A transformer overheat protection method and system

Abstract

The present application provides a transformer overheating protection method and system, the method comprising: collecting data collected during transformer operation; obtaining a cooling activation index by analyzing the time series data change rate of the cooling system operating power, preprocessing, and forming a standard input tensor; calculating a unit heat rise efficiency index based on the standard input tensor, obtaining a dynamic residual index based on the deviation of the unit heat rise efficiency index and its historical mean value, and generating a cooling state label; inputting the standard input tensor and cooling state label into a pre-constructed temperature prediction model, generating a hotspot temperature prediction value of a future period by integrating the cooling state label into the backbone network structure; and based on the hotspot temperature prediction value of the future period and the cooling state label, matching with a preset thermal risk classification logic table, and feeding back the execution result to optimize the system closed loop.

Description

Technical Field

[0001] This invention belongs to the field of transformer overheat protection, and particularly relates to a transformer overheat protection method and system. Background Technology

[0002] With the continuous growth of power system load and the increasing instability of new energy integration, the safe and stable operation of transformers, as core equipment in power transmission and distribution systems, is receiving increasing attention. In particular, transformers generate a large amount of heat during operation; if this heat is not released in a timely and effective manner, it will lead to increased temperatures in the windings, core, and insulation system, ultimately causing insulation aging, thermal breakdown, and even serious power accidents. Therefore, transformer overheat protection technology, as a crucial component of ensuring power grid safety, has become a key focus for equipment manufacturers and operation and maintenance units.

[0003] Existing transformer overheat protection methods can be broadly categorized into two types: one is a passive protection method based on temperature threshold judgment, typically using temperature sensors installed near the tank or windings to trigger an alarm or trip command when the measured temperature exceeds a set threshold; the other is a predictive method based on thermal equivalent models to estimate hot spots, commonly found in IEC standard thermal modeling methods. This type of method calculates the winding hot spot temperature using input variables such as load current and oil temperature, thereby assessing the transformer's thermal state. However, these methods have significant drawbacks: First, the limited sensor placement and the inability to directly measure some hot spot areas lead to inaccurate hot spot temperature estimates; second, traditional thermal models rely on static parameters and empirical coefficients, making it difficult to adapt to the dynamic operating environment of transformers, such as external climate changes and cooling equipment performance degradation; third, these methods often rely solely on whether the temperature exceeds the limit, lacking the ability to distinguish the cause of the temperature rise, and cannot differentiate between abnormal load, cooling degradation, or internal structural deterioration, making them prone to misjudgment or missed judgment. Furthermore, current systems typically lack foresight, meaning they cannot predict when the temperature will exceed a threshold and can only take delayed responses after the temperature has already exceeded the limit, thus losing a valuable window of opportunity for prevention.

[0004] More complexly, in actual operation, the temperature rise of transformers is often affected by non-electrical factors such as cooling system degradation, localized load concentration, and oil circuit blockage. These factors can lead to hot spot accumulation or even thermal runaway even when the electrical load is within the normal range. Current overheat protection systems cannot effectively identify and handle these latent overheat risks, often relying on manual experience for judgment, which is inefficient and unreliable. Therefore, there is an urgent need for a comprehensive system that can dynamically identify the status of the cooling system, accurately predict the winding hot spot temperature, analyze the causes of overheating, and promptly output effective protection strategies, thereby achieving a leap from passive response to active identification and intelligent early warning. Summary of the Invention

[0005] The purpose of this invention is to propose a transformer overheat protection method and system, which effectively overcomes the core problems of existing technologies, such as passive monitoring, high false alarm rate, lack of interpretability and intervention capability. It is applicable to modern smart grid scenarios with high requirements for transformer operating status.

[0006] To achieve the above objectives, a transformer overheat protection method is provided in a first aspect of the present invention, the method comprising the following steps:

[0007] Data collected during transformer operation includes top oil temperature, ambient temperature, load power, cooling system operating power, and a three-minute temperature difference sequence. A cooling activation index is obtained by performing time-series data change rate analysis on the cooling system operating power. This data, combined with the top oil temperature, ambient temperature, load power, cooling system operating power, three-minute temperature difference sequence, and cooling activation index, is preprocessed to form a standard input tensor. The cooling activation index reflects the current state of the cooling system.

[0008] The unit heat rise efficiency index is calculated based on the standard input tensor. A dynamic residual index is obtained based on the deviation of the unit heat rise efficiency index and its historical mean. Combined with the cooling activation index, a cooling status label is generated. The cooling status label is used to characterize the degree of degradation of the cooling system.

[0009] The standard input tensor and cooling state label are input into the pre-built temperature prediction model. By integrating the cooling state label into the backbone network structure, the prediction path is dynamically adjusted to generate hotspot temperature prediction values ​​for future periods.

[0010] Based on the predicted hotspot temperature and cooling status labels for the future period, the risk level is obtained and a protection strategy is generated by matching it with a preset thermal risk classification logic table. The protection strategy includes cooling commands, load adjustment and alarm behavior.

[0011] The protection strategy is executed by sending control commands to the cooling system and load scheduling module, and the execution results are fed back to optimize the system closed loop.

[0012] Furthermore, the preprocessing includes: performing time alignment and noise reduction on the multi-frequency sampled data, and generating a short-term temperature difference sequence through a sliding window. The temperature difference sequence of the past three minutes is used to reflect the dynamic offset trend between oil temperature and ambient temperature.

[0013] Furthermore, the calculation of the unit heat rise efficiency index based on the standard input tensor specifically includes:

[0014] The oil top layer temperature, ambient temperature, and load power are obtained, and performance index analysis is performed to obtain the unit heat rise efficiency index, which is used to measure the degree of temperature rise of the system under unit load.

[0015] Furthermore, the cooling status label includes normal cooling status, slight degradation, and severe degradation.

[0016] Furthermore, the structure of the temperature prediction model includes:

[0017] The backbone modeling layer processes the standard input tensor with a lightweight recursive network to learn the underlying nonlinear dynamic laws;

[0018] The cooling state adjustment layer integrates the cooling state label into the backbone network structure and dynamically adjusts the prediction path, enabling the model to adaptively select different internal prediction structures or adjust prediction weights under different cooling states.

[0019] The prediction function of the temperature prediction model is expressed as:

[0020]

[0021] in, It predicts the winding hot spot temperature δ minutes from now; f θ (X t This is the baseline output of the temperature prediction model under default cooling conditions, with the input being the standard input tensor; It is the cooling state adjustment coefficient, when the cooling state label S c When (t) = 0, take the value 0, S c When (t) = 2, it takes the value 1, indicating the residual adjustment strength; R(X) t ) is a residual network used to represent the temperature rise offset modeling during cooling degradation.

[0022] Furthermore, the loss function L of the temperature prediction model is expressed as:

[0023]

[0024] Among them, T hotspot (t+δ) represents the actual future hot spot temperature, which is derived from the field winding hot spot estimation system or the IEEE thermal model. The cooling state response regularization term measures the model's prediction sensitivity to changes in the cooling state label. The main loss is calculated using mean squared error to measure prediction bias; λ is the regularization weighting parameter.

[0025] Furthermore, the predicted hotspot temperature and cooling status label based on the future time period are matched with a preset thermal risk classification logic table to obtain the risk level and generate a protection strategy, specifically including:

[0026] Obtain the safe upper limit temperature for the current equipment model, and calculate the over-limit ratio of the predicted temperature relative to the threshold based on the predicted hotspot temperature for future periods and the safe upper limit temperature;

[0027] Based on the over-limit ratio of the predicted temperature relative threshold and the cooling status label, the risk level is obtained by matching with the preset thermal risk classification logic table, including no risk, load impact risk, cooling degradation risk and coupling risk.

[0028] A protection strategy is generated based on the risk level.

[0029] Furthermore, the execution of the protection strategy involves sending control commands to the cooling system and load scheduling module, and feeding back the execution results to optimize the system closed loop, specifically including:

[0030] The protection strategy is obtained and mapped to a device control signal through a mapping function; if the response is successful, the device control signal is executed; if no confirmation feedback is received from the executing device within the set confirmation window after the device control signal is sent, the system enters the protection rollback process and triggers the default action.

[0031] A second aspect of the present invention provides a transformer overheat protection system, the system comprising:

[0032] The data acquisition module is used to collect data during transformer operation, including top oil temperature, ambient temperature, load power, cooling system operating power, and the temperature difference sequence over the past three minutes. By performing time-series data change rate analysis on the cooling system operating power, a cooling activation index is obtained. This index, combined with the top oil temperature, ambient temperature, load power, cooling system operating power, the temperature difference sequence over the past three minutes, and the cooling activation index, undergoes preprocessing to form a standard input tensor. The cooling activation index reflects the current state of the cooling system.

[0033] The cooling status identification module is used to calculate the unit heat rise efficiency index based on the standard input tensor, obtain the dynamic residual index based on the deviation of the unit heat rise efficiency index and its historical mean, and generate a cooling status label by combining the cooling activation index. The cooling status label is used to characterize the degree of degradation of the cooling system.

[0034] The prediction module is used to input the standard input tensor and cooling state label into the pre-built temperature prediction model. By integrating the cooling state label into the backbone network structure, the prediction path is dynamically adjusted to generate hotspot temperature prediction values ​​for future periods.

[0035] The strategy generation module is used to obtain the risk level and generate a protection strategy based on the predicted hot spot temperature value and cooling status label of the future time period by matching it with a preset thermal risk classification logic table. The protection strategy includes cooling instructions, load adjustment and alarm behavior.

[0036] The execution module is used to execute the protection strategy, send control commands to the cooling system and load scheduling module, and provide feedback on the execution results to optimize the system closed loop.

[0037] The beneficial technical effects of the present invention are at least as follows:

[0038] To address the shortcomings of existing overheat protection methods, such as low accuracy in hotspot temperature prediction, difficulty in identifying cooling system anomalies, unexplained causes of overheating, and lack of proactive protection capabilities, this invention proposes a transformer overheat protection system based on a multi-dimensional state modeling and thermal behavior fusion reasoning mechanism. This system aims for unified thermal risk identification and designs multiple synergistic modules capable of dynamically modeling the changing trends of heat dissipation performance. Combined with equipment operating parameters, it accurately infers the temperature rise behavior of winding hotspots, thereby identifying potential thermal risks in advance. This invention establishes a key intermediate variable reflecting changes in cooling capacity and integrates it into the hotspot temperature prediction mechanism, achieving the classification and identification of temperature rise causes and dynamic correction of hotspot prediction results. The system not only provides early warning before temperatures exceed limits but also identifies the main source of risk, whether it is cooling degradation or load fluctuations, thus providing a scientific basis for subsequent protection actions (such as cooling scheduling, load reallocation, or fault alarms). The design of this invention ensures a logical closed loop in the technical path, enabling prediction, judgment, and response to be coupled together. This effectively overcomes the core problems of existing technologies, such as passive monitoring, high false alarm rate, and lack of interpretability and intervention capabilities. It is suitable for modern smart grid scenarios with high requirements for transformer operating status. Attached Figure Description

[0039] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0040] Figure 1 This is a flowchart of a transformer overheat protection method according to the present invention. Detailed Implementation

[0041] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0042] like Figure 1 As shown in the figure, an embodiment of the present invention provides a transformer overheat protection method, the method comprising:

[0043] S1. Collect data collected during transformer operation, including top oil temperature, ambient temperature, load power, cooling system operating power, and temperature difference sequence over the past three minutes; perform time series data change rate analysis on the cooling system operating power to obtain the cooling activation index; combine the top oil temperature, ambient temperature, load power, cooling system operating power, temperature difference sequence over the past three minutes, and cooling activation index, perform preprocessing, and construct a standard input tensor; the cooling activation index is used to reflect the current state of the cooling system.

[0044] Specifically, the goal of this step is to construct a unified, structured input tensor X for transformer operation monitoring. t It serves as the underlying input structure for modeling and decision-making in the entire overheat protection system.

[0045] The data collected during transformer operation includes: top oil temperature T oil (t), ambient temperature T env (t), load power P(t), cooling system operating power C(t), temperature difference sequence ΔT over the past three minutes t-k (k = 1, 2, 3) and cooling response indicators Since different sensors have different sampling frequencies, such as oil temperature every 1 minute and ambient temperature every 5 minutes, linear interpolation and moving average are required for time alignment and noise reduction.

[0046] The short-term temperature difference characteristic sequence is calculated in the following way, forming the core input channel for the thermal inertia trend:

[0047] ΔT t-k =T oil (tk)-T env (tk), k = 1, 2, 3

[0048] This variable differs from the common first-order difference. It more directly expresses the thermal offset state of the transformer under environmental conditions, and can avoid model instability caused by the amplification of small temperature measurement errors. For example, during a certain operation, the ambient temperature is 30℃, while the oil temperature rises from 54℃ to 58℃. At this time, the continuous increase of ΔT is a better indicator of the system being in a thermal accumulation trend than the temperature difference.

[0049] Meanwhile, this step introduces an innovative input variable for modeling cooling dynamics—the cooling activation index. Defined as:

[0050]

[0051] This variable quantifies the relative intervention level of the cooling system at the current moment, i.e., whether the cooling equipment has just been activated, increased its power, or is in stable operation at a certain point in time. For example, in a real-world scenario, when the transformer load suddenly increases, the oil temperature rises, and the system activates the backup fan, causing C(t) to experience a step increase. A value significantly greater than 0 indicates that cooling intervention has occurred, and this characteristic is highly explanatory for judging whether the cooling response is timely.

[0052] Finally, the constructed standard input tensor is:

[0053]

[0054] This tensor is not merely a simple concatenation of variables, but a feature combination optimized through thermophysical structural design. All variables have uniform units (temperature in °C, power in kW), consistent sampling time steps, and compatibility with different transformer models. Through this structural input method, this invention not only solves the problem of heterogeneous data sources, but also explicitly models the cooling responsiveness of the heat capacity effect during transformer operation as an input feature—a part neglected in most existing AI models. It provides a consistent, physically interpretable, and engineering-deployable data foundation for all subsequent modules of the patent.

[0055] S2. Calculate the unit heat rise efficiency index based on the standard input tensor, obtain the dynamic residual index based on the deviation of the unit heat rise efficiency index and its historical mean, and generate a cooling status label by combining the cooling activation index. The cooling status label is used to characterize the degree of degradation of the cooling system.

[0056] Specifically, this step is based on the standard input tensor X constructed in the previous step. tThe aim is to extract and quantify the effectiveness of the transformer's current cooling capacity, thereby assessing potential non-load-driven overheating risks. Traditional protection systems typically only focus on whether the temperature exceeds a threshold, neglecting a crucial fact: under the same load, decreased cooling performance can lead to faster temperature rises or even thermal runaway, phenomena that are quite subtle in oil temperature data. Therefore, this step aims not only to identify the temperature rise state but also to construct a propagable, interpretable, and adjustable cooling state label S. c (t) to serve subsequent prediction and response decisions.

[0057] From X in step one t In this invention, three key variables are selected as the core inputs for this step: oil temperature T. oil (t), ambient temperature T env Given load power P(t), construct the unit heat rise efficiency index η. t It is used to measure the degree of temperature rise exhibited by a system under a unit load. Its basic definition is:

[0058]

[0059] Among them, T oil (t) is the temperature of the top layer of the fuel tank, which is collected in real time by a temperature sensor; T env (t) is the ambient temperature, measured by an outdoor monitoring point; P(t) is the current transformer load power, in kW; ∈ is a minimum constant, usually taken as 0.1, to prevent the denominator from being zero under low load.

[0060] This indicator reflects the temperature rise effect per unit load. If the load remains constant or changes slowly, η... t A continuous increase indicates that the cooling system may be experiencing a decline in efficiency or a slow response.

[0061] But due to η t As an instantaneous indicator, it is greatly affected by external disturbances (such as sudden changes in ambient temperature, load spikes, etc.) and is difficult to use directly for system judgment. Therefore, this invention introduces a trend residual control mechanism to construct a dynamic residual indicator δ. t :

[0062]

[0063] Where, μ t =mean(η) t-w:t-1 ) represents the moving average over the past w time points (e.g., 15 minutes); The time-varying rate of thermal efficiency (which can be expressed as η) t -η t-1(Approximately), reflecting its growth trend; λ is the trend adjustment coefficient, which determines the sensitivity of the control system to accelerated temperature rise, and is usually taken as 0.5 to 1.0.

[0064] By comparing the current efficiency with its historical average and upward trend, δ t It can more stably and proactively identify cooling performance degradation.

[0065] To enhance the physical explanation capability, this invention introduces the cooling activation index constructed in step one. This indicates whether the cooling system is currently activated or enhanced. When... Very low and δ t A significantly positive value indicates that the cooling system is not responding; when Very high and δ t A positive value indicates that the cooling system is responding but ineffective. Both scenarios essentially indicate cooling degradation, but with different causes. This ability to identify these issues will be helpful for downstream modeling. θ Provide clear guidance.

[0066] Finally, this step generates the cooling status label S. c (t), with the following possible values:

[0067] S c (t) = 0: Cooling is normal, δ t Approaching zero Stablize;

[0068] S c (t) = 1: slight degradation, δ t There is an offset. There was a response, but it was insufficient;

[0069] S c (t) = 2: Severe degradation, δ t Continued significant positive skewness Invalid or no response.

[0070] This tag will be directly passed to the hotspot prediction model f in the next step. θ This is used as part of the model input for conditional modeling. Compared to traditional methods, this step does not rely on manually set temperature rise thresholds, but instead uses structural variables η. t Trend residual δ t With response indicators By modeling the causal relationships between these factors, the system truly achieves the organic integration of dynamic identification, classification interpretation, and data-driven control of cooling states. This mechanism also enables the system to have adaptive interpretation capabilities, allowing for explicit modeling of changes in cooling mechanisms in thermal behavior, supporting subsequent predictive logic and closed-loop control of protection response strategies.

[0071] S3. Input the standard input tensor and cooling state label into the pre-built temperature prediction model. By integrating the cooling state label into the backbone network structure, dynamically adjust the prediction path to generate hotspot temperature prediction values ​​for future periods.

[0072] Specifically, this step uses X t and S c Using (t) as input, construct a cooling-state-guided temperature prediction model f. θ The output target is This involves predicting the hot spot temperature of the winding after δ minutes. The core difficulty of this prediction task lies in the fact that temperature changes are affected not only by the load but also by cooling performance (which is usually not directly measurable). If the model cannot perceive the differences in cooling status, it may mistakenly assume that all temperature increases are caused by the load, leading to misjudgments or even missed judgments. However, in step two of this invention, the cooling status label S has been explicitly marked. c (t), which provides a key support for building a state-aware prediction model.

[0073] To achieve a state-aware prediction mechanism, this invention proposes a conditionally adjustable prediction network structure f θ Its overall architecture consists of two parts: (a) a backbone modeling layer, which uses a lightweight recurrent network (such as GRU or temporal convolutional module) to process the thermal behavior sequence X. t (a) Learning potential nonlinear dynamic laws; (b) Cooling state adjustment layer, which will adjust the S c (t) Integrated into the backbone network structure, the prediction path is dynamically adjusted, allowing the model to adaptively select different internal prediction structures or adjust prediction weights under different cooling states. For example, when S c When (t) = 0, the model should trust that the cooling system is working effectively; while when S c When (t) = 2, the model should pay more attention to the temperature rise trend and be biased towards early warning.

[0074] To specifically model this structural adjustment effect, this invention introduces a state-guided gating structure, such that the prediction function f θ In mathematics, this is represented as follows:

[0075]

[0076] in, It predicts the winding hot spot temperature δ minutes from now; f θ (X t The output is the baseline output of the model under the default cooling state, with the input being the tensor X from step one. t , including T oil (t), T env (t), P(t), C(t), ΔT t-k, wait; It is the cooling state adjustment coefficient, when S c When (t) = 0, take the value 0, S c When (t) = 2, it takes the value 1, indicating the residual adjustment strength; R(X) t ) is a residual network used to represent the temperature rise offset modeling during cooling degradation.

[0077] To improve the stability of the predicted output and prevent the model from blindly relying on historical temperature trends while ignoring cooling signals, this invention introduces an innovative loss regularization term, called the state variation consistency constraint, to encourage the model to be more sensitive to temperature predictions when cooling conditions change drastically and more conservative when the state is stable. Its loss function form is as follows:

[0078]

[0079] Among them, T hotspot (t+δ) represents the actual future hotspot temperature, derived from the field winding hotspot estimation system or the IEEE thermal model; the first term is the main loss, with mean square error used to measure the prediction deviation; the second term is the cooling state response regularization term, measuring the model prediction's impact on S. c (t) Sensitivity to change; λ is the regularization weight parameter, usually set to 0.05 to 0.2.

[0080] This structure enables the model to possess a thermophysical regulation channel, thereby achieving sensitivity and interpretability of temperature prediction behavior to cooling status. For example, in a real-world engineering scenario, this invention discovered two oil temperature rise events with similar load changes. However, one was due to the cooling system not restarting promptly after a short power outage, while the other was due to an increase in external ambient temperature during normal operation. Traditional black-box models cannot distinguish between these two scenarios, but this patented method can achieve this through S... c By changing the label of (t), the first type of heat rise is correctly identified as cooling degradation type, thus making a judgment on the rise of hot spot temperature earlier and triggering the early warning mechanism in advance.

[0081] S4. Based on the predicted hotspot temperature and cooling status label for the future time period, the risk level is obtained by matching it with a preset thermal risk classification logic table, and a protection strategy is generated. The protection strategy includes cooling commands, load adjustment, and alarm behavior.

[0082] Specifically, the goal of this step is to classify and determine the system's thermal risks and generate response strategies based on hotspot temperature prediction results and cooling status labels. This step will... and S c The two structural variables (t) are input into the rule judgment system to form an interpretable risk level output R(t) and an actual executable protection strategy A(t).

[0083] Input includes: The model output from step three represents the system's predicted future hotspot temperature; S c (t): Cooling status label from step two, indicating the current cooling capacity status, with values ​​of 0 (normal), 1 (slight degradation), and 2 (severe degradation); T safe The upper limit of safe temperature is determined by the equipment model; for example, it is 110°C for oil-immersed transformers.

[0084] To determine the risk level, the ratio of the predicted temperature to the relative threshold exceeding the limit, ρ(t), is first calculated:

[0085]

[0086] Where: ρ(t) represents the proportion by which the predicted temperature exceeds the safe upper limit; if ρ(t) ≤ 0, it indicates no risk of over-temperature; if ρ(t) > 0, it indicates a risk of predicted temperature exceeding the limit, requiring further investigation to determine its source. This is based on the predicted temperature rise intensity (i.e., the magnitude of ρ(t)) and the cooling state S. c In this invention, a rule-driven thermal risk classification logic table is designed to output the risk level R(t), whose values ​​have the following meanings:

[0087] R(t) = 0: No risk, ρ(t) ≤ 0;

[0088] R(t) = 1: Load shock type risk, ρ(t) > 0 and S c (t) = 0;

[0089] R(t) = 2: cooling degradation risk, ρ(t) > 0 and S c (t) = 2;

[0090] R(t) = 3: Coupled risk, ρ(t) > 0 and S c (t) = 1.

[0091] The advantages of this structure are that the judgment process is reproducible, the logic is clear, and it is easy to adjust parameters or make manual revisions, making it particularly suitable for the deployment needs of power system engineering.

[0092] Based on this, the system outputs a response strategy A(t) that can be recognized by the control system, which has a structure of a triple:

[0093] A(t) = [Cooling command, load strategy, alarm behavior]

[0094] For example:

[0095] If R(t) = 0, then A(t) = [cooling maintenance, normal load, no alarm];

[0096] If R(t) = 1, A(t) = [Auxiliary cooling activated, load reduced, operation monitoring alert];

[0097] If R(t) = 2, then A(t) = [forced fan start, load maintenance, alarm + manual maintenance];

[0098] If R(t) = 3, A(t) = [cooling enhancement, load reduction, record and alarm].

[0099] These actions are not merely suggestive textual descriptions, but rather configurable output commands for actual control systems (such as via remote terminals, PLC controllers, and SCADA systems), possessing physical executable capability. Ultimately, R(t) and A(t), as the outputs of this step, will be fed into step five for actual control response, completing the crucial transition from intelligent prediction to engineering execution and ensuring that the entire patented system forms a prediction-judgment-response closed loop.

[0100] S5. Execute the protection strategy, send control commands to the cooling system and load scheduling module, and provide feedback on the execution results to optimize the system closed loop.

[0101] Specifically, the goal of this step is to translate the response strategy A(t) output in the previous step into actual control actions U(t), forming an executable and feedback-enabled protection response mechanism, thereby achieving a closed-loop system from prediction and judgment to control execution.

[0102] The input is A(t), which is a ternary control strategy from step four, defined as: A(t) = [cooling control command, load adjustment strategy, alarm behavior];

[0103] Each element is a structured action selected from a finite set of actions, for example:

[0104] Cooling commands: ∈{Maintain cooling, increase fan speed, start oil pump, switch to standby cooling};

[0105] Load strategy: ∈{normal operation, short-term load limit, fixed load limit of 10%, tiered load reduction};

[0106] Alarm behaviors: ∈{No alarm, Log record, Manual prompt, Forced alarm, Emergency shutdown}.

[0107] This step constructs a protection execution function that maps the strategy structure A(t) to the device control signal U(t):

[0108] U(t)=Ψ(A(t),Ω dev )

[0109] Where U(t) is the control layer execution command (such as starting the wind turbine, issuing a load limit request, or triggering a SCADA alarm); A(t) is the structural response strategy input; Ω dev It represents the execution capability and constraint parameters of the device layer (such as a certain type of fan cannot be repeatedly started and stopped, the maximum load reduction limit is 20%, etc.); Ψ(·) is a mapping function that, based on the rule table or control logic, converts the policy action into hardware instructions to ensure that each type of action can be implemented and executed.

[0110] To enhance the robustness and practicality of the control, this step introduces a response acknowledgment mechanism. If U(t) is issued and an acknowledgment window τ is set... ack If no confirmation feedback is received from the executing device, the system enters the protection rollback process and triggers the default action U. fail (t):

[0111]

[0112] Where, τ ack The response confirmation time limit is typically 10–30 seconds; U fail (t) can be set to force alarm + activate cooling backup channel + limit load by 20% to ensure system stability.

[0113] In practical applications, the system can distribute actions through the following interface:

[0114] The control layer output U(t) can be connected to a DCS control module, SCADA system, intelligent PLC or IEC61850 protocol interface;

[0115] The system timestamps and records each action and feedback, serving as the data foundation for subsequent operation and maintenance audits and fault analysis.

[0116] Finally, the control execution result U(t) will be synchronously written to the protection system log, completing the closed-loop response logic of the present invention and realizing a truly intelligent and executable overheat protection system.

[0117] This invention also provides a transformer overheat protection system, the system comprising:

[0118] The data acquisition module is used to collect data during transformer operation, including top oil temperature, ambient temperature, load power, cooling system operating power, and the temperature difference sequence over the past three minutes. By performing time-series data change rate analysis on the cooling system operating power, a cooling activation index is obtained. This index, combined with the top oil temperature, ambient temperature, load power, cooling system operating power, the temperature difference sequence over the past three minutes, and the cooling activation index, undergoes preprocessing to form a standard input tensor. The cooling activation index reflects the current state of the cooling system.

[0119] The cooling status identification module is used to calculate the unit heat rise efficiency index based on the standard input tensor, obtain the dynamic residual index based on the deviation of the unit heat rise efficiency index and its historical mean, and generate a cooling status label by combining the cooling activation index. The cooling status label is used to characterize the degree of degradation of the cooling system.

[0120] The prediction module is used to input the standard input tensor and cooling state label into the pre-built temperature prediction model. By integrating the cooling state label into the backbone network structure, the prediction path is dynamically adjusted to generate hotspot temperature prediction values ​​for future periods.

[0121] The strategy generation module is used to obtain the risk level and generate a protection strategy based on the predicted hot spot temperature value and cooling status label of the future time period by matching it with a preset thermal risk classification logic table. The protection strategy includes cooling instructions, load adjustment and alarm behavior.

[0122] The execution module is used to execute the protection strategy, send control commands to the cooling system and load scheduling module, and provide feedback on the execution results to optimize the system closed loop.

[0123] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0124] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of apparatuses or units may be electrical, mechanical, or other forms.

[0125] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0126] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for overheat protection of a transformer, characterized in that, The method includes: Data collected during transformer operation includes top oil temperature, ambient temperature, load power, cooling system operating power, and a three-minute temperature difference sequence. A cooling activation index is obtained by performing time-series data change rate analysis on the cooling system operating power. This data, combined with the top oil temperature, ambient temperature, load power, cooling system operating power, three-minute temperature difference sequence, and cooling activation index, is preprocessed to form a standard input tensor. The cooling activation index reflects the current state of the cooling system. The unit heat rise efficiency index is calculated based on the standard input tensor. A dynamic residual index is obtained based on the deviation of the unit heat rise efficiency index and its historical mean. Combined with the cooling activation index, a cooling status label is generated. The cooling status label is used to characterize the degree of degradation of the cooling system. The standard input tensor and cooling state label are input into the pre-built temperature prediction model. By integrating the cooling state label into the backbone network structure, the prediction path is dynamically adjusted to generate hotspot temperature prediction values ​​for future periods. Based on the predicted hotspot temperature and cooling status labels for the future period, the risk level is obtained and a protection strategy is generated by matching it with a preset thermal risk classification logic table. The protection strategy includes cooling commands, load adjustment and alarm behavior. The protection strategy is executed by sending control commands to the cooling system and load scheduling module, and the execution results are fed back to optimize the system closed loop.

2. The transformer overheat protection method according to claim 1, characterized in that, The preprocessing includes: performing time alignment and noise reduction on the multi-frequency sampled data, and generating a short-term temperature difference sequence through a sliding window. The temperature difference sequence of the past three minutes is used to reflect the dynamic offset trend between oil temperature and ambient temperature.

3. The transformer overheat protection method according to claim 1, characterized in that, The calculation of the unit heat rise efficiency index based on the standard input tensor specifically includes: The oil top layer temperature, ambient temperature, and load power are obtained, and performance index analysis is performed to obtain the unit heat rise efficiency index, which is used to measure the degree of temperature rise of the system under unit load.

4. The transformer overheat protection method according to claim 1, characterized in that, The cooling status labels include normal cooling status, slight degradation, and severe degradation.

5. A transformer overheat protection method according to claim 1, characterized in that, The structure of the temperature prediction model includes: The backbone modeling layer processes the standard input tensor with a lightweight recursive network to learn the underlying nonlinear dynamic laws; The cooling state adjustment layer integrates the cooling state label into the backbone network structure and dynamically adjusts the prediction path, enabling the model to adaptively select different internal prediction structures or adjust prediction weights under different cooling states. The prediction function of the temperature prediction model is expressed as: in, It predicts the winding hot spot temperature δ minutes from now; f θ (X t This is the baseline output of the temperature prediction model under default cooling conditions, with the input being the standard input tensor; It is the cooling state adjustment coefficient, when the cooling state label S c When (t) = 0, take the value 0, S c When (t) = 2, it takes the value 1, indicating the residual adjustment strength; R(X) t ) is a residual network used to represent the temperature rise offset modeling during cooling degradation.

6. A transformer overheat protection method according to claim 5, characterized in that, The loss function L of the temperature prediction model is expressed as: Among them, T hotspot (t+δ) represents the actual future hot spot temperature, which is derived from the field winding hot spot estimation system or the IEEE thermal model. The cooling state response regularization term measures the model's prediction sensitivity to changes in the cooling state label. The main loss is calculated using mean squared error to measure prediction bias; λ is the regularization weighting parameter.

7. A transformer overheat protection method according to claim 1, characterized in that, The predicted hotspot temperature and cooling status labels based on the future time period are matched with a preset thermal risk classification logic table to obtain the risk level and generate a protection strategy, specifically including: Obtain the safe upper limit temperature for the current equipment model, and calculate the over-limit ratio of the predicted temperature relative to the threshold based on the predicted hotspot temperature for future periods and the safe upper limit temperature; Based on the over-limit ratio of the predicted temperature relative threshold and the cooling status label, the risk level is obtained by matching with the preset thermal risk classification logic table, including no risk, load impact risk, cooling degradation risk and coupling risk. A protection strategy is generated based on the risk level.

8. A transformer overheat protection method according to claim 1, characterized in that, The execution of the protection strategy involves sending control commands to the cooling system and load scheduling module, and feeding back the execution results to optimize the system closed loop. Specifically, this includes: The protection strategy is obtained and mapped to a device control signal through a mapping function; if the response is successful, the device control signal is executed; if no confirmation feedback is received from the executing device within the set confirmation window after the device control signal is sent, the system enters the protection rollback process and triggers the default action.

9. A transformer overheat protection system, characterized in that, The system includes: The data acquisition module is used to collect data during transformer operation, including top oil temperature, ambient temperature, load power, cooling system operating power, and the temperature difference sequence over the past three minutes. By performing time-series data change rate analysis on the cooling system operating power, a cooling activation index is obtained. This index, combined with the top oil temperature, ambient temperature, load power, cooling system operating power, the temperature difference sequence over the past three minutes, and the cooling activation index, undergoes preprocessing to form a standard input tensor. The cooling activation index reflects the current state of the cooling system. The cooling status identification module is used to calculate the unit heat rise efficiency index based on the standard input tensor, obtain the dynamic residual index based on the deviation of the unit heat rise efficiency index and its historical mean, and generate a cooling status label by combining the cooling activation index. The cooling status label is used to characterize the degree of degradation of the cooling system. The prediction module is used to input the standard input tensor and cooling state label into the pre-built temperature prediction model. By integrating the cooling state label into the backbone network structure, the prediction path is dynamically adjusted to generate hotspot temperature prediction values ​​for future periods. The strategy generation module is used to obtain the risk level and generate a protection strategy based on the predicted hot spot temperature value and cooling status label of the future time period by matching it with a preset thermal risk classification logic table. The protection strategy includes cooling instructions, load adjustment and alarm behavior. The execution module is used to execute the protection strategy, send control commands to the cooling system and load scheduling module, and provide feedback on the execution results to optimize the system closed loop.

Images