A method for intelligently evaluating health state of transformer dynamically

Abstract

The application relates to the technical field of electric performance detection, and discloses a kind of intelligent transformer health state dynamic evaluation methods, the method comprises the following steps: collecting transformer current time measured operating parameter and measured state parameter;Filtering historical data meeting steady-state conditions and statistics, build transformer own baseline parameter set;Read the nameplate rated parameter, combine current high-voltage side current, ambient temperature and multi-physical field coupling mapping relationship, obtain current working condition physical anchor point parameter set.Physical deviation of measured state parameter relative to physical anchor point, baseline deviation of measured parameter relative to own baseline are calculated respectively, and finally, according to the preset judging rule, the transformer health state evaluation result is obtained by comprehensively considering the two kinds of deviations;The application can improve the efficiency of a kind of intelligent transformer health state dynamic evaluation.

Description

Technical Field

[0001] This invention relates to the field of electrical performance testing technology, and in particular to an intelligent method for dynamic assessment of the health status of transformers. Background Technology

[0002] Currently, transformer health status assessment mainly employs a single reference system method based on threshold discrimination. This method typically involves setting fixed alarm thresholds for each status parameter according to factory design values ​​or recommended values ​​in regulations, comparing real-time monitoring data with these fixed thresholds, and determining the transformer's health status based on whether these limits are exceeded. The core premise of this method is that the preset fixed thresholds can effectively represent the safety boundaries of the equipment under all operating conditions. The fundamental problem with this approach is that it uses a static reference system as the assessment benchmark, failing to consider the dynamic changes in equipment operating conditions and individual differences, thus leading to an inherent deviation between the assessment results and the actual health status.

[0003] In practical applications, the aforementioned static reference system based on fixed thresholds will experience dynamic amplification of the deviation between the fixed threshold and the actual safety boundary of the equipment due to periodic fluctuations in grid load, seasonal changes in ambient temperature, and the continuous evolution of the equipment's aging process. For transformers operating under light or heavy load conditions for extended periods, the fixed threshold may lead to the suppression of actual abnormal changes or frequent false alarms of normal fluctuations, respectively. Furthermore, individual differences in manufacturing processes, operating history, and aging levels among different transformers mean that a uniform fixed threshold cannot accurately characterize the personalized health boundary of each device. This results in a lag in the identification of progressive faults such as early insulation degradation, and may even lead to missed or incorrect assessments. Summary of the Invention

[0004] This invention provides an intelligent dynamic assessment method for the health status of transformers to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an intelligent dynamic assessment method for the health status of transformers, comprising:

[0006] Pt.1. Obtain the measured operating parameter set and measured status parameter set of the transformer to be evaluated at the current moment. The measured operating parameter set includes the high-voltage side current value, the low-voltage side voltage value, and the ambient temperature value. The measured status parameter set includes the top oil temperature value, the winding hot spot temperature value, and the concentration value of the characteristic gas dissolved in the oil.

[0007] Pt.2. Select historical data that meet the steady-state criteria from the historical operation records of the transformer to be evaluated, perform statistical processing on the selected historical data, and generate a set of baseline parameters corresponding to the transformer to be evaluated.

[0008] Pt.3. Read the rated parameter values ​​from the nameplate parameter library of the transformer to be evaluated, and generate the physical anchor point parameter set under the current operating condition based on the current high-voltage side current value and the ambient temperature value, combined with the preset multi-physics field coupling mapping relationship.

[0009] Pt.4. Compare the measured state parameter set with the physical anchor point parameter set to obtain the physical deviation, and compare the measured operating parameter set and the measured state parameter set with the baseline parameter set to obtain the baseline deviation.

[0010] Pt.5. Based on the physical deviation and the baseline deviation, and in accordance with the preset judgment rules, generate the health status assessment result of the transformer to be evaluated.

[0011] Compared with the prior art, the present invention has the following beneficial effects:

[0012] 1. This invention constructs a dual reference system consisting of physical mechanism anchor points and historical statistical baselines. Real-time monitoring data is compared with both the physical anchor point parameter set and the baseline parameter set to obtain physical deviation and baseline deviation. A health status assessment result is generated based on the collaborative determination of these two deviations. This allows the assessment benchmark to simultaneously reflect the theoretical safety boundary of the equipment under given operating conditions and its personalized normal range based on its own operating history, overcoming the inherent defect of a single static threshold reference system in terms of benchmark mismatch under operating condition fluctuations and individual equipment differences. Furthermore, the collaborative determination of the two types of deviations can mutually verify and complement physical anomalies and statistical anomalies, thereby improving the sensitivity of the assessment results to early signs of degradation and its resistance to non-fault fluctuations, fundamentally improving the accuracy and robustness of the condition assessment.

[0013] 2. Based on the years of operation and historical load accumulation, this invention reads the corresponding control limit compensation offset from a preset aging drift mapping table, and superimposes the compensation offset onto the control limits of each parameter under each operating condition group to obtain the aging-compensated baseline control boundary. This allows the baseline to adaptively update along with the aging process of the equipment. Building upon the dual-reference system evaluation framework established in the independent claims, this further enhances the continuous effectiveness of the baseline reference system throughout the entire lifecycle, avoiding misjudgments caused by ignoring aging trend drift. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating an intelligent transformer health status dynamic assessment method according to an embodiment of the present invention.

[0015] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0016] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0017] This application provides an intelligent method for dynamically assessing the health status of transformers. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the intelligent method for dynamically assessing the health status of transformers can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0018] Reference Figure 1 The diagram shown is a flowchart illustrating an intelligent transformer health status dynamic assessment method according to an embodiment of the present invention. In this embodiment, the intelligent transformer health status dynamic assessment method includes:

[0019] Pt.1. Obtain the measured operating parameter set and measured status parameter set of the transformer to be evaluated at the current moment. The measured operating parameter set includes the high-voltage side current value, the low-voltage side voltage value, and the ambient temperature value. The measured status parameter set includes the top oil temperature value, the winding hot spot temperature value, and the concentration value of the characteristic gas dissolved in the oil.

[0020] In this embodiment of the invention, the timing of acquiring measured parameters directly affects the reliability of subsequent evaluation results during transformer operation status monitoring. Due to the thermal inertia of transformer windings, the temperature change of their copper conductors is delayed by a thermal time constant on the order of tens of seconds to several minutes relative to the change in load current, while the thermal time constant of insulating oil can even reach tens of minutes to several hours. If status parameters are acquired during the transition process of drastic load current fluctuations, the top oil temperature and winding hot spot temperature are still dynamically changing, and the measured temperature value at this time lacks a stable correspondence with the current high-voltage side current value. This time-domain mismatch will cause the calculation basis for physical deviations in subsequent steps to become invalid. Therefore, it is necessary to determine an appropriate acquisition timing to ensure the synchronization and representativeness between the measured operating parameter set and the measured status parameter set.

[0021] In one feasible implementation, the measured operating parameter set and the measured state parameter set are synchronously acquired through a sensor set. The high-voltage side current value is measured by a current transformer connected to the high-voltage side bushing lead-out line of the transformer; the low-voltage side voltage value is measured by a voltage transformer connected to the low-voltage side output terminal of the transformer; and the ambient temperature value is measured by a temperature sensor located outside the transformer body, avoiding heat-generating components and direct sunlight. The top oil temperature value is measured by a platinum resistance thermometer or thermocouple thermometer inserted into the top oil layer of the transformer tank; and the winding hot spot temperature value is measured by a fiber Bragg grating temperature sensor pre-embedded between the winding coils or a thermal simulation temperature measuring element placed in the winding hot spot prediction area. The concentration values ​​of characteristic gases are measured by an online dissolved gas monitoring device in the oil. This device extracts dissolved gases from the transformer oil through an oil-gas separation membrane or dynamic headspace degassing method, and then quantitatively analyzes each component gas using gas chromatography or photoacoustic spectroscopy. The characteristic gases include at least one of hydrogen, methane, ethane, ethylene, acetylene, carbon monoxide, and carbon dioxide.

[0022] In another feasible implementation, the high-voltage side current and low-voltage side voltage values ​​can be directly obtained from the digital measurements already collected and synchronized by the intelligent electronic equipment or protection and control devices installed on-site at the transformer. The ambient temperature value can be obtained by reading data from the meteorological monitoring system of the substation where the transformer is located, or by a temperature and humidity transmitter located in a Stevenson screen near the transformer. For the top oil temperature and winding hot spot temperature, on transformers without fiber optic temperature measurement, the winding hot spot temperature value can be calculated by adding a preset hot spot temperature rise coefficient to the measured top oil temperature.

[0023] In another feasible implementation, the sampling frequency can be adjusted according to the monitoring objective. For example, for long-term trend monitoring, the sampling interval is set to five to fifteen minutes; for short-term state assessment, the sampling interval is set to one to five minutes. The measured value at the current moment refers to the value obtained at the most recent sampling moment. If no sampling event occurs at the current moment, the value at the most recent sampling moment is used as the measured value at the current moment.

[0024] Pt.2. Select historical data that meet the steady-state criteria from the historical operation records of the transformer to be evaluated, perform statistical processing on the selected historical data, and generate a set of baseline parameters corresponding to the transformer to be evaluated.

[0025] In this embodiment of the invention, during the actual operation of the transformer, the high-voltage side current value fluctuates continuously with changes in grid load dispatching commands and electricity demand. This fluctuation drives changes in winding copper losses and stray losses of structural components through electromagnetic induction and Joule heating effects, thereby causing dynamic changes in heat source intensity. Since the thermal time constant of transformer windings is typically on the order of seconds to minutes, while the thermal time constant of insulating oil can reach tens of minutes or even hours, when the load current fluctuates frequently, the top oil temperature and winding hot spot temperature are always in the process of catching up with thermal equilibrium, making it impossible to establish a stable correspondence with the current load conditions. If historical data from the non-steady-state transition process are included in the baseline statistics, a large number of time-domain mismatch noise samples will be introduced, making the control limits of the baseline parameter set too wide, thus weakening the sensitivity of anomaly identification during subsequent baseline deviation comparison. Therefore, it is necessary to screen data samples from historical operating records where both the electrical and thermal domains have tended to a steady state.

[0026] In one feasible implementation, the steady-state criteria include load steady-state criteria and temperature steady-state criteria. The load steady-state criteria refer to the fact that the fluctuation range of the high-voltage side current historical sequence within a preset time window does not exceed a preset allowable fluctuation range. The temperature steady-state criteria refer to the fact that the temperature change of the top oil temperature historical sequence within a continuous time window does not exceed a preset allowable temperature gradient range. The length of the preset time window is set based on the transformer's thermal time constant; for oil-immersed power transformers, the preset time window length can be selected as 20 to 60 minutes. The upper limit of the allowable fluctuation range can be determined comprehensively based on the current transformer's range and the transformer's rated current value, for example, taking 1% to 5% of the rated current value. The upper limit of the allowable temperature gradient range can be taken as 0.1 degrees Celsius to 0.5 degrees Celsius per minute. Historical data satisfying the steady-state criteria refers to all historical data corresponding to the time window that simultaneously satisfies both the load and temperature steady-state criteria. Through the above steady-state screening, mismatched samples collected under non-steady-state conditions are eliminated, ensuring that the data samples for subsequent statistical processing have a consistent operating condition basis.

[0027] In another feasible implementation, the steady-state criterion can also be a comprehensive criterion of current fluctuation amplitude and oil temperature change rate. The current fluctuation amplitude is calculated as follows: within a preset time window, the difference between the maximum and minimum high-voltage side current values ​​within the window is taken, and then divided by the arithmetic mean of the high-voltage side current values ​​within the window to obtain the relative fluctuation amplitude. When the relative fluctuation amplitude is less than a preset allowable fluctuation amplitude threshold, the load steady-state criterion is satisfied. The allowable fluctuation amplitude threshold can be any value between 2% and 8%. The oil temperature change rate is calculated as follows: within a continuous time window, the top oil temperature value at the end of the window is subtracted from the top oil temperature value at the beginning of the window, and then divided by the window duration to obtain the average temperature change rate. When the absolute value of the average temperature change rate is less than a preset temperature change rate threshold, the temperature steady-state criterion is satisfied. The temperature change rate threshold can be any value between 0.05 degrees Celsius per minute and 0.3 degrees Celsius per minute.

[0028] In another feasible implementation, the steady-state criterion can also incorporate an auxiliary criterion: the rate of change of winding hot spot temperature. The calculation method for the rate of change of winding hot spot temperature is the same as that for the rate of change of top-layer oil temperature. Only when the load steady-state criterion, the top-layer oil temperature steady-state criterion, and the winding hot spot temperature steady-state criterion are all satisfied simultaneously is the historical data for that period marked as meeting the steady-state criterion. This triple steady-state criterion is suitable for evaluation scenarios with higher data quality requirements.

[0029] Regardless of which steady-state criterion is used, after extracting the selected historical data that meet the steady-state criterion from the historical operation records, this data is statistically processed to generate its own baseline parameter set. The statistical processing method is as follows: for each type of monitored parameter, the arithmetic mean and standard deviation of that parameter in the selected historical data are calculated. The arithmetic mean is used as the center value of the parameter, and the center value plus or minus three times the standard deviation is used as the upper and lower control limits of the parameter. The types of monitored parameters include at least one of the following: top oil temperature, winding hot spot temperature, and concentration of various characteristic gases. The upper and lower control limits together constitute the individualized fluctuation range of the parameter under normal operating conditions.

[0030] In one feasible implementation, the method further includes an aging compensation step for the self-baseline parameter set. The construction method and specific content of the aging drift mapping table are crucial for achieving adaptive aging compensation during the generation of the aging compensation self-baseline parameter set. During long-term operation of a transformer, the degree of polymerization of the insulating paper continuously decreases due to thermal aging, leading to a gradual reduction in the mechanical and dielectric strength of the insulating material. This aging process causes a slow, trend-like drift in the statistical distribution characteristics of various equipment state parameters as the years of operation and cumulative load increase. If the self-baseline parameter set remains fixed after generation, parameter deviations caused by normal aging may be misjudged as abnormal states in the middle and later stages of the equipment's lifespan. Therefore, it is necessary to establish an aging drift mapping table to quantify the compensation amount for this trend-like drift.

[0031] In one feasible implementation, the aging drift mapping table is constructed as follows: Historical operating data of multiple transformers of the same model or voltage level under multiple combinations of service life and cumulative load are pre-collected. For each transformer, its own baseline parameter set is generated according to the method described in claims 1 to 5. This own baseline parameter set is then compared with the initial baseline parameter set established during the initial operation of the transformer, and the control limit offset of each parameter under each operating condition group is calculated. After summarizing the data from the multiple transformers, a mapping relationship is established using the service life and historical cumulative load as input variables, and the control limit compensation offset as the output variable, through multiple linear regression or interpolation fitting methods. The service life is measured in years and is an integer between zero and fifty. The historical cumulative load is measured in per-unit values ​​multiplied by years. It is calculated by multiplying the load rate of the transformer at each moment by the corresponding time element and integrating on the time axis. The load rate is defined as the ratio of the measured high-voltage side current to the rated current. The time element is measured in years. The historical cumulative load is a real number between zero and fifty.

[0032] In one feasible implementation, the granularity of the service life intervals in the aging drift mapping table is as follows: 0 to 5 years, 5 to 10 years, 10 to 15 years, 15 to 20 years, 20 to 30 years, and 30 to 50 years, for a total of six intervals. The granularity of the load accumulation intervals is also as follows: 0 to 5 years, 5 to 10 years, 10 to 15 years, 15 to 20 years, 20 to 30 years, and 30 to 50 years, for a total of six intervals. The service life intervals and load accumulation intervals are paired to form 36 mapping cells, each recording a control limit compensation offset. The control limit compensation offset is a numerical pair, including an upper compensation offset and a lower compensation offset, in the measurement units corresponding to each parameter. For the top oil temperature parameter, the upper and lower compensation offsets range from 0.5 degrees Celsius to 5 degrees Celsius; positive values ​​indicate that aging causes the overall control limit to drift upwards. For the winding hot spot temperature parameter, the upper and lower compensation offset values ​​range from one to eight degrees Celsius. For the characteristic gas concentration parameter, the upper and lower compensation offset values ​​range from one to five percent of the full-scale concentration measurement.

[0033] In another feasible implementation, the granularity of the commissioning period interval and the load accumulation interval can be finer or coarser, depending on the richness of the sample data. When the sample data is sufficient, the commissioning period interval can be divided into intervals of one to two years, and the load accumulation interval can also be divided into intervals of one to two years, to provide a finer compensation granularity. When the sample data is limited, the commissioning period interval can be divided into intervals of ten years, and the load accumulation interval can be divided into intervals of ten, forming a total of twenty-five mapping grids.

[0034] In another feasible implementation, the aging drift mapping table can be constructed using a calculation method based on a physical aging model instead of a data-based statistical method. Specifically, based on the Arrhenius equation and historical data of the transformer's actual operating temperature, the decrease curve of the degree of polymerization of the insulating paper over time is calculated. Based on the relationship between the decrease in the degree of polymerization and the mechanical and dielectric strength of the insulating paper, the theoretical drift of the control limits of each state parameter is calculated. The theoretical drift is then organized into a mapping table according to two dimensions: years of operation and cumulative load. This physical model-based construction method is suitable for scenarios where historical data from multiple transformers of the same model is lacking.

[0035] In another feasible implementation, the control limit compensation offset in the aging drift mapping table can be further differentiated into varying compensation amounts for different types of parameters. For thermal characteristic parameters, namely top oil temperature and winding hot spot temperature, the control limit compensation offset is only related to the years of operation and has a weak correlation with the accumulated load. Therefore, in the compensation offset for this type of parameter, the weight of the years of operation is set to 70% to 90%, and the weight of the accumulated load is set to 10% to 30%. For characteristic gas concentration parameters, the control limit compensation offset has a strong correlation with the accumulated load. Therefore, in the compensation offset for this type of parameter, the weight of the accumulated load is set to 60% to 80%, and the weight of the years of operation is set to 20% to 40%. Through these differentiated weight settings, the compensation offset for each type of parameter more accurately reflects its respective aging drift pattern.

[0036] With the specific construction method and content supplement of the above-mentioned aging drift mapping table, those skilled in the art can establish the mapping table and calculate the aging compensation control limit based on the description in the specification and in combination with conventional testing and statistical methods in the field.

[0037] In one feasible implementation, when the amount of historical data selected is insufficient, such as when the number of selected data samples is less than thirty, statistical processing can use a moving range-based method instead of the standard deviation method. Specifically, the selected historical data are arranged chronologically by collection time, and the absolute difference between any two adjacent data points is calculated sequentially. The arithmetic mean of all absolute differences is taken as the moving average range. The arithmetic mean of all data is used as the center value, and the center value plus or minus 2.66 times the moving average range is used as the upper and lower control limits. This alternative scheme is suitable for newly commissioned transformers with limited historical data.

[0038] Pt.3. Read the rated parameter values ​​from the nameplate parameter library of the transformer to be evaluated, and generate the physical anchor point parameter set under the current operating condition based on the current high-voltage side current value and the ambient temperature value, combined with the preset multi-physics field coupling mapping relationship.

[0039] In transformer operating condition assessment, the measured state parameter set reflects the actual physical response of the equipment under current operating conditions. However, this response is mixed with variations caused by changes in normal operating conditions and variations caused by the degradation of the equipment's own health. Specifically, the top oil temperature and winding hot spot temperature depend not only on the thermal conductivity and heat dissipation conditions of the insulation medium, but also on the combined constraints of the copper loss heat source intensity generated by the current load current and the heat dissipation boundary conditions determined by the ambient temperature. The gas generation rate of characteristic gases is controlled by the thermal aging degradation rate of the insulation paper and influenced by chemical environmental factors such as dissolved oxygen concentration and moisture content in the oil. Without establishing a theoretical reference benchmark that considers the coupling relationship of multiple physical fields such as electricity, heat, flow, and aging, it is impossible to isolate the influence component caused by changes in operating conditions from the measured state parameters, resulting in a lack of physical reference for subsequent health status judgments. Therefore, it is necessary to deduce the theoretical values ​​of each state parameter under the current operating conditions based on the transformer's design parameters and current operating conditions through multi-physical field coupling mapping relationships.

[0040] In some embodiments, the nameplate parameter library is stored in the transformer's intelligent electronic device or station control layer data server. Rated parameter values ​​include rated capacity, rated voltage, rated current, short-circuit impedance, no-load loss, and load loss. Specifically, the rated capacity and rated voltage are used to determine the transformer's rated apparent power and voltage level; the rated current is the base current calculated based on the rated capacity and rated voltage; the short-circuit impedance is used to calculate the short-circuit impedance voltage drop and additional losses under load current; and the no-load loss and load loss are used to determine the transformer's total heat loss sources. After the current high-voltage side current value is determined, the load loss value is calculated according to a relationship proportional to the square of the high-voltage side current value, i.e., the current load loss equals the rated load loss multiplied by the square of the ratio of the current current to the rated current. The no-load loss value is taken as a constant loss value under rated voltage. The ambient temperature value serves as an input parameter for the heat dissipation boundary conditions, used to determine the base temperature of the cooling medium.

[0041] In some embodiments, the multiphysics coupling mapping relationship is implemented through the following cascaded inference chain. The first stage is the electrothermal correlation mapping, whose inputs are the high-voltage side current value, ambient temperature value, and rated parameter value, and the output is the theoretical value of the top oil temperature. The electrothermal correlation mapping is established based on the thermal circuit model. In the heat transfer path from the total heat loss of the transformer to the surrounding environment through the tank wall and radiator, the difference between the top oil temperature and the ambient temperature is equal to the total heat loss source divided by the equivalent thermal conductivity. The total heat loss source is equal to the sum of the no-load loss value and the current load loss value. The reference value of the equivalent thermal conductivity is taken as the quotient obtained by dividing the rated load loss value by the rated temperature rise value. The rated temperature rise value is the steady-state temperature rise of the top oil temperature of the transformer under rated load conditions. For oil-immersed power transformers, the rated temperature rise value is set to 45 degrees Celsius to 55 degrees Celsius. For oil-immersed self-cooled transformers, the equivalent thermal conductivity is taken as the above reference value. For oil-immersed air-cooled transformers and forced oil circulation air-cooled transformers, the equivalent thermal conductivity varies with the operating conditions of the fan and oil pump. Correspondingly, correction factors are introduced for different cooling equipment operation combinations. The correction factor is determined as follows: During the initial operation of the transformer, a load step test is conducted under each cooling equipment operation combination. The steady-state rise in top oil temperature is recorded. The quotient obtained by dividing the rated load loss value by the difference between this steady-state rise and the current ambient temperature is taken as the actual equivalent thermal conductivity under that cooling combination. Then, this actual equivalent thermal conductivity is divided by a reference value to obtain the correction factor. The correction factor ranges from 0.8 to 1.5. The theoretical value of the top oil temperature obtained in this way reflects the steady-state theoretical level of the transformer's top oil temperature under the current load and ambient temperature.

[0042] In other embodiments, the electrothermal correlation mapping can also employ a calculation method based on loss separation and thermal resistance networks. Specifically, the total loss is divided into two components: no-load loss and load loss. The no-load loss is applied as a constant heat source to the core region, while the load loss is distributed to each heat-generating region according to the ratio of winding copper loss and stray losses from structural components. The thermal resistance network consists of the thermal resistance from the winding to the top oil, the thermal resistance from the top oil to ambient air, and the thermal resistance from the core to the top oil. The values ​​of each thermal resistance are read from a preset typical value database based on the transformer's structural dimensions and cooling method, or obtained through parameter identification based on temperature rise test data during initial commissioning. The theoretical value of the top oil temperature equals the sum of the ambient temperature and the temperature rise generated by the total loss heat source at each thermal resistance. This alternative approach is suitable for evaluation scenarios where detailed requirements for the temperature distribution in the winding region are not critical.

[0043] In some embodiments, the electrothermal correlation mapping can also employ a dynamic thermal circuit model based on the thermal time constant. This dynamic thermal circuit model further introduces thermal capacity elements into the thermal resistance network, forming an RC network. The winding thermal capacity is taken as the product of the copper conductor mass and the specific heat capacity of copper, while the insulating oil thermal capacity is taken as the product of the transformer oil mass and the oil's specific heat capacity. The differential equations of the RC network are numerically solved using the fourth-order Runge-Kutta method, with a time step set to one to ten seconds. The dynamic thermal circuit model can output not only the theoretical value of the top oil temperature under steady-state conditions but also the trajectory of the top oil temperature change during transient processes. This alternative is suitable for evaluation scenarios with frequent load changes and the need to track thermal dynamics.

[0044] The second stage is the heat flow correlation mapping, which takes the theoretical value of the top oil temperature as input and outputs the theoretical values ​​of the winding axial temperature distribution and oil flow velocity. The heat flow correlation mapping is based on the axial heat conduction and oil convection heat transfer equations of the winding region. Along the axial height of the winding, the winding is discretized into multiple coil units. The number of coil units can be selected between ten and fifty, depending on the structural complexity of the transformer winding. For each coil unit, the copper loss heat generation is equal to the product of the conductor resistance within that coil unit and the square of the high-voltage side current. The conductor resistance increases with temperature, and its temperature coefficient is taken as 0.00393 per degree Celsius. Part of the heat generated by this coil unit is transferred axially to adjacent coil units through heat conduction, and the other part is carried away by convection heat into the oil flow. The axial heat transfer term equals the thermal conductivity multiplied by the temperature difference between adjacent coil units, then divided by the axial spacing between the coils. The thermal conductivity is taken as the weighted average of the thermal conductivity of copper and the thermal conductivity of the insulating paper, where the thermal conductivity of copper is taken as 380 watts per Kelvin per meter, and the thermal conductivity of the insulating paper is taken as 0.2 watts per Kelvin per meter. The convective heat transfer term equals the oil heat transfer coefficient multiplied by the surface heat transfer area of ​​the coil unit, then multiplied by the difference between the temperature of that coil unit and the local oil temperature. The oil heat transfer coefficient is taken as 200 watts per Kelvin per square meter to 600 watts per Kelvin per square meter, with the specific value determined based on the oil flow rate. The oil flow rate is determined by the balance between the driving head of the oil pump and the oil circuit resistance. The driving head is taken as the rated head of the oil pump, and the oil circuit resistance is proportional to the square of the oil flow rate, with the proportionality coefficient determined based on the oil tank structure and radiator piping parameters. The aforementioned heat balance equations are written for each coil unit, forming a system of linear equations containing the same number of unknowns as the number of coil units, namely the temperature of each coil. This system of equations is solved using Gaussian elimination or the pursuit method to obtain the temperature values ​​of each coil unit. The distribution of these temperature values ​​along the axial direction constitutes the theoretical value of the winding's axial temperature distribution. The theoretical value of the oil flow velocity is obtained by back-calculation using the convective heat transfer coefficient. When the oil flow heat transfer coefficient is 300 watts per square meter per Kelvin, the corresponding theoretical oil flow velocity is 0.1 m / s to 0.5 m / s. Through the aforementioned heat flow correlation mapping, the theoretical value of the top layer oil temperature is used as a boundary condition and passed downwards to deduce the temperature distribution and oil flow velocity at various positions along the winding's axial direction.

[0045] In other embodiments, the heat-fluid correlation mapping can also be implemented using a reduced-order model from computational fluid dynamics simulation. Specifically, during the transformer design phase, based on the transformer's accurate geometric model and material properties, the Navier-Stokes equations and energy equations are solved using the finite volume method to obtain the winding region temperature field and oil flow field distribution under different combinations of top oil temperature and oil pump speed. The simulation results are then constructed into a polynomial response surface model, with top oil temperature and oil pump speed as inputs and winding axial temperature distribution and oil flow velocity as outputs. The response surface model uses a quadratic polynomial form, and its coefficients are obtained through least-squares fitting. During actual operation, the theoretical values ​​of the top oil temperature and the current oil pump speed are input into the response surface model to directly calculate the theoretical values ​​of the winding axial temperature distribution and oil flow velocity. The computation time of the response surface model is much shorter than that of a complete three-dimensional simulation, making it suitable for real-time evaluation scenarios.

[0046] In some embodiments, the heat flow correlation mapping can also employ a simplified estimation method based on winding temperature distribution characteristic parameters. Specifically, a reference distribution curve is selected from a pre-defined library of typical temperature distribution curves according to the transformer's cooling method. The reference distribution curve uses the relative axial position of the winding as the abscissa and the relative temperature rise as the ordinate. The relative temperature rise is defined as the difference between the temperature at each location of the winding and the top oil temperature. The theoretical value of the top oil temperature is superimposed onto the reference distribution curve to obtain the theoretical value of the winding axial temperature distribution. The reference distribution curve library stores several typical curves corresponding to different cooling methods and different load ranges. Each curve is defined by multiple discrete points, and the discrete points are interpolated using cubic spline interpolation to form a continuous curve. This simplified estimation method is suitable for rapid evaluation scenarios that do not require precise temperature distribution details.

[0047] The third level is the thermal aging correlation mapping, whose input is the theoretical value of the winding axial temperature distribution, and output is the theoretical value of the rate of decrease in the degree of polymerization of the insulating paper. The thermal aging correlation mapping is based on the Arrhenius equation, which states that there is a linear relationship between the logarithm of the rate of decrease in the degree of polymerization of the insulating paper and the reciprocal of the thermodynamic temperature. For each axial temperature value in the theoretical value of the winding axial temperature distribution, the theoretical rate of decrease in the degree of polymerization of the insulating paper at that location is calculated according to the Arrhenius equation. In this equation, the pre-exponential factor for the thermal aging response of the insulating paper in oil-immersed transformers ranges from 2.5% x 10^12 to 3.5% x 10^12 per day, the activation energy for the thermal aging degradation response of the insulating paper ranges from 100 kilojoules per mole to 120 kilojoules per mole, and the ideal gas constant is 8.314 joules per mole per Kelvin. The thermodynamic temperature value is equal to the sum of the Celsius temperature value at that location and 273.15.

[0048] In other embodiments, when precise activation energy and pre-exponential factor data for the insulating paper are unavailable, a simplified aging rate formula based on the IEEE standard can be used for alternative calculations. In the IEEE standard, the relationship between aging rate and hotspot temperature is expressed exponentially, with the temperature constant for oil-immersed transformer insulating paper taken as 15,000 Kelvin, and the reference temperature value taken as 383 Kelvin (110 degrees Celsius). Multiplying the relative aging rate by the baseline aging rate yields the theoretical value of the rate of decrease in the degree of polymerization of the insulating paper. The baseline aging rate is taken as the rate of decrease in the degree of polymerization of the insulating paper at the reference temperature of 110 degrees Celsius. For ordinary insulating paper, the baseline aging rate is taken as one ten-thousandth to one thousandth of the degree of polymerization decrease per second. Converted to a percentage per day, the baseline aging rate is taken as 0.05% to 0.15% per day.

[0049] In some embodiments, the thermal aging correlation mapping can also consider the accelerated aging effects of moisture and oxygen. The moisture acceleration factor is calculated as an exponential function with the natural constant e as the base and the product of the moisture acceleration coefficient and the moisture content in the insulating paper as the exponent. The moisture acceleration coefficient ranges from 0.5% to 1.0% per percentage of moisture content, and the moisture content in the insulating paper, expressed as a mass percentage, ranges from 0.5% to 2.0%. The oxygen acceleration factor is calculated as an exponential function with the natural constant e as the base and the product of the oxygen acceleration coefficient and the dissolved oxygen concentration in the oil as the exponent. The oxygen acceleration coefficient ranges from 0.1% to 0.3% per microliter, and the dissolved oxygen concentration in the oil ranges from 5,000 to 30,000 microliters per liter. Multiplying the moisture acceleration factor and oxygen acceleration factor by the theoretical value of the rate of decrease in the degree of polymerization of the insulating paper calculated using the Arrhenius equation yields the theoretical value of the rate of decrease in the degree of polymerization of the insulating paper after correction for moisture and oxygen. In this alternative, the moisture content in the insulating paper is calculated by converting the trace water concentration in the oil into the moisture balance curve of the insulating paper, and the dissolved oxygen concentration in the oil is read by an online dissolved gas monitoring device.

[0050] The theoretical values ​​of top-layer oil temperature, winding axial temperature distribution, oil flow velocity, and insulation paper polymerization degree decrease rate are summarized and used as the physical anchor point parameter set under the current operating condition. In this embodiment, through a three-level cascaded reasoning of electrothermal correlation mapping, heat flow correlation mapping, and thermal aging correlation mapping, the high-voltage side current value and ambient temperature value are gradually transformed into a complete physical anchor point parameter set including top-layer oil temperature, winding axial temperature distribution, oil flow velocity, and insulation paper polymerization degree decrease rate. Regardless of which electrothermal correlation mapping method is used, the obtained theoretical value of top-layer oil temperature is used as the input of heat flow correlation mapping. Regardless of which heat flow correlation mapping method is used, the obtained theoretical value of winding axial temperature distribution is used as the input of thermal aging correlation mapping. Through the above multi-physics coupling mapping, a theoretical reference benchmark with clear physical mechanism support is provided for the subsequent calculation of physical deviations.

[0051] Pt.4. Compare the measured state parameter set with the physical anchor point parameter set to obtain the physical deviation, and compare the measured operating parameter set and the measured state parameter set with the baseline parameter set to obtain the baseline deviation.

[0052] In this embodiment of the invention, the difference between the measured state parameter set and the theoretical value in transformer health status assessment may originate from two different types of causes. The first cause is changes in operating conditions, such as seasonal fluctuations in ambient temperature, daily cycle changes in load current, and switching of cooling equipment operation status. These factors cause a systematic deviation of the state parameters within the normal range. The second cause is the degradation of the equipment's own health status, such as thermal aging of insulation paper leading to increased dielectric loss, winding deformation leading to local overheating, and abnormal growth of dissolved characteristic gases in the oil. These factors cause the state parameters to deviate from the theoretically expected values. If these two types of deviations are judged together, it is impossible to distinguish between normal operating condition fluctuations and abnormal state degradation, resulting in insufficient sensitivity of the assessment results in identifying early fault symptoms. Therefore, it is necessary to independently compare the measured state parameter set with the physical anchor point parameter set and its own baseline parameter set to obtain deviation information in two different dimensions: physical deviation and baseline deviation, providing differentiated decision-making basis for subsequent comprehensive judgment.

[0053] In one feasible implementation, the specific method for comparing the measured state parameter set with the physical anchor point parameter set to obtain the physical deviation is as follows: Extract the measured top-layer oil temperature from the measured state parameter set, extract the theoretical top-layer oil temperature from the physical anchor point parameter set, and subtract the theoretical top-layer oil temperature from the measured top-layer oil temperature to obtain the oil temperature deviation. Extract the measured winding hot spot temperature from the measured state parameter set, extract the theoretical winding axial temperature distribution from the physical anchor point parameter set, determine the axial position corresponding to the measured winding hot spot temperature, extract the theoretical temperature value corresponding to that axial position from the theoretical winding axial temperature distribution, and subtract the theoretical temperature value from the measured winding hot spot temperature to obtain the hot spot deviation. Extract the measured characteristic gas concentration from the measured state parameter set, and simultaneously obtain the characteristic gas concentration value at the previous sampling time. Subtract the previous concentration value from the current measured concentration value and divide by the time interval between the two samplings to obtain the measured gas generation rate. Extract the theoretical value of the rate of decrease in the degree of polymerization of the insulating paper from the physical anchor point parameter set, and subtract the theoretical value of the rate of decrease in the degree of polymerization of the insulating paper from the measured gas generation rate to obtain the gas generation deviation. The oil temperature deviation, hot spot deviation, and gas production deviation are summarized as the physical deviation. Through the above comparison, the physical deviation reflects the degree to which the measured physical response deviates from the theoretical design behavior, and this degree of deviation is related to the degradation of the equipment's health status at the physical mechanism level.

[0054] In another feasible implementation, the time interval for calculating the measured gas production rate can be adaptively adjusted based on the rate of change of the characteristic gas concentration. When the characteristic gas concentration changes drastically, the time interval is shortened to four to twelve hours. When the characteristic gas concentration changes gradually, the time interval is extended to twenty-four to seventy-two hours. The unit of the measured gas production rate is microliters per liter per day. The unit of the theoretical rate of decrease in the degree of polymerization of the insulating paper is percentage per day. When the measured gas production rate is greater than the theoretical rate of decrease in the degree of polymerization of the insulating paper, the gas production deviation is positive, indicating a possible abnormal overheating or discharge fault. When the measured gas production rate is less than or equal to the theoretical rate of decrease in the degree of polymerization of the insulating paper, the gas production deviation is negative or zero, indicating that the gas production during thermal aging is within the theoretically expected range.

[0055] In another feasible implementation, the specific method for comparing the measured operating parameter set and the measured state parameter set with its own baseline parameter set to obtain the baseline deviation is as follows: Extract the measured high-voltage side current and ambient temperature from the measured operating parameter set; extract the measured top oil temperature and winding hot spot temperature from the measured state parameter set. Determine the current seasonal attribute and current operating mode identifier; search the own baseline parameter set for the operating condition group that matches the seasonal attribute and operating mode identifier; read the upper and lower control limits of each parameter under this operating condition group, including the upper and lower control limits of the high-voltage side current, the top oil temperature, and the winding hot spot temperature. Compare the measured high-voltage side current value with the upper and lower control limits of this parameter. When the measured high-voltage side current value falls between the upper and lower control limits, a normal current indicator is generated. When the measured high-voltage side current value exceeds the upper control limit, a current over-limit indicator is generated. When the measured value of the high-voltage side current is less than the control lower limit, a current limit exceedance indicator is generated. The same interval comparison operation is performed on the measured values ​​of the top-layer oil temperature and the winding hot spot temperature, generating either a normal oil temperature indicator or an oil temperature limit exceedance indicator and the direction of the exceedance, as well as a normal hot spot indicator or a hot spot limit exceedance indicator and the direction of the exceedance. The normal or exceedance indicators and the direction of the exceedance for each of the above parameters are summarized as the baseline deviation. Through the above comparison, the baseline deviation reflects the degree to which the measured value deviates from its historical normal fluctuation range, and this degree of deviation is statistically correlated with the change in the equipment's operating state relative to its historical behavior.

[0056] In another feasible implementation, the specific values ​​of the upper and lower control limits for the high-voltage side current, the top oil temperature, and the winding hot spot temperature vary depending on the operating condition group. For the summer high-temperature operating condition group, the upper control limit for the top oil temperature is set to 90 to 95 degrees Celsius. For the winter low-temperature operating condition group, the upper control limit for the top oil temperature is set to 75 to 80 degrees Celsius. For the heavy-load operating condition group, the upper control limit for the winding hot spot temperature is set to 110 to 120 degrees Celsius. For the light-load operating condition group, the upper control limit for the winding hot spot temperature is set to 85 to 90 degrees Celsius. These specific differences in control limits reflect the different normal fluctuation ranges of the transformer's thermal state under different operating conditions. If a uniform fixed threshold is used instead of the above-mentioned differentiated control limits, it may lead to false alarms at high ambient temperatures or missed alarms at low ambient temperatures. By binding control limits to operating condition groups, the baseline deviation interval comparison benchmark adapts to the operating conditions, thus avoiding misjudgment problems under the fixed threshold method.

[0057] Pt.5. Based on the physical deviation and the baseline deviation, and in accordance with the preset judgment rules, generate the health status assessment result of the transformer to be evaluated.

[0058] In this embodiment of the invention, after obtaining the physical deviation and baseline deviation, it is necessary to integrate these two types of deviation information, which have different physical meanings and dimensions, into a unified health status conclusion. The physical deviation reflects the degree to which the actual physical response of the equipment deviates from its theoretical design behavior. Its value has a specific physical unit; for example, the unit for oil temperature deviation is degrees Celsius, and the unit for gas production deviation is microliters per liter per day. The baseline deviation reflects the degree to which the measured value deviates from its historical normal fluctuation range. It is expressed as a normal or over-limit indicator for each parameter and the direction of the over-limit, belonging to qualitative or semi-quantitative deviation information. Because the two types of deviations have different properties and expressions, they cannot be directly algebraically calculated or simply compared. Furthermore, the importance of physical deviation and baseline deviation in indicating the health status of the equipment varies under different load conditions. For example, under heavy load conditions, thermal characteristic-related physical deviations have a more significant impact on equipment safety, while under light load conditions, baseline over-limits in gas production characteristics may better reflect insulation abnormalities. Therefore, it is necessary to establish pre-defined judgment rules, convert the physical deviation and baseline deviation into quantifiable deviation indices, and then combine them according to their respective weights to finally map them into a health status level.

[0059] In one feasible implementation, the preset judgment rules first obtain the current value of the high-voltage side of the transformer to be evaluated at the current moment, and determine the current load range based on the high-voltage side current value. The load range is divided according to the percentage relationship between the high-voltage side current value and the rated current value. When the high-voltage side current value is less than 60% of the rated current value, it is judged as a light load range. When the high-voltage side current value is greater than or equal to 60% and less than or equal to 90% of the rated current value, it is judged as a medium load range. When the high-voltage side current value is greater than 90% of the rated current value, it is judged as a heavy load range. The physical deviation weight value and baseline deviation weight value corresponding to the current load range are read from the preset weight allocation table. The physical deviation weight value includes oil temperature deviation weight, hot spot deviation weight, and gas generation deviation weight, and the sum of the three is the first total. The baseline deviation weight value includes current over-limit weight, oil temperature over-limit weight, and hot spot over-limit weight, and the sum of the three is the second total. The first total and the second total can be the same or different. For example, in one weight allocation method, the first total is greater than the second total in the heavy load range to highlight the importance of the thermal characteristic physical deviation. In another weighting method, the second sum is greater than the first sum in the light load interval to highlight the sensitivity of baseline statistical exceedances. Through the above-mentioned adaptive weighting of load intervals, the contribution ratio of physical deviation and baseline deviation to the final evaluation result under different operating conditions is reasonably adjusted.

[0060] In one feasible implementation, the physical deviation index is calculated as follows: The oil temperature deviation, hot spot deviation, and gas production deviation values ​​in the physical deviation quantities are multiplied by their respective weights (oil temperature deviation weight, hot spot deviation weight, and gas production deviation weight), and then summed to obtain the physical deviation index. The physical deviation index is a dimensionless quantity, and its value reflects the comprehensive degree to which the measured physical response of the equipment deviates from the theoretical design behavior. The specific values ​​of the oil temperature deviation weight, hot spot deviation weight, and gas production deviation weight are determined according to the load range. For example, in the heavy load range, the oil temperature deviation weight is set to 3.0 to 5.0, the hot spot deviation weight to 4.0 to 6.0, and the gas production deviation weight to 1.0 to 3.0. In the medium load range, the oil temperature deviation weight is set to 2.0 to 4.0, the hot spot deviation weight to 2.0 to 4.0, and the gas production deviation weight to 2.0 to 4.0. Under light load conditions, the weighting for oil temperature deviation is set to 1.0 to 2.0, the weighting for hot spot deviation is set to 1.0 to 2.0, and the weighting for gas production deviation is set to 4.0 to 6.0. The absolute values ​​of oil temperature deviation and hot spot deviation are taken before being substituted into the calculation. The gas production deviation value is directly substituted according to its sign; a positive value indicates that the gas production rate is higher than the theoretical expectation, and a negative value indicates that the gas production rate is lower than the theoretical expectation.

[0061] In other embodiments, since the units for oil temperature deviation, hot spot deviation, and gas production deviation are all in degrees Celsius, while the baseline deviation index is dimensionless, to ensure that the physical deviation index and the baseline deviation index have consistent dimensionless dimensions before weighted summation, the physical deviation index uses a dimensionless normalization calculation method instead of the original unit-based summation method. For oil temperature deviation, the control limit width obtained by subtracting the control lower limit from the upper control limit of the top oil temperature under this operating condition group is used as the normalization benchmark. The absolute value of the oil temperature deviation is divided by this control limit width to obtain the oil temperature deviation degree. For hot spot deviation, the control limit width obtained by subtracting the control lower limit from the upper control limit of the winding hot spot temperature under this operating condition group is used as the normalization benchmark. The absolute value of the hot spot deviation is divided by this control limit width to obtain the hot spot deviation degree. For the gas production deviation value, the gas production rate reference value obtained by dividing the difference between the upper and lower control limits of the characteristic gas concentration under this operating condition group by the sampling time interval is used as the normalization benchmark. The gas production deviation value is then divided by this gas production rate reference value to obtain the gas production deviation degree. A positive value retains a positive sign, and a negative value retains a negative sign to reflect the direction of the gas production rate deviation. The three deviation degrees are multiplied by their corresponding oil temperature deviation weight, hot spot deviation weight, and gas production deviation weight, respectively, and then summed to obtain the physical deviation index. All three deviation degrees are dimensionless, and the resulting physical deviation index is also dimensionless. The values ​​of each weight are determined based on the load range, and their specific numerical ranges are the same as those of the corresponding weights mentioned earlier.

[0062] In some embodiments, the physical deviation index can also use the threshold percentage method instead of the control limit width normalization method. The oil temperature deviation value is obtained by dividing the oil temperature deviation value by a preset maximum permissible oil temperature deviation value, which is the difference between the upper and lower control limits of the oil temperature, divided by two, and then multiplied by a redundancy coefficient, with a redundancy coefficient ranging from 1.2 to 1.5. The hot spot deviation value is obtained by dividing the hot spot deviation value by a preset maximum permissible hot spot deviation value, and its setting method is the same as that for the maximum permissible oil temperature deviation value. The gas production deviation value is obtained by dividing the gas production deviation value by a preset maximum permissible gas production deviation value, which is the theoretical expected value of the gas production rate multiplied by a safety factor, with a safety factor ranging from 2 to 5.

[0063] In some other embodiments, when the gas production deviation value is negative (i.e., the measured gas production rate is lower than the theoretical rate of decrease in the degree of polymerization of the insulating paper), the gas production deviation value is taken as zero rather than negative, to avoid the negative value offsetting the contribution of other positive deviations in the weighted summation. This alternative approach is suitable for evaluation logics that only focus on gas production deviations in the abnormally rising direction, and do not deduct points for gas production rates lower than theoretical expectations.

[0064] In one feasible implementation, the baseline deviation index is calculated as follows: The current over-limit indicator and its direction, the oil temperature over-limit indicator and its direction, and the hot spot over-limit indicator and its direction in the baseline deviation are each converted into their corresponding over-limit deviation values. The conversion rules for the over-limit deviation values ​​are as follows: When a parameter is marked as normal, the corresponding over-limit deviation value is zero. When a parameter is marked as upper over-limit, the corresponding over-limit deviation value is the quotient obtained by subtracting the upper control limit of the parameter from its measured value and then dividing by the difference between the upper and lower control limits. When a parameter is marked as lower over-limit, the corresponding over-limit deviation value is the quotient obtained by subtracting the measured value of the parameter from its lower control limit and then dividing by the difference between the upper and lower control limits. The over-limit deviation value is a dimensionless positive number, and its magnitude reflects the relative degree to which the measured value deviates from the control limit. The baseline deviation index is obtained by multiplying the over-limit deviation values ​​of each parameter by their corresponding current over-limit weight, oil temperature over-limit weight, and hot spot over-limit weight, and then summing the results. The specific values ​​for the current over-limit weight, oil temperature over-limit weight, and hot spot over-limit weight are determined based on the load range. For example, in the heavy load range, the current over-limit weight is set to 1.0 to 2.0, the oil temperature over-limit weight to 3.0 to 5.0, and the hot spot over-limit weight to 4.0 to 6.0. In the light load range, the current over-limit weight is set to 3.0 to 5.0, the oil temperature over-limit weight to 1.0 to 2.0, and the hot spot over-limit weight to 1.0 to 2.0. Each over-limit weight is coordinated with the aforementioned physical deviation weight within the same load range, ensuring that the physical deviation index and the baseline deviation index are comparable on a numerical scale.

[0065] In another feasible implementation, the conversion of out-of-limit deviation values ​​can also employ methods other than linear mapping. For example, when a measured value exceeds the limit, the out-of-limit deviation value is taken as the ratio of the absolute value of the measured value's deviation from the control limit to the width of the control limit, multiplied by a preset out-of-limit amplification factor to amplify the contribution of the out-of-limit degree to the baseline deviation index. The out-of-limit amplification factor ranges from 1.5 to 3.0. This alternative method is suitable for application scenarios that are more sensitive to out-of-limit events.

[0066] In another feasible implementation, the comprehensive deviation index is calculated as follows: the physical deviation index and the baseline deviation index are weighted and summed to obtain the comprehensive deviation index. The weighted sum of the weights of the physical deviation index and the baseline deviation index equals one. The weight of the physical deviation index is set to 40% to 60%, and the weight of the baseline deviation index is set to 40% to 60%.

[0067] In some embodiments, the weighted sum of the physical deviation index and the baseline deviation index is achieved by first normalizing them separately and then summing them. The physical deviation index and the baseline deviation index are normalized separately, each mapped to a range of zero to one. The upper limit of normalization for the physical deviation index is set to five standard deviations of the mean physical deviation index under normal operating conditions, or it can be a constant of three. The upper limit of normalization for the baseline deviation index is also set to five standard deviations of the mean baseline deviation index under normal operating conditions, or it can be a constant of three. During normalization, the physical deviation index is divided by its upper limit of normalization, and the smaller of one and the quotient is taken as the normalized physical deviation index. The baseline deviation index is processed in the same way. The normalized physical deviation index is multiplied by its comprehensive weight, and the normalized baseline deviation index is multiplied by its comprehensive weight. The two are then added together to obtain the comprehensive deviation index. The comprehensive weight of the physical deviation index ranges from 0.4 to 0.6, and the comprehensive weight of the baseline deviation index is one minus the comprehensive weight of the physical deviation index. Through the above normalization process, the physical deviation index and the baseline deviation index are weighted and summed under the premise of consistent dimensions. The resulting comprehensive deviation index is a dimensionless quantity with a value range of zero to one.

[0068] In some other embodiments, when either the physical deviation index or the baseline deviation index exceeds a preset individual alarm threshold, the weight of that individual item in the overall deviation index is automatically increased to over 70%. The individual alarm threshold is set as follows: for the physical deviation index, it is set to three standard deviations of the mean physical deviation index under normal operating conditions. For the baseline deviation index, it is set to three standard deviations of the mean baseline deviation index under normal operating conditions.

[0069] In one feasible implementation, the health status assessment result is obtained by mapping the comprehensive deviation index according to a pre-set health level mapping table. The health level mapping table records the correspondence between the numerical range of the comprehensive deviation index and the health status level. The health status levels include at least four levels: normal, attention, abnormal, and severe. The comprehensive deviation index range for the normal level is 0 to 0.3, indicating that all status parameters of the equipment are within the normal range and there are no abnormal signs. The comprehensive deviation index range for the attention level is greater than 0.3 and less than or equal to 0.6, indicating that the equipment has a slight deviation, and it is recommended to increase the monitoring frequency. The comprehensive deviation index range for the abnormal level is greater than 0.6 and less than or equal to 0.9, indicating that the equipment has a significant abnormal deviation, and it is recommended to arrange a power outage for maintenance. The comprehensive deviation index range for the severe level is greater than 0.9, indicating that the equipment has a serious abnormality, and it is recommended to immediately shut it down. The boundary values ​​of each level range can be adaptively adjusted according to the transformer's voltage level, capacity level, and importance. For example, for transformers with voltage levels of 500 kV and above, the boundary values ​​for each voltage level range are shifted downward by 10% to 20% to apply more stringent evaluation standards to this type of equipment. For transformers with voltage levels of 110 kV and below, the boundary values ​​for each voltage level range are shifted upward by 10% to 20% to balance the sensitivity of the evaluation and the economics of operation.

[0070] In another feasible implementation, the health level mapping table can also have only three or five levels. The three levels are normal, abnormal, and severe, with corresponding comprehensive deviation index ranges of 0 to 0.4, greater than 0.4 and less than or equal to 0.8, and greater than 0.8, respectively. With five levels, a good level is added between the normal and attention levels, and a dangerous level is added between the abnormal and severe levels. The corresponding comprehensive deviation index ranges are 0 to 0.2, greater than 0.2 and less than or equal to 0.4, greater than 0.4 and less than or equal to 0.6, greater than 0.6 and less than or equal to 0.8, and greater than 0.8, respectively. The number of levels is determined based on the specific application scenario's need for refined health status grading.

[0071] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0072] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0073] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for dynamic assessment of the health status of an intelligent transformer, characterized in that, The method includes: Pt.

1. Obtain the measured operating parameter set and measured status parameter set of the transformer to be evaluated at the current moment. The measured operating parameter set includes the high-voltage side current value, the low-voltage side voltage value, and the ambient temperature value. The measured status parameter set includes the top oil temperature value, the winding hot spot temperature value, and the concentration value of the characteristic gas dissolved in the oil. Pt.

2. Select historical data that meet the steady-state criteria from the historical operation records of the transformer to be evaluated, perform statistical processing on the selected historical data, and generate a set of baseline parameters corresponding to the transformer to be evaluated. Pt.

3. Read the rated parameter values ​​from the nameplate parameter library of the transformer to be evaluated, and generate the physical anchor point parameter set under the current operating condition based on the current high-voltage side current value and the ambient temperature value, combined with the preset multi-physics field coupling mapping relationship. Pt.

4. Compare the measured state parameter set with the physical anchor point parameter set to obtain the physical deviation, and compare the measured operating parameter set and the measured state parameter set with the baseline parameter set to obtain the baseline deviation. Pt.

5. Based on the physical deviation and the baseline deviation, and in accordance with the preset judgment rules, generate the health status assessment result of the transformer to be evaluated.

2. The intelligent transformer health status dynamic assessment method as described in claim 1, characterized in that, The acquisition of the measured operating parameter set and measured status parameter set of the transformer to be evaluated at the current moment includes: Obtain the high-voltage side current sampling value of the transformer to be evaluated at the current moment, and trace back from the current moment to extract the high-voltage side current historical sequence from the historical operation record of the transformer to be evaluated; The fluctuation amplitude of the current value in the high-voltage side current history sequence is detected, and the fluctuation amplitude is compared with the preset fluctuation allowable range. When the fluctuation amplitude of the current value exceeds the fluctuation allowable range, it is determined that the current moment is in an unsteady state. Starting from the current moment, the time length of the high-voltage side current history sequence is extended until the fluctuation amplitude of the current value in the high-voltage side current history sequence falls into the fluctuation allowable range. The end moment of the sequence that falls into the fluctuation allowable range is determined as the reference acquisition moment. Starting from the reference acquisition time, a set of operating parameter sampling values ​​and a set of state parameter sampling values ​​are acquired according to a preset sampling interval. The operating parameter sampling values ​​include high-voltage side current sampling values, low-voltage side voltage sampling values, and ambient temperature sampling values. The state parameter sampling values ​​include top oil temperature sampling values, winding hot spot temperature sampling values, and characteristic gas concentration sampling values. The set of sampled operating parameters and the set of sampled state parameters obtained at the same time are used together as the measured operating parameter set and the measured state parameter set.

3. The intelligent transformer health status dynamic assessment method as described in claim 1, characterized in that, The process of selecting historical data that meet the steady-state criteria from the historical operating records of the transformer to be evaluated includes: Extract the high-voltage side current history sequence from the historical operation records of the transformer to be evaluated, and detect the current fluctuation amplitude between adjacent current values ​​in the high-voltage side current history sequence. The current fluctuation amplitude corresponding to each adjacent time moment is compared with the preset fluctuation allowable range, and the time period in which the current fluctuation amplitude continuously falls within the fluctuation allowable range is marked as the load stable period. Extract the top oil temperature history sequence corresponding to the load stable period from the historical operation record, and detect the temperature change of the top oil temperature history sequence within a continuous time window; The temperature change is compared with a preset temperature gradient allowable range, and the time period in which the temperature change continuously falls within the temperature gradient allowable range is marked as a temperature stable period. The overlapping period between the load stable period and the temperature stable period is selected, and the duration of the overlapping period is determined. When the duration of the overlapping period meets the preset duration criterion, all historical data corresponding to the overlapping period is extracted as historical data that meets the steady-state criterion.

4. The intelligent transformer health status dynamic assessment method as described in claim 3, characterized in that, Before performing statistical processing on the selected historical data, the following steps are included: Read the collection timestamp corresponding to each data point in the historical data that satisfies the steady-state criterion, and determine the seasonal attribute of the data point based on the collection timestamp. The seasonal attribute is determined by the date range of the collection timestamp in a year. Read the operating mode identifier corresponding to each data point in the historical data that meets the steady-state criteria. The operating mode identifier is extracted from the historical operating records and is used to indicate the power supply operating mode of the transformer to be evaluated when the data point was collected. Historical data with the same seasonal attribute and the same operating mode identifier are grouped into the same operating condition group. After traversing all historical data, operating condition groups are obtained. Each operating condition group corresponds to a combination of seasonal attribute and operating mode identifier. All operating conditions are grouped together and their included historical data are aggregated to form unprocessed data with operating condition category labels.

5. The intelligent transformer health status dynamic assessment method as described in claim 4, characterized in that, The step of statistically processing the selected historical data to generate a baseline parameter set corresponding to the transformer to be evaluated includes: Read the historical data within any working condition group of the data to be processed, and arrange the historical data under the same parameter name within the working condition group in chronological order of collection time to form a parameter value sequence for that parameter under that working condition group; Taking adjacent data points in the parameter value sequence as units, the difference between the next data point and the previous data point is calculated in turn. The absolute value of each difference is taken and then the average value of all absolute differences is calculated as the mean moving range of the parameter under the working condition group. Read the arithmetic mean of each data point in the parameter value sequence and use it as the center value of the parameter under this working condition group; Using the center value as a reference and the mean value of the moving range as a scale, the mean value of the moving range is shifted upward and downward by a set multiple to obtain the upper control limit and lower control limit of the parameter under the working condition group. Read the skewness coefficient and kurtosis coefficient of each data point in the parameter value sequence, and compare the skewness coefficient and kurtosis coefficient with the normal distribution reference value respectively. When the deviation of the skewness coefficient from the normal distribution reference value exceeds the first allowable range or the deviation of the kurtosis coefficient from the normal distribution reference value exceeds the second allowable range, it is determined that the parameter value sequence of the parameter under the working condition group follows a non-normal distribution. For a parameter value sequence that follows a non-normal distribution, the value located at the preset percentile position in the parameter value sequence is read, and the value is used to replace the control upper limit value and control lower limit value, as the corrected control upper limit value and corrected control lower limit value of the parameter under the working condition group. For a sequence of parameter values ​​that follows a normal distribution, the upper and lower control limits remain unchanged. Traverse all operating condition groups and all parameters under each operating condition group in the data to be processed to obtain the upper control limit and lower control limit or the modified upper control limit and lower control limit for each parameter under each operating condition group. Summarize all the obtained control limits as the baseline parameter set of the transformer to be evaluated.

6. The intelligent transformer health status dynamic assessment method as described in claim 5, characterized in that, After generating the baseline parameter table corresponding to the transformer to be evaluated, the process further includes: The commissioning years and historical load accumulation of the transformer to be evaluated are read. The commissioning years are extracted from the equipment history file of the transformer to be evaluated, and the historical load accumulation is obtained by accumulating the load rate at each moment in the historical operation record of the transformer to be evaluated. The commissioning age value and the historical load accumulation are entered into a preset aging drift mapping table for retrieval. The aging drift mapping table records the control limit compensation offset that corresponds to both the commissioning age range and the load accumulation range. The control limit compensation offset that matches the current commissioning age value and the historical load accumulation is read from the aging drift mapping table. Read the upper control limit and lower control limit or the modified upper control limit and lower control limit of each parameter under each working condition group in the baseline parameter set, and add the control limit compensation offset to each upper control limit and each lower control limit respectively to obtain the aging compensation control upper limit and aging compensation control lower limit of the parameter under the working condition group. Iterate through all operating condition groups and all parameters, and summarize the upper limit value and lower limit value of aging compensation control for each parameter under each operating condition group, as the output of the aging compensation baseline parameter set of the transformer to be evaluated.

7. The intelligent transformer health status dynamic assessment method as described in claim 1, characterized in that, The process involves reading the rated parameter values ​​from the transformer nameplate parameter library to be evaluated, and generating a set of physical anchor point parameters for the current operating condition based on the current high-voltage side current value and the ambient temperature value, combined with a preset multi-physics coupling mapping relationship. This includes: Read the rated parameter values ​​from the nameplate parameter library of the transformer to be evaluated. The rated parameter values ​​include rated capacity, rated voltage, rated current, short-circuit impedance, no-load loss, and load loss. The electrothermal correlation map, heat flow correlation map, and thermal aging correlation map are extracted from the preset multiphysics coupling mapping relationship. The electrothermal correlation map records the theoretical output value of the top oil temperature corresponding to the high-voltage side current value and the ambient temperature value. The heat flow correlation map records the theoretical output value of the winding axial temperature distribution and the theoretical output value of the oil flow velocity corresponding to the top oil temperature value. The thermal aging correlation map records the theoretical output value of the insulation paper polymerization degree decrease rate corresponding to the winding hot spot temperature value. Input the current high-voltage side current value and the ambient temperature value into the electrothermal correlation mapping, and read the theoretical output value of the top oil temperature that matches the current high-voltage side current value and the current ambient temperature value from the electrothermal correlation mapping, and use it as the theoretical value of the top oil temperature. The theoretical value of the top oil temperature is input into the heat flow correlation mapping. The theoretical output value of the winding axial temperature distribution and the theoretical output value of the oil flow velocity that match the theoretical value of the top oil temperature are read from the heat flow correlation mapping and used as the theoretical value of the winding axial temperature distribution and the theoretical value of the oil flow velocity. The theoretical value of the axial temperature distribution of the winding is input into the thermal aging correlation mapping. The theoretical output value of the rate of decrease in the degree of polymerization of the insulating paper, which matches the temperature value of each axial position in the theoretical value of the axial temperature distribution of the winding, is read from the thermal aging correlation mapping and used as the theoretical value of the rate of decrease in the degree of polymerization of the insulating paper. The theoretical values ​​of the top layer oil temperature, the winding axial temperature distribution, the oil flow velocity, and the insulation paper polymerization degree decrease rate are summarized and used as the physical anchor point parameter set under the current working condition.

8. The intelligent transformer health status dynamic assessment method as described in claim 1, characterized in that, The step of comparing the measured state parameter set with the physical anchor point parameter set to obtain the physical deviation includes: Read the measured state parameter group, and extract the measured value of the top oil temperature, the measured value of the winding hot spot temperature, and the measured concentration value of the characteristic gas from the measured state parameter group; Read the physical anchor point parameter set, and extract the theoretical values ​​of top oil temperature, winding axial temperature distribution, and insulation paper polymerization degree decrease rate from the physical anchor point parameter set; The oil temperature deviation value is obtained by subtracting the measured value of the top oil temperature from the theoretical value of the top oil temperature. Determine the axial position corresponding to the measured value of the hot spot temperature of the winding, extract the theoretical temperature value corresponding to the axial position from the theoretical value of the axial temperature distribution of the winding, and subtract the measured value of the hot spot temperature of the winding from the theoretical temperature value to obtain the hot spot deviation value; The current measured concentration value and the previous sampled concentration value of the characteristic gas are obtained, and the time interval between the current sampling and the previous sampling is obtained. The difference between the current measured concentration value and the previous sampled concentration value is divided by the time interval to obtain the measured gas production rate. The theoretical value of the rate of decrease in the degree of polymerization of the insulating paper is extracted from the set of physical anchor point parameters. The measured value of the gas generation rate is subtracted from the theoretical value of the rate of decrease in the degree of polymerization of the insulating paper to obtain the gas generation deviation value. The oil temperature deviation, the hot spot deviation, and the gas production deviation are summed up as the physical deviation.

9. The intelligent transformer health status dynamic assessment method as described in claim 1, characterized in that, The step of comparing the measured operating parameter set and the measured state parameter set with the baseline parameter set to obtain the baseline deviation includes: Read the measured operating parameter set, and extract the measured value of the high-voltage side current and the measured value of the ambient temperature from the measured operating parameter set; read the measured state parameter set, and extract the measured value of the top oil temperature and the measured value of the winding hot spot temperature from the measured state parameter set; Read the upper and lower control limits corresponding to the working condition group that matches the current seasonal attribute and current operating mode identifier in the baseline parameter set. Extract the upper and lower control limits of the current control corresponding to the high voltage side current value, the upper and lower control limits of the oil temperature corresponding to the top oil temperature value, and the upper and lower control limits of the hot spot temperature corresponding to the winding hot spot temperature value from the upper and lower control limits. The measured value of the high-voltage side current is compared with the upper limit value and the lower limit value of the current control. When the measured value of the high-voltage side current falls between the upper limit value and the lower limit value of the current control, a normal current indicator is generated. When the measured value of the high-voltage side current exceeds the upper limit value of the current control or is lower than the lower limit value of the current control, a current over-limit indicator and the over-limit direction are generated. Perform the same interval comparison operation as the measured value of the top oil temperature and the measured value of the winding hot spot temperature on the same basis as the measured value of the high voltage side current to generate an oil temperature normal mark or an oil temperature over-limit mark and the over-limit direction, as well as a hot spot normal mark or a hot spot over-limit mark and the over-limit direction. The normal current indicator or the current exceeding the limit indicator and the direction of exceeding the limit, the normal oil temperature indicator or the oil temperature exceeding the limit indicator and the direction of exceeding the limit, and the normal hot spot indicator or the hot spot exceeding the limit indicator and the direction of exceeding the limit are summarized as the baseline deviation.

10. The intelligent transformer health status dynamic assessment method as described in claim 1, characterized in that, The step of generating a health status assessment result for the transformer to be evaluated based on the physical deviation and the baseline deviation, according to a preset judgment rule, includes: Obtain the high-voltage side current value of the transformer to be evaluated at the current moment, determine the current load range based on the high-voltage side current value, and read the physical deviation weight value and baseline deviation weight value corresponding to the load range from the preset weight allocation table. The physical deviation weight value includes oil temperature deviation weight, hot spot deviation weight and gas production deviation weight, and the baseline deviation weight value includes current over-limit weight, oil temperature over-limit weight and hot spot over-limit weight. The physical deviation index is obtained by multiplying the oil temperature deviation value, hot spot deviation value and gas production deviation value in the physical deviation quantity by their respective oil temperature deviation weight, hot spot deviation weight and gas production deviation weight, and then summing them up. The current over-limit indicator and over-limit direction, oil temperature over-limit indicator and over-limit direction, and hot spot over-limit indicator and over-limit direction in the baseline deviation are converted into corresponding over-limit deviation values. Each of the over-limit deviation values ​​is multiplied by the corresponding current over-limit weight, oil temperature over-limit weight, and hot spot over-limit weight, and then summed to obtain the baseline deviation index. The physical deviation index and the baseline deviation index are weighted and summed to obtain the comprehensive deviation index. According to the preset health level mapping table, the comprehensive deviation index is mapped to the corresponding health status level as the health status assessment result of the transformer to be evaluated.

Images