A preventive test method for transformers in power grid renovation

Abstract

This invention belongs to the field of power system technology and relates to a preventive testing method for transformers used in power grid upgrades. The invention acquires information on transformer family defects and equipment status time-series data, determines primary and secondary factor parameters based on historical fault statistics, and uses a multi-source information fusion strategy combined with a dual-model approach to calculate the baseline value of remaining lifespan, dynamically correcting and generating lifespan assessment results. When the primary factor parameters exceed the lifespan risk threshold, an early warning signal is output; otherwise, based on the lifespan assessment results, the maximum relative proximity of the primary factor parameters, the load level of the power supply area, and the risk of family defects, a multi-dimensional hierarchical sorting process is used to generate a upgrade priority sequence. The lifespan assessment results are continuously monitored, and upgrades are initiated when the upgrade trigger threshold is reached. This method effectively improves the accuracy of transformer remaining lifespan prediction, optimizes equipment maintenance decisions, reduces power grid operation risks, extends equipment lifespan, and enhances the safe and stable operation level of the power grid.

Description

Technical Field

[0001] This invention belongs to the field of power system technology and relates to a preventive testing method for transformers used in power grid renovation. Background Technology

[0002] As a key piece of equipment in the power system, the reliable operation of transformers is crucial to grid security. With the continuous expansion of the power grid and the increasing service life of equipment, conducting preventive testing and upgrades of transformers in a scientific and rational manner has become a major challenge for power companies. Currently, traditional preventive testing mainly relies on fixed cycles, failing to fully consider individual differences in equipment and actual operating conditions, making it difficult to support precise and differentiated upgrade decisions.

[0003] In transformer life assessment, existing technologies mostly rely on single methods. Some methods establish life models based on thermal aging theory, but due to the simplification assumptions in the theoretical models, they are difficult to accurately reflect the actual aging process under complex operating conditions. Other methods use artificial intelligence algorithms for state prediction, which have good data fitting capabilities, but lack physical mechanism support and have insufficient interpretability.

[0004] In terms of power grid upgrade decisions, existing methods do not comprehensively consider multi-dimensional factors such as differences in load levels across power supply areas. These load level differences directly affect power supply reliability requirements, and failure to prioritize upgrades in critical load areas will increase the operational risks of the power grid. Furthermore, existing methods have not established an effective correlation between family defect information and upgrade decisions in transformer preventative testing. This information can be used to screen key parameters that significantly impact transformer lifespan. The lack of effective utilization of such information weakens the accuracy of remaining life analysis and leads to a lack of targeted allocation of upgrade resources, making it difficult to maximize safety benefits. Summary of the Invention

[0005] In view of this, in order to solve the problems mentioned in the background technology, a preventive test method for power grid transformation transformers is proposed.

[0006] The objective of this invention can be achieved through the following technical solution: a preventive test method for transformers used in power grid renovation, comprising: acquiring family defect information and current equipment status time sequence data of the target transformer.

[0007] Based on information about family-related defects, the main and auxiliary factors that affect the lifespan of transformers were determined.

[0008] By integrating primary factor parameters, auxiliary factor parameters, and current equipment status time-series data, a multi-source information fusion strategy is used to analyze the remaining life of the transformer, obtain a baseline value for the remaining life, and dynamically correct this baseline value to generate a life assessment result that includes the median and confidence interval.

[0009] If any primary factor parameter exceeds its corresponding life risk threshold, an equipment warning signal will be output. If it does not exceed the threshold, a renovation priority sequence will be generated based on the life assessment results, combined with the maximum relative proximity of the primary factor parameters, the load level of the power supply area, and the defect risk of the equipment family, through multi-dimensional hierarchical sorting.

[0010] The life assessment results are continuously monitored, and when the results reach the transformation trigger threshold, the transformer transformation start signal is triggered.

[0011] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention adopts a multi-source information fusion strategy that integrates family defect information and real-time status data, and combines thermal aging theory and neural network model to conduct preventive tests on transformer remaining life analysis, which solves the problem that the traditional method relies on a single model, resulting in one-sided or inaccurate evaluation results. This method not only generates a remaining life benchmark value, but also dynamically corrects the output evaluation results, realizing a more comprehensive, reliable and quantitative evaluation of transformer life, providing a solid risk basis for transformation decisions.

[0012] (2) This invention automatically identifies the main and auxiliary factors that have the greatest impact on transformer lifespan by statistically analyzing the family-related defect history of transformers of the same model and batch. This method enables the focus on key risk factors from massive defect information, providing precise target dimensions for subsequent life assessment and risk warning, and improving the accuracy and pertinence of condition evaluation.

[0013] (3) In the retrofit decision-making stage, this invention introduces a four-dimensional hierarchical ranking mechanism that integrates remaining lifespan, maximum relative proximity, load level of the power supply area, and family defect risk. This solves the problem that existing retrofit priority ranking standards are singular and fail to take into account both the equipment's own condition and the grid's operational needs. This method can generate a scientific retrofit sequence, ensuring that retrofit resources are prioritized for equipment with short lifespan, high risk, located in important load areas, and with severe family defects, thereby achieving a refined and differentiated retrofit strategy. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart illustrating the preventive testing method for a power grid renovation transformer according to the present invention.

[0016] Figure 2This is a flowchart of the method for obtaining life assessment results in this invention.

[0017] Figure 3 This is a schematic diagram of the modified priority sequence rule in this invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 As shown, the present invention provides a preventive test method for transformers in power grid transformation, including: Step 1, obtaining family defect information and current equipment status time sequence data of the target transformer.

[0020] In one embodiment of the present invention, considering that the assessment of the remaining life of the transformer and the determination of the priority of renovation need to be based on comprehensive and reliable basic data, historical defect information and real-time operating status data are collected through multiple channels.

[0021] Meanwhile, family-related defect information reveals potential inherent risks in equipment of the same model or batch due to design, materials, or processes, providing prior failure modes and probability distributions for life assessment; while current equipment status time-series data reflects the real-time operational health and performance degradation trajectory of the target transformer, serving as a direct basis for assessing its individual aging degree and remaining lifespan. The fusion of these two methods effectively overcomes the limitations of relying on a single data source, ensuring the accuracy of subsequent analysis results.

[0022] Specifically, this step includes: extracting structured family-based defect information from the power grid production management system, equipment manufacturer defect reports, and industry defect sharing platforms. This information refers to records of common defects caused by design, materials, or processes in equipment of the same model, batch, or manufacturer, such as insulating oil deterioration, winding deformation, and cooling system failure.

[0023] Simultaneously, online monitoring devices installed on the transformer collect current equipment status time-series data. This data includes oil chromatography data such as the content of gases like H2, CH4, and C2H2, temperature data such as winding temperature, top oil temperature, and partial discharge data.

[0024] As an example, familial defect information is updated monthly or quarterly; device status time-series data is collected every 15 minutes.

[0025] Preferably, after data acquisition, the process also includes data preprocessing and automatic verification: First, the raw data is cleaned by using a moving window-based box plot method or a time-series anomaly detection algorithm to identify and remove abnormal values ​​caused by instantaneous sensor interference.

[0026] Secondly, data alignment is performed by using linear interpolation to unify familial defect information with high-frequency acquired status data onto the same analysis time series.

[0027] Next, automatic verification is performed: the system will automatically verify the integrity, temporal continuity and physical rationality of the data. In the integrity verification stage, the system automatically counts the number of valid data points in a single monitoring period and calculates the percentage of the theoretically required number of data points. When the percentage is less than 95%, the verification is deemed to have failed.

[0028] In the time continuity verification stage, the system detects the timestamp interval between adjacent data points, allowing normal jitter within ±1 minute. However, if five or more consecutive data points are missing, i.e., five or more consecutive gaps appear in the time series, the verification is deemed to have failed.

[0029] In the physical rationality verification stage, the system compares the values ​​of each monitoring parameter with the preset physical rationality range. For example, it verifies whether the oil temperature monitoring value is within the reasonable range of -20℃ to 100℃, and whether the acetylene (C2H2) content is lower than the safety threshold of 5μL / L. Any parameter value that exceeds the preset reasonable range will result in the verification failing.

[0030] The reasonable range can be configured in the system parameters according to the transformer model and operating environment.

[0031] Furthermore, the specific process for data supplementation or alarm processing is as follows: If the verification fails, the system first automatically triggers a data supplementation request. If the data is still unqualified after supplementation, an alarm work order containing the device ID, the failed verification item, and the current data value is immediately generated and pushed to the alarm center of the power grid production management system through the message middleware to remind maintenance personnel to handle it in a timely manner.

[0032] Through the above operations, all verified data will be stored in the historical database and a mapping relationship with the transformer ID will be established, providing a data basis for determining the parameters of the main factors and auxiliary factors in step two.

[0033] Step 2: Based on information about family-related defects, determine the main and auxiliary factors that affect the transformer's lifespan.

[0034] In one embodiment of the present invention, considering that different defect factors have different degrees of influence on the life of transformers, the core influencing factors are screened by statistical analysis of the probability of defect occurrence.

[0035] By identifying the top N types of defects with the highest cumulative occurrence rate as primary factor parameters and the rest as auxiliary factor parameters, this method aims to focus limited analytical resources on the key parameters that have the greatest impact on equipment lifespan, thereby effectively improving subsequent remaining lifespan calculations.

[0036] Specifically, the steps include: statistically analyzing historical failure records in the family defect information, calculating the cumulative occurrence ratio of each type of defect factor in the same batch of equipment based on the ratio of the number of failures caused by the defect factor to the total number of equipment in the same batch; then, determining the parameters corresponding to the top N types of defect factors with the highest cumulative occurrence ratio as the main factor parameters, and the rest as auxiliary factor parameters.

[0037] As an example, assuming N is 3, the system will automatically select the three most frequent defect factors, such as insulating oil deterioration, winding deformation, and cooling system failure, as the main factor parameters based on historical data, and other factors as auxiliary factor parameters. The implementer can adjust the value of N according to the equipment type and importance, and it can generally be set to 2 to 5.

[0038] Furthermore, in the subsequent remaining life analysis, the main factor parameters will serve as the core basis for judging the life risk threshold and ranking in multiple dimensions, while the auxiliary factor parameters will be mainly used to dynamically correct the remaining life baseline value and assess the confidence level of the results.

[0039] Preferably, in one embodiment of the present invention, the method for obtaining the main factor parameters and auxiliary factor parameters further includes a dynamic adjustment mechanism for parameter weights. When the system detects that an auxiliary factor parameter exceeds its historical normal range for multiple consecutive periods, such as three periods, for example, exceeding the mean ± three times the standard deviation, its weight can be temporarily increased and included in the analysis range of the main factor parameters.

[0040] Through the above operations, a system of key factors affecting transformer lifespan can be obtained, providing parameter input for lifespan analysis in step three.

[0041] Step 3: Integrate main factor parameters, auxiliary factor parameters, and current equipment status time series data to perform remaining life analysis on the transformer using a multi-source information fusion strategy, obtain a baseline value for remaining life, and dynamically correct this baseline value to generate a life assessment result including the median and confidence interval.

[0042] In one embodiment of the present invention, considering the inherent limitations of a single lifetime assessment method: if only an equivalent lifetime assessment method based on the principle of thermal aging is used, although the physical meaning is clear, it is difficult to reflect the actual degradation process under the coupling effect of multiple factors; while if only data-driven models such as long short-term memory networks are relied upon, the results are highly dependent on the quality and coverage of training data, and the generalization ability is limited when data is scarce or when there are no previously seen patterns.

[0043] Therefore, a method is proposed to integrate the mechanistic model and the data-driven model. The mechanistic model provides a life decay benchmark that conforms to physical laws, while the data-driven model captures the nonlinear degradation characteristics in time series data. Then, the two estimates derived from different principles are arithmetically averaged to combine the advantages of both, improve the uncertainty of a single model, and obtain a more robust remaining life benchmark value.

[0044] Specifically, this step includes: employing an equivalent life assessment method based on the principle of thermal aging, the core of which lies in utilizing the exponential relationship between the aging rate of insulation materials and hot spot temperature. First, based on the transformer's real-time load rate and cooling method, the winding hot spot temperature is calculated using a dynamic thermal circuit model. This model equates the heat transfer process from the transformer winding to the environment to a time-varying system composed of thermal resistance and thermal capacity. The inputs are the real-time load rate, top oil temperature, and ambient temperature, and the output is the winding hot spot temperature. The thermal resistance and time constant in the model are identified from historical temperature rise test data or online monitoring data of the same type of equipment.

[0045] Secondly, the hot spot temperature is substituted into the Arrhenius form of the aging rate expression to calculate the instantaneous aging rate at each moment; then, the instantaneous aging rate is integrated over time over the operating period to obtain the cumulative aging amount.

[0046] Finally, the cumulative aging amount is considered as the equivalent operating time consumed by the equipment, and a first estimate of the remaining lifespan is derived by reverse calculation. The material-related parameters in the aging rate expression are determined from historical aging test data of the same model of equipment.

[0047] Simultaneously, a Long Short-Term Memory (LSTM) network model is adopted, using the feature sequence composed of main factor parameters, auxiliary factor parameters, and equipment status time series data as input, and outputting a second estimate.

[0048] Preferably, the method for constructing and training the long short-term memory network model specifically includes: extracting a large amount of historical operating data of transformers of the same type from the power grid historical database as a training set, wherein the data includes, but is not limited to, equipment status time series data, family defect records, and the final failure scrapping time or the remaining life value determined by disassembly inspection; using status data and defect factors as input features, and the remaining life as a label, the network is pre-trained through supervised learning; when applied to the target transformer, the model parameters are fine-tuned online using the transformer's own monitoring data to better adapt it to the individual characteristics of the equipment.

[0049] It should be understood that the arithmetic mean is a basic and effective way to fuse the results of two models. In other alternative embodiments, implementers may also use other fusion algorithms such as weighted averaging or dynamic weight allocation based on model confidence, based on the historical performance of the models or expert experience, all of which fall within the protection scope of this invention.

[0050] See Figure 2 As shown, the method for obtaining the life assessment results is as follows: the baseline value is dynamically corrected according to the changing trend of the main factor parameters. If any main factor parameter shows a monotonically increasing trend in three or more consecutive monitoring periods, the remaining life baseline value is multiplied by a preset correction coefficient.

[0051] Based on the dispersion of the main factor parameters of the same model of equipment in the historical database, such as the standard deviation, it is assumed that the remaining life follows a log-normal distribution. The corrected life value is used as the median of the distribution, i.e. the 50th percentile. Then, the life confidence interval at a pre-set confidence level, such as 90%, is calculated as the final life assessment result.

[0052] It should be explained that the log-normal distribution is used because the remaining lifetime is a positive value and its distribution usually has a right-skewed characteristic. Therefore, the log-normal distribution is more in line with engineering practice than the normal distribution.

[0053] The reason for using a probability distribution to output the results here is that the transformer's lifespan degradation process itself is uncertain, influenced by various random factors such as equipment materials, operating conditions, and unforeseen events. A single lifespan estimate cannot reflect this uncertainty. Providing a probability distribution that includes the median and confidence intervals can more comprehensively and scientifically characterize the possible range of remaining lifespan. For example, a 90% confidence interval of [5.8 years, 8.6 years] means there is a 90% certainty that the transformer's true remaining lifespan falls within this range. The lower bound of 5.8 years is particularly important for formulating conservative renovation plans.

[0054] As an example, suppose the first estimate for a 110kV transformer is 8.5 years and the second estimate is 7.2 years, then the baseline value is 7.85 years. When the system detects that the main factor parameter is strictly monotonically increasing over three consecutive monitoring periods, the baseline value is multiplied by a correction factor of 0.9 to obtain a corrected value of 7.065 years. Furthermore, the system will generate a lifespan range with a 90% confidence level, such as [5.8 years, 8.6 years], based on the dispersion of historical data and using the corrected value as the median.

[0055] Furthermore, in the method for obtaining the life assessment results, the preset correction coefficient can be dynamically adjusted according to the increasing rate of the main factor parameters. Its core construction logic lies in the fact that the deterioration of the main factor parameters is not uniform; their increasing slope, i.e., the rate of change, is a key indicator for predicting accelerated life reduction. The larger the slope, the faster the parameter deteriorates, and the more significant the reduction in remaining lifespan.

[0056] Based on this logic, the specific dynamic adjustment method is as follows: the slope k of the parameter change curve in the most recent three periods is calculated by linear regression. In order to eliminate the influence of dimensions between different parameters, the slope k is first normalized to convert it into a dimensionless value.

[0057] Subsequently, the correction factor was set to α is a sensitivity factor, whose value range is set to 0.05 to 0.2 based on expert experience. For example, it can be 0.1, which is used to adjust the intensity of the effect of slope on life reduction.

[0058] This formula ensures that the faster the parameters deteriorate (i.e., the larger the |k| value), the smaller the correction factor, thus allowing for a more significant reduction in the remaining lifetime baseline. At the same time, to avoid overly inaccurate assessment results when parameters deteriorate rapidly, a lower limit is set for this correction factor, for example, not less than 0.7.

[0059] Through the above operations, a life assessment result including the median and confidence interval was obtained, providing a core basis for the early warning judgment and renovation priority ranking in step four.

[0060] Step 4: If any primary factor parameter exceeds its corresponding life risk threshold, an equipment warning signal will be output. If it does not exceed the threshold, a renovation priority sequence will be generated based on the life assessment results, combined with the maximum relative proximity of the primary factor parameters, the load level of the power supply area, and the defect risk of the equipment family, through multi-dimensional hierarchical sorting.

[0061] In one embodiment of the present invention, considering that traditional transformer preventive testing methods usually rely on fixed cycles and single thresholds, it is difficult to accurately assess the true aging state and urgency of the equipment, which leads to the risk of delayed or overly conservative maintenance and renovation decisions.

[0062] Therefore, by using the life risk threshold, we can achieve immediate early warning of emergency risks and ensure operational safety. For equipment that has not triggered an early warning, we can build a quantitative comprehensive scoring system by integrating four dimensions: median remaining life, maximum relative proximity, load level of the power supply area, and family defect risk. This system will map factors such as the health status of the transformer, parameter degradation trend, power supply importance, and historical reliability to a sortable priority sequence, and ultimately provide accurate data support for prioritizing the renovation of the highest-risk and most impactful equipment within a limited budget.

[0063] Specifically, the steps include: first, determining whether the main factor parameters exceed the life risk threshold; if they do, triggering an early warning; if they do not exceed the threshold, calculating four priority indicators and ranking them hierarchically.

[0064] As an example, suppose the dissolved gas content in a transformer oil reaches 105% of its lifespan risk threshold as one of the main factor parameters, an equipment warning signal will be immediately issued. For equipment that has not exceeded the limit, the first indicator will be calculated based on the ratio of the median remaining lifespan to the design lifespan. If the ratio is 0.35, the first indicator will score 65 points. The second indicator will be calculated based on the ratio of the current value of the main factor parameter to its threshold value. If the maximum relative proximity is 0.8, the second indicator will score 80 points. The third indicator will be determined based on the load level of the power supply area. If it is a level two load, the third indicator will score 85 points. The fourth indicator will be determined based on the number of main factor parameters. If there are two main factor parameters, the fourth indicator will score 75 points. The number of main factor parameters is determined based on statistics from the same batch of equipment.

[0065] Preferably, in one embodiment of the present invention, the method for obtaining the life risk threshold is based on historical data mapping, that is, the minimum acceptable life is determined according to the power grid transformation planning cycle, such as 10 years, and then the corresponding main factor parameter values ​​are found in the historical mapping relationship.

[0066] Further, see Figure 3 As shown, the method for generating the renovation priority sequence includes: sorting all transformers to be evaluated first by the first indicator from high to low, then sorting by the second indicator if they are the same, then sorting by the third indicator if they are still the same, and finally sorting by the fourth indicator, thereby forming the renovation priority sequence.

[0067] This hierarchical ranking strategy ensures that the device with the shortest remaining lifespan has the highest ranking weight. Based on this, its parameter degradation rate, power supply importance, and inherent defect risk are compared in turn, forming a multi-level decision-making logic from core security to comprehensive impact.

[0068] In other embodiments, a weighted comprehensive scoring method can be used instead of hierarchical ranking, based on the actual operating conditions of the power grid, that is, weights are assigned to the four indicators respectively.

[0069] For example, the first indicator, the relative value of remaining life, directly reflects the core dimension of the equipment's own health status and is the fundamental basis for the necessity of modification, so it is given the highest weight, such as 40%. The second indicator, the maximum relative proximity, characterizes the immediate deterioration risk of key parameters and is an important supplement to the remaining life trend, so it is given a secondary weight, such as 30%.

[0070] The third indicator, the load level of the power supply area, reflects the functional importance of the equipment in the power grid. It is a key external factor from the perspective of system safety, and therefore it is given a third weight, for example, 20%. The fourth indicator, family defect risk, reflects the inherent reliability defects of the equipment family. It is used as a background risk factor for final fine-tuning, and therefore it is given a minimum weight, for example, 10%. Based on this, a linear weighting is performed to obtain the priority ranking of the renovation. The specific weight values ​​can be calibrated and adjusted according to the operation and maintenance strategies of different power grid companies.

[0071] Through the above operations, the risk warning of equipment operation or the priority ranking of transformation was completed, providing a decision basis for triggering the transformation start signal in step five.

[0072] Step 5: Continuously monitor the life assessment results. When the results reach the transformation trigger threshold, trigger the transformer transformation start signal.

[0073] In one embodiment of the present invention, considering that the remaining life of a transformer is a probabilistic assessment value that evolves dynamically with the operating conditions, by continuously monitoring and combining it with the critical value for triggering the transformation, it can be ensured that the transformation is initiated in time before the equipment life approaches the acceptable lower limit, thereby ensuring the safe and stable operation of the power grid.

[0074] Specifically, this step includes: setting a minimum acceptable lifespan based on the power grid renovation planning cycle, such as 8 years; and, to ensure the conservatism and reliability of the decision, selecting a high-confidence quantile, such as 95%, as the renovation trigger threshold from the remaining lifespan probability distribution. This means that when the lifespan assessment results trigger this condition, there is a 95% certainty that the transformer's actual remaining lifespan is already below the minimum acceptable lifespan.

[0075] Then, the life assessment results are continuously tracked. When the median remaining life obtained from monitoring is less than or equal to the critical value within multiple consecutive monitoring cycles, such as 3 cycles, the modification start signal is triggered. This continuous judgment mechanism can effectively avoid false triggering caused by fluctuations in single monitoring data or instantaneous errors in the model.

[0076] When the system detects that the median remaining lifespan of a transformer is less than or equal to 8.5 years for three consecutive monitoring cycles, such as one week, it will automatically generate a modification start signal and push a standard work order containing information such as transformer ID, current lifespan assessment result, and trigger basis to the operation and maintenance management system.

[0077] Preferably, the modification triggering conditions can be dynamically and adaptively adjusted according to the power grid operation status. For example, during the expected peak load period, such as the summer peak electricity consumption period, the implementer can temporarily increase the number of continuous monitoring cycles required for the triggering conditions from 3 to 5, or temporarily adjust the confidence level from 95% to 90%, thereby strategically postponing some non-extremely urgent modification projects to ensure the reliability of power supply and operational safety during the peak summer season.

[0078] For example, once the upgrade start signal is triggered, the system sends an upgrade start notification to the operation and maintenance management system. The implementer can then finalize the upgrade plan and schedule the timeline based on the actual operating status of the power grid. Through this process, a scientific decision-making process for the timing of transformer upgrades is achieved, effectively balancing equipment safety and upgrade costs.

[0079] In summary, this invention integrates family-related defect information with current equipment status time-series data, combines a multi-source information fusion strategy with a multi-dimensional hierarchical sorting mechanism, and achieves intelligent management of preventive testing and retrofitting decisions for power grid transformers. This improves the accuracy and efficiency of power equipment operation and maintenance, effectively reduces power grid operation risks, and optimizes the allocation of retrofitting resources.

[0080] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0081] Those skilled in the art will recognize that the algorithmic steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0082] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0083] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0084] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method of preventive test of a grid-modification transformer, characterized in that, The application relates to a transformer remaining life assessment method and device. The method comprises the following steps: acquiring family defect information and current equipment state time series data of a target transformer; determining main factor parameters and auxiliary factor parameters related to transformer life based on the family defect information; performing remaining life analysis on the transformer by fusing the main factor parameters, the auxiliary factor parameters and the current equipment state time series data through a multi-source information fusion strategy, obtaining a remaining life benchmark value, dynamically correcting the benchmark value, and generating a life assessment result containing a median value and a confidence interval; if any main factor parameter exceeds a corresponding life risk critical value, outputting a device warning signal, and if the main factor parameter does not exceed the life risk critical value, generating a transformation priority sequence through multi-dimensional hierarchical sorting based on the life assessment result and in combination with the maximum relative proximity of the main factor parameter, a power supply area load level and a device family defect risk; continuously monitoring the life assessment result, and triggering a transformer transformation start signal when the result reaches a transformation triggering critical value; 2. A preventive test method for a grid-modification transformer according to claim 1, characterized in that, the transformation priority sequence generation method comprises the following steps: acquiring four priority indexes of each target transformer; a first index is calculated according to the ratio of the output remaining life median value to the device design life, and the first index linearly increases with the decrease of the remaining life median value; a second index is calculated according to the maximum relative proximity of all main factor parameters in a current monitoring period, wherein the relative proximity is the ratio of the current value of the parameter to the corresponding life risk critical value, and the ratio is linearly converted into a score value of 0-100 points, and the larger the ratio is, the higher the second index is; a third index is obtained from a preset mapping table according to the load level of the power supply area in the power grid production management system, wherein the highest score value corresponds to a first-level load area, and the lowest score value corresponds to a third-level load area; and a fourth index is set according to the number of identified main factor parameters, and the more the main factor parameters are, the higher the fourth index is; specifically, when the number of main factor parameters is not less than two, the fourth index takes a first preset score value; when the number of main factor parameters is one, the fourth index takes a second preset score value which is lower than the first preset score value; and when no main factor parameter is identified, the fourth index takes a third preset score value which is lower than the second preset score value. The method for acquiring the family defect information and the current equipment state time series data comprises the following steps: acquiring the family defect information from a power grid production management system, a device manufacturer defect report and an industry defect sharing platform; time marking the family defect information according to a set sampling period; collecting current equipment state time series data at a preset time interval through an online monitoring device installed on the transformer, wherein the data comprises oil chromatogram data, temperature data and partial discharge data; 3. A preventive test method for grid-modification transformers according to claim 1, characterized in that, after the data is collected, automatically checking the integrity, time continuity and physical rationality of the data, and if the checking is passed, the data is used for subsequent analysis, and if the checking is not passed, data resampling or alarm processing is started. The method for acquiring the main factor parameters and the auxiliary factor parameters comprises the following steps: statistically analyzing historical failure records in the family defect information, and calculating the cumulative occurrence proportion of each type of defect factor in the same batch of equipment; The parameters corresponding to the top N defect factors with the highest cumulative occurrence ratio are determined as the main factor parameters, and the rest are determined as the auxiliary factor parameters, where N is a preset positive integer.

4. The preventive test method for grid-modification transformer according to claim 1, wherein The method for obtaining the remaining life benchmark value is: An equivalent life evaluation method based on the thermal aging principle is adopted, and the first estimated value of the remaining life is calculated by combining the main factor parameters, the auxiliary factor parameters, and the current equipment state time series data; A long short-term memory network model is adopted, and the second estimated value of the remaining life of the equipment is directly output by taking the main factor parameters, the auxiliary factor parameters, and the current equipment state time series data as inputs; The first estimated value and the second estimated value are arithmetically averaged to obtain the remaining life benchmark value.

5. A preventive test method for a grid-modification transformer according to claim 1, wherein The method for obtaining the life evaluation result is: If the numerical sequence of any main factor parameter in the continuous three and more monitoring sampling periods shows a monotonic increasing trend, the remaining life benchmark value is multiplied by a preset correction coefficient to obtain a corrected life value; The probability distribution range of the remaining life is determined according to the dispersion degree of the historical data of the main factor parameters, the corrected life value is taken as the median of the probability distribution, and the life evaluation result is output in the form of probability distribution.

6. A preventive test method for a grid-modification transformer according to claim 1, wherein The method for obtaining the life risk critical value is: Based on the fault records of the historical same-batch equipment, a mapping relationship between the main factor parameter value and the median of the remaining life is established, and the minimum acceptable life is determined according to the power grid transformation planning period; When the main factor parameter value corresponding to the median of the remaining life equaling the minimum acceptable life is found in the mapping relationship, the value is taken as the life risk critical value of the current equipment.

7. A preventive test method for a grid-modification transformer according to claim 1, wherein After the device warning signal is output, the method further comprises: checking whether the auxiliary factor parameters satisfy a stability condition in the latest preset number of continuous monitoring sampling periods, the stability condition being that the absolute value of the numerical change rate between any adjacent periods does not exceed a preset proportion of the historical mean value, and the difference between the maximum value and the minimum value in the whole period does not exceed a preset proportion of the historical mean value; If the stability condition is not satisfied, an emergency maintenance process is started; if the stability condition is satisfied, the routine monitoring is continued.

8. A power grid retrofit transformer preventive test method as claimed in claim 7, characterized in that, The method for obtaining the transformation priority sequence is: All target transformers are sorted according to the following priority order: first, sorted from high to low according to the first index; for transformers with the same first index, sorted from high to low according to the second index; if the second index is still the same, sorted from high to low according to the third index; if the third index is still the same, sorted from high to low according to the fourth index, thereby generating the transformation priority sequence.

9. A power grid retrofit transformer preventive test method as claimed in claim 6, characterized in that, The transformer transformation start signal is triggered, specifically: The minimum acceptable life is set based on the power grid transformation planning period, a life quantile with a preset high confidence is selected as the transformation trigger critical value in the output remaining life probability distribution; when the median of the remaining life obtained by continuous monitoring is less than or equal to the transformation trigger critical value, the transformer transformation start signal is triggered.

Images