A method and apparatus for dynamic temperature measurement and calibration of heat generating elements

Abstract

The application provides a method and device for dynamic temperature measurement and calibration of a heating element, and relates to the technical field of temperature measurement, which synchronously extracts the measured temperature of a target electromagnetic element, the first temperature of a conductive coil and the second temperature of a magnetic conductive medium in response to a temperature rise monitoring period, and obtains surface radiation of adjacent equipment and a physical distance mapping; based on spatial distance, an external additional temperature rise is analyzed to eliminate measured interference, and a double-material heating reference curve is constructed in combination with a double heat source theory; then, an internal abnormal risk value is calculated through waveform fitting comparison to generate calibration parameters of the inside and outside, and data feedback correction is performed on the measured track and the theoretical reference. External heat crosstalk and internal different heating mechanisms are effectively decoupled, high-fidelity dynamic calibration and tracking of the real temperature of the heating element under complex working conditions are realized.

Description

Technical Field

[0001] This invention relates to the field of temperature measurement technology, specifically to a method and apparatus for dynamic temperature measurement and calibration of heating elements. Background Technology

[0002] With the deep evolution of Industrial Internet of Things (IIoT) and smart grid technologies, refined thermal management of high-power electromagnetic components (such as transformers and reactors) has become a core foundation for ensuring equipment safety and dispatch reliability. In complex and overlapping industrial cabinet environments, the temperature of heat-generating components exhibits highly nonlinear and spatiotemporally coupled characteristics. This requires temperature measurement systems to not only acquire transient values ​​but also possess the quantitative capability to dynamically track the actual thermodynamic evolution trajectory. Existing temperature measurements in such scenarios suffer from the following problems: Traditional technologies tend to directly regard the external surface values ​​acquired by sensors as the body temperature, failing to effectively isolate external heat crosstalk conducted from adjacent devices in the physical space, which can easily lead to environmentally dependent "artificially high" temperature results. When constructing temperature measurement benchmarks, existing solutions approximate the target component as a single homogeneous heat-generating body, ignoring the different properties of the internal conductive coil (current-dominated copper loss) and the magnetic medium (voltage-dominated iron loss) in terms of heat conduction efficiencies, resulting in severe distortion of the theoretical benchmark under dynamic loads. For example, Chinese patent CN117824854A discloses a method for early warning of abnormal temperature rise in transformer bodies. This method improves model compatibility and estimation accuracy to some extent by introducing historical offset criticality to correct the estimated values. However, this approach still relies on purely data-driven statistical empirical fitting and lacks a physical decoupling mechanism for "external environmental crosstalk" and "shifts in the heating law of internal heterogeneous materials under aging" in real physical fields. Such purely open-loop or single-dimensional statistical architectures still exhibit irreversible drift limitations in their dynamic temperature evolution trajectory when facing material degradation or localized heat accumulation in the cabinet during long-term operation. Summary of the Invention

[0003] The purpose of this invention is to provide a method and apparatus for dynamic temperature measurement and calibration of heating elements. It uses multi-source sensing data as input to the core processing center and incorporates spatial physical topological distance and electromagnetic loss models. By spatially damping and filtering external environmental heat and performing phase alignment and waveform fitting on internal dual-material heat conduction, it achieves a shift from single-data estimation to multi-dimensional physical parameter collaborative decision-making, thereby solving the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] A method for dynamic temperature measurement and calibration of heating elements, comprising the following steps:

[0006] S1: In response to the triggering of the temperature rise monitoring cycle, the temperature data acquisition terminal synchronously extracts the measured temperature time series set of the target electromagnetic element, which is the entity representing the heating element, as well as the first temperature data of the conductive coil and the second temperature data of the magnetic medium, which are the internal structures of the target electromagnetic element; at the same time, it extracts the surface radiation temperature set of the surrounding adjacent heating devices and the physical distance mapping table between the devices.

[0007] S2: Import the set of surface radiation temperatures of the surrounding adjacent heating devices and the mapping table of physical distances between the devices into the pre-configured external environment heat filtering module, and parse to generate the environmental additional temperature rise value conducted by the adjacent devices;

[0008] Based on the environmental temperature rise value, an interference removal operation is performed on the measured temperature time series set to generate the true temperature rise trajectory data after deducting external heat interference.

[0009] The first temperature data of the conductive coil and the second temperature data of the magnetic medium are imported into a preset dual-heat-source theoretical curve generation module to jointly construct a dual-material theoretical heating reference curve.

[0010] The actual heating trajectory data and the theoretical heating reference curve of the dual material are sent to the waveform dynamic fitting module to perform waveform shape comparison and calculate the internal abnormal temperature rise risk value.

[0011] S3: Based on the environmental additional temperature rise value and the internal abnormal temperature rise risk value, generate external temperature measurement error compensation parameters for eliminating external temperature measurement interference, and internal temperature measurement reference calibration coefficients for correcting the dual-material theoretical temperature rise reference curve.

[0012] Based on the external temperature measurement error compensation parameter and the internal temperature measurement reference calibration coefficient, data feedback correction is performed on the measured temperature time series set and the dual-material theoretical heating reference curve to generate dynamic temperature evolution trajectory data of the target electromagnetic element after measurement and calibration.

[0013] Based on the dynamic temperature evolution trajectory data, the current effective temperature value of the dynamic temperature evolution trajectory data at the end node of the current temperature rise monitoring cycle, after completing various error deductions and compensations, is extracted and used as the real-time calibration temperature value output of the target electromagnetic component.

[0014] A device for dynamic temperature measurement and calibration of a heating element, the device being used to perform the method for dynamic temperature measurement and calibration of a heating element, comprising:

[0015] The multi-source heterogeneous parameter sensing module is configured to: in response to the triggering of the temperature rise monitoring cycle, synchronously extract the measured temperature time series set of the target electromagnetic element, which is the entity representing the heating element, and the first temperature data of the conductive coil and the second temperature data of the magnetic medium, which are the internal structures of the target electromagnetic element; at the same time, extract the surface radiation temperature set of the surrounding adjacent heating devices and the physical distance mapping table between the devices;

[0016] The spatiotemporal decoupling and feature fitting module is configured to: import the surface radiation temperature set of the surrounding adjacent heating devices and the physical distance mapping table between the devices into a pre-configured external environment heat filtering module to parse and generate the environmental additional temperature rise value conducted by the adjacent devices; perform interference removal operation on the measured temperature time series set based on the environmental additional temperature rise value to generate the true temperature rise trajectory data after deducting external heat interference; import the first temperature data of the conductive coil and the second temperature data of the magnetic medium into a pre-set dual heat source theoretical curve generation module to jointly construct a dual-material theoretical temperature rise reference curve; and send the true temperature rise trajectory data and the dual-material theoretical temperature rise reference curve into the waveform dynamic fitting module to perform waveform shape comparison and calculate the internal abnormal temperature rise risk value.

[0017] The closed-loop calibration and status output module is configured to: generate external temperature measurement error compensation parameters for eliminating external temperature measurement interference and internal temperature measurement reference calibration coefficients for correcting the dual-material theoretical temperature rise reference curve, based on the environmental additional temperature rise value and the internal abnormal temperature rise risk value; perform data feedback correction on the measured temperature time series set and the dual-material theoretical temperature rise reference curve based on the external temperature measurement error compensation parameters and the internal temperature measurement reference calibration coefficients, generating dynamic temperature evolution trajectory data of the target electromagnetic element after measurement and calibration; and extract the current effective temperature value of the dynamic temperature evolution trajectory data at the end node of the current temperature rise monitoring cycle after completing various error deductions and compensations, and output it as the real-time calibration temperature value of the target electromagnetic element.

[0018] Compared with the prior art, the beneficial effects of the present invention are:

[0019] This invention extracts a mapping table between the surface radiation temperature of surrounding devices and the physical distance between devices through steps S1 and S2, and introduces a spatial thermal damping and conduction delay model to analytically generate environmental additional temperature rise values. It quantitatively eliminates interference from spurious environmental baking heat attached to the target's outer shell, improving the anti-interference stability of surface temperature measurement results in densely deployed environments.

[0020] In steps S1 and S2, this invention reduces the sequential current and voltage during operation to a first temperature of the conductive coil and a second temperature of the magnetic medium, respectively, and performs weighted compensation and phase alignment to construct a theoretical temperature rise reference curve for the dual materials. It decouples the asynchronous heat transfer aging of copper and iron losses at a deeper physical mechanism level, and captures insulation degradation characteristics through waveform morphology fitting, thereby quantifying the risk of internal abnormal temperature rise and effectively curbing the distortion of the measurement baseline.

[0021] In step S3 of this invention, based on the analyzed additional temperature rise and internal risk values, external temperature measurement error compensation parameters and internal temperature measurement reference calibration coefficients are calculated and generated respectively. Surface error elimination is performed on the measured trajectory, and dynamic calibration is performed on the theoretical reference amplitude of the dual heat sources. This ensures that the final output is an accurate real-time calibrated temperature value. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the technical principle of the present invention;

[0023] Figure 2 This is a combined view of the technical roadmap of the present invention;

[0024] Figure 3 This is a schematic diagram of the dual-track error decoupling and dynamic fitting calibration of the present invention;

[0025] Figure 4 The feedback linkage and intervention logic diagram provided in the embodiments of the present invention;

[0026] Figure 5 This is a schematic diagram of the numerical verification experiment results for the adaptive feature decoupling algorithm for the insulation aging risk index. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0028] It is understood that the terms “first,” “second,” etc., used in this application may be used herein to describe various elements, but unless otherwise stated, these elements are not limited by these terms. These terms are used only to distinguish one element from another.

[0029] Example 1:

[0030] Please see Figures 1 to 5 The present invention provides a technical solution:

[0031] A method for dynamic temperature measurement and calibration of heating elements includes the following steps:

[0032] In a preferred embodiment, the heating element / electromagnetic element of this embodiment includes, but is not limited to, transformers in industrial power grids.

[0033] Step S1-A: Extract the internal multi-dimensional operating parameters and basic temperature time-series characteristics of the heating element represented by the target electromagnetic component: In response to the triggering of a preset temperature rise monitoring cycle, the state sensing process for the target electromagnetic component is initiated by the temperature data acquisition terminal. The temperature rise monitoring cycle (denoted as...) This represents the fixed time length from the completion of one round of complete data extraction, calculation, and processing to status feedback in this embodiment. In this embodiment, the temperature rise monitoring cycle... The optimal parameter boundaries were defined as a time range of 500ms to 2000ms.

[0034] It should be noted that, in this embodiment, the electrical sensing component refers to a collection of physical hardware modules deployed at the front end of the power supply circuit of the target electromagnetic element for high-precision, non-invasive (or electrically isolated) acquisition of electrical signals. This component includes, but is not limited to: a high-frequency Hall current sensor or current transformer for capturing transient currents in conductive coils, and a voltage transformer or high-impedance voltage divider measurement network for synchronously monitoring the excitation voltage of the magnetic medium. This electrical sensing component aims to step down, isolate, and linearly scale down the high-voltage analog signal in the primary power supply circuit, converting it into a low-voltage analog electrical signal or initial digital sequence for subsequent processing, thereby providing electromagnetic physical quantities for the physical conversion calculation unit to evaluate copper and iron losses.

[0035] The operating parameters of the internal structure of the target electromagnetic component are synchronously extracted using electrical sensing components. Based on preset multi-channel data synchronous acquisition rules, the raw electrical signal sampling sequence output by the electrical sensing components is obtained. Timing channel alignment is performed on the raw electrical signal sampling sequence. A first-order low-pass digital filter operator is invoked based on pre-configured frequency domain cutoff boundary parameters to perform low-pass feature truncation. Its core time-domain mapping recursive logic is as follows: ;in The original electrical signal input value for the current sampling period. This is the output of the smoothed features from the previous cycle. For filtering and smoothing weights. Specifically, extract preset frequency domain cutoff boundary parameters. and system sampling frequency The system sampling frequency This represents the data acquisition rate of the electrical sensing component when performing analog-to-digital conversion on the power grid waveform. In this embodiment, its preferred value is 10000Hz, and the system sampling frequency is configured within the range of [5000Hz, 20000Hz] to adapt to sensors with different high-frequency precisions. This parameter ensures the subsequent filtering and smoothing weights. The acquisition of the filtering and smoothing weights is determined by the following logic: The filtering and smoothing weights are used to perform point-by-point weighted truncation of the sampled sequence to eliminate sensor sampling interference spikes. In this embodiment, the frequency domain cutoff boundary parameter... The highest frequency of the electrical fundamental wave that can be allowed to pass without loss is preferably set in the range of 400Hz to 800Hz. If the frequency domain cutoff boundary parameter is set to deviate from the lower limit of 400Hz, the peak of the actual transient current change may be forcibly flattened, resulting in false negatives in the subsequent heat power calculation. If it is greater than 800Hz, the high-frequency carrier noise of the frequency converter may be mixed in, causing the cutoff operation to fail.

[0036] This embodiment introduces a real-time power supply load characterizing the conductive coil and denoted as . The runtime sequence current data, and the real-time excitation state representing the magnetic medium, are recorded as follows: The runtime timing voltage data is then reconstructed after filtering and denoising to generate runtime timing current data with standard timestamps. With runtime timing voltage data It is then imported into the inter-device data communication channel to perform phase state locking of the acquisition, ensuring the alignment of subsequent multidimensional data under the same time slice.

[0037] This embodiment will use runtime sequence current data. With runtime timing voltage data Import the physical conversion calculation unit to perform a heat power conversion operation based on the electromagnetic loss physical model. Specifically, extract runtime sequence current data. The copper heat dissipation power (watts) is calculated based on the proportional mapping between the square of the current and the static resistance of the coil. This copper heat dissipation power is then compared with the pre-configured static thermal resistance coefficient of the coil. Perform multiplicative dimensionality reduction calculations to generate a first temperature data (degrees Celsius) with unified dimensions, denoted as . Simultaneously, the iron loss heat power and the pre-configured static thermal resistance coefficient of the magnetic core are calculated based on voltage / frequency characteristics. Perform multiplicative dimensionality reduction calculations to generate dimensionally aligned second temperature data, denoted as . .

[0038] The copper heat dissipation power is calculated based on the proportional mapping between the square of the current and the static resistance of the coil. The specific execution logic is as follows: extract the effective value parameter of the current data during operation and perform a square exponentiation operation on it; extract the DC static resistance value of the conductive coil obtained in advance from the equipment nameplate or offline measurement; perform a product operation on the square of the effective value of the current and the static resistance value and calculate the copper heat dissipation power in watts, which characterizes the thermal effect of the coil current.

[0039] The iron loss heat power is calculated based on voltage / frequency characteristics. The specific execution logic is as follows: The timing voltage data is parsed, and the current grid drive frequency and magnetic flux density characteristics are extracted. These characteristics are then imported into a pre-configured Steinmetz electromagnetic loss separation model for dimensionality reduction calculation. Specifically, the material-specific hysteresis loss coefficient, eddy current loss coefficient, and Steinmetz constant are obtained from the standard electromagnetic component material handbook, and a nonlinear multiplication-addition fusion operation is performed: The exponential terms of the frequency and magnetic flux density characteristics, along with the hysteresis loss coefficient, are multiplied together to calculate the first power component (hysteresis loss component); the squared terms of the frequency and magnetic flux density characteristics, along with the eddy current loss coefficient, are multiplied together to calculate the second power component (eddy current loss component); and the first and second power components are linearly summed and aggregated to calculate the iron loss heat power, representing the core excitation loss and measured in watts.

[0040] The static thermal resistance coefficient of the coil With the static thermal resistance coefficient of the magnetic core All values ​​represent the steady-state temperature rise caused by each 1W of power dissipated by the corresponding heating element. In this embodiment, based on the standard electromagnetic component material handbook, the coil's static thermal resistance coefficient... The preferred quantification value is defined as being in the range of 0.15℃ / W to 0.35℃ / W. The standard electromagnetic component material handbook here specifically refers to publicly available industry technical specifications or factory data sheets that record the basic physical properties (including but not limited to resistivity, thermal conductivity, specific heat capacity, and loss characteristic curves) of industrial-grade electromagnetic materials (such as silicon steel sheets, ferrite cores, pure copper wires, etc.) at nominal temperatures.

[0041] Static thermal resistance coefficient of magnetic core This characterizes the surface temperature rise caused by each 1W of electromagnetic power dissipated by the magnetic core material. In this embodiment, the preferred value of the static thermal resistance coefficient of the magnetic core is defined as being in the range of 0.10℃ / W to 0.30℃ / W. This parameter is determined in advance through an offline pure iron loss heat treatment calibration experiment (injecting high-frequency open-circuit voltage and physically blocking the coil current) and is fixedly written into the configuration library of the temperature measurement data access node to ensure the second temperature data. The mapping transformation possesses both numerical closed-loop and physical self-consistency.

[0042] The temperature characteristics of the above isomorphic components, along with the time series set of measured temperatures of the target electromagnetic components acquired through conventional probes, are synchronously reported to the temperature data acquisition terminal.

[0043] Step S1-B: Extract spatial topology information and radiation temperature set of surrounding adjacent devices; trigger the perception of the external environment status in parallel. Specifically, perform the following spatial topology feature extraction actions:

[0044] In response to the rack node temperature measurement and scanning command, the current rack equipment operating status table is parsed to extract the set of operating heat-generating devices within the target rack area. This embodiment introduces the physical radius of thermal radiation interference, which characterizes the effective intervention range of spatial heat conduction. .

[0045] In this preferred embodiment, the physical radius of thermal radiation interference The quantization threshold was set to a physical spatial distance range of 0.5 meters to 1.2 meters. This was based on the physical radius of the thermal radiation interferometry. Perform spatial relative position matching on the set of operating heat-generating devices, filter out adjacent heat-generating devices within the radius, and designate them as key adjacent heat-generating devices for monitoring.

[0046] Establish data communication channels with adjacent heat-generating devices in the surrounding environment, extract real-time infrared temperature measurement data streams from the corresponding devices' casings, and generate a structured set of surface radiation temperatures after standardizing the temperature measurement data format and synchronizing it with timestamps. Furthermore, introduce a heat dissipation attenuation coefficient, representing the intensity of heat dissipation in the airflow environment within the cabinet. In this embodiment, an offline spatial thermal simulation sandbox environment is constructed during the formal online or initial deployment phase of the rack, and the equipment operation data stream under historical full-load extreme heating conditions is injected into the sandbox. The individual adjacent devices in the rack are controlled to simulate continuous full-load heating operation conditions. The temporal fluctuation status of the surface radiation temperature of each node in the space is continuously acquired and monitored through array-type logic probes.

[0047] Extract the heat generation power parameter when adjacent devices are the sole heat source, and dynamically record the extreme values ​​of the thermal radiation amplitude received by the target electromagnetic component's surface. In the calibration simulation, the environmental detection convergence rule is set as follows: when the heat conduction increment on the target electromagnetic component's outer shell surface tends to converge within three consecutive preset polling cycles, and the fluctuation amplitude is lower than the preset physical tolerance (less than 0.1℃), it is determined that the cabinet airflow obstruction steady-state boundary has been reached. Extract the ratio of the initial extreme value of the adjacent heat source's amplitude to the amplitude received by the target component, and combine this with the physical installation spacing parameter to perform a spatial damping gradient calculation. The specific preset spatial thermal damping exponent mapping formula is as follows: ;in Characterizing the extreme values ​​of the received wave amplitude on the surface of the target electromagnetic element. Characterizing the initial extreme values ​​of the amplitudes of adjacent heat sources, This parameter characterizes the physical mounting distance between the target electromagnetic component and adjacent heat sources. Among them, the heat dissipation attenuation coefficient... This is a dimensionless physical state attenuation coefficient representing the intensity of heat transfer across space due to air convection within the cabinet. In this embodiment, its preferred value is set within the range of 0.15 to 0.35. If this coefficient deviates from 0.15, the system may mistakenly deduct a large amount of heat energy that should be dissipated by the air under normal airflow conditions, resulting in a lower subsequent internal temperature calibration value. If the coefficient is below 0.35, the system may assume that the environment has strong heat dissipation capacity, thus failing to adequately remove external crosstalk heat when actual heat accumulation occurs, causing the target electromagnetic component to bear the false temperature rise of adjacent equipment.

[0048] Analyze the physical installation distance parameters between each adjacent heat-generating device and the target electromagnetic component, and extract the heat dissipation attenuation coefficient calibrated above. The mapping generates a structured physical distance mapping table between devices; Table 1 below is a joint representation of the "Current Cabinet Device Operating Status Table" and the "Physical Distance Mapping Table Between Devices";

[0049] Table 1: Joint Mapping Table of Rack Space Topology and Physical Distance Between Devices

[0050]

[0051] The topology node identifier represents the unique hardware address mapping code of each independent heat-generating entity within the industrial cabinet in three-dimensional physical space. The operating heat-generating equipment status identifier indicates the power-on activation status of the equipment as captured in real-time by the communication gateway. Extracting this identifier allows for the early rejection of "cold-state equipment" that is close to the physical distance but not generating any heat exchange through logic gating. The physical installation spacing parameter is denoted as... : Represents the linear spatial physical span between the outer casing of the corresponding topological node device and the temperature probe of the target electromagnetic component, in standard meters (m). Physical radius of thermal radiation interference calibrated based on historical data. The boundaries of the intervals were defined, and the physical installation spacing parameters were extracted for comparison and interception: nodes greater than 1.2m or not in operation were removed, and only the actual heat sources within the intervals were extracted. As shown in Table 1, dual-track interception was performed based on status indicators and physical spacing. Node N2 was blocked because it was not energized, and node N4 was blocked because the spatial distance reached 1.8m. Ultimately, only nodes N1 and N3 were locked to generate a "comparison distance mapping table".

[0052] Step S2-A: Perform spatiotemporal stripping and real trajectory extraction of external environmental heat crosstalk: Import the generated surface radiation temperature set and the physical distance mapping table between devices into the pre-configured external environmental heat filtering module. Perform amplitude feature extraction on the surface radiation temperature set to extract the surface radiation temperature fluctuation values. It is used to characterize the original heat emission intensity of adjacent heat-generating devices in the current monitoring cycle. For the amplitude feature extraction operation, the calculation logic is as follows: within the time axis sliding window of the current temperature rise monitoring cycle, traverse and read all discrete temperature sampling points in the surface radiation temperature set; optimize and extract the absolute temperature maximum value in the set and the initial steady-state temperature reference value of the cycle; then perform a difference operation on the two, and use the calculated difference as the surface radiation temperature fluctuation value.

[0053] Combined with the physical installation spacing parameters recorded in the physical distance mapping table between equipment Compared with the previously calibrated heat dissipation attenuation coefficient , for surface radiation temperature fluctuation value Perform spatial thermal damping attenuation calculation: Dynamically calculate the estimated temperature rise due to external environmental radiation interference applied to the surface of the target electromagnetic component's casing. Among them, the physical installation spacing parameters The absolute linear spatial distance between the outer shell of the interference source and the target probe is represented, and in this embodiment, its dimension is uniformly converted to the standard meter (m).

[0054] This embodiment introduces a physical time hysteresis property to characterize the time consumed by heat energy transfer across media. This parameter represents the actual physical transmission time of heat radiation generated by adjacent heat-generating devices penetrating the airflow field inside the cabinet and ultimately reaching the surface of the target electromagnetic component. In this preferred embodiment, the physical time hysteresis attribute is calculated based on the rated convection velocity of the cooling fan inside the standard cabinet (standard 3 m / s normal wind speed) and the equipment installation spacing. The preferred value is 3.5s, but in practical applications it can be configured within the range of [1.5s, 10s]. It can align the external heat calculation model with the actual physical spatiotemporal delay, avoiding data misjudgment caused by time differences.

[0055] This embodiment introduces a nonlinear transformation rule for the (time domain and temperature domain), specifically combined with the physical time hysteresis property. Estimated temperature rise due to external environmental radiation interference Perform lag compensation mapping operations on the time axis, based on the following compensation mapping logic: ; where exp is short for exponential function, which represents backtracking on the historical timeline and extracting the phase after hysteresis offset ( Estimated temperature rise due to external environmental radiation interference at (time) Then extract the current temperature rise monitoring period. And compare it with the physical time hysteresis property A ratio mapping is performed to eliminate the time dimension, and the time decay correction multiplier is generated by substituting it into the exponential inverse compensation function; a thermal inertia mapping factor is introduced. This is then combined with the aforementioned time decay correction multiplier and hysteresis offset temperature rise value to extract the ambient additional temperature rise value that is absolutely aligned with the current temperature measurement phase. Among them, the thermal inertia mapping factor This value, representing the actual absorption and retention ratio of external airflow heat by the metal casing of the heating element inside the cabinet, is preferably set to the range of 0.82 to 0.91 in this embodiment. Further, the measured temperature time series is subtracted from the ambient additional temperature rise value to filter out the influence of ambient baking, generating the true temperature rise trajectory data after deducting external heat interference, denoted as... .

[0056] Step S2-B involves jointly constructing the internal dual-material theoretical temperature rise benchmark: After eliminating external spatial interference, the theoretical temperature rise of coil copper loss represented by the first temperature data generated in step S1 and the theoretical temperature rise of magnetic core iron loss represented by the second temperature data are imported into a preset dual-heat-source theoretical curve generation module. This embodiment performs weighted compensation and phase alignment aggregation operations based on the physical heat conduction mechanism on both. Since the magnetic core is located inside the coil, there is a physical lag in the time it takes for heat to conduct to the surface. This embodiment introduces a mapping formula to perform dimensionality reduction fusion, as shown below: It obtains the first temperature data at the current temperature measurement time point t. and compare it with the pre-configured coil thermal conductivity bias weight. Perform weighted multiplication; simultaneously extract the phase sequence delay parameter on the time axis. Second temperature data after backtracking offset This is compared with the pre-configured core thermal conductivity bias weight. A weighted multiplication is performed; then, a linear summation and aggregation operation is performed on the two sets of product results to jointly construct a theoretical temperature rise reference curve for dual materials that conforms to the actual dual-source heating law. The coil thermal conductivity bias weight is among them. With magnetic core thermal conductivity bias weight This is a constant used to represent the proportional distribution of heat conduction from two different materials to the outer casing. In this embodiment, its preferred quantization values ​​are configured as 0.65 and 0.35, respectively, and the sum of the two is 1. If the coil thermal conductivity bias weight... Setting it below 0.5 will overemphasize the heat generation of the internal magnetic core, resulting in a theoretical baseline value calculated under high current and low magnetic loss conditions that is lower than the actual expected value. Phase sequence delay parameter The physical time it takes for heat energy to be transferred from the internal magnetic core across the medium to the external sensing node is preferably set in the range of 150ms to 300ms.

[0057] Step S2-C, perform condition-driven waveform dynamic fitting and risk feature adaptive decoupling: by generating real heating trajectory data With the theoretical temperature rise reference curve of dual materials The waveform is fed into a dynamic waveform fitting module to perform waveform morphology comparison aimed at identifying the aging of insulating materials. This embodiment extracts the time-series temperature rise rate and peak characteristics of the two curves mentioned above. A permissible range for heat conduction time hysteresis, denoted as , is introduced to absorb normal physical delays. This parameter represents the legal heat transfer buffer time window that must be experienced for internal heat energy to penetrate the magnetic core and insulation layer to reach the surface after a sudden change in coil current. In this preferred embodiment, its range is preferably [-20ms, +150ms], which is selected and configured within the range of [-50ms, +300ms] based on the physical volume of the heated object being measured.

[0058] Within the allowable range of heat conduction time hysteresis A discretized sliding computation window is introduced to perform translational fitting optimization along the time axis. The configuration for this fitting operation is as follows: Where argmin represents the independent variable that minimizes the function. The theoretical heating reference curve of the dual-material system is controlled by setting a discretized translation-sliding step size. Data relative to the actual temperature rise trajectory on the timeline The translation and offset are performed step by step, and the offset is... ; Extract two sets of curves at each offset state during the current temperature rise monitoring cycle ( to The corresponding discrete sampling points within the range are used, and the absolute difference summation operation is performed to obtain the total area residual under this offset state; the entire allowable range of heat conduction time hysteresis is traversed. The interval is compared and selected to find the optimal offset time value that makes the total area residual reach the global minimum. This value is then directly output as a waveform time difference value, denoted as . And extract the time difference of the waveform. The absolute ordinate difference between the peak points of the two sets of curves under aligned conditions is used to analytically extract the temperature amplitude deviation, denoted as... This embodiment introduces a translational sliding step size. In this preferred embodiment, the translation sliding step size is preset to a range of 10ms to 50ms.

[0059] Existing technologies use a fixed static coefficient to measure temperature amplitude deviation. Waveform time difference This comprehensive calculation method cannot distinguish between "normal thermal inertia fluctuations caused by transient surges in grid load" and "abnormal divergence caused by actual insulation aging". In this embodiment, an adaptive feature fusion strategy based on dynamic load gradient is invoked.

[0060] During waveform morphology comparison, the sequential current data of the conductive coil during the current temperature rise monitoring cycle is extracted simultaneously. The rate of change of current (Amperes / second) between adjacent sampling points is calculated, and the rated reference current of the target electromagnetic component in the configuration library (denoted as ) is extracted. Divide the rate of change of current by the rated reference current of the equipment. Perform a baseline ratio calculation to extract the electrical load change rate parameter expressed as a percentage, denoted as... It is used to indicate the degree of fluctuation in the actual heat source driving state of the equipment. The rated reference current of the equipment is included. This represents the stable full-load current value of a component at its factory-rated power. The specific value depends on the hardware nameplate parameters of the device under test. Dividing by this static physical constant converts the absolute fluctuation of the scalar quantity into a relative percentage index that is universally applicable across devices.

[0061] This embodiment introduces a gating condition for distinguishing between steady-state and transient states: set as... The load fluctuation threshold. This parameter is pre-set as a static operation configuration parameter in the global state registry of the edge node and is dynamically read and invoked with each round of electrical load change rate calculation. This parameter directly represents the maximum legal jump gradient of the grid current per unit time. Load fluctuation threshold The value is defined as ranging from 6% / s to 12% / s. If the load fluctuation threshold... Setting the threshold too low can easily misjudge normal, minor fluctuations in the power grid as severe sudden changes, masking the true early fault characteristics under steady-state conditions; if the load fluctuation threshold is too low... Setting the value too high can easily treat real power grid surges as a stable state, leading to mismatches in the subsequent judgment matrix weights and miscalculations of risks. Furthermore, after comparison and verification, a multi-branch adaptive evaluation logic is triggered:

[0062] Branch 1 is defined as a transient surge defense mechanism: responding to the electrical load change rate parameter. Greater than the preset load fluctuation threshold The system determines that it is currently in a period of heat surge caused by external power grid impact. At this point, the load mutation assessment branch is triggered and invoked to perform a dynamic weighted fusion operation to reduce the calculation bias of component temperature conduction hysteresis. Specifically, it retrieves preset historical extreme value boundary detection parameters to assess temperature amplitude deviation. Time difference with waveform Perform range mapping normalization processing separately. Use the temperature amplitude deviation... For example, its range mapping formula is as follows: ;

[0063] It extracts the historical safety lower limit extreme values ​​of amplitude from the configuration library. and the upper limit extreme value of amplitude collapse Among them, the historical safety lower limit extreme value of amplitude. This parameter represents the baseline of normal temperature fluctuations of the target component during normal operation, caused by reasonable physical thermal inertia and ambient background noise. In this embodiment, the preferred value for this parameter is 1°C, but in actual configurations, it can be any real number within the range of [0.2°C, 3°C]. The deviation of the current measured temperature amplitude is then calculated. The difference between the original physical data and the safety lower limit is divided by the global range space formed by the two historical extreme values ​​mentioned above, thereby mapping the original physical data to generate an amplitude deviation comparison coefficient unified to the dimensionless interval [0,1], denoted as . Similarly, extract the extreme values ​​of the time hysteresis safety lower limit from the configuration library. With time delay limit collapse upper limit extreme value Through formula Complete the waveform time difference The range mapping is used to calculate and generate time difference comparison coefficients within the interval of 0 to 1, denoted as . Among them, the extreme value of the time delay limit collapse upper limit. This value represents the maximum tolerable hysteresis time before physical obstruction or insulation failure occurs during internal heat conduction across the medium. Its preferred value is 200 ms, and its range is configured within the interval [100 ms, 500 ms]. The extreme value of the safe lower limit for time hysteresis is... The preferred value is absolute zero, 0ms. Wherein, the amplitude limit collapse upper limit extreme value... This represents the maximum traceable temperature difference peak that can be recorded before insulation breakdown occurs. In this embodiment, its preferred calibration value is set to 15°C to 25°C.

[0064] This embodiment is based on the electrical load change rate parameter. The current value triggers and invokes the nonlinear suppression and dimensionality reduction mechanism to perform a numerical inverse mapping operation. This mapping logic is configured as follows: It extracts the mutation load increment beyond the stationary boundary ( and The difference is multiplied and amplified by the pre-configured control parameters and substituted into the inverse proportional suppression structure to generate a convergence result, thereby generating an amplitude ratio weighting coefficient in the range of 0 to 1. This indicates that the more drastic the sudden change in electrical load, the more exponentially the amplitude ratio decreases relative to the weighting factor. Due to thermal inertia, short-term temperature amplitude fluctuations exhibit hysteresis and distortion, thus reducing the reliability of amplitude deviations. This is the load sensitivity suppression constant, which determines the ability to desensitize to load surges on the grid side. In this embodiment, its preferred value is limited to the range of 1.5 to 2.5. If the load sensitivity suppression constant is lower than 1.5, it is easy to fail to quickly reduce the comparison weight of the amplitude under strong current surges, causing thermal inertia deviation to be misjudged as insulation damage; if the load sensitivity suppression constant is higher than 2.5, the ability of this embodiment to monitor and capture amplitude anomalies during load fluctuations is weaker.

[0065] Compare the amplitude to the weighting coefficient Assigned to amplitude deviation comparison coefficient And extract the corresponding complementary residual weights ( Assignment to time difference comparison coefficient Further, a dimensionality reduction and weighted aggregation operation is performed, the specific logic of which is as follows: ; Obtain the dimensionless amplitude deviation comparison coefficient Compare it with the currently dynamically generated amplitude and reduce the weighting coefficient. Perform the multiplication operation; simultaneously obtain the dimensionless time difference comparison coefficients. This is then multiplied by the complementary residual weights; subsequently, the two sets of dimensionality-reduced products are summed and aggregated, and the underlying safety margin constant is added. The output represents the insulation aging risk index, indicating the actual insulation aging risk under load surge conditions. Among them, the basic safety margin constant The inherent reference offset constant used to represent the heating element in a factory-undamaged state is preferably preset to 0.02 to 0.05 in this embodiment.

[0066] Branch 2 is defined as a steady-state deep verification mechanism: responding to the electrical load change rate parameter. Not greater than the preset load fluctuation threshold The system determines that the current operating condition is stable. At this point, the load stability assessment branch is triggered and invoked to perform feature comparison and aggregation. Specifically, a preset stability state determination weight table is retrieved, denoted as... This table is generated via offline calibration pre-configuration based on boundary flow probing of extreme load test data. An offline, real-world historical fault data playback environment is pre-built, and historical test log streams generated during stable load periods by electromagnetic components known to have experienced insulation failure are batch-injected into this environment; the temperature amplitude deviation indicated by amplitude anomalies on the fault evolution link is continuously monitored. The time difference between the waveform represented by the time delay anomaly The mapping hit rate of the final fault confirmation results is statistically analyzed. Specifically, in the offline calibration experiment, the preset sample size is at least 1000 sets of historical test log streams covering the entire life cycle of the target electromagnetic component. Two key continuous sequences are extracted from the test data: the temperature amplitude deviation set and the waveform time difference set; at the same time, the set of physical parameters representing the severity of the actual fault (e.g., insulation resistance attenuation rate or partial discharge measured under shutdown conditions) that are measured synchronously with these test data are extracted. The first Pearson correlation coefficient between the temperature amplitude deviation sequence and the actual fault parameter sequence, and the second Pearson correlation coefficient between the waveform time difference sequence and the actual fault parameter sequence are calculated respectively. The temperature amplitude deviation represents the local thermal resistance anomaly under steady state. The corresponding first Pearson correlation coefficient (greater than 0.85) is higher than the waveform time difference. The corresponding second Pearson correlation coefficient (between 0.3 and 0.5). Further normalization is performed based on the ratio mapping of the above Pearson correlation coefficients to solidify and generate a stationary state determination weight table. In this preferred embodiment, the temperature amplitude deviation is included in the steady-state determination weight table. The baseline weighting was fixed in the range of 0.75 to 0.85, while the waveform time difference... The baseline allocation weight corresponds to the range of 0.15 to 0.25. If the temperature amplitude deviates by... Setting the weights deviating from the lower limit by 0.75 will weaken the sensitivity of this embodiment to heat leakage from minor damage, leading to missed or delayed steady-state monitoring. Under stable grid load conditions, the specific "Steady-State Judgment Weight Table" is retrieved and locked. This table is generated based on verification using an offline standard dataset and has a locked baseline weight allocation.

[0067] Table 2: Objective Configuration Record of the Stationary State Judgment Weight Table Based on Offline Standard Dataset

[0068]

[0069] The electrical load steady-state indicator indicates that, in this embodiment, the power grid load has not exceeded the surge boundary during the current temperature rise monitoring period, indicating a normal heating physical condition. The load increment change rate range is used as a mathematical constraint to determine "steady," and the currently sampled electrical load change rate parameter is within the safe range of the load fluctuation threshold. The amplitude deviation comparison coefficient allocation weight and the time difference comparison coefficient allocation weight respectively represent the static confidence distribution law for the heat overflow phenomenon caused by internal insulation deterioration after excluding the external electrical interference of "severe thermal inertia lag." Table 2 discloses the "Steady State Judgment Weight Table" under steady-state routing. Based on the statistical distribution law of historical fault data playback, when the load increment does not change abruptly, the obvious characteristics caused by insulation damage are highly concentrated in temperature amplitude anomalies rather than conduction time differences. The table records that the dominant weight of 0.8 is locked to the amplitude feature, while the auxiliary weight of 0.2 is allocated to the time feature.

[0070] Step S3-A involves performing conditional mapping and generation of dual-track error compensation parameters; specifically, it receives the environmental additional temperature rise value output from step S2. With insulation aging risk index The internal abnormal temperature rise risk value is represented, and parameter mapping logic based on boundary judgment is executed respectively.

[0071] To address external environmental crosstalk, this embodiment introduces an environmental benchmark threshold for determining whether the background heat of the server rack reaches the level of intervention, denoted as . It represents the maximum harmless background radiant heat that a target component within the physical rack space can withstand. In this embodiment, it is based on the historical environmental monitoring statistics of the rack cooling gateway and the environmental baseline threshold. The preferred value is 5℃, and its effective configuration range is limited to [3℃, 8℃].

[0072] Response to ambient temperature rise values Greater than the preset environmental reference threshold It was determined that the external crosstalk had exceeded the safety margin, and therefore an additional environmental temperature rise value was added. As a benchmark for environmental thermal error compensation, and combined with the physical material properties of the outer shell, the pre-configured surface heat transfer attenuation ratio is retrieved. The compensation operation logic for performing productized numerical calculations is as follows: It generates the external temperature measurement error compensation parameter that ultimately approaches the internal probe by performing a dimension-reduced multiplication of the aforementioned environmental benchmark with the attenuation ratio, denoted as... This parameter represents the "artificially high" temperature rise that needs to be forcibly deducted from the internal core measured curve, and is comprehensively described as the environmental heat error compensation benchmark tolerance range.

[0073] Response to ambient temperature rise values Not greater than the preset environmental benchmark threshold If the current cabinet background crosstalk is determined to be weak and within the range of the environment's safe heat dissipation capacity, then the above attenuation calculation is skipped, and the external temperature measurement error compensation parameter is applied. Calibrated to absolute zero. This effectively avoids the risk of false negative drift caused by excessive subtraction of the target probe temperature under normal, interference-free conditions. The surface heat transfer attenuation ratio is also considered. This represents the percentage of energy dissipated by external heat waves as they penetrate the equipment casing and reach the temperature probe node. In this embodiment, for a standard industrial flame-retardant plastic casing, the preferred value is fixed in the range of 0.65 to 0.85.

[0074] This embodiment, based on a historical waveform evolution database of healthy components, calibrates and introduces a safety reference range for constraining heating states, denoted as... It is used to represent the normal tolerance band of the amplitude deviation of the actual heat flow conduction curve between the target element coil and magnetic core under conditions of no insulation damage and no severe aging.

[0075] In this embodiment, the safety reference range The optimal quantization value is defined as an asymmetric fluctuation range of [-5%, +8%]. If the lower limit of the safety reference range is set to 0%, it is easy to misjudge the normal small fluctuations in thermal inertia caused by sudden load increases as internal degradation, thus frequently triggering erroneous data correction feedback, resulting in the theoretical temperature rise reference curve of the dual materials being infinitely suppressed. If the upper limit of the safety reference range is too large, it is easy to treat the weak heating of the actual early insulation dendritic discharge as "normal drift" and allow it to pass. This corresponds to the internal abnormal temperature rise risk value represented by the insulation aging risk index calculated in the previous step. The preset asymmetric safety benchmark range was not met. The internal theoretical heating evolution law of the heating element has been irreversibly deviated due to substantial material aging or insulation damage. This embodiment further invokes an internal correction mechanism to calculate and generate an internal temperature measurement reference calibration coefficient, denoted as... It is used as a dynamic compensation factor for the initial theoretical heating slope.

[0076] Furthermore, this embodiment introduces an adaptive calibration strategy based on state deviation gradients. Specifically, it constructs an internal material calibration sensitivity factor. This is used to represent the severity of compensation adjustments when abnormal temperature rise is caused by internal insulation degradation. The insulation aging risk index calculated in the previous step is extracted. And dynamically extract the specific boundary limit value that is substantially breached, denoted as (When the insulation aging risk index overflows upwards, this parameter is taken from the safe reference range.) The upper bound of the collapse extreme value is +8%; when the risk index falls below the lower limit, this parameter is taken as the lower bound of the collapse extreme value -5%), and further relative deviation feature alignment calculation is performed, specifically configured as follows: It extracts the absolute overflow ratio caused by the current insulation aging risk index exceeding the benchmark extreme value and compares it with the internal material calibration sensitivity factor. The feature fusion and amplification operation is performed, and finally the basic slope state constant 1 is superimposed to calculate the internal temperature measurement reference calibration coefficient. Furthermore, the internal material calibration sensitivity factor. Offline calibration was performed in a historical fault sandbox environment. Specifically, multiple sets of operational log features of heating components with known degrees of damage were injected into the sandbox. The statistical distribution of the deviation between actual temperature drift and theoretical curves was extracted, thereby determining the internal material calibration sensitivity factor. The quantization boundary is defined in the range of 0.1 to 0.3.

[0077] This embodiment extracts historical operating data and imports it into the configured system judgment model. Based on the verification mechanism of offline standard datasets, core sensor logs of the same batch of industrial high-power transformers are extracted at different physical aging stages throughout their entire life cycle. The physical boundary parameters configured in this verification scenario are as follows: coil thermal conductivity bias weight. The curing distribution is 0.65, and the core thermal conductivity bias weight is [not specified]. The curing ratio is 0.35. Safety baseline range. upper limit of the collapse Calibration set at 0.08, internal material calibration sensitivity factor. The value was fixed at 0.2. External environmental crosstalk did not exceed the threshold within the extraction range under this operating condition, and the external temperature measurement error compensation parameter... The value is constant at 0. The above historical operational data is retrieved, and a feature calculation based on the internal material degradation timeline is executed. The deterministic state response data calculated by the system is as follows:

[0078] Table 3: Data Records of Core Internal Heating Benchmark Reconstruction and Dynamic Calibration Indicators Based on Offline Standard Dataset

[0079]

[0080] Based on the internal thermal runaway compensation increment defined in Table 3, it is generated by extracting the real-time calibration temperature value output in this embodiment and subtracting the absolute difference calculated from the dual-material theoretical temperature rise benchmark. This parameter quantitatively characterizes the absolute temperature difference of irreversible abnormal heat released substantially additionally to the equipment surface by the target heating element under grid-driven load due to the physical obstruction and deterioration fracture of its internal insulation material, resulting in an increase in the temperature measurement baseline.

[0081] Extracting records 1 and 2 reveals that, under healthy system conditions (where the insulation aging risk index does not exceed the collapse threshold of 0.08), this embodiment performs multiplication and addition operations based on the first and second temperature data, respectively superimposed with thermal conductivity bias weights of 0.65 and 0.35, and outputs dual-material theoretical temperature rise benchmarks of 63 and 83. Simultaneously, the internal temperature measurement benchmark calibration coefficient is anchored to the benchmark value of 1 to ensure that the real-time calibrated temperature value conforms to the physical baseline, demonstrating the absolute stability of this embodiment under healthy operating conditions. Records 2, 3, and 4 constitute the key comparative benchmarks for material heterogeneous degradation under the same driving load. In these three sets of extracted data, the first temperature data calculated from the sensed electrical parameters is consistently 90, the second temperature data is consistently 70, and the dual-material theoretical temperature rise benchmark is objectively locked at 83. Relying on conventional temperature measurement schemes that do not include a closed-loop feedback architecture, this embodiment will always output the theoretical value of 83 as a constant benchmark, losing the ability to capture internal thermal field accumulation and causing false negatives. In contrast, this embodiment, by pre-extracting the continuously deteriorating insulation aging risk index (0.06, 0.12, and 0.2 respectively), keenly captures the fact that the index has exceeded the boundary extreme value of 0.08 in records 3 and 4. An adaptive calibration strategy is triggered: taking record 4 as an example, the actual insulation aging risk index 0.2 is extracted and the limit value 0.08 is subtracted to obtain a difference of 0.12; this difference is divided by the limit value 0.08 to obtain a deviation multiplier of 1.5; multiplied by the internal material calibration sensitivity factor 0.2, a gain multiplier of 0.3 is obtained, thereby pushing the internal temperature measurement reference calibration coefficient to 1.3. Finally, the initial reference of 83 is multiplied by the internal temperature measurement reference calibration coefficient of 1.3, outputting the real-time calibration temperature value of 107.9.

[0082] Record 5 further confirms that when approaching the critical state of aging failure, based on the extracted extreme risk index of 0.36, the calibration coefficient is nonlinearly pushed up to 1.7, and the real-time output of 158.1℃ far exceeds the physical theoretical baseline of 93℃.

[0083] Step S3-B, in this embodiment, in response to the above-mentioned additional environmental temperature rise exceeding the environmental reference threshold... And generate external temperature measurement error compensation parameters. Following the steps, it is determined that there is localized heat buildup in the physical cabinet space where the target electromagnetic component is located, and a global heat dissipation linkage intervention process is immediately triggered in parallel. Specifically, a heat buildup anti-jitter comparison mechanism is introduced. The temporal maintenance characteristics of the environmental heat error compensation benchmark within the historical sliding window (the continuous survival time of the compensation benchmark being greater than 0) are extracted, and a preset heat buildup warning time tolerance is introduced, denoted as... The time sequence maintenance characteristic is continuously greater than or equal to the siltation warning time tolerance if and only if the time sequence maintenance characteristic is continuously greater than or equal to the siltation warning time tolerance. A transition state flag is triggered and generated to indicate the presence of localized thermal congestion in the physical cabinet space, with the activation intervention flag set to 1. In this embodiment, the preferred value for the congestion warning time tolerance is set to the range of 5s to 15s. If the congestion warning time tolerance is set too small, the instantaneous airflow swirling caused by opening the cabinet door or the thermal noise fluctuations of the sensor may be misjudged as thermal congestion, leading to frequent and rapid acceleration of the exhaust fan. If the congestion warning time tolerance is set too large, the ability to identify thermal stress damage to the core circuit board caused by high ambient temperatures will be reduced.

[0084] When performing intervention, the environmental heat error compensation benchmark generated above is used as the basis, and the physical distance mapping table between devices established in step S1 is extracted to perform reverse tracing of the heat source distribution location. Specifically, the topology nodes in the physical distance mapping table between devices are traversed, and the adjacent heat-generating devices with the highest heat interference weight and causing current heat accumulation within the current monitoring cycle are locked. After locking the target topology node, a safe conversion from temperature characteristics to electromechanical execution parameters is established. Specifically, the pre-configured stepped temperature control mapping matrix is ​​retrieved, denoted as... This embodiment extracts historical operating data and provides a detailed data-driven disclosure of the "step temperature control mapping matrix" mentioned in step S3-B, which is derived from the environmental heat error compensation benchmark and the fan duty cycle. See Table 4 below for details:

[0085] Table 4: Data Structure of the Step Temperature Control Mapping Matrix Based on Hardware Intervention Link

[0086]

[0087] The environmental thermal error compensation baseline tolerance range represents the external disturbance temperature rise amplitude represented by the external temperature measurement error compensation parameter. This range is divided into three discrete threshold segments. The rack exhaust fan array transition command identifier represents the control bit parameters sent to the temperature control gateway, which are truncated into three discrete states: level 1, level 2, and level 3, by reducing the continuously fluctuating temperature scalar to a lower dimension. The basic speed-up duty cycle parameter (PWM) represents the percentage of the output pulse width modulation signal, and its value directly determines the instantaneous rotational angular velocity of the physical fan. This table is based on the external temperature measurement error compensation parameter. The different tolerance ranges it falls into will be matched with three fixed duty cycle parameters of 40%, 75% and 100%.

[0088] By extracting the currently generated external temperature measurement error compensation parameters This is then imported into a stepped temperature control mapping matrix to perform a hierarchical correlation transformation of heat dissipation control features. The mapping rules for these levels are specifically defined as follows: when the external temperature measurement error compensation parameter... When within the first tolerance range (0.5℃ to 2℃), the first gear (representing a 40% PWM base speed-up duty cycle) is calculated; when the external temperature measurement error compensation parameter... When the temperature is within the second accumulation range (2.1℃ to 5℃), the transition is mapped to the second gear (representing a 75% strong exhaust duty cycle); when the upper limit of the second range is exceeded, the output is mapped to the extreme value third gear (representing a 100% full-speed emergency response duty cycle). Based on the above matrix rules, after obtaining the determined gear data, a dynamic heat dissipation coordination command is generated, denoted as... It contains embedded identifiers for the specific physical rack slots that are locked and the calculated target speed-up level. This embodiment will use dynamic heat dissipation coordination commands. The internal communication protocol is transmitted to the rack space temperature control gateway to trigger the rack exhaust fan array that matches the adjacent heat-generating equipment to increase its speed, thereby controlling the rack exhaust fan array to perform physical exhaust cooling.

[0089] Step S3-C involves the final feedback correction of the parameter mapping and hardware scheduling execution data. Specifically, this includes adjusting the external temperature measurement error compensation parameters. The feedback link is connected to the data link of the measured temperature time series set, and the external temperature measurement error compensation parameter is executed. The surface temperature measurement error elimination operation specifically involves subtracting the compensation parameter from the values ​​of each time-series node in the original acquisition set to eliminate the thermal radiation adhesion effect; simultaneously, the internal temperature measurement reference calibration coefficient is adjusted. The feedback loop is linked to the generation chain of the dual-material theoretical temperature rise reference curve, performing a data calibration operation to correct the reference amplitude of the dual-heat-source theoretical curve. Specifically, the time-series temperature rise value of the original theoretical curve is directly multiplied by the compensation factor to achieve a dynamic rise or drop in the theoretical reference line. Through the above-mentioned internal and external dual-track feedback correction, dynamic temperature evolution trajectory data characterizing the actual heating characteristics of the target high-power transformer, after measurement and calibration, is generated and denoted as... .

[0090] Further execution involves extracting dynamic temperature evolution trajectory data. The final state time point within the current temperature rise monitoring cycle is denoted as... (This represents the last polling timestamp after the current temperature measurement algorithm has completed all external deductions and internal multiplier corrections). Analysis and this final state time node. The absolutely aligned current effective temperature value. This value represents the latest effective temperature measurement result obtained in this embodiment after processing all interference data within the current monitoring cycle. It reflects the instantaneous calibration temperature of the transformer at this moment. This current effective temperature value is directly assigned to the target variable, and the real-time calibration temperature value of the target high-power transformer is calculated and output to the external industrial control bus, denoted as... .

[0091] Furthermore, external temperature measurement error compensation parameters When the output value is equal to or close to 0, the radiative temperature rise conducted by adjacent external heating devices has not penetrated the dissipation defense of the air duct convection. This embodiment directly cuts off the compensation link to effectively prevent excessive subtraction of the target probe temperature data and avoids the physical failure risk of negative drift of the measured temperature.

[0092] When the output value of the external temperature measurement error compensation parameter approaches its positive maximum, it indicates that "heat accumulation in the air duct" has occurred in the physical cabinet space. At this time, the strong heat wave generated by adjacent equipment has substantially exceeded the thermal damping attenuation of the space and adheres in large quantities to the outer shell of the target electromagnetic components. A downward stripping calculation based on the measured trajectory data is performed, triggering a full-load emergency cooling action of the cabinet's exhaust fan array at the physical hardware layer.

[0093] Insulation aging risk index Approaching the basic safety margin constant This indicates that the conductive coil and magnetic medium of the target electromagnetic component are in a healthy operating period, at which time internal heat can be conducted to the surface within a standard aging time. Insulation aging risk index. Approaching the maximum value 1: This indicates a blockage or breakage. At this point, the sudden increase in current may cause heat energy to be trapped by the aging insulation layer, resulting in an irreversible collapse of the measured curve in terms of amplitude or time axis. This state will trigger the calibration feedback link of the temperature measurement reference outside the asymmetric safety reference range.

[0094] The heat dissipation attenuation coefficient and the estimated temperature rise due to external environmental radiation interference exhibit an exponentially negative correlation. With fixed physical installation spacing parameters, stronger airflow in the duct results in a larger heat dissipation attenuation coefficient, leading to an exponentially sharp decrease in the residual radiated energy reaching the target enclosure. Designing the heat dissipation attenuation coefficient to have an exponentially negative correlation ensures that this embodiment actively suppresses the estimated external temperature rise under excellent heat dissipation conditions with the cabinet fans operating normally.

[0095] The correlation between the electrical load change rate and the amplitude ratio weighting coefficient is inversely proportional and nonlinearly negatively correlated. The greater the gradient of the grid current jump, the more distorted the instantaneous characterization of surface temperature amplitude becomes, thus depriving amplitude characteristics of their dominant role in the comprehensive heat transfer time delay assessment. This correlation design constructs a reverse isolation defense between the severity of load abrupt changes and amplitude confidence. Upon sensing a strong current surge, it autonomously reduces the amplitude ratio weighting coefficient to a low range, thereby avoiding the false positive logic defect of misjudging normal thermal inertia fluctuations as insulation damage.

[0096] This embodiment is configured in the following industrial cabinet node digital twin monitoring scenario. The preconditions set in this monitoring scenario are as follows: the physical installation distance parameters between the target heating element and adjacent heating sources. Set to 0.8, heat dissipation attenuation coefficient The static calibration is 0.2, representing the surface heat transfer attenuation ratio. The calibration value is 0.75. The preset environmental baseline threshold is set to 5, and the load fluctuation threshold is... Limited to 8, load-sensitive inhibition constant Calibrated to 2, basic safety margin constant The calibration is set to 0.02. The steady-state determination weight table under steady-state conditions is assigned to the amplitude deviation comparison coefficient. The baseline weighting is 0.8, which is allocated to the time difference comparison coefficient. The baseline allocation weight is 0.2.

[0097] The front-end temperature measurement data acquisition terminal aggregates the ambient additional temperature rise value and the electrical load change rate; when the load increment change rate does not exceed the threshold, the conventional stable weight is invoked to perform feature correlation mapping; when the load surges and exceeds the limit, the nonlinear suppression and dimensionality reduction logic is driven to reduce the amplitude correlation weight. Based on the above constraints and mapping logic, the calculated deterministic system response data is shown in Table 5 below:

[0098] Table 5: Example of Calculation of Core Temperature and Risk Response Indicators under the Conditions of Heat Accumulation in Industrial Cabinet Air Ducts and Sudden Changes in Power Grid Load

[0099]

[0100] The amplitude baseline allocation weight of 0.8 is extracted from the steady-state judgment weight table. This is subtracted from the amplitude comparison weighting coefficient dynamically generated under the load mutation assessment branch. The difference between the two values ​​is then divided by the amplitude baseline allocation weight to calculate the amplitude false positive suppression rate. This parameter quantitatively characterizes the intervention and defense ratio strength of this embodiment in actively reducing thermal inertial hysteresis interference and correcting and suppressing false alarm actions under power grid load mutation conditions.

[0101] Extracting scenarios 1 and 2, the ambient additional temperature rise value increases from 2.5 to 4.5. In this embodiment, based on the unidirectional threshold that this value does not exceed the preset environmental reference threshold 5, the external temperature measurement error compensation parameter is anchored to 0. When the ambient additional temperature rise value input reaches 8 in scenario 3, this parameter exceeds the environmental reference threshold 5. Therefore, 8 is extracted and compared with the surface heat transfer attenuation ratio. 0.75 performs a dimension reduction multiplication operation to calculate the external temperature measurement error compensation parameter of 6, thereby achieving efficient start-up of environmental redundant temperature stripping under thermal accumulation pollution.

[0102] In scenario 4, the electrical load change rate increases to 13, causing the amplitude to deviate from the comparison coefficient. The amplitude deviation is passively increased to 0.9. The amplitude deviation comparison coefficient of 0.9 is multiplied by the fixed amplitude baseline weight of 0.8, and then the product of the time difference comparison coefficient of 0.05 and the fixed time weight of 0.2 is added. Finally, the basic safety margin constant of 0.02 is summed, resulting in an insulation aging risk index of 0.75. The difference of 5 between the actual electrical load change rate of 13 and the load fluctuation threshold of 8 is extracted. This difference is multiplied by the load sensitivity suppression constant of 2 to obtain the intermediate value of 10. The constant 1 is further extracted and divided by the aggregation base of 11, which includes the baseline offset of 1, thus converging the amplitude comparison weighting coefficient to 0.091. Subsequently, the complementary residual weight of 0.909 is extracted and allocated to the time difference comparison coefficient of 0.05. The final synthesized insulation aging risk index is only 0.147.

[0103] Further comparison of scenarios 4 and 5 reveals that, under the constraint of experiencing the same dramatic increase in load (both amplitude comparison weighting coefficients are locked at 0.091), scenario 5 exhibits a higher comparison coefficient due to actual physical aging, resulting in a lower time difference. The value climbed to 0.85. Using the aggregation operation logic that smoothly transitions the dominant complementary weight of 0.909 and applies it to the time dimension feature, an insulation aging risk index of 0.875 was calculated.

[0104] Figure 1 Within the industrial cabinet space, the original electrical signal sampling sequence of the target electromagnetic component is synchronously extracted using electrical sensing components. A first-order low-pass digital filter is then performed on the original electrical signal sampling sequence to reconstruct and generate runtime sequence current data with standard timestamps. With runtime timing voltage data Calling dimensional alignment will restore the runtime sequence current data. With runtime timing voltage data Import the physical conversion calculation unit, and combine it with the pre-configured coil static thermal resistance coefficient. With the static thermal resistance coefficient of the magnetic core Performing a heat power conversion operation based on the electromagnetic loss physical model, the first temperature data of the isomorphic conductive coil is finally generated. Second temperature data with magnetic medium .

[0105] Figure 2A confined three-dimensional cabinet space environment subject to heat crosstalk is constructed. The diagram illustrates the dual-heat-source physical architecture of the internal conductive coil and magnetic medium overlapping. A measured temperature probe is anchored to the surface of the transformer body, and an electrical sensor is embedded inside. The signal lines of both are connected to the data processing hub at the center of the cabinet. Adjacent heat-generating devices scattered around continuously release external radiant heat waves through the dotted ripples in the diagram. The "multi-source heterogeneous parameter acquisition" method, triggered by the temperature rise monitoring cycle, involves the temperature data acquisition end simultaneously extracting the measured temperature time series set of the target electromagnetic component, the first temperature data of the conductive coil, and the second temperature data of the magnetic medium. Simultaneously, it extracts the surface radiation temperature set of adjacent heat-generating devices and the physical distance mapping table between devices. The "external environment interference removal" method imports relevant information from adjacent heat-generating devices into the external environment heat filtering module, analyzes and generates the environmental additional temperature rise value conducted by adjacent devices, and performs interference removal operations on the measured temperature time series set accordingly to generate the true temperature rise after deducting external heat interference. The steps for obtaining temperature trajectory data include: In the "Internal Dynamic Fitting Evaluation" method, the first and second temperature data are imported into the dual-heat-source theoretical curve generation module to jointly construct a dual-material theoretical temperature rise reference curve, and then sent to the waveform dynamic fitting module to perform waveform morphology comparison and calculate the internal abnormal temperature rise risk value; In the "Feedback Correction and Calibration Output" method, based on the environmental additional temperature rise value and the internal abnormal temperature rise risk value, external temperature measurement error compensation parameters and internal temperature measurement reference calibration coefficients are generated respectively, and feedback correction is performed on the data to generate dynamic temperature evolution trajectory data after measurement calibration. Finally, this data is output as the current effective temperature value at the end of the cycle of the dynamic temperature evolution trajectory data, while simultaneously triggering the heat dissipation linkage hardware intervention process in parallel.

[0106] Figure 3 Specifically, this corresponds to "spatiotemporal stripping of external environmental thermal crosstalk" and "dynamic fitting of waveforms driven by operating conditions". Figure 3 As shown in (A), based on the physical time hysteresis property The calculated ambient temperature rise value is then subtracted from the measured temperature time series data to generate the true temperature rise trajectory data after removing external heat interference. .like Figure 3 As shown in (B), the first temperature data With phase sequence delay parameter Second temperature data after backtracking offset We performed weighted compensation and phase alignment polymerization operations based on physical heat conduction mechanisms to jointly construct a theoretical temperature rise reference curve for both materials. .like Figure 3 As shown in (C), within the preset allowable range of heat conduction time hysteresis. Inside, execute real temperature rise trajectory data. With the theoretical temperature rise reference curve of dual materials The waveform morphology was compared, and the deviation of the fitted temperature amplitude was calculated comprehensively. Time difference with waveform This allows us to calculate the risk value of abnormal internal temperature rise.

[0107] Figure 4 Based on the environmental additional temperature rise value and the internal abnormal temperature rise risk value output in step S2, parameter mapping based on condition judgment is performed respectively: in response to the environmental additional temperature rise value being greater than the preset environmental reference threshold. Mapping to generate external temperature measurement error compensation parameters ; In response to the internal abnormal temperature rise risk value not meeting the preset safety benchmark range Generate internal temperature measurement reference calibration coefficients External temperature measurement error compensation parameters Calibration coefficient with internal temperature measurement reference Feedback is sent back to the data link to perform data corrections to generate the final real-time calibrated temperature value; in response to external temperature measurement error compensation parameters. If the value is greater than zero, the rack-wide heat dissipation linkage intervention process is triggered in parallel. The heat source distribution location is traced in reverse through the environmental heat error compensation benchmark, and the pre-configured stepped temperature control mapping matrix is ​​retrieved. Through the rack space temperature control gateway, the rack exhaust fan array is controlled to increase its speed.

[0108] Figure 5 The X-axis represents the rate of change of electrical load increment, and the two Y-axis represent the amplitude ratio weighting coefficient and the calculated insulation aging risk index, respectively. Here, marker L1 actually represents the steady-state value of the load fluctuation threshold, 8; visual state marker W1 represents the extreme weighting value of 0.091 after nonlinear suppression routing is triggered; markers R1 and B1 represent the risk index of 0.147 and the risk value of 0.75 under the extreme load change, respectively. It should be noted that this appendix... Figure 5 The green solid curve and red dashed line in the diagram are intended to isolate the physical variables of insulation aging. Figure 5 At the point where the X-axis equals 18% / s, the risk index of the closed mapping drops to 0.113 (this pure anti-disturbance, no-aging control value is not shown in Table 5); this value represents the anti-disturbance effect of suppressing the static theoretical extreme value of 0.79 to 0.113 when the time difference comparison coefficient is constant at 0.05 under the load impact equivalent to scenario 6 in Table 5.

[0109] Example 2:

[0110] A device for dynamic temperature measurement and calibration of a heating element, the device being used to perform the method for dynamic temperature measurement and calibration of a heating element, comprising:

[0111] The multi-source heterogeneous parameter sensing module is configured to: in response to the triggering of the temperature rise monitoring cycle, synchronously extract the measured temperature time series set of the target electromagnetic element, which is the entity representing the heating element, and the first temperature data of the conductive coil and the second temperature data of the magnetic medium, which are the internal structures of the target electromagnetic element; at the same time, extract the surface radiation temperature set of the surrounding adjacent heating devices and the physical distance mapping table between the devices;

[0112] The spatiotemporal decoupling and feature fitting module is configured to: import the surface radiation temperature set of the surrounding adjacent heating devices and the physical distance mapping table between the devices into a pre-configured external environment heat filtering module to parse and generate the environmental additional temperature rise value conducted by the adjacent devices; perform interference removal operation on the measured temperature time series set based on the environmental additional temperature rise value to generate the true temperature rise trajectory data after deducting external heat interference; import the first temperature data of the conductive coil and the second temperature data of the magnetic medium into a pre-set dual heat source theoretical curve generation module to jointly construct a dual-material theoretical temperature rise reference curve; and send the true temperature rise trajectory data and the dual-material theoretical temperature rise reference curve into the waveform dynamic fitting module to perform waveform shape comparison and calculate the internal abnormal temperature rise risk value.

[0113] The closed-loop calibration and status output module is configured to: generate external temperature measurement error compensation parameters for eliminating external temperature measurement interference and internal temperature measurement reference calibration coefficients for correcting the dual-material theoretical temperature rise reference curve, based on the environmental additional temperature rise value and the internal abnormal temperature rise risk value; perform data feedback correction on the measured temperature time series set and the dual-material theoretical temperature rise reference curve based on the external temperature measurement error compensation parameters and the internal temperature measurement reference calibration coefficients, generating dynamic temperature evolution trajectory data of the target electromagnetic element after measurement and calibration; and extract the current effective temperature value of the dynamic temperature evolution trajectory data at the end node of the current temperature rise monitoring cycle after completing various error deductions and compensations, and output it as the real-time calibration temperature value of the target electromagnetic element.

[0114] The computational logic involved in this application can be constructed using machine learning algorithms such as regression analysis to build mathematical models, and implemented using computational tools such as Python and R. To eliminate the influence of physical dimensions, the input parameters of each formula can be dimensionlessly processed using techniques such as maximum-minimum normalization or Z-score standardization. Those skilled in the art should understand that the embodiments of this application can be provided as methods, systems, or computer program products, and can be in the form of pure hardware, pure software, or a combination of hardware and software, and can be implemented on storage media (including but not limited to disks, optical disks, etc.) containing computer-usable program code.

[0115] This application is described with reference to relevant flowcharts and / or block diagrams. It should be understood that each flowchart / block and combination thereof in the flowcharts and / or block diagrams can be implemented by computer program instructions. These instructions can be stored in a computer-readable storage device or executed by a processor of a general-purpose computer, a special-purpose computer, or other programmable devices to achieve the functions specified in the figures. Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope thereof. If such modifications and variations fall within the protection scope of the claims of this application and their equivalents, this application is also intended to cover such modifications and variations.

Claims

1. A method for dynamic temperature measurement and calibration of heating elements, characterized in that, The specific steps include: S1: In response to the triggering of the temperature rise monitoring cycle, the temperature data acquisition terminal synchronously extracts the measured temperature time series set of the target electromagnetic element, which is the entity representing the heating element, as well as the first temperature data of the conductive coil and the second temperature data of the magnetic medium, which are the internal structures of the target electromagnetic element; at the same time, it extracts the surface radiation temperature set of the surrounding adjacent heating devices and the physical distance mapping table between the devices. S2: Import the set of surface radiation temperatures of the surrounding adjacent heating devices and the mapping table of physical distances between the devices into the pre-configured external environment heat filtering module, and parse to generate the environmental additional temperature rise value conducted by the adjacent devices; Based on the environmental temperature rise value, an interference removal operation is performed on the measured temperature time series set to generate the true temperature rise trajectory data after deducting external heat interference. The first temperature data of the conductive coil and the second temperature data of the magnetic medium are imported into a preset dual-heat-source theoretical curve generation module to jointly construct a dual-material theoretical heating reference curve. The actual heating trajectory data and the theoretical heating reference curve of the dual material are sent to the waveform dynamic fitting module to perform waveform shape comparison and calculate the internal abnormal temperature rise risk value. S3: Based on the environmental additional temperature rise value and the internal abnormal temperature rise risk value, generate external temperature measurement error compensation parameters for eliminating external temperature measurement interference, and internal temperature measurement reference calibration coefficients for correcting the dual-material theoretical temperature rise reference curve. Based on the external temperature measurement error compensation parameter and the internal temperature measurement reference calibration coefficient, data feedback correction is performed on the measured temperature time series set and the dual-material theoretical heating reference curve to generate dynamic temperature evolution trajectory data of the target electromagnetic element after measurement and calibration. Based on the dynamic temperature evolution trajectory data, the current effective temperature value of the dynamic temperature evolution trajectory data at the end node of the current temperature rise monitoring cycle, after completing various error deductions and compensations, is extracted and used as the real-time calibration temperature value output of the target electromagnetic component.

2. The method for dynamic temperature measurement and calibration of a heating element according to claim 1, characterized in that: Before the step of simultaneously extracting the first temperature data of the conductive coil and the second temperature data of the magnetic medium from the temperature data acquisition terminal, there is also a process of performing physical parameter acquisition and conversion, specifically: The timing current data of the conductive coil and the timing voltage data of the magnetic medium are extracted synchronously through the electrical sensing components. The runtime current data and the runtime voltage data are imported into the physical conversion calculation unit, and a heat power conversion operation based on the electromagnetic loss physical model is performed. The first temperature data of the conductive coil and the second temperature data of the magnetic medium are generated and reported to the temperature data acquisition terminal, respectively. The step of synchronously extracting the operating sequence current data of the conductive coil and the operating sequence voltage data of the magnetic medium through the electrical sensing component includes: Based on preset multi-channel data synchronous acquisition rules, the original electrical signal sampling sequence output by the electrical sensing component is obtained; The original electrical signal sampling sequence is subjected to timing channel alignment and high-frequency filtering operations to eliminate sensor sampling interference spikes and reconstruct the runtime timing current data and runtime timing voltage data with standard timestamps. The runtime timing current data and the runtime timing voltage data are imported into the inter-device data communication channel to perform phase state locking.

3. The method for dynamic temperature measurement and calibration of a heating element according to claim 2, characterized in that: The simultaneous extraction of the surface radiation temperature set of adjacent heating devices and the physical distance mapping table between devices includes: In response to the rack node temperature measurement and scanning command, the current rack equipment operation status table is parsed to extract the set of operating heat-generating devices in the target rack area; Based on the spatial relative position matching of the set of operating heat-generating devices, adjacent heat-generating devices located within a preset thermal radiation interference physical radius are selected and regarded as the surrounding adjacent heat-generating devices. Establish a data communication channel between the device and each of the surrounding adjacent heat-generating devices, extract the real-time outer shell infrared temperature measurement data stream of the corresponding device, and generate the surface radiation temperature set by unifying the temperature measurement data format and synchronizing the timestamp. The physical installation distance parameters between each of the surrounding adjacent heat-generating devices and the target electromagnetic component are analyzed, and the heat dissipation attenuation coefficient of the airflow environment in the cabinet is extracted and mapped to generate a physical distance mapping table between the devices.

4. The method for dynamic temperature measurement and calibration of a heating element according to claim 3, characterized in that: The specific processing logic of the external environment heat filtering module, the dual heat source theoretical curve generation module, and the waveform dynamic fitting module includes: The process of generating the environmental additional temperature rise value conducted by adjacent devices includes: performing a heat conduction time delay conversion based on physical distance and heat dissipation duct resistance, and extracting the environmental additional temperature rise value. Based on the environmental additional temperature rise value, an interference removal operation is performed on the measured temperature time series set, including: performing a difference subtraction operation between the measured temperature time series set and the environmental additional temperature rise value to generate the real temperature rise trajectory data after deducting external heat interference; The joint construction of the dual-material theoretical heating reference curve includes: performing weighted compensation and phase alignment aggregation operations on the first temperature data and the second temperature data to generate the dual-material theoretical heating reference curve. The calculation of the internal abnormal temperature rise risk value includes: comparing the waveform shape of the actual temperature rise trajectory data with the theoretical temperature rise reference curve of the dual materials, and calculating the internal abnormal temperature rise risk value that characterizes the probability of material aging or insulation damage of the target electromagnetic component.

5. The method for dynamic temperature measurement and calibration of a heating element according to claim 4, characterized in that: The calculation of heat conduction time delay based on physical distance and heat dissipation duct resistance is used to extract the additional environmental temperature rise value, including: An amplitude feature extraction operation is performed on the surface radiation temperature set to generate surface radiation temperature fluctuation values, as well as physical installation spacing parameters and heat dissipation attenuation coefficients recorded in the physical distance mapping table between devices; Combining the physical installation spacing parameters and the heat dissipation attenuation coefficient, spatial thermal damping attenuation calculation is performed on the surface radiation temperature fluctuation value to calculate the estimated value of external environmental radiation interference temperature rise applied to the surface of the target electromagnetic component housing. Combining the physical time lag property of the heat conduction process, the estimated temperature rise of the external environmental radiation interference is subjected to time axis lag compensation mapping to extract the additional environmental temperature rise value that is aligned with the current temperature measurement phase. By comparing the waveform of the actual temperature rise trajectory data with the theoretical temperature rise reference curve of the dual-material system, the internal abnormal temperature rise risk value, which characterizes the probability of material aging or insulation damage of the target electromagnetic component, is calculated, including: The time-series temperature rise rate and peak characteristics of the actual heating trajectory data and the theoretical heating reference curve of the dual materials are extracted respectively; Within the preset allowable range of heat conduction time hysteresis, perform time axis translation fitting on both, and compare the differences in waveform morphology after fitting. By comprehensively calculating the temperature amplitude deviation and waveform time difference after fitting, a comprehensive heat transfer time delay evaluation index characterizing the degree of heat conduction hindrance of the coil and magnetic core is calculated, and the comprehensive heat transfer time delay evaluation index is used as the internal abnormal temperature rise risk value.

6. The method for dynamic temperature measurement and calibration of a heating element according to claim 5, characterized in that: Based on the environmental additional temperature rise value and the internal abnormal temperature rise risk value, external temperature measurement error compensation parameters and internal temperature measurement reference calibration coefficients are generated, including: In response to the additional environmental temperature rise being greater than a preset environmental reference threshold, the additional environmental temperature rise is used as an environmental heat error compensation reference, and the external temperature measurement error compensation parameter is generated by mapping it; in response to the additional environmental temperature rise not being greater than the preset environmental reference threshold, the external temperature measurement error compensation parameter is calibrated to a reference zero value. In response to the internal abnormal temperature rise risk value not meeting the preset safety benchmark range, it is determined that the theoretical heating evolution law of the heating element has deviated, and the internal temperature measurement benchmark calibration coefficient is calculated and generated. Perform data feedback correction on the measured temperature time series set and the dual-material theoretical heating reference curve, including: The external temperature measurement error compensation parameter is fed back and linked to the measured temperature time series set, and a surface temperature measurement error elimination operation based on the external temperature measurement error compensation parameter is performed. The internal temperature measurement reference calibration coefficient is fed back to the dual-material theoretical temperature rise reference curve to perform a data calibration operation to correct the reference amplitude of the dual-heat source theoretical curve. Based on the dynamic temperature evolution trajectory data, the current effective temperature value of the dynamic temperature evolution trajectory data at the end node of the current temperature rise monitoring cycle, after completing various error deductions and compensations, is extracted and used as the real-time calibration temperature value of the target electromagnetic component, including: Extract the final state time node of the dynamic temperature evolution trajectory data within the current temperature rise monitoring cycle; The current effective temperature value matching the final state time node is analyzed, and the real-time calibration temperature value is calculated, generated, and output.

7. The method for dynamic temperature measurement and calibration of a heating element according to claim 6, characterized in that: After the step of using the environmental additional temperature rise value as the environmental heat error compensation benchmark and mapping it to generate the external temperature measurement error compensation parameter in response to the environmental additional temperature rise value being greater than the preset environmental reference threshold, the process further includes a step of triggering the rack global heat dissipation linkage intervention process in parallel, specifically: Based on the comparison results of the time-series maintenance characteristics of the environmental heat error compensation benchmark and the preset accumulation warning tolerance, a transition state identifier is triggered and generated to characterize the local air duct heat accumulation in the physical cabinet space where the target electromagnetic component is located. Based on the environmental heat error compensation benchmark and the physical distance mapping table between the devices, reverse tracking of the heat source distribution location is performed to lock the surrounding adjacent heat-generating devices that cause heat accumulation in the cabinet. A dynamic heat dissipation coordination command is generated and configured to flow to the rack space temperature control gateway to trigger the rack exhaust fan array that matches the adjacent heat-generating equipment to perform a speed increase action, thereby controlling the rack exhaust fan array to perform physical exhaust cooling action.

8. The method for dynamic temperature measurement and calibration of a heating element according to claim 7, characterized in that: The combined calculation of the temperature amplitude deviation and waveform time difference after fitting yields a comprehensive heat transfer time delay evaluation index characterizing the degree of heat conduction hindrance in both the coil and the magnetic core, including: Synchronously extract the operating sequence current data of the conductive coil during the current temperature rise monitoring cycle, calculate the current change rate between adjacent sampling points, and extract the electrical load change rate parameter characterizing the load fluctuation amplitude; In response to the electrical load change rate parameter being greater than a preset load fluctuation threshold, the load mutation assessment branch is triggered and invoked to perform a dynamic weighted fusion operation on the temperature amplitude deviation and the waveform time difference to reduce the calculation deviation of component temperature conduction hysteresis, thereby generating the comprehensive heat transfer time delay assessment index. In response to the electrical load change rate parameter not being greater than the load fluctuation threshold, the load stability assessment branch is triggered and invoked to perform feature comparison and aggregation operations on the temperature amplitude deviation and the waveform time difference to generate the comprehensive heat transfer time delay assessment index.

9. A method for dynamic temperature measurement and calibration of a heating element according to claim 8, characterized in that: The load mutation assessment branch is invoked to perform a dynamic weighted fusion operation on the temperature amplitude deviation and the waveform time difference to reduce the calculation deviation of component temperature conduction hysteresis, generating the comprehensive heat transfer hysteresis assessment index, including: Numerical normalization is performed on the temperature amplitude deviation and the waveform time difference respectively, and dimensionless amplitude deviation comparison coefficient and time difference comparison coefficient are generated. Based on the current value of the electrical load change rate parameter, perform a numerical inverse mapping operation to calculate and generate an amplitude ratio weighting coefficient in the range of zero to one. The amplitude comparison weighting coefficient is assigned to the amplitude deviation comparison coefficient, and the corresponding remaining weight is extracted and assigned to the time difference comparison coefficient. The weighted aggregation operation of the two is performed to calculate the insulation aging risk index, which characterizes the actual insulation aging risk under load surge conditions, and it is used as the comprehensive heat transfer time delay evaluation index. The load stability assessment branch is invoked to perform feature comparison and aggregation operations on the temperature amplitude deviation and the waveform time difference to generate the comprehensive heat transfer time delay assessment index, including: A preset steady state determination weight table is retrieved, and the steady state determination weight table is configured to assign a fusion determination weight to the temperature amplitude deviation being higher than the waveform time difference. The temperature amplitude deviation and the waveform time difference are imported into the steady state judgment weight table for feature correlation mapping, and the insulation aging risk index that characterizes the internal thermal conductivity abnormality under the load steady state is calculated and used as the comprehensive heat transfer time delay evaluation index.

10. A device for dynamic temperature measurement and calibration of heating elements, characterized in that: The device is used to perform the method for dynamic temperature measurement and calibration of a heating element as described in any one of claims 1-9, comprising: The multi-source heterogeneous parameter sensing module is configured to: in response to the triggering of the temperature rise monitoring cycle, synchronously extract the measured temperature time series set of the target electromagnetic element, which is the entity representing the heating element, and the first temperature data of the conductive coil and the second temperature data of the magnetic medium, which are the internal structures of the target electromagnetic element; at the same time, extract the surface radiation temperature set of the surrounding adjacent heating devices and the physical distance mapping table between the devices; The spatiotemporal decoupling and feature fitting module is configured to: import the surface radiation temperature set of the surrounding adjacent heating devices and the physical distance mapping table between the devices into a pre-configured external environment heat filtering module to parse and generate the environmental additional temperature rise value conducted by the adjacent devices; perform interference removal operation on the measured temperature time series set based on the environmental additional temperature rise value to generate the true temperature rise trajectory data after deducting external heat interference; import the first temperature data of the conductive coil and the second temperature data of the magnetic medium into a pre-set dual heat source theoretical curve generation module to jointly construct a dual-material theoretical temperature rise reference curve; and send the true temperature rise trajectory data and the dual-material theoretical temperature rise reference curve into the waveform dynamic fitting module to perform waveform shape comparison and calculate the internal abnormal temperature rise risk value. The closed-loop calibration and status output module is configured to: generate external temperature measurement error compensation parameters for eliminating external temperature measurement interference and internal temperature measurement reference calibration coefficients for correcting the dual-material theoretical temperature rise reference curve, based on the environmental additional temperature rise value and the internal abnormal temperature rise risk value; perform data feedback correction on the measured temperature time series set and the dual-material theoretical temperature rise reference curve based on the external temperature measurement error compensation parameters and the internal temperature measurement reference calibration coefficients, generating dynamic temperature evolution trajectory data of the target electromagnetic element after measurement and calibration; and extract the current effective temperature value of the dynamic temperature evolution trajectory data at the end node of the current temperature rise monitoring cycle after completing various error deductions and compensations, and output it as the real-time calibration temperature value of the target electromagnetic element.

Images