Transformer Lifecycle Management System and Method Based on Big Data Analytics

Abstract

This invention relates to the field of transformer technology, specifically disclosing a transformer lifecycle management system and method based on big data analysis. The method includes the following steps: periodically collecting transformer health parameters within sub-time periods; performing Z-score standardization on the health parameters to obtain target parameters; constructing a covariance matrix for the same type of target parameters, obtaining the statistical energy of the corresponding type of target parameters, and obtaining a fitted curve based on the statistical energy; differentiating the fitted curve, recording the position of the derivative sign change and determining the sub-interval, calculating the curvature of the fitted curve within the sub-interval, and taking the absolute value of the curvature to obtain the target value; calculating the dynamic weights at each time node throughout the transformer's lifecycle based on the target value; and calculating the transformer's health score based on the target parameters at the current time and the dynamic weights. This invention improves the reliability of health score calculation.

Description

Technical Field

[0001] This invention relates to the field of transformer technology, and more specifically to a transformer lifecycle management system and method based on big data analysis. Background Technology

[0002] Condition assessment is an integral part of transformer lifecycle management. During the design and manufacturing phase, simulation and testing are used to control materials and processes, reducing initial risks. In the initial installation and commissioning phase, installation accuracy and initial deterioration are monitored to establish baseline data. During stable operation, online and offline data are integrated for dynamic monitoring to predict faults. During maintenance and overhaul, the effectiveness is verified and strategies are optimized. Before decommissioning, safety and recycling value are assessed.

[0003] A Chinese patent application with publication number CN107992678A discloses a method for evaluating the health status of a transformer based on its entire life cycle. The method mainly includes calculating the health status evaluation score of the transformer based on its initial state parameters, aging parameters, and operating parameters; and generating the health status of the transformer based on the health status evaluation score. When calculating the health status assessment score, the weights of the three parameters in the health status assessment scoring model are fixed, ignoring the different degrees of influence of these three parameters on the health status score at different times. This leads to a distortion of the health status score because: In the early stage of operation, most potential faults come from design and manufacturing defects, and the information contained in the initial state parameters, such as winding geometric deviation and insulation breakdown voltage, has the greatest explanatory power for the health status. In the middle stage, chemical aging and thermo-electric coupling fatigue continue to accumulate, and the fluctuation range and degradation rate of aging parameters (such as the degree of polymerization of insulating paper and the composition of dissolved gases in oil) directly determine the remaining life. If the same low weight is maintained as in the early stage, it will suppress the perception of the true degree of deterioration. When the equipment is close to the end of its design life, the instantaneous extreme values ​​of current-heat dissipation-electric field become the dominant triggering factors for inducing faults. The immediate anomalies of operating parameters (such as hot spot temperature, load inrush current, and partial discharge pulse number) can best reveal the impending fault. However, the fixed weight model cannot amplify these high-frequency signals in time within a short window, resulting in a lack of sensitivity of the score to the impending failure and affecting the decisions made based on the health status score. Summary of the Invention

[0004] The purpose of this invention is to provide a transformer lifecycle management system and method based on big data analysis, thereby solving the aforementioned technical problems.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] The transformer lifecycle management method based on big data analytics includes the following steps:

[0007] The entire life cycle of the transformer is divided into several sub-periods. Within a single sub-period, the transformer's health parameters are periodically collected. The health parameters include initial state parameters, aging parameters, and operating value parameters. The health parameters are then standardized using Z-scores to obtain the target parameters.

[0008] Construct covariance matrices for the same type of target parameters, calculate the trace of the covariance matrices, and use it as the statistical energy of the target parameters of the corresponding type to generate coordinate points (t). i X i ), t i Let X represent the i-th sub-time period. i Let represent the statistical energy corresponding to the i-th sub-time period. Fit the coordinate points to obtain the fitted curve.

[0009] The fitted curve is differentiated, and the position of the derivative sign change is recorded. The time interval between the positions of two adjacent reciprocal sign changes is taken as a sub-interval. The curvature of the fitted curve within the sub-interval is calculated, and the target value is obtained by taking the absolute value of the curvature.

[0010] The dynamic weights at each time point throughout the transformer's entire life cycle are calculated based on the target value, and the transformer's health score is calculated based on the target parameters and dynamic weights at the current time.

[0011] Preferably, calculating the transformer's health score includes:

[0012] Throughout the entire life cycle of the transformer, several time nodes are set at preset time intervals. At a single time node, the curvature corresponding to each target parameter is obtained, and the dynamic weight Yk=Qk / Q' of the k-th target parameter is calculated. Qk and Q' represent the target values ​​of the k-th target parameters and the sum of the target values ​​of all types of target parameters at that time node, respectively.

[0013] Obtain the target parameters and corresponding dynamic weights for the current time, and then perform a weighted summation of the real-time parameters and dynamic weights to obtain the current time transformer health score.

[0014] Preferably, after obtaining the health score, the process further includes:

[0015] A transformer capacity expansion and upgrade plan is generated based on health scores and grid load forecasts.

[0016] Preferably, if the curvature is zero in a certain sub-interval, then a preset value is assigned to all target parameters of that sub-interval.

[0017] Preferably, the method for fitting the coordinate points includes cubic smoothing splines or locally weighted regression.

[0018] Preferably, in the process of calculating the trace of the covariance matrix:

[0019] When the initial state parameters of a transformer are constant or approximately constant in the time dimension, if the trace value of the covariance matrix tends to zero, the lateral variance of the initial state parameters of transformers in the same batch is introduced as a substitute trace to avoid the statistical energy being always zero.

[0020] Preferably, when the health score is less than a preset health score threshold, an early warning message is sent to a preset management personnel.

[0021] A transformer lifecycle management system based on big data analytics, characterized by including:

[0022] Data Acquisition Module: Divides the entire life cycle of the transformer into several sub-periods. Within a single sub-period, the health parameters of the transformer are periodically acquired. The health parameters include initial state parameters, aging parameters, and operating value parameters. The health parameters are then standardized using Z-scores to obtain the target parameters.

[0023] Analysis module: Constructs the covariance matrix of the same type of target parameters, calculates the trace of the covariance matrix as the statistical energy of the target parameters of the corresponding type, and generates coordinate points (t). i X i ), t i Let X represent the i-th sub-time period. i Let represent the statistical energy corresponding to the i-th sub-time period. Fit the coordinate points to obtain the fitted curve.

[0024] Optimization module: Calculate the derivative of the fitted curve, record the position of the derivative sign change, take the time interval between two adjacent reciprocal sign change positions as a sub-interval, calculate the curvature of the fitted curve within the sub-interval, and take the absolute value of the curvature to obtain the target value;

[0025] Scoring module: Calculates the dynamic weights at each time point throughout the transformer's entire life cycle based on the target value, and calculates the transformer's health score based on the target parameters and dynamic weights at the current time.

[0026] The beneficial effects of this invention are:

[0027] 1. This invention obtains the curvature of three types of target parameters at continuous time points throughout the entire life cycle, and transforms the absolute value of curvature into a dynamic weight that changes over time. This highlights the influence of initial state parameters in the early stage, and gradually increases the proportion of aging parameters and operating parameters in the middle and later stages. The health status evaluation score can keep pace with the actual degradation rhythm of the transformer, avoiding the deviation caused by the fixed weight model ignoring time series differences, and thus making the scoring results closer to the actual operating conditions.

[0028] 2. This invention first performs Z-score standardization on the collected data, then uses the trace value of the covariance matrix to represent the statistical energy, and introduces the transverse variance of the same batch as a substitute when the trace value approaches zero. Subsequently, smooth splines or local weighted regression are used to complete the fitting. The numerical continuity is maintained by the absolute value of curvature and zero zone compensation. Thus, a robust data processing chain is formed in terms of dimension unification, noise suppression and division by zero protection, which improves the reliability and comparability of the health status evaluation model. Attached Figure Description

[0029] The invention will now be further described with reference to the accompanying drawings.

[0030] Figure 1 This is a flowchart illustrating the transformer lifecycle management method based on big data analysis proposed in this invention. Detailed Implementation

[0031] 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.

[0032] Please see Figure 1 As shown, this invention is a transformer full life cycle management method based on big data analysis, including the following steps:

[0033] Step 1: Divide the entire life cycle of the transformer into several sub-periods. Periodically collect the health parameters of the transformer within a single sub-period and perform Z-score standardization on the health parameters to obtain the target parameters.

[0034] In one specific embodiment, the entire operation process is continuously divided into multiple interconnected time periods based on the expected service life and operation stage characteristics of the transformer. Each time period is called a sub-time period, which is used to define a complete sampling window.

[0035] Within each sub-period, three types of data are collected sequentially at uniform time intervals: initial state parameters, aging parameters, and operational value parameters. After the three types of raw data are collected, similar data are assembled into column vectors within a sub-period. Then, similar historical samples are extracted from a large database accumulated over multiple years. The overall mean and variance of this type of sample are calculated as a normalization benchmark. The standardized target parameters are generated by subtracting the mean from each raw observation and then dividing by the square root of the variance. This transformation allows measurements of different dimensions to present a zero-center, unit-variance distribution on the same statistical scale, thus laying a unified baseline for subsequent covariance calculations and time-series comparisons.

[0036] Step 2: Construct the covariance matrix of the same type of target parameters, calculate the trace of the covariance matrix as the statistical energy of the target parameters of the corresponding type, and generate coordinate points (t). i X i ), t i Let X represent the i-th sub-time period. i Let represent the statistical energy corresponding to the i-th sub-time period. Fit the coordinate points to obtain the fitted curve.

[0037] In a specific embodiment, taking the target parameter (aging target parameter) corresponding to the aging parameter as an example, the values ​​of the aging target parameter obtained continuously at a fixed sampling interval within a sub-period are arranged in chronological order as a single column vector. Each component in this vector corresponds to an independent observation, so there are natural differences between the components caused by fluctuations in equipment operation.

[0038] Using this single column vector as input, according to the mathematical definition of the covariance matrix, the vector itself is multiplied by itself, and then the expectation of the result is taken. The resulting covariance matrix has only one diagonal and its order is one. Because the matrix elements degenerate into the average of the squares of the deviations of all observations of the vector from its arithmetic mean, the trace of the matrix is ​​the same as the only non-zero element in the matrix, which is essentially equivalent to the variance of the vector.

[0039] This variance is considered as the statistical energy of the aging target parameter within the current sub-period. The start time of the sub-period is then mapped to the horizontal axis, and the variance is mapped to the vertical axis, forming a coordinate pair.

[0040] The coordinate pairs formed by multiple consecutive sub-periods are grouped into the same sequence according to the chronological order. In order to avoid the accidental spikes from causing excessive pull on the overall trend, a smoothing step based on the weighted average of adjacent coordinate pairs is introduced when processing the sequence. The effect of short-term drastic fluctuations is reduced by the temporal proximity, while the overall shape of the statistical energy changing over time is preserved. After this continuous smoothing process, a fitting curve describing the change of the statistical energy of the aging target parameter with the life cycle is obtained.

[0041] It should be noted that methods for fitting coordinate points include cubic smoothing splines or locally weighted regression.

[0042] In a preferred embodiment of the present invention, during the calculation of the trace of the covariance matrix:

[0043] When calculating the trace of the covariance matrix for the initial state parameter sequence, all observations of the parameter for the same transformer within a sub-period are grouped into a single column vector along the time axis. Then, according to the statistical definition, the vector is outwardly producted with itself and the expected value is taken. If the variance is found to approach zero, it means that the parameter is almost unchanged in the time series of this equipment. At this time, directly using the variance as the statistical energy will cause subsequent steps to ignore the initial state information. In order to preserve the quality characteristics brought about by the differences in the manufacturing process, it is necessary to switch the perspective from the vertical time dimension to the horizontal batch dimension.

[0044] The specific approach is as follows: retrieve all equipment belonging to the same production batch as the target transformer from the original production records, extract the same initial state parameters obtained from the corresponding factory inspection process of these equipment, arrange these lateral observations into a set and calculate their variance. This variance comes from the slight differences in raw material purity, winding geometric tolerance, insulation dryness, etc. under the same production conditions, and can reflect the inherent dispersion of the parameter due to manufacturing process fluctuations. Therefore, this lateral variance is used to replace the time variance and written into the trace position of the covariance matrix. Using this replacement value to continue the subsequent statistical energy process can avoid the initial state information being completely diluted due to zero variance, while still maintaining the consistency of the statistical energy calculation caliber with other sub-periods.

[0045] Step 3: Take the derivative of the fitted curve, record the position of the derivative sign change, take the time interval between two adjacent reciprocal sign change positions as a sub-interval, calculate the curvature of the fitted curve within the sub-interval, and take the absolute value of the curvature to obtain the target value.

[0046] In a specific embodiment, after obtaining the fitted curve of the descriptive statistical energy changing with time, the mathematical expression of the curve is first written in the form of time variable expansion, and then the expression is differentiated once. By differentiating, the slope information of the curve at any time can be obtained. A positive slope indicates that the curve is changing upward at that time, while a negative slope indicates that it is changing downward.

[0047] The slope sign is calculated point by point along the time axis by sampling at equal intervals, and the time coordinates of each change from positive to negative or from negative to positive are recorded sequentially. These time coordinates are the zero points of the slope. Several consecutive time periods are defined by pairwise adjacent zero points, and each consecutive time period is defined as a sub-interval.

[0048] Within each subinterval, the fitted curve is differentiated a second time. The second derivative gives the instantaneous rate of change of the curve's curvature. Geometrically, the curvature formula can be used to connect the second derivative with the first derivative to obtain the curvature expression. Then, by performing an integral average or dense sampling average on the curvature expression, the average curvature within that subinterval can be obtained. Curvature measures the strength of the curve's curvature within this interval. To avoid the cancellation of positive and negative values ​​due to sign, the absolute value of the average curvature is taken and recorded as the target value for that subinterval.

[0049] For example, when the fitted curve exhibits an upward convex shape over a certain period of time, the second derivative is positive and the absolute value of curvature is large. If it exhibits a nearly straight shape over another period of time, the second derivative approaches zero and the absolute value of curvature is small. By using such absolute values ​​of curvature, the differences in the rate of change of statistical energy within different sub-intervals can be compared.

[0050] It should be noted that after curvature extraction, if the second derivative of the three types of fitted curves in a certain sub-interval is zero at all sampling points in the interval, it can be determined that the statistical energy of that interval changes linearly or approximately linearly with time, and the calculation result of the absolute value of curvature is all zero.

[0051] Upon detecting this situation, the system first queries a preset non-zero constant. This constant is derived from the lower limit empirical value of the absolute value of curvature in the transformer's historical samples and has the same dimension as curvature.

[0052] Write the constant into the curvature position corresponding to the initial state target parameter, aging target parameter and running value target parameter in the sub-interval, respectively, to form a new set of absolute curvature values; thereby raising the three curvatures from zero to the same minimum positive value, maintaining the relative balance of the curvature set, and avoiding numerical pauses caused by taking the denominator of the dynamic weight to zero in the next step.

[0053] Step 4: Calculate the dynamic weights at each time point throughout the transformer's entire life cycle based on the target value, and calculate the transformer's health score based on the target parameters and dynamic weights at the current time.

[0054] In a preferred embodiment of the present invention, calculating the health score of a transformer includes:

[0055] Based on the planned maintenance cycle or the computing load of the monitoring platform, a fixed time interval is set. A series of equally spaced time nodes are marked along the time axis for the entire life cycle. At each time node, the absolute values ​​of curvature previously calculated in the sub-interval of the node are read from the initial state target parameters, aging target parameters, and operating value target parameters. The three absolute values ​​of curvature are recorded as Q1, Q2, and Q3 respectively, and the sum of the three is recorded as Q'. Then, the dynamic weights Y1, Y2, and Y3 are obtained by dividing each absolute value of curvature by Q' using the proportional relationship.

[0056] In a geometric sense, the absolute value of curvature corresponds to the degree of curvature of the fitted curve. The higher the degree of curvature, the faster the statistical energy of the target parameter changes at this stage, reflecting its greater contribution to the time-varying sensitivity of the health state. Therefore, using the absolute value of curvature as the numerator allows the weights to be adjusted in real time according to the contribution, while introducing the sum of the absolute values ​​of curvature into the denominator can constrain the three weights to maintain their normalization properties within the same comparable proportional framework.

[0057] After the weight calculation is completed, the real-time initial state target parameters, real-time aging target parameters, and real-time running value target parameters that were just stored in the monitoring system are called synchronously. These three real-time target parameters are arranged in the same order as a column vector P, and the three dynamic weights are arranged in the same order as a column vector Y. The corresponding component products of vector P and vector Y are multiplied and summed to obtain the health score at the current time node. This summation step is equivalent to weighting the contributions of the three real-time target parameters within the same statistical scale. Therefore, the current state of the three types of parameters can be condensed into a comprehensive index of a single scale for subsequent decision-making logic to call.

[0058] In another preferred embodiment of the present invention, after obtaining the health score, the method further includes:

[0059] A transformer capacity expansion and upgrade plan is generated based on health scores and grid load forecasts.

[0060] In one specific embodiment, the health scores are arranged into a curve in chronological order. By observing the downward trend of this curve in several future prediction windows, the remaining lifespan of the equipment can be determined. Then, the power grid load prediction curve is mapped to the same time coordinate, and the peak-to-valley difference of the load curve shows the fluctuation range of future operating stress.

[0061] Within each prediction window, the vertical distance between the lowest point of the health score curve and the highest point of the load curve is calculated. This distance is then compared with the safety margin of the transformer's current rated capacity. The smaller the vertical distance, the closer the health margin is to the upcoming load peak, indicating that the rated capacity is no longer sufficient to meet subsequent growth demands.

[0062] When the comparison results trigger the capacity insufficiency judgment, the thermal-electric-mechanical bidirectional expansion coefficient reserved in the equipment production file during the factory design stage is retrieved. These coefficients reflect the upper limit potential of the winding conductor cross-section, core magnetic density and oil channel cooling capacity under the design redundancy. Combined with the space constraints of the site layout and the possible additional modification range of the heat dissipation circuit, the expandable conductor cross-section increase, core magnetic density increase value and oil pump flow rate increase ratio are compiled into a list of alternative capacity expansion schemes.

[0063] To determine the optimal capacity expansion scheme, the peak duration of the load forecast curve is statistically analyzed. A longer peak duration indicates a need for higher heat dissipation stability after the upgrade, while a shorter peak duration allows for a localized enhanced cooling scheme. Based on this, an expansion scheme that meets both thermal stability requirements and can be implemented within the operational space is selected from the list. Construction steps such as winding replacement process, rerouting of insulating oil circulation path, and adding cooler group locations are listed according to the scheme. Finally, a capacity expansion and upgrade implementation plan including technical route, construction period, and raw material configuration is formed.

[0064] A transformer lifecycle management system based on big data analytics includes:

[0065] Data Acquisition Module: Divides the entire life cycle of the transformer into several sub-periods. Within a single sub-period, the health parameters of the transformer are periodically acquired. The health parameters include initial state parameters, aging parameters, and operating value parameters. The health parameters are then standardized using Z-scores to obtain the target parameters.

[0066] Analysis module: Constructs the covariance matrix of the same type of target parameters, calculates the trace of the covariance matrix as the statistical energy of the target parameters of the corresponding type, and generates coordinate points (t). i X i ), t i Let X represent the i-th sub-time period. i Let represent the statistical energy corresponding to the i-th sub-time period. Fit the coordinate points to obtain the fitted curve.

[0067] Optimization module: Calculates the derivative of the fitted curve, records the position of the derivative sign change, takes the time interval between two adjacent reciprocal sign change positions as a sub-interval, calculates the curvature of the fitted curve within the sub-interval, and obtains the target value by taking the absolute value of the curvature.

[0068] Scoring module: Calculates the dynamic weights at each time point throughout the transformer's entire life cycle based on the target value, and calculates the transformer's health score based on the target parameters and dynamic weights at the current time.

[0069] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A transformer lifecycle management method based on big data analysis, characterized in that, Includes the following steps: The entire life cycle of the transformer is divided into several sub-periods. Within a single sub-period, the transformer's health parameters are periodically collected. The health parameters include initial state parameters, aging parameters, and operating value parameters. The health parameters are then standardized using Z-scores to obtain the target parameters. Construct covariance matrices for the same type of target parameters, calculate the trace of the covariance matrices, and use it as the statistical energy of the target parameters of the corresponding type to generate coordinate points (t). i X i ), t i Let X represent the i-th sub-time period. i Let represent the statistical energy corresponding to the i-th sub-time period. Fit the coordinate points to obtain the fitted curve. The fitted curve is differentiated, and the position of the derivative sign change is recorded. The time interval between the positions of two adjacent reciprocal sign changes is taken as a sub-interval. The curvature of the fitted curve within the sub-interval is calculated, and the target value is obtained by taking the absolute value of the curvature. The dynamic weights at each time point throughout the transformer's entire life cycle are calculated based on the target value, and the transformer's health score is calculated based on the target parameters and dynamic weights at the current time. During the process of generating coordinate points: The target parameters within the sub-period are arranged into a single column vector according to the time axis. The single column vector is then subjected to an outer product operation to obtain the outer product result. The expectation of the outer product result is then taken to obtain the covariance matrix. The trace of the covariance matrix is ​​used as the statistical energy. The start time of the sub-period is used as the horizontal axis and the corresponding statistical energy is used as the vertical axis, thus generating the coordinate points. If the trace of the covariance matrix tends to 0, then the equipment from the same production batch as the target transformer is selected, denoted as the target equipment, and the variance of the target parameters of the target equipment is calculated as the statistical energy. In the process of obtaining the fitted curve: The coordinate points are sorted in chronological order along the time axis. A weighted average is taken between two adjacent coordinate points in the sorting, and the overall shape of the statistical energy change over time is preserved.

2. The transformer lifecycle management method based on big data analysis according to claim 1, characterized in that, The calculation of the transformer's health score includes: Throughout the entire life cycle of the transformer, several time nodes are set at preset time intervals. At a single time node, the curvature corresponding to each target parameter is obtained, and the dynamic weight Yk=Qk / Q' of the k-th target parameter is calculated. Qk and Q' represent the target values ​​of the k-th target parameters and the sum of the target values ​​of all types of target parameters at that time node, respectively. Obtain the target parameters and corresponding dynamic weights for the current time, and then perform a weighted summation of the real-time parameters and dynamic weights to obtain the current time transformer health score.

3. The transformer lifecycle management method based on big data analysis according to claim 1, characterized in that, After obtaining a health score, the following is also included: A transformer capacity expansion and upgrade plan is generated based on health scores and grid load forecasts.

4. The transformer lifecycle management method based on big data analysis according to claim 1, characterized in that, If the curvature is zero in a certain sub-interval, then a preset value is assigned to all target parameters in that sub-interval.

5. The transformer lifecycle management method based on big data analysis according to claim 1, characterized in that, Methods for fitting coordinate points include cubic smoothing splines or locally weighted regression.

6. The transformer lifecycle management method based on big data analysis according to claim 1, characterized in that, In the process of calculating the trace of the covariance matrix: When the initial state parameters of a transformer are constant or approximately constant in the time dimension, if the trace value of the covariance matrix tends to zero, the lateral variance of the initial state parameters of transformers in the same batch is introduced as a substitute trace to avoid the statistical energy being always zero.

7. The transformer lifecycle management method based on big data analysis according to claim 1, characterized in that, When the health score is lower than the preset health score threshold, an alert message is sent to the preset management personnel.

8. A transformer lifecycle management system based on big data analytics, characterized in that: include: Data Acquisition Module: Divides the entire life cycle of the transformer into several sub-periods. Within a single sub-period, the health parameters of the transformer are periodically acquired. The health parameters include initial state parameters, aging parameters, and operating value parameters. The health parameters are then standardized using Z-scores to obtain the target parameters. Analysis module: Constructs the covariance matrix of the same type of target parameters, calculates the trace of the covariance matrix as the statistical energy of the target parameters of the corresponding type, and generates coordinate points (t). i X i ), t i Let X represent the i-th sub-time period. i Let represent the statistical energy corresponding to the i-th sub-time period. Fit the coordinate points to obtain the fitted curve. Optimization module: Calculate the derivative of the fitted curve, record the position of the derivative sign change, take the time interval between two adjacent reciprocal sign change positions as a sub-interval, calculate the curvature of the fitted curve within the sub-interval, and take the absolute value of the curvature to obtain the target value; Scoring module: Calculates the dynamic weights at each time point throughout the transformer's entire life cycle based on the target value, and calculates the transformer's health score based on the target parameters and dynamic weights at the current time; During the process of generating coordinate points: The target parameters within the sub-period are arranged into a single column vector according to the time axis. The single column vector is then subjected to an outer product operation to obtain the outer product result. The expectation of the outer product result is then taken to obtain the covariance matrix. The trace of the covariance matrix is ​​used as the statistical energy. The start time of the sub-period is used as the horizontal axis and the corresponding statistical energy is used as the vertical axis, thus generating the coordinate points. If the trace of the covariance matrix tends to 0, then the equipment from the same production batch as the target transformer is selected, denoted as the target equipment, and the variance of the target parameters of the target equipment is calculated as the statistical energy. In the process of obtaining the fitted curve: The coordinate points are sorted in chronological order along the time axis. A weighted average is taken between two adjacent coordinate points in the sorting, and the overall shape of the statistical energy change over time is preserved.

Images