresponding underlying feature indicators under each comprehensive indicator. Taking into account the scores of the underlying features, the calculated weights, and expert experience, the terminal experimentally sets the timeliness parameters of each feature in fault assessment during prediction, ultimately completing the calculation of the status scores and operational status information for all comprehensive fault indicators.

[0071] The confidence level of an association rule can be expressed as: confidence(P) i >L j )=P(L j |P i )=f(P i Lj ) / f(L j )×100%. Among them, confidence(P i >L j ) represents the association rule P i >L j The confidence level, i=1,2,…,N, j=1,2,…,M, f(P) i L j ) indicates the characteristic indicator (abnormal operation characteristic) P i When in an abnormal fluctuation state, fault (abnormal type) L j Frequency of occurrence (second occurrence frequency), f(L) j ) indicates fault L j The frequency of occurrence (first occurrence frequency). The lower the confidence level, the lower the reliability of the association rule.

[0072] The terminal further determines the weight coefficients (first weight) of the underlying feature indicators. Specifically, this can be expressed as: w ji =confidence(P i >L j ) / (∑ l=1 Kj confidence(P i >L j ))×100%, i=1,2,…,K j Among them, w ji (j=1,2,…,M, i=1,2,…,N) represents the weight of the underlying feature index under a specific comprehensive index mapping, K j This represents the total number of all feature indicators (such as abnormal sample features) under a specific comprehensive indicator mapping, j=1,2,…,M.

[0073] The terminal can determine the second weight based on the first weight and the determined number of current anomaly types. Specifically, the terminal first determines the status score value of the comprehensive fault category index (the score for the current anomaly level), which can be expressed as follows:

[0074] .

[0075] Where, k j 1 and k j 2These represent the number of test items and real-time monitoring items (such as the number of current anomaly types) under the comprehensive index mapping. ρ∈(0,1) is used to characterize the timeliness of test items in the comprehensive index scoring; a larger value indicates stronger effectiveness. Test items refer to equipment status data that can only be obtained through periodic, planned preventative tests or maintenance. This type of data has a longer update cycle, but the measurements are usually more comprehensive and accurate, reflecting the inherent, fundamental performance status of the equipment. Examples include laboratory physicochemical analysis data of transformer oil, winding DC resistance test values, and insulation resistance test values. Real-time monitoring items refer to operating status data continuously or automatically collected at high frequency through online monitoring sensors installed on the transformer. This type of data has strong real-time performance and can dynamically capture abnormal changes and trends in the equipment. Examples include online monitoring data of dissolved gases in transformer oil, online monitoring of partial discharge, core grounding current, top oil temperature, and winding hot spot temperature.

[0076] The terminal further determines the current operating status information of the transformer. For example... Figure 5 As shown, Figure 5 This is a flowchart illustrating the steps for determining the operating status in one embodiment. The terminal collects relevant characteristic data of the transformer to be evaluated, performs data normalization processing sequentially, and calculates the scores of its underlying characteristic indicators and comprehensive indicators. Simultaneously, it determines the weight (second weight) of each comprehensive indicator. The score p of the comprehensive indicator... j This refers to the severity score of a specific type of fault. It represents the transformer's status in a specific fault dimension. It is calculated from the underlying characteristic indicators associated with this type of fault and their dynamic weights (first weight). The score of the operating status of the assessed object refers to the comprehensive health score of the transformer as a whole. It represents the overall health level of the transformer. It is obtained by weighted summation of the comprehensive index scores of all fault categories and their secondary adjusted weights (second weight).

[0077] Based on the above results, the terminal calculates a comprehensive score of the overall operating status of the device (current operating status information), thereby determining its current operating status and clarifying the corresponding management and control level. For power transformers in abnormal or non-normal states, the terminal can combine the scores of its comprehensive indicators and underlying characteristic indicators to further analyze and infer the possible fault location range and its severity. The weighting formula for the comprehensive indicators can be expressed as: Q j =Q* j p α-1 j / ∑ i=1 M Q* i p α-1 iWhere α is the balancing parameter, 0≤α≤1. The smaller the value, the more emphasis is placed on the importance of severe deterioration fault information in the evaluation of the transformer's overall operating status. When the degree of damage caused to the equipment by the fault is relatively balanced, α can be set to 0. j The initial weights are typically set to Q*. j =1 / M. The formula for calculating the score (current operating status information) of the evaluated object's running status can be expressed as: R=∑ j=1 M Q j p j Among them, p j Q represents the status score (score for each current anomaly level) of each type of fault obtained in the first stage based on this evaluation method. j The corresponding weights (second weights) are used to describe the importance of various fault information in the overall transformer operation status assessment, j=1,2,…,M. The scores of the above-mentioned assessment objects' operation status can be used as input parameters for the deterioration prediction model to determine the transformer's operation status level (normal, warning, minor, abnormal, severe, etc.).

[0078] The score value of the operating status of the assessed object in the previous prediction period can also serve as a general trigger signal to initiate the prediction and maintenance decision-making process. For example, during the prediction period or in real-time monitoring, the terminal continuously calculates the above score value. Once the score value falls below a certain threshold, or a key real-time monitoring indicator suddenly becomes abnormal, the terminal will trigger an alarm and initiate subsequent analysis.

[0079] The scores for the operational status of the aforementioned assessed objects can also serve as a quantitative dashboard for the overall health management of transformers. For example, based on the scores and their historical trends, the terminal can evaluate the effectiveness of maintenance and optimize model parameters. After maintenance is performed, the R-value calculated from the re-collected data should show an increase. By comparing the changes in scores before and after maintenance, the terminal can quantitatively assess the effectiveness of the maintenance action. This data can be used to reverse-correct and optimize the weights and parameters in the status assessment model, as well as the empirical constants in the fault prediction model, making the system increasingly accurate with use.

[0080] Through the above embodiments, the terminal combines the confidence level of the association rules to determine the first weight of the abnormal operation characteristics, and combines the first weight, the abnormality level and the second weight to predict the current operating status information of the transformer, thereby enabling timely and accurate prediction of the transformer's operating status and improving the accuracy and timeliness of abnormal prediction of the transformer.

[0081] In one embodiment, the method further includes: acquiring abnormal operation sample features and an abnormal sample type set; determining a target normalization function based on the type of the abnormal operation sample features; normalizing the abnormal operation sample features according to the target normalization function to obtain normalized abnormal operation sample features; acquiring the support corresponding to the target abnormal sample type in the normalized abnormal operation sample features and the abnormal sample type set; the support characterizes the effectiveness of the association between the abnormal operation sample features and the abnormal sample type; if the support is greater than the support threshold, then establishing the association between the abnormal operation sample features and the target abnormal sample type.

[0082] In this embodiment, the terminal can pre-construct association rules between abnormal operation sample features and abnormal sample types. The terminal includes a sample library. The terminal can retrieve abnormal operation sample features and a set of abnormal sample types from the sample library. The set of abnormal sample types includes multiple abnormal sample types. The terminal can normalize each abnormal operation sample feature. Different types of abnormal operation sample features correspond to different normalization functions. For example, the terminal determines a target normalization function based on the type of the abnormal operation sample feature, and normalizes the abnormal operation sample features according to the target normalization function to obtain the normalized abnormal operation sample features.

[0083] The terminal can obtain the normalized features of the aforementioned abnormal operation samples and the support corresponding to the target abnormal sample type in the aforementioned set of abnormal sample types. The support represents the effectiveness of the association between the aforementioned abnormal operation sample features and the abnormal sample type. Therefore, the terminal determines whether to establish an association between the abnormal operation sample features and the target abnormal sample type based on the support. If the terminal detects that the support is greater than a support threshold, then an association between the aforementioned abnormal operation sample features and the aforementioned target abnormal sample type is established. The terminal can obtain the aforementioned association rules based on multiple association relationships.

[0084] Through this embodiment, the terminal can normalize the features of various types of abnormal operation samples by combining the normalization function, and use the normalized abnormal operation sample features and the support corresponding to the abnormal sample type to determine whether to establish a correlation relationship, thereby improving the accuracy of the established correlation relationship. Based on the accurate correlation relationship, the terminal can perform abnormal prediction of transformers, thereby improving the timeliness of abnormal prediction.

[0085] In one embodiment, obtaining the normalized abnormal operation sample features and the support corresponding to the target abnormal sample type in the abnormal sample type set includes: obtaining the normalized abnormal operation sample features and the third occurrence frequency of the abnormal sample type occurring simultaneously; and determining the support based on the third occurrence frequency and the number of abnormal sample types in the abnormal sample type set.

[0086] In this embodiment, the terminal can obtain the normalized third occurrence frequency of the aforementioned abnormal operation sample features and the aforementioned abnormal sample types occurring simultaneously. The third occurrence frequency represents the number of times the abnormal operation sample features and the aforementioned abnormal sample types occur simultaneously. The terminal can then determine the aforementioned support level based on the third occurrence frequency and the number of abnormal sample types in the aforementioned abnormal sample type set. For example, the support level can be determined by a ratio.

[0087] Specifically, the support of an association rule can be expressed as: Support(P i >L j )=P(P i ,L j )=F(P i ,L j ) / F(Ψ)×100%. Where, P i This represents the abnormal fluctuation event (abnormal operation sample feature) of the i-th feature indicator, i=1,2,…,N; L j Let Ψ represent the occurrence event of a specific type of fault (abnormal sample type), j=1,2,…,M. Ψ represents the total set of historical fault samples of the transformer to be evaluated (abnormal sample type set), and F(Ψ) is the total number of corresponding samples (the number of abnormal sample types in the abnormal sample type set); F(P) i ,L j ) represents event P in the fault sample. i and L j The number of samples occurring simultaneously (third occurrence frequency).

[0088] Through this embodiment, the terminal can use the frequency of occurrence of the third event and the number of abnormal samples to determine the support, thereby improving the reliability of the determined association.

[0089] In one embodiment, determining the target normalization function based on the type of the abnormal operation sample features includes: determining the target normal threshold corresponding to the abnormal operation sample features based on the type of the abnormal operation sample features; the target normal threshold characterizes the normal range of the operation parameters corresponding to the abnormal operation sample features; and determining the target normalization function based on the target normal threshold.

[0090] In this embodiment, different types of abnormal operating sample features correspond to different normalization functions to adapt to the normal thresholds of different abnormal operating sample features. The normal threshold represents the threshold at which the operating parameters corresponding to the abnormal operating sample feature are normal. The terminal can determine the target normal threshold corresponding to the abnormal operating sample feature based on its type. The target normal threshold represents the normal range of the operating parameters corresponding to the abnormal operating sample feature. Therefore, the terminal can determine the target normalization function based on the target normal threshold.

[0091] Specifically, when constructing a transformer operation status assessment system, the terminal can normalize all feature data (abnormal operation sample features), calculate the support of association rules between abnormal fluctuations of each feature and different fault types (abnormal sample types), extract effective association rules based on a set minimum support threshold, and then construct a mapping relationship between comprehensive fault indicators and underlying feature indicators. Since the normal thresholds for different feature data are inconsistent, the normalization formula can include multiple methods. For example, the normal threshold can be limited to [x* i -a i ,x* i +b i The characteristic index of ] has a normalization function as follows:

[0092] v i ={(x i -a i 0 ) / ((x* i -a i )-a i 0 )×0.8,d i 0 ≤x i <x* i -a i ;(x i -(x* i -a i )) / a i ×0.2+0.8,x* i -a i ≤x i <x* i ;((x* i +b i )-x i ) / b i ×0.2+0.8,x* i ≤x i ≤x* i +b i ;(d i 1 -x i) / (d i 1 -(x* i +b i ))×0.8,x* i +b i <x i ≤d i *}, i=1,2,…,N.

[0093] The normal threshold is limited to [x] i a ,x i b The characteristic index of ] has a normalization function as: v i ={(x i -d i 0 ) / (x i a -d i 0 )×0.8,d i 0 ≤x i <x i ^a;(x i -x i a ) / (x i a -x i b x × 0.2 + 0.8 i a ≤x i <x* i ;(x i b -x i ) / (x i b -x i a x* 0.2 + 0.8 i ≤x i ≤x i b ;(d i 1 -x i ) / (d i 1 -x i b )×0.8, x i b <x i ≤d i *}, i=1,2,…,N. Where x* i =(xi a +x i b ) / 2; For the normal threshold limitation method, x i ≤x* i The characteristic index of is normalized by the function: v i ={(x* i -x i ) / (x* i -d i 0 )×0.6+0.4,d i 0 ≤x i ≤x* i ;(d i 1 -x i ) / (d i 1 -x* i x* i <x i ≤d i *}, i=1,2,…,N. Where, d i 0 and d i * These represent the minimum and maximum values ​​of the characteristic limit, respectively.

[0094] Through this embodiment, the terminal can determine the corresponding normal threshold by combining the type of abnormal operation characteristics, and then determine the normalization function based on the corresponding normal threshold for normalization, so that the normalized abnormal operation characteristics can accurately express the changes in operation parameters, thereby improving the accuracy of the normalized abnormal operation characteristics.

[0095] In one exemplary embodiment, such as Figure 6 As shown, Figure 6 This is a flowchart illustrating a predictive maintenance method for transformers in another embodiment. In this embodiment, the terminal includes a data acquisition module, a condition assessment module, a fault prediction module, a maintenance decision module, and an execution feedback module. These modules work collaboratively to form a closed-loop management system. Wherein:

[0096] Data acquisition module: Through sensor networks and information sharing platforms, it collects real-time monitoring data of the transformer (such as oil chromatography data, partial discharge quantity, temperature, etc.), periodic pre-inspection test data (such as insulation resistance, winding DC resistance, etc.), and historical fault and maintenance records. The collected data is normalized, mapping indicators of different dimensions to the [0,1] interval for easier subsequent analysis.

[0097] Status assessment module: Based on multi-source data, an assessment model is built. First, the correlation between feature indicators and faults is mined through association rules to determine the weight of the underlying indicators. Then, fault status information is integrated to obtain a panoramic operation status score of the transformer.

[0098] Fault prediction module: Based on the severity level of the fault (such as level 1, level 2, level 3) and the current operating status of the equipment, a fault degradation model is constructed to predict the state change trend of potential faults during the maintenance phase.

[0099] Maintenance decision module: Based on the current status, predicted status, and severity of the fault, combined with limited maintenance resources and downtime constraints, an optimized decision model is constructed to realize the division of maintenance priorities and the rational allocation of maintenance resources.

[0100] Execution Feedback Module: Executes the maintenance plan and tracks its effects, feeding the maintenance results back to the system for updating the evaluation model and prediction parameters, thereby achieving continuous optimization.

[0101] Specifically, the terminal achieves quantitative measurement of the current state of the equipment by mining the correlation between characteristic indicators and faults and combining it with a dynamic weight adjustment mechanism.

[0102] The terminal can organize a sample database. For example, the terminal can access a fault sample database of transformers, number and classify each fault, and then categorize all operating states of the transformers into five levels: Normal, Attention, Minor, Abnormal, and Severe. Normal indicates that the transformer's performance in all aspects is good and it can maintain long-term safe and stable operation. Attention indicates that although a certain indicator has slight abnormal fluctuations, it does not affect its stable operation, but its development trend needs to be monitored. Minor indicates that the transformer has a fault, but it will not affect the safety and stability of the equipment in the short term. Abnormal indicates that the fault affects the stability of the transformer's operation and requires shutdown and maintenance. Severe indicates that the transformer's fault level poses a safety threat to the equipment and heating system, requiring immediate shutdown and maintenance.

[0103] The terminal can build a transformer operation status assessment system. For example, the terminal can normalize all feature data (abnormal operation sample features), calculate the support of association rules between abnormal fluctuations of each feature and different fault types (abnormal sample types), extract effective association rules based on a set minimum support threshold, and then construct a mapping relationship between comprehensive fault indicators and underlying feature indicators. Since the normal thresholds for different feature data are inconsistent, the normalization formula can include multiple methods. For example, the normal threshold can be limited to [x* i -a i ,x* i +b i The characteristic index of ] has a normalization function as follows:

[0104] vi ={(x i -a i 0 ) / ((x* i -a i )-a i 0 )×0.8,d i 0 ≤x i <x* i -a i ;(x i -(x* i -a i )) / a i ×0.2+0.8,x* i -a i ≤x i <x* i ;((x* i +b i )-x i ) / b i ×0.2+0.8,x* i ≤x i ≤x* i +b i ;(d i 1 -x i ) / (d i 1 -(x* i +b i ))×0.8,x* i +b i <x i ≤d i *}, i=1,2,…,N.

[0105] The normal threshold is limited to [x] i a ,x i b The characteristic index of ] has a normalization function as: v i ={(x i -d i 0 ) / (x i a -d i 0 )×0.8,d i 0 ≤x i <x i ^a;(x i -x i a) / (x i a -x i b x × 0.2 + 0.8 i a ≤x i <x* i ;(x i b -x i ) / (x i b -x i a x* 0.2 + 0.8 i ≤x i ≤x i b ;(d i 1 -x i ) / (d i 1 -x i b )×0.8, x i b <x i ≤d i *}, i=1,2,…,N. Where x* i =(x i a +x i b ) / 2; For the normal threshold limitation method, x i ≤x* i The characteristic index of is normalized by the function: v i ={(x* i -x i ) / (x* i -d i 0 )×0.6+0.4,d i 0 ≤x i ≤x* i ;(d i 1 -x i ) / (d i 1 -x* i x* i <x i ≤d i *}, i=1,2,…,N. Where, d i 0 and d i* These represent the minimum and maximum values ​​of the characteristic limit, respectively.

[0106] The support of an association rule can be expressed as: Support(P) i >L j )=P(P i ,L j )=F(P i ,L j ) / F(Ψ)×100%. Where, P i This represents the abnormal fluctuation event (abnormal operation sample feature) of the i-th feature indicator, i=1,2,…,N; L j Let Ψ represent the occurrence event of a specific type of fault (abnormal sample type), j=1,2,…,M. Let Ψ represent the total set of historical fault samples of the transformer to be evaluated (abnormal sample type set), and F(Ψ) be the total number of corresponding samples (the number of abnormal sample types in the abnormal sample type set); F(P) i ,L j ) represents event P in the fault sample. i and L j The number of samples occurring simultaneously (third occurrence frequency).

[0107] The terminal can determine the confidence level of association rules, the weight of abnormal operation features, the score of anomaly level, and the operational status information. For example, based on a classified sample set and normalized feature score data, the terminal mines meaningful association rules between underlying feature indicators (abnormal operation features) and comprehensive fault indicators (abnormal information) and calculates their confidence levels. Based on the obtained association rule confidence levels, the terminal further determines the weight coefficients (first weights) of the corresponding underlying feature indicators under each comprehensive indicator. Taking into account the scores of the underlying features, the calculated weights, and expert experience, the terminal experimentally sets the timeliness parameters of each feature in fault assessment during prediction, ultimately completing the calculation of the status scores and operational status information for all comprehensive fault indicators.

[0108] The confidence level of an association rule can be expressed as: confidence(P) i >L j )=P(L j |P i )=f(P i L j ) / f(L j )×100%. Among them, confidence(P i >L j ) represents the association rule P i >L j The confidence level, i=1,2,…,N, j=1,2,…,M, f(P)i L j ) indicates the characteristic indicator (abnormal operation characteristic) P i When in an abnormal fluctuation state, fault (abnormal type) L j Frequency of occurrence (second occurrence frequency), f(L) j ) indicates fault L j The frequency of occurrence (first occurrence frequency). The lower the confidence level, the lower the reliability of the association rule.

[0109] The terminal further determines the weight coefficients (first weight) of the underlying feature indicators. Specifically, this can be expressed as: w ji =confidence(P i >L j ) / (∑ l=1 Kj confidence(P i >L j ))×100%, i=1,2,…,K j Among them, w ji (j=1,2,…,M, i=1,2,…,N) represents the weight of the underlying feature index under a specific comprehensive index mapping, K j This represents the total number of all feature indicators (such as abnormal sample features) under a specific comprehensive indicator mapping, j=1,2,…,M.

[0110] The terminal can determine the second weight based on the first weight and the determined number of current anomaly types. Specifically, the terminal first determines the status score value of the comprehensive fault category index (the score for the current anomaly level), which can be expressed as follows:

[0111] .

[0112] Where, k j 1 and k j 2These represent the number of test items and real-time monitoring items (such as the number of current anomaly types) under the comprehensive index mapping. ρ∈(0,1) is used to characterize the timeliness of test items in the comprehensive index scoring; a larger value indicates stronger effectiveness. Test items refer to equipment status data that can only be obtained through periodic, planned preventative tests or maintenance. This type of data has a longer update cycle, but the measurements are usually more comprehensive and accurate, reflecting the inherent, fundamental performance status of the equipment. Examples include laboratory physicochemical analysis data of transformer oil, winding DC resistance test values, and insulation resistance test values. Real-time monitoring items refer to operating status data continuously or automatically collected at high frequency through online monitoring sensors installed on the transformer. This type of data has strong real-time performance and can dynamically capture abnormal changes and trends in the equipment. Examples include online monitoring data of dissolved gases in transformer oil, online monitoring of partial discharge, core grounding current, top oil temperature, and winding hot spot temperature.

[0113] The terminal further determines the current operating status information of the transformer. The terminal collects relevant characteristic data of the transformer to be evaluated, performs data normalization processing sequentially, and calculates the scores of its underlying characteristic indicators and comprehensive indicators, while determining the weight (second weight) of each comprehensive indicator. The score p of the comprehensive indicator... j This refers to the severity score of a specific type of fault. It represents the transformer's status in a specific fault dimension. It is calculated from the underlying characteristic indicators associated with this type of fault and their dynamic weights (first weight). The score of the operating status of the assessed object refers to the comprehensive health score of the transformer as a whole. It represents the overall health level of the transformer. It is obtained by weighted summation of the comprehensive index scores of all fault categories and their secondary adjusted weights (second weight).

[0114] Based on the above results, the terminal calculates a comprehensive score of the overall operating status of the device (current operating status information), thereby determining its current operating status and clarifying the corresponding management and control level. For power transformers in abnormal or non-normal states, the terminal can combine the scores of its comprehensive indicators and underlying characteristic indicators to further analyze and infer the possible fault location range and its severity. The weighting formula for the comprehensive indicators can be expressed as: Q j =Q* j p α-1 j / ∑ i=1 M Q* i p α-1 i Where α is the balancing parameter, 0≤α≤1. The smaller the value, the more emphasis is placed on the importance of severe deterioration fault information in the evaluation of the transformer's overall operating status. When the degree of damage caused to the equipment by the fault is relatively balanced, α can be set to 0.j The initial weights are typically set to Q*. j =1 / M. The formula for calculating the score (current operating status information) of the evaluated object's running status can be expressed as: R=∑ j=1 M Q j p j Among them, p j Q represents the status score (score for each current anomaly level) of each type of fault obtained in the first stage based on this evaluation method. j The corresponding weights (second weights) are used to describe the importance of various fault information in the overall operation status assessment of transformers, j=1,2,…,M.

[0115] Taking anomalies as an example, the terminal can accurately predict the evolution of potential faults in a transformer during maintenance, ensuring the effective execution of predictive maintenance strategies. The terminal can classify faults by severity level. For instance, based on the impact of a fault on equipment stability, the terminal divides transformer faults into three levels, clearly defining the degradation characteristics and maintenance strategies for each level. Level 1 faults represent the most severe type, requiring immediate shutdown and repair upon occurrence; Level 2 faults represent moderate severity but will affect transformer stability, requiring repair within a short period; Level 3 faults represent minor severity with a long degradation cycle, and can be included in rational maintenance.

[0116] The terminal can also construct a degradation prediction model. For example, based on parameters such as the current operating status of the transformer, the severity of the fault, and the equipment's operating time, combined with the transformer fault degradation patterns, the terminal can establish a degradation trend prediction model. The transformer fault state degradation prediction model can be expressed as: H z,l (t)~F z,l (t)=a z,l tb z,l Among them, the terminal definition function H z,l (t) describes the change in state score over time for different levels of faults under a specific operating condition. The terminal sets the time of the first occurrence of the fault signal as t0 (with t0=0 as the starting reference point for the maintenance plan). Variables z=1,2,3 represent the hazard level of the fault, where smaller values ​​indicate a greater threat to the safe operation of the system. Simultaneously, the operating state level of the equipment (operating state information, l=1,2,…,5) is also characterized in a similar way; lower level values ​​indicate better equipment health. The time point when the transformer fault deteriorates to its most severe state (target maintenance time point) is TF. z,l a z,l and b z,lThese represent empirical constants for a transformer in a specific state at fault level z. Their values ​​can be determined based on statistical analysis of historical operating data and the experience of domain experts. During prediction, the terminal inputs the transformer's current fault degradation time (current time), empirical constants, fault type, fault hazard level (current anomaly level), and equipment operating status level (current operating status information) into the degradation prediction model. The degradation prediction model then quantitatively describes the degradation trend of the fault from its current state to the degradation stage.

[0117] In determining maintenance strategies, the terminal combines fixed prediction cycles with fault development status maintenance methods based on fault degradation time. The terminal can pre-set prediction cycles. For example, the terminal divides the entire lifecycle of a transformer into three stages: initial, intermediate, and late. In the initial stage, the transformer begins operation, and the operation of new equipment, as well as its connection and coordination with other related equipment, may be unstable, requiring frequent monitoring of its operating status; therefore, the prediction cycle interval is relatively short. In the intermediate stage, the transformer's operation has reached a stable state, and the prediction cycle interval is longer. In the late stage, the transformer's components and functions begin to age, and the frequency of faults continuously increases, requiring a continuous reduction in the prediction cycle interval.

[0118] The terminal can also perform fault development status maintenance based on the fault deterioration time (target maintenance time). For example, within the prediction period, when the terminal detects abnormal fluctuations in a real-time monitorable indicator (operating parameter) (such as abnormal operating characteristics), the terminal performs an online assessment of the potential faults in the transformer and its current operating status. Based on this, it predicts the deterioration trend of the potential fault during the maintenance phase, and then allocates maintenance resources and arranges maintenance work.

[0119] For example, at the (i-1)th prediction time, the terminal did not detect any abnormal fluctuations in the transformer's operating parameters. After time interval T1, the i-th prediction time begins. The time interval between this prediction and the (i+1)th prediction is T2. In this prediction, the terminal detects abnormal fluctuations in the operating parameters at time T21. At this point, the terminal can perform fault diagnosis (identify abnormal information), status assessment (determine the transformer's operating status information), and fault prediction (determine the target maintenance time and maintenance strategy) for the transformer. The terminal completes the fault analysis at time T22 and predicts that the fault will deteriorate to its most severe state at time T23. Therefore, the terminal can perform maintenance on the transformer within time T4 and complete the maintenance work before time T24 (the target maintenance time).

[0120] Through the above embodiments, the terminal determines the current abnormal information by combining abnormal operation characteristics and association rules, predicts the deterioration trend and maintenance strategy of the transformer through the deterioration prediction model, and completes the maintenance of the transformer within the target maintenance time before abnormal deterioration based on the maintenance strategy, thereby improving the timeliness of transformer maintenance.

[0121] 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 in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0122] Based on the same inventive concept, this application also provides a predictive maintenance device for a transformer to implement the predictive maintenance method for transformers described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations of one or more embodiments of the predictive maintenance device for transformers provided below can be found in the limitations of the predictive maintenance method for transformers described above, and will not be repeated here.

[0123] In one exemplary embodiment, such as Figure 7 As shown, a predictive maintenance device for a transformer is provided, comprising: an acquisition module 500, a determination module 502, a prediction module 504, and a maintenance module 506, wherein:

[0124] The acquisition module 500 is used to acquire the corresponding abnormal operating characteristics of the transformer when the prediction conditions are met; the above-mentioned abnormal operating characteristics represent the operating characteristics corresponding to the abnormality of the operating parameters.

[0125] The determination module 502 is used to determine the current abnormal information corresponding to the above-mentioned transformer based on the above-mentioned abnormal operation characteristics and preset association rules; the above-mentioned preset association rules include the association relationship between each abnormal operation characteristic and the abnormal information; the above-mentioned current abnormal information includes the current time, current abnormality type, current abnormality level and current operation status information.

[0126] The prediction module 504 is used to input the above-mentioned current abnormal information into the deterioration prediction model; the above-mentioned deterioration prediction model is used to determine the target maintenance time and maintenance strategy corresponding to the above-mentioned transformer based on the above-mentioned current time, current abnormality type, current abnormality level and current operating status information in the above-mentioned abnormal information.

[0127] Maintenance module 506 is used to perform maintenance on the transformer before the target maintenance time according to the above maintenance strategy.

[0128] In one embodiment, the determining module 502 is configured to determine the current abnormality type corresponding to the abnormal operation feature and the current abnormality level corresponding to the current abnormality type according to the preset association rules; determine the second weight corresponding to the current abnormality level according to the first weight of the abnormal operation feature and the number of the current abnormality types; and determine the current operation status information according to the current abnormality level and the second weight.

[0129] In one embodiment, the determining module 502 is configured to: obtain the abnormal operation sample features corresponding to the abnormal operation features and the abnormal sample types corresponding to the abnormal operation sample features; obtain the first occurrence frequency of the abnormal sample types and the second occurrence frequency of the abnormal operation sample features and the abnormal sample types occurring simultaneously; determine the confidence level of the target association relationship corresponding to the abnormal operation features based on the first occurrence frequency and the second occurrence frequency; and determine the first weight based on the confidence level.

[0130] In one embodiment, the above apparatus further includes: an association module, configured to acquire abnormal operation sample features and an abnormal sample type set; determine a target normalization function based on the type of the abnormal operation sample features; normalize the abnormal operation sample features according to the target normalization function to obtain normalized abnormal operation sample features; acquire the support corresponding to the target abnormal sample type in the normalized abnormal operation sample features and the abnormal sample type set; the support characterizes the effectiveness of the association between the abnormal operation sample features and the abnormal sample type; if the support is greater than a support threshold, then establish an association relationship between the abnormal operation sample features and the target abnormal sample type.

[0131] In one embodiment, the aforementioned association module is used to obtain the normalized abnormal operation sample characteristics and the third occurrence frequency of the aforementioned abnormal sample types occurring simultaneously; and to determine the aforementioned support based on the aforementioned third occurrence frequency and the number of abnormal sample types in the aforementioned abnormal sample type set.

[0132] In one embodiment, the aforementioned association module is configured to determine a target normal threshold corresponding to the aforementioned abnormal operating sample feature based on the type of the aforementioned abnormal operating sample feature; the aforementioned target normal threshold characterizes the normal range of the operating parameters corresponding to the aforementioned abnormal operating sample feature; and determine the aforementioned target normalization function based on the aforementioned target normal threshold.

[0133] The modules in the aforementioned predictive maintenance device for transformers can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0134] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 8 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When executed by the processor, the computer program implements a predictive maintenance method for transformers. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0135] Those skilled in the art will understand that Figure 8 The 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.

[0136] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described predictive maintenance method for transformers.

[0137] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the above-described predictive maintenance method for transformers.

[0138] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the above-described predictive maintenance method for transformers.

[0139] 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.

[0140] Those skilled in the art will understand that all or part of the processes in the methods of 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, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory 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, artificial intelligence (AI) processors, etc., and are not limited to these.

[0141] 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 application.

[0142] 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 predictive maintenance method for transformers, characterized in that, The method includes: When the prediction conditions are met, the corresponding abnormal operation characteristics of the transformer are obtained; the abnormal operation characteristics represent the operation characteristics corresponding to the presence of abnormal operating parameters. Based on the abnormal operation characteristics and preset association rules, the current abnormal information corresponding to the transformer is determined; the preset association rules include the association relationship between each abnormal operation characteristic and the abnormal information; the current abnormal information includes the current time, current abnormal type, current abnormal level, and current operating status information. The current anomaly information is input into the deterioration prediction model; the deterioration prediction model is used to determine the target maintenance time and maintenance strategy for the transformer based on the current time, current anomaly type, current anomaly level and current operating status information in the anomaly information. According to the maintenance strategy, the transformer shall be maintained before the target maintenance time.

2. The method according to claim 1, characterized in that, The steps for obtaining the current operating status information include: Based on the preset association rules, determine the current anomaly type corresponding to the abnormal operation feature, and the current anomaly level corresponding to the current anomaly type; Based on the first weight of the abnormal operation feature and the number of the current abnormality type, determine the second weight corresponding to the current abnormality level; The current operating status information is determined based on the current anomaly level and the second weight.

3. The method according to claim 2, characterized in that, The steps for obtaining the first weight include: Obtain the abnormal operation sample features corresponding to the abnormal operation features and the abnormal sample types corresponding to the abnormal operation sample features; Obtain the first occurrence frequency of the abnormal sample type, and the second occurrence frequency of the abnormal running sample feature and the abnormal sample type occurring simultaneously; Based on the first occurrence frequency and the second occurrence frequency, determine the confidence level of the target association corresponding to the abnormal operation feature; The first weight is determined based on the confidence level.

4. The method according to any one of claims 1 to 3, characterized in that, The method further includes: Obtain the characteristics and types of abnormal operation samples; The target normalization function is determined based on the type of the abnormal operation sample characteristics; The abnormal operation sample features are normalized according to the target normalization function to obtain the normalized abnormal operation sample features. Obtain the normalized features of the abnormal operation samples and the support corresponding to the target abnormal sample type in the set of abnormal sample types; the support characterizes the effectiveness of the association between the features of the abnormal operation samples and the abnormal sample types. If the support is greater than the support threshold, then an association is established between the abnormal operation sample features and the target abnormal sample type.

5. The method according to claim 4, characterized in that, The process of obtaining the normalized features of the abnormal running samples and the support corresponding to the target abnormal sample type in the set of abnormal sample types includes: Obtain the third occurrence frequency of the normalized abnormal operation sample features and the abnormal sample type occurring simultaneously; The support is determined based on the third occurrence frequency and the number of abnormal sample types in the abnormal sample type set.

6. The method according to claim 4, characterized in that, The step of determining the target normalization function based on the type of the abnormal operation sample characteristics includes: Based on the type of the abnormal operation sample feature, a target normal threshold corresponding to the abnormal operation sample feature is determined; the target normal threshold represents the normal range of the operation parameter corresponding to the abnormal operation sample feature. The target normalization function is determined based on the target normal threshold.

7. A predictive maintenance device for a transformer, characterized in that, The device includes: The acquisition module is used to acquire the corresponding abnormal operating characteristics of the transformer when the prediction conditions are met; the abnormal operating characteristics represent the operating characteristics corresponding to the presence of abnormal operating parameters; The determination module is used to determine the current abnormal information corresponding to the transformer based on the abnormal operation characteristics and preset association rules; the preset association rules include the association relationship between each abnormal operation characteristic and the abnormal information; the current abnormal information includes the current time, the current abnormal type, the current abnormal level, and the current operating status information. The prediction module is used to input the current anomaly information into the deterioration prediction model; the deterioration prediction model is used to determine the target maintenance time and maintenance strategy for the transformer based on the current time, current anomaly type, current anomaly level and current operating status information in the anomaly information. A maintenance module is used to perform maintenance on the transformer before the target maintenance time, according to the maintenance strategy.

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 6.

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 6.

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 6.

Images