A transformer overload capacity dynamic evaluation method and system

Abstract

This invention relates to the field of transformer overload control, specifically to a dynamic assessment method and system for transformer overload capacity. First, high-precision data acquisition and sliding window differential verification technology are used to accurately extract load step characteristics and eliminate interference. Then, a finite state machine is run based on fluid dynamics principles, switching between three modes—steady-state coupling, inertial decoupling, and linear recovery—according to load changes. Particularly during load surges, the system enters decoupling mode, using adiabatic temperature rise estimation to replace the delayed measured top-layer oil temperature, effectively solving the calculation distortion problem of traditional models during oil flow lag. Furthermore, the system performs parallel dual-thread calculations using a standard model and an inertial sensing model, outputting the final hotspot temperature through maximum value arbitration logic, and using this to estimate the remaining allowable operating time to generate protection control signals. This invention significantly improves the assessment accuracy and operational safety of transformers under short-term overload conditions.

Description

Technical Field

[0001] This invention relates to the field of transformer overload control, and specifically to a method and system for dynamic evaluation of transformer overload capacity. Background Technology

[0002] As a key component of the power system, the dynamic assessment of a transformer's overload capacity directly impacts the reliability and security of the power grid. In transformer operation monitoring, accurately monitoring and predicting winding hotspot temperatures is crucial for assessing insulation life loss, preventing thermal breakdown damage, and developing emergency load shedding strategies.

[0003] Existing methods for calculating transformer hot spot temperatures primarily employ the thermal circuit differential equation model recommended by the IEC 60076-7 standard. This general model, based on the lumped parameter method, treats the transformer as a thermal system. It typically uses real-time acquired load factor and top oil temperature as key input variables, and simulates the dynamic process of winding temperature rise relative to oil by solving first-order differential equations, thereby estimating the current winding hot spot temperature.

[0004] However, this traditional model has significant drawbacks when dealing with sudden impact loads. Due to the hydrodynamic inertia of the flow and heat conduction of the insulating oil inside the transformer, the change in oil temperature lags significantly behind the change in heat generated by the windings. In the initial stage of a load surge, the top-layer oil temperature collected by external sensors has not yet responded to the rise. If this delayed measured top-layer oil temperature is directly used as a boundary condition in the differential equation, the calculated hotspot temperature rise slope will be much smaller than the actual physical process, failing to accurately reflect the rapid temperature rise of the windings under near-insulation conditions. This calculation distortion will cause the system to severely underestimate the thermal risk of the transformer in the initial stage of overload, potentially leading to delayed protection actions. Summary of the Invention

[0005] To address the problem that traditional models cannot accurately reflect the rapid temperature rise of windings under near-insulation conditions, this invention proposes a dynamic evaluation method for transformer overload capacity in its first aspect. The method includes: acquiring real-time transformer load data and analyzing the load data based on a time-series sliding window to extract load step characteristics; running a finite state machine model, which includes a steady-state coupling mode, an inertial decoupling mode, and a linear recovery mode; when the load step characteristic is detected, switching the finite state machine from the steady-state coupling mode to the inertial decoupling mode; setting a reference oil temperature for the thermal circuit differential equation based on the current operating mode of the finite state machine, whereby the thermal circuit differential equation characterizes the functional relationship between the rate of change of winding hot spot temperature over time and the load factor and the reference oil temperature: in the steady-state coupling mode, the measured top-layer oil temperature of the transformer is used as the reference oil temperature; in the inertial decoupling mode, the input correlation of the measured top-layer oil temperature is severed, and the adiabatic extrapolation oil temperature or the oil temperature locked before load impact is used as the reference oil temperature; and solving the thermal circuit differential equation to obtain the winding hot spot temperature.

[0006] This invention solves the problem in existing technologies where transformer winding hot spot temperature calculations heavily rely on measured top-layer oil temperature, leading to an inaccurate reflection of transient winding temperature rises during sudden load changes due to the thermal inertia lag of oil temperature. By introducing a finite state machine model, the calculation mode is switched from steady-state coupling to inertial decoupling mode when a load step is detected. This disconnects the lagging input of measured top-layer oil temperature and uses adiabatic extrapolation or locked oil temperature as a reference, eliminating calculation errors caused by sensor thermal lag. This enables rapid and accurate dynamic assessment of transformer winding hot spot temperature under overload impact, ensuring the safe operation of the equipment.

[0007] Furthermore, extracting the load step feature includes: determining that the difference between the first and last data of the sliding window is greater than a preset step determination threshold; and determining that the proportion of points in the sliding window where the load data shows a monotonically increasing trend exceeds a preset monotonicity threshold.

[0008] This invention, by combining the determination of the difference between the first and last values ​​of the sliding window with the determination of the proportion of monotonically increasing data, can effectively distinguish between real load step changes and normal load fluctuations or data noise. It significantly improves the anti-interference ability and robustness of load feature extraction, ensures the accuracy of the timing of finite state machine mode switching, and avoids frequent model switching or calculation failure caused by misjudgment.

[0009] Furthermore, the method for obtaining the adiabatic simulation oil temperature includes: based on the load factor at the current moment and the preset initial temperature rise rate constant under the rated current, the measured top oil temperature or the simulation oil temperature at the moment before the impact occurs is accumulated and calculated.

[0010] This invention uses the load factor and initial temperature rise rate constant to simulate the theoretical temperature rise behavior of the winding in the absence of oil cooling feedback, providing a physically consistent reference temperature input for the inertial decoupling mode. This ensures the continuity and thermodynamic rationality of the calculation process during the disconnection of measured sensor data, and truly reflects the heat accumulation process inside the winding.

[0011] Furthermore, the duration of the finite state machine in the inertial decoupling mode depends on the transformer's cooling method; for natural oil circulation cooling, the duration is set to 15 to 20 minutes; for forced oil circulation cooling, the duration is set to 3 to 5 minutes.

[0012] Furthermore, in the linear recovery mode, the method for obtaining the reference oil temperature includes: performing a time-varying weighted average of the adiabatic extrapolation oil temperature or the pre-impact locked oil temperature and the measured top-layer oil temperature, wherein the weighting coefficient changes linearly with the increase of recovery time until it is completely switched to the measured top-layer oil temperature.

[0013] In the process of the system returning from transient decoupling mode to steady-state mode, the present invention achieves a smooth transition of reference oil temperature through time-varying weighted averaging, effectively avoiding non-physical abrupt changes or oscillations in the calculation results caused by switching of calculation models or re-intervention of sensor data, and ensuring the smoothness, continuity and correctness of the temperature evaluation curve.

[0014] Furthermore, it also includes running a standard thermal circuit calculation process in parallel, wherein the standard thermal circuit calculation process always uses the measured top oil temperature as the reference oil temperature to calculate the standard winding hot spot temperature; the winding hot spot temperature calculated by the finite state machine model is compared with the standard winding hot spot temperature, and the larger value is selected as the final evaluation result.

[0015] This invention employs a dual-track parallel computing strategy, combining a finite state machine model that considers thermal inertia with a standard thermal circuit model that is always based on the measured top oil temperature, and selecting the larger value of the calculation result as the final evaluation basis. This not only retains the finite state machine model's ability to sensitively capture transient temperature rises, but also utilizes the conservatism of traditional methods to construct a safety baseline, preventing temperature underestimation due to model parameter deviations, and ensuring that the transformer does not overheat to the greatest extent possible.

[0016] Furthermore, the difference between the winding hot spot temperature obtained from the finite state machine model and the standard thermal circuit calculation process is calculated. If the difference continues to exceed the preset temperature difference alarm threshold, an alarm signal for thermal model mismatch or sensor failure is generated.

[0017] Furthermore, based on the final evaluation results, the remaining allowable operating time of the transformer is estimated by querying the pre-stored insulation material aging curves to determine the life loss rate at the current temperature.

[0018] Furthermore, when the remaining allowable running time is less than the preset protection action time threshold, an emergency load reduction or tripping command is sent to the scheduling system.

[0019] In a second aspect, the present invention provides a dynamic evaluation system for transformer overload capacity, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a dynamic evaluation method for transformer overload capacity of the present invention is implemented.

[0020] The technical effects of this invention are as follows:

[0021] The core of this invention lies in proposing a dynamic evaluation method for transformer overload capacity based on a finite state machine. Addressing the problem of inaccurate winding hot spot temperature calculations due to the lag in the response of the top-level oil temperature sensor during sudden load changes in transformers, this invention innovatively introduces an inertial decoupling mode. At the moment of impact, the correlation between the measured top-level oil temperature and the actual load is severed, and adiabatic extrapolation calculation is used instead. Simultaneously, by combining standard thermal circuit parallel calculation with a dual verification mechanism, this method not only solves the transient errors caused by thermal inertia but also ensures the conservative safety of the evaluation results, achieving accurate thermal protection and lifespan management under short-term emergency load conditions for transformers. Attached Figure Description

[0022] Figure 1 This is a schematic flowchart illustrating a method for dynamically evaluating the overload capacity of a transformer according to an embodiment of the present invention.

[0023] Figure 2 This is a schematic diagram illustrating the time-series change of the load factor and the waveform of the impact load detection range in an embodiment of the present invention;

[0024] Figure 3 This is a schematic diagram illustrating the temperature response curve of the improved thermal circuit model after introducing fluid inertial sensing in an embodiment of the present invention.

[0025] Figure 4 This is a schematic graph illustrating the comparison of calculation results between the standard model and the improved model in an embodiment of the present invention, as well as the final arbitration decision.

[0026] Figure 5 This is a schematic diagram illustrating the monitoring curve of the deviation value of the dual-thread calculation results changing over time in an embodiment of the present invention;

[0027] Figure 6 This is a schematic diagram illustrating the structural block of a transformer overload capacity dynamic evaluation system according to an embodiment of the present invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] The technical solution of this invention is based on a highly reliable transformer intelligent monitoring unit (IED) or relay protection device. This device includes a core processing module, a data acquisition module, a communication module, and a storage module.

[0030] In this embodiment, the core processing module uses an embedded microcontroller, whose main frequency is preferably set to 400MHz to 600MHz. The controller's storage space is divided into a buffer for storing real-time sampled values ​​and a static parameter library for storing thermal circuit model parameters.

[0031] The data acquisition module includes a high-precision A / D converter with a sampling accuracy of no less than 16 bits and a sampling frequency preferably set to 1kHz to 10kHz, used to convert the analog current signal transmitted by the current transformer (CT) into a digital signal. The communication module supports CAN-FD bus and industrial Ethernet interface based on the IEC 61850 protocol for bidirectional data interaction with the upper-level SCADA scheduling system.

[0032] It should be noted that in other embodiments, the core processing module may also be a digital signal processor (DSP), a field-programmable gate array (FPGA), or an industrial-grade x86 architecture industrial control computer, as long as it has floating-point operation capability and corresponding I / O interface, it is within the protection scope of this invention.

[0033] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0034] like Figure 1 As shown, the present invention provides a method for dynamic evaluation of transformer overload capacity, comprising:

[0035] S101. Use a high-precision acquisition module to acquire real-time load data and perform differential and monotonicity checks based on a sliding window to accurately extract load step features.

[0036] The system first constructs a time-series sliding window to capture load changes. In this embodiment, the length of the sliding window... The value of should cover the electrical time constant but be much smaller than the thermal time constant of the transformer. In this preferred embodiment, Set the number of sampling points corresponding to 10 to 30 seconds. For example, if the sampling frequency is 1kHz, the queue length is 10,000 to 30,000 data points.

[0037] Specifically, in order to obtain a load factor that accurately reflects the thermal state of the transformer The processor controls the high-precision A / D acquisition module to synchronously sample the current signal at the three-phase bushings of the transformer. The processor performs root mean square calculation on the raw sampled waveform of each phase and selects the maximum value among the effective values ​​of the three-phase current. This serves as the characteristic current at the current moment. Subsequently, the processor applies the formula... Real-time calculation of load factor, where The rated current specified on the transformer's nameplate.

[0038] To eliminate false spikes caused by electromagnetic interference, the system performs the following dual verification mechanism:

[0039] First, the processor calculates the difference between the first and last data points of the window in real time. If this difference exceeds a preset threshold... If this occurs, a preliminary judgment is triggered. In this embodiment, Defined as the percentage of significant change in rated load, it is preferably set to 0.3 pu to 0.5 pu. However, in practical engineering applications, this threshold can be flexibly adjusted within the range of 0.2 pu to 0.8 pu, depending on the transformer capacity.

[0040] Further statistical analysis is performed on the monotonicity of the data within the window. In one embodiment, the data within the statistical window satisfies... The percentage of data points. If this percentage exceeds 90% and the amplitude verification is also met, the system determines that an impact load event has occurred.

[0041] In another embodiment, the mean of the first derivatives of the data within the window or the slope fitting error is calculated. If the slope is positive and the root mean square error (RMSE) is less than a preset noise tolerance, such as 0.05, it is determined to be a valid rise, thereby avoiding misjudgments caused by sensor jitter.

[0042] like Figure 2 The figure illustrates the fluctuations in the load factor collected by the system over time. The originally smooth curve represents the real-time load factor, and the horizontal dashed line represents the rated load reference of the transformer. When the curve shows a steep step increase and enters a specific vertically marked interval, it indicates that the system has successfully identified the impact load event based on the differential and monotonicity verification algorithm using a sliding window. Within these identified intervals, the rate of change of the load exceeds a preset threshold, triggering a switch in the subsequent thermal circuit calculation mode.

[0043] S102. Run a finite state machine based on the principle of fluid dynamics and perform deterministic switching between different thermal circuit calculation modes according to the load impact signal.

[0044] The processor runs a finite state machine that switches between three thermal circuit calculation modes based on the impact load event signal calculated in step S101, in order to solve the calculation distortion problem of the traditional model during the oil flow hysteresis period.

[0045] State I: Steady-state coupling mode. When no impact load is detected, i.e., the load is stable or changes slowly, the system determines that the oil flow circulation has been fully established. At this time, the system adopts the standard thermal circuit equation and uses the real-time measured top oil temperature as the boundary condition to calculate the winding hot spot temperature.

[0046] State II: Inertial Decoupling Mode. When an impact load is detected, due to the inertia of hydrodynamics causing the cooling system to be unable to respond immediately to the surge in heat generation, the system immediately disconnects the input association of the real-time oil temperature sensor and enters the decoupling calculation state.

[0047] State III: Linear Recovery Mode. After the duration of State II reaches the preset fluid hysteresis time, the system enters the recovery period. To prevent non-physical oscillations in the model output, the system does not immediately switch back to the measured top-layer oil temperature, but instead smoothly transitions using a time-varying weighted algorithm.

[0048] S103. Adaptively adjust the parameters of the thermal circuit differential equation according to the current operating status and solve the winding hot spot temperature to simulate the hysteresis effect of fluid heat dissipation.

[0049] Under any of the above conditions, the winding hot spot temperature The evolution of all follows the following thermal differential equation, which is derived based on the IEC 60076-7 standard:

[0050] ;

[0051] in This refers to the hot spot temperature of the winding. The winding time constant; This is the load factor; This is the rated hot spot temperature rise; For reference oil temperature; These are model constants.

[0052] It should be noted that the constant parameters in the above differential equations are pre-written into the controller's memory, such as Flash or EEPROM, by engineers during the system initialization phase, based on the factory temperature rise test report or technical specifications of the specific transformer, using a host computer configuration tool. This ensures that the thermal circuit model accurately matches the thermal characteristics of the physical transformer, and that the winding time constant is also accurately set. It is a non-zero physical constant.

[0053] From the above formula, we can see the rate of change of hotspot temperature. It depends directly on the difference between the current hot spot temperature and the reference oil temperature, as well as the load factor. When the load factor... When there is a sudden increase, refer to the oil temperature. Keeping the temperature low means good heat dissipation, resulting in a slower calculated temperature rise; if the reference oil temperature is artificially increased or locked, the temperature rise slope will become larger.

[0054] Based on the above principles, this embodiment performs the following substitutions and corrections on the parameters under different conditions:

[0055] In one embodiment, the parameter correction under state II is as follows:

[0056] Reference oil temperature No longer based on the real-time measured top oil temperature Instead of relying on the temperature directly above the impact point, the reference oil temperature is either fixed at the value just before the impact or derived using the adiabatic temperature rise formula. Specifically, the selection of either of these reference oil temperatures depends on whether the system possesses a calibrated temperature rise rate constant. When the controller parameter library already contains the temperature rise rate constant calibrated according to the factory temperature rise test report. When the oil temperature is under normal conditions, the system prioritizes the adiabatic simulation scheme to obtain dynamic temperature tracking that more closely resembles the actual physical process; conversely, if the controller parameter library does not contain a calibrated temperature rise rate constant... The system uses the pre-impact locked oil temperature as the reference oil temperature, that is, the reference oil temperature is kept unchanged at the measured top oil temperature value at the moment of entering state II.

[0057] In this embodiment, the adiabatic deduction logic is not a qualitative description, but rather a specific execution of the following recursive formula:

[0058] ;

[0059] in The adiabatic oil temperature calculated at the current moment; The value is the calculated value of the previous moment, and the first moment is the measured top oil temperature of the moment before the impact occurred; The average temperature rise rate constant is the initial temperature rise slope under rated current, and its value is usually taken in the range of 0.05K / s to 0.15K / s. The step size is calculated to be consistent with the sampling period; This represents the load factor at the current moment.

[0060] Among them, the average temperature rise rate constant in the adiabatic derivation formula This represents the initial temperature rise slope of the transformer under rated current. Upon entering state II, due to the significant lag in oil temperature dissipation feedback caused by hydrodynamic inertia, the system disconnects the dissipation feedback correlation of the measured top-layer oil temperature in the model. At this point, the load factor is used... and constant The above cumulative simulation reflects the heat accumulation in the early stages of overload impact.

[0061] The duration of final state II, i.e., the fluid lag time. The value is determined by referring to a table based on the transformer's cooling method. For ONAN (self-cooling) mode, the value ranges from 15 to 20 minutes; for ODAF (forced oil-air cooling) mode, the value ranges from 3 to 5 minutes, thus causing the calculated winding temperature to rise linearly at a slope close to that of pure resistance heating.

[0062] In another embodiment, the transition calculation in state III is as follows:

[0063] The system executes a time-varying weighted algorithm to synthesize a reference oil temperature. The processor calculates the normalized percentage of the current time within the recovery window. ,in The summation increases linearly from 0 to 1 over time, and the following weighted summation is performed:

[0064] When the oil temperature is adiabatic and calculated in State II...

[0065] ;

[0066] in To adiabatically extrapolate oil temperature, This is to measure the top layer oil temperature.

[0067] When State II uses pre-impact oil temperature locking, then the formula above will be... Replace with pre-impact lock-in oil temperature. If impact load is detected again during recovery, the system will immediately reset. Set to 0 and revert to state II.

[0068] like Figure 3 The figure shows the calculation results of the winding hot spot temperature after applying the improved thermal circuit model of this invention. The undulating curve at the bottom of the figure represents the change of the reference oil temperature, and the solid curve at the top represents the calculated winding hot spot temperature. It can be seen that at the moment the impact load is detected, the model cuts off the correlation with the real-time oil temperature, and the temperature shows a rapid upward trend similar to an adiabatic process; while in the recovery phase after the impact, the temperature curve transitions smoothly without any non-physical oscillations. The horizontal dotted line represents the Class A insulation limit, and this figure intuitively reflects the sensitive response characteristics of the improved model in dealing with fluid heat dissipation lag.

[0069] S104. Perform parallel dual-thread hotspot temperature calculations and output the final decision result based on the maximum value arbitration logic to ensure system safety assessment.

[0070] To ensure the absolute safety of the power system, the following decision-making mechanism is adopted in this embodiment.

[0071] Thread A always runs the unmodified standard thermal circuit model, and its input directly depends on real-time sensor data.

[0072] Thread B runs the state machine model with fluid inertial awareness described in steps S102 and S103 above.

[0073] Ultimately, the system executes high-level arbitration logic, comparing the calculation results of the two threads in real time, and always adopting the value with the higher temperature as the final decision basis. That is:

[0074] ;

[0075] It ensures that the system is always in the safest evaluation state in extreme cases of sensor failure or model mismatch.

[0076] like Figure 4 As shown, the system runs two computation threads simultaneously and compares them. The relatively flat curve in the figure represents the calculation result of the uncorrected standard thermal path model, while the curve with a distinct peak represents the calculation result of the improved model that incorporates fluid inertial sensing. The thick dashed line at the top represents the final decision temperature after maximum value arbitration, and this trajectory always runs along the higher of the two values. The shaded area between the two curves visually demonstrates the additional temperature rise calculation provided by the improved model during shock loads, i.e., the safety margin reserved by the system to offset the heat dissipation hysteresis effect, ensuring the conservatism and safety of the evaluation results.

[0077] It is important to note that, to improve system robustness, the system also calculates the deviation between the two threads in parallel while performing maximum value arbitration. If the deviation value continues to exceed the preset temperature difference (e.g., 20K) for a preset time threshold (e.g., for 3 minutes or 5 consecutive sampling cycles), the system will still output the maximum value for protection, but at the same time it will generate a self-test alarm signal for thermal model mismatch or sensor failure, prompting maintenance personnel to check the oil temperature probe.

[0078] like Figure 5As shown, the system calculates and monitors the difference between the two-thread results in real time. The solid peak areas in the figure represent the change in the absolute value of the temperature difference calculated by the two models over time. The horizontal dashed line represents the preset temperature difference alarm threshold. When the peak height of the shaded area exceeds this horizontal dashed line, the system not only performs maximum value arbitration but also simultaneously triggers a self-check alarm for thermal model mismatch or potential sensor failure, thereby verifying the robustness of the model operation.

[0079] S105. Based on the final hot spot temperature, the remaining insulation life is calculated in real time and an emergency load reduction or trip control signal is generated to prevent equipment damage.

[0080] The final hotspot temperature obtained from the arbitration in step S104 The system queries the pre-stored insulation life loss curve in real time. The processor calculates the remaining allowable operating time of the transformer under the current temperature and load trends.

[0081] The specific deduction logic for the remaining allowed running time is as follows:

[0082] The processor first establishes a virtual forecast timeline and assumes future load factors. The current level will be maintained. Subsequently, the processor uses the thermal differential equation from step S103 to perform rapid forward iterative calculations to solve for future time steps. Hotspot temperature prediction When the predicted value When the insulation material first reaches its critical temperature, for example, 105°C for Class A insulation or 140°C under short-term overload conditions, the time difference between that moment and the current moment is calculated as the remaining allowable operating time. Here, short-term overload conditions refer to abnormal operating conditions where the overload duration is less than 30 minutes.

[0083] If the calculated remaining allowable running time is less than the protection action threshold The system will send an emergency load shedding or tripping request to the dispatching terminal via the communication bus. In this embodiment, it is recommended that the protection action threshold be set to 5 to 10 minutes to allow the dispatcher sufficient reaction margin and avoid irreversible thermal breakdown damage to the transformer.

[0084] An embodiment of a dynamic evaluation system for transformer overload capacity:

[0085] On the other hand, the present invention also provides a dynamic evaluation system for transformer overload capacity. For example... Figure 6 As shown, a transformer overload capacity dynamic evaluation system includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a transformer overload capacity dynamic evaluation method according to the first aspect of the present invention.

[0086] A dynamic evaluation system for transformer overload capacity also includes other components well known to those skilled in the art, such as communication interfaces. Their settings and functions are known in the art and will not be described in detail here.

[0087] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device. Any application or module described in this invention can be implemented using computer-readable / executable instructions stored or otherwise maintained by such a computer-readable medium.

Claims

1. A method for dynamic assessment of overload capability of a transformer, characterized in that, The method includes: Real-time load data of the transformer is acquired, and the load data is analyzed based on a time-series sliding window to extract load step characteristics; Run a finite state machine model, which includes a steady-state coupled mode, an inertial decoupled mode, and a linear recovery mode. When a load step characteristic is detected, switch the finite state machine from the steady-state coupled mode to the inertial decoupled mode. The reference oil temperature is set based on the current operating mode of the finite state machine and the thermal differential equation. The thermal differential equation is used to characterize the functional relationship between the rate of change of the winding hot spot temperature with time, the load factor, and the reference oil temperature. In steady-state coupling mode, the measured top oil temperature of the transformer is used as the reference oil temperature; In inertial decoupling mode, the input correlation of the measured top oil temperature is cut off, and the adiabatic extrapolation oil temperature or the oil temperature locked before load impact is used as the reference oil temperature. Solving the thermal differential equation yields the winding hot spot temperature; Steady-state coupling mode: When no impact load is detected, i.e., the load is stable or changes slowly, the system determines that the oil flow circulation has been fully established; Inertial decoupling mode: When an impact load is detected, due to the inertia of fluid dynamics causing the heat dissipation system to be unable to respond immediately to the surge in heat generation, the system immediately disconnects the input association of the real-time oil temperature sensor and enters the decoupling calculation state; Linear recovery mode: When the duration of the inertial decoupling mode reaches the preset fluid hysteresis time, the system enters the recovery period; The method for obtaining the adiabatic simulation oil temperature includes: based on the load factor at the current moment and the preset initial temperature rise rate constant under the rated current, the measured top oil temperature or the simulated oil temperature at the moment before the impact occurs is accumulated and calculated. In linear recovery mode, the method for obtaining the reference oil temperature includes: performing a time-varying weighted average of the adiabatic simulation oil temperature or the pre-impact locked oil temperature and the measured top oil temperature, wherein the weighting coefficient changes linearly with the increase of recovery time until it is completely switched to the measured top oil temperature.

2. The method of claim 1, wherein, Extracting load step features includes: The difference between the first and last data points of the sliding window is determined to be greater than a preset step threshold. And determine that the percentage of points in the sliding window where the load data shows a monotonically increasing trend exceeds the preset monotonicity threshold.

3. The method of claim 1, wherein, The sustaining time of the finite state machine in inertial decoupling mode depends on the cooling method of the transformer; For natural oil circulation cooling, the duration is set to 15 to 20 minutes. For forced oil circulation cooling, the duration is set to 3 to 5 minutes.

4. The method of claim 1, wherein, It also includes running a standard thermal circuit calculation process in parallel, which always uses the measured top oil temperature as the reference oil temperature to calculate the standard winding hot spot temperature; The hot spot temperature of the winding calculated by the finite state machine model is compared with the hot spot temperature of the standard winding, and the larger value is selected as the final evaluation result.

5. The method of claim 4, wherein, The difference between the winding hot spot temperature obtained from the finite state machine model and the standard thermal circuit calculation process is calculated. If the difference continues to exceed the preset temperature difference alarm threshold, an alarm signal for thermal model mismatch or sensor failure is generated.

6. The method of claim 4, wherein, It also includes querying pre-stored insulation material aging curves based on the final evaluation results to determine the life loss rate at the current temperature and extrapolate the remaining allowable operating time of the transformer.

7. The method for dynamic evaluation of transformer overload capacity according to claim 6, characterized in that, When the remaining allowable running time is less than the preset protection action time threshold, an emergency load reduction or trip command is sent to the scheduling system.

8. A dynamic evaluation system for transformer overload capacity, characterized in that, It includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, it implements the dynamic evaluation method for transformer overload capacity as described in any one of claims 1 to 7.

Images