An amorphous transformer core loss prediction system

Abstract

The application discloses an amorphous transformer core loss prediction system, which comprises a batch process fingerprint modeling module, a running characteristic synchronous acquisition module, a phase locking stress disturbance module, a standard no-load conversion module, a loss component inversion module, a prediction and early warning output module and a baseline update control module; the batch process fingerprint modeling module is used for acquiring manufacturing end data and factory test data of a single amorphous transformer core, establishing a batch process fingerprint according to the core number, and obtaining an initial loss baseline through no-load loss conversion; and the application relates to the technical field of power transformer loss prediction. The amorphous transformer core loss prediction system binds the manufacturing end data, the factory test data and subsequent running loss prediction by establishing the batch process fingerprint of the single amorphous core, so that the initial loss baseline is single-device targeted, and the judgment deviation caused by the average parameters of the same model is reduced.

Description

Technical Field

[0001] This invention relates to the field of power transformer loss prediction, and more specifically, to an amorphous transformer core loss prediction system. Background Technology

[0002] Amorphous alloy cores are widely used in distribution transformers due to their low no-load loss. However, their losses are affected by factors such as strip batch, winding tension, overlap structure, annealing process, clamp preload, operating temperature, harmonics, and DC bias. In the prior art, patent CN102360768B discloses an amorphous core and its manufacturing method, focusing on wide-width amorphous strips, stepped heat treatment, and low-noise structures; patent CN104502738A discloses a transformer loss detection device, mainly detecting input and output voltage, current, and energy consumption; patent CN112180299A uses standard core loss as an auxiliary verification method in inter-turn short-circuit monitoring; and patent CN202794326U focuses on online monitoring of core grounding current. While the above solutions can improve manufacturing performance or identify some faults, they do not bind the manufacturing fingerprint of a single iron core to the multi-source characteristics of operation, nor do they perform component inversion on voltage frequency disturbances, harmonics, DC bias, thermal coupling, stress anomalies, and iron core degradation. Therefore, it is difficult to determine the true source of increased losses. Hence, it is necessary to propose this solution. Summary of the Invention

[0003] The purpose of this invention is to provide an amorphous transformer core loss prediction system, which solves the problem that existing core loss prediction systems do not bind the manufacturing fingerprint of a single core to the multi-source characteristics of operation, nor do they perform component inversion on voltage frequency disturbances, harmonics, DC bias, thermal coupling, stress anomalies and core body deterioration, making it difficult to determine the true source of increased loss and failing to meet the application requirements.

[0004] This invention achieves the above objective through the following technical solution: an amorphous transformer core loss prediction system, the system comprising: The module includes a batch process fingerprint modeling module, a synchronous acquisition module for operational features, a phase-locked stress disturbance module, a standard no-load conversion module, a loss component inversion module, a prediction and early warning output module, and a baseline update control module. The batch process fingerprint modeling module is used to obtain the manufacturing end data and factory test data of a single amorphous transformer core, establish the batch process fingerprint according to the core number, and obtain the initial loss baseline by converting the factory no-load loss. The synchronous acquisition module for operational features is used to acquire electrical features, temperature features, and vibration features during operation, and to form a real-time operational feature set with a unified timestamp. The phase-locked stress disturbance module is used to determine the phase offset of the main frequency of core vibration with the fundamental phase of the excitation current as a reference, and to generate the core stress disturbance factor by combining the harmonic content, the vibration amplitude of the clamp and the local temperature rise gradient. The standard no-load conversion module is used to convert the current operating state to a virtual standard no-load condition to obtain a virtual standard no-load loss value. The loss component inversion module is used to determine the loss increment based on the difference between the virtual standard no-load loss value and the initial loss baseline, and decomposes the loss increment into operating disturbance loss component, harmonic additional loss component, DC bias loss component, thermal coupling loss component, stress anomaly loss component and core body deterioration loss component. The prediction and early warning output module is used to output the predicted value of the iron core body loss, the source of the loss anomaly and the early warning level; The baseline update control module is used to control the updating, pausing, or freezing of the initial loss baseline based on the early warning results.

[0005] Furthermore, the batch process fingerprint modeling module includes: Manufacturing parameter binding unit, factory status binding unit, and baseline conversion unit; The manufacturing parameter binding unit is used to bind the amorphous strip thickness, strip width, strip batch number, winding tension, number of winding layers, overlap length, number of overlap layers, annealing temperature, holding time, cooling rate, and clamping preload. The factory status binding unit is used to bind the factory no-load loss, factory excitation current, factory no-load voltage, factory frequency, factory core temperature and factory vibration frequency. The baseline conversion unit is used to convert the factory no-load loss to the reference state corresponding to the standard voltage, standard frequency and standard temperature, and the conversion result is used as the initial loss baseline of the single amorphous iron core.

[0006] Furthermore, the operational feature synchronous acquisition module includes: Electrical data acquisition unit, temperature data acquisition unit, vibration data acquisition unit, and time alignment unit; The electrical acquisition unit is used to acquire three-phase voltage, three-phase current, frequency, load rate, total harmonic distortion, harmonic current content, DC bias, and excitation current waveform. The temperature acquisition unit is used to acquire the core temperature, winding temperature, oil temperature, ambient temperature, lap area temperature, and clamping area temperature. The vibration acquisition unit is used to acquire the vibration amplitude of the clamp, the vibration spectrum of the iron core, and the main frequency of the iron core vibration. The time alignment unit is used to perform timestamp correction, outlier removal, and window averaging on the above data according to the same sampling window.

[0007] Furthermore, the phase-locked stress disturbance module includes: Phase reference establishment unit, vibration phase calculation unit, temperature rise gradient calculation unit, and disturbance factor generation unit; The phase reference establishment unit is used to extract the fundamental phase from the excitation current waveform; The vibration phase calculation unit is used to extract the main frequency phase of the core vibration from the core vibration spectrum and calculate its phase offset relative to the fundamental phase of the excitation current. The temperature rise gradient calculation unit is used to calculate the temperature rise gradient of the overlapping area relative to the average temperature of the iron core, and the temperature rise gradient of the clamping area relative to the average temperature of the iron core, respectively. The disturbance factor generation unit is used to generate core stress disturbance factors based on phase offset, total harmonic distortion, clamp vibration amplitude, temperature rise gradient of lap joint area and temperature rise gradient of clamping area.

[0008] Furthermore, when generating the core stress disturbance factor, the disturbance factor generation unit first converts the phase offset, total harmonic distortion rate, clamp vibration amplitude, temperature rise gradient of the overlapping area and temperature rise gradient of the clamping area into dimensionless deviations, and then normalizes them according to the historical stable operating range of the same amorphous core. When the phase offset deviates from the historical stable range, and the temperature rise gradient of the overlapping area or the temperature rise gradient of the clamping area increases synchronously, the disturbance factor generation unit marks this state as a stress disturbance enhancement state. When the vibration amplitude of the clamp increases but the temperature gradient does not increase synchronously, the disturbance factor generation unit marks this state as an external vibration transmission state.

[0009] Furthermore, the standard no-load conversion module includes: Voltage-frequency conversion unit, temperature conversion unit, operating disturbance deduction unit, and standard loss generation unit; The voltage-frequency conversion unit is used to convert the current voltage and current frequency to a preset standard voltage and a preset standard frequency. The temperature conversion unit is used to convert the core temperature, winding temperature, oil temperature and ambient temperature to a preset standard temperature. The operation disturbance deduction unit is used to deduct the effects of load rate, total harmonic distortion rate, DC bias and ambient temperature changes on the estimated no-load loss. The standard loss generation unit is used to output the virtual standard no-load loss value after excluding operating condition disturbances.

[0010] Furthermore, the loss component inversion module includes: The system includes an incremental determination unit, a first deduction unit, a second deduction unit, and a body degradation determination unit. The increment determination unit is used to calculate the difference between the virtual standard no-load loss value and the initial loss baseline to obtain the loss increment; The first deduction unit is used to determine the operating disturbance loss component based on the voltage deviation, frequency deviation and load rate, and deduct the operating disturbance loss component from the loss increment; The second subtraction unit is used to subtract the harmonic additional loss component, the DC bias loss component, and the thermal coupling loss component in sequence based on the total harmonic distortion rate, the content of each harmonic current, the DC bias amount, and the temperature change. The core degradation determination unit is used to determine the remaining loss as the core degradation loss component after further deducting the stress anomaly loss component.

[0011] Furthermore, the loss component inversion module also includes a component reliability verification unit; The component reliability verification unit is used to calculate the correlation between the operating disturbance loss component and the voltage frequency deviation, the correlation between the harmonic additional loss component and the total harmonic distortion rate, the correlation between the DC bias loss component and the DC bias quantity, the correlation between the thermal coupling loss component and the temperature change, and the correlation between the stress anomaly loss component and the core stress disturbance factor; when When any correlation is lower than the corresponding component confidence threshold, the component confidence verification unit triggers the re-inversion of the loss component and prohibits the use of the remaining loss before the re-inversion as the core body deterioration loss component.

[0012] Furthermore, the prediction and early warning output module includes: Loss prediction unit, source identification unit, and graded early warning unit; The loss prediction unit is used to output the current core body loss prediction value and the change trend within the prediction period based on the initial loss baseline, the core body deterioration loss component and the core stress disturbance factor. The source discrimination unit is used to determine the source of loss anomaly based on the proportion of each loss component in the loss increment, and output the corresponding source labels in voltage frequency disturbance, harmonic distortion disturbance, DC bias disturbance, thermal coupling disturbance, stress disturbance and core body degradation. The graded early warning unit is used to output an abnormal core body loss warning when the core body deterioration loss component continuously exceeds the core body deterioration threshold and the abnormal stress loss component and the phase offset of the core vibration main frequency meet the same direction enhancement condition. When the increase in loss is dominated by operating disturbance loss components, harmonic additional loss components, DC bias loss components, or thermal coupling loss components, an operating condition disturbance warning will be output.

[0013] Furthermore, the baseline update control module includes: Stable state determination unit, baseline slow update unit, and abnormal freeze unit; The stability determination unit is used to determine that the amorphous transformer is in a stable operating state when the voltage deviation, frequency deviation, total harmonic distortion rate, DC bias, load rate, core temperature change rate, core stress disturbance factor and core body deterioration loss component are all within the corresponding stability threshold range. The baseline slow update unit is used to update the initial loss baseline by a single update magnitude of less than one percent of the current initial loss baseline after the preset observation period has been reached in a stable operating state. The abnormal freezing unit is used to stop updating the initial loss baseline when the prediction and early warning output module outputs an abnormal loss warning of the iron core body, an enhanced stress disturbance state, or an operating condition disturbance prompt, and uses the initial loss baseline before stopping the update as the comparison benchmark for subsequent loss component inversion.

[0014] The beneficial effects of this invention are as follows: 1. This invention establishes a batch process fingerprint for a single amorphous iron core, binding manufacturing data, factory test data, and subsequent operational loss predictions together, making the initial loss baseline specific to a single device and reducing judgment bias caused by average parameters of the same model.

[0015] 2. This invention simultaneously collects electrical, temperature, and vibration characteristics, and forms a real-time operating feature set through a unified timestamp. This reduces the impact of different sampling periods on loss component inversion and improves the completeness of the operating status characterization.

[0016] 3. This invention uses the fundamental phase of the excitation current as a reference and combines the phase offset of the main frequency of core vibration, the vibration amplitude of the clamping member, and the local temperature rise gradient to generate a core stress disturbance factor, which can quantitatively characterize the additional losses caused by changes in clamping force, micro-gap of the lap joint, release of residual stress, and local magnetic flux distortion.

[0017] 4. This invention uses virtual standard no-load conditions to calculate and deduct the effects of voltage deviation, frequency deviation, load thermal coupling, harmonic distortion, DC bias, and ambient temperature changes on the loss results, making the output results closer to the loss state of the iron core itself.

[0018] 5. This invention decomposes the loss increment into six types of loss components: operational disturbance, harmonic addition, DC bias, thermal coupling, stress anomaly, and core degradation. Based on the early warning results, it controls the baseline update, pauses, or freezes the system, which can reduce false alarms and baseline contamination and improve the stability of subsequent predictions. Attached Figure Description

[0019] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a system block diagram of the present invention; Figure 2 This is a flowchart illustrating the overall method of the present invention; Figure 3 This is a schematic diagram of the loss component inversion of the present invention; Figure 4 This is a schematic diagram of the anomaly detection logic of the present invention. Detailed Implementation

[0020] The present application will now be described in further detail with reference to the accompanying drawings. It should be noted that the following specific embodiments are only used to further illustrate the present application and should not be construed as limiting the scope of protection of the present application. Those skilled in the art can make some non-essential improvements and adjustments to the present application based on the above application content.

[0021] Example 1: Please see Figure 1-4 This invention provides an amorphous transformer core loss prediction system. This system is used to predict the core loss of an amorphous transformer in operation and identify the sources of increased loss. The system includes a batch process fingerprint modeling module, an operational feature synchronous acquisition module, a phase-locked stress disturbance module, a standard no-load conversion module, a loss component inversion module, a prediction and early warning output module, and a baseline update control module.

[0022] In this embodiment, the rated capacity is... The rated voltage on the high-voltage side is The rated voltage on the low-voltage side is Rated frequency is The oil-immersed amorphous alloy distribution transformer is used as the monitoring object. The system uses standard voltage... Set as the rated voltage on the high-voltage side , standard frequency Set as Standard temperature Set as The electrical characteristic sampling period is... The temperature feature sampling period is The vibration feature sampling period is The system is based on This serves as a unified analysis window, and the data within the same window undergoes timestamp alignment, outlier removal, and window averaging. The advantage of this design is that it reduces the impact of transient noise and single-point anomalies on the loss component inversion results while ensuring data real-time performance.

[0023] The batch process fingerprint modeling module is used to acquire manufacturing data and factory test data for a single amorphous transformer core, and to establish a batch process fingerprint according to the core number. Manufacturing data includes amorphous strip thickness, strip width, strip batch number, winding tension, number of winding layers, overlap length, overlap layer count, annealing temperature, holding time, cooling rate, and clamp preload. Factory test data includes factory no-load loss, factory excitation current, factory no-load voltage, factory frequency, factory core temperature, and factory dominant vibration frequency.

[0024] In practice, the batch process fingerprint modeling module first acquires the aforementioned data through the manufacturing management system, factory testing system, or manual input. Then, it binds the manufacturing data and factory test data corresponding to the same core number to form the batch process fingerprint for a single amorphous core. The advantage of this design is that the system no longer relies solely on the general parameters of transformers of the same model for loss judgment, but instead uses the actual manufacturing differences and factory condition of a single core as the basis for prediction, improving the specificity of the initial benchmark.

[0025] The batch process fingerprint modeling module also includes a baseline conversion unit. This unit converts the factory no-load loss to a reference state corresponding to standard voltage, standard frequency, and standard temperature, and uses the conversion result as the initial loss baseline for that single amorphous core. In this embodiment, the factory no-load loss... for Factory open-circuit voltage for Factory frequency for Factory core temperature for Initial loss baseline Determined according to the following formula:

[0026] in, This is the initial loss baseline, in units of ; This refers to the no-load loss at the factory, in units of... ; and These are the standard voltage and the factory open-circuit voltage, respectively, in units of... ; and These are the standard frequency and the factory default frequency, respectively, in units of... ; and These are the standard temperature and the factory core temperature, respectively, in units of... ; The voltage correction index is calculated using the same type of amorphous iron core. , , The no-load loss was determined by experimental fitting at three voltage points. In this embodiment, we take... ; For frequency correction exponent, by , , The no-load loss was determined by fitting tests at three frequency points. In this embodiment, we take... ; This is a temperature correction factor, determined by the same batch of prototypes. and The no-load loss test results are fitted to determine the value. In this embodiment, we take... The indices and coefficients mentioned above are all derived from type test or factory test data of the same product model, and are not set subjectively by humans. This formula ensures that the factory no-load loss and the subsequent operational equivalent loss are under the same reference conditions, avoiding reference deviations caused by differences in test temperature, voltage, or frequency.

[0027] The synchronous acquisition module for operational characteristics is used to collect electrical, temperature, and vibration characteristics during the operation of amorphous transformers, forming a real-time operational characteristic set with a unified timestamp. The module includes an electrical acquisition unit, a temperature acquisition unit, a vibration acquisition unit, and a time alignment unit. The electrical acquisition unit collects three-phase voltage, three-phase current, frequency, load rate, total harmonic distortion, harmonic current content, DC bias, and excitation current waveform. The temperature acquisition unit collects core temperature, winding temperature, oil temperature, ambient temperature, lap area temperature, and clamping area temperature. The vibration acquisition unit collects clamp vibration amplitude, core vibration spectrum, and core vibration dominant frequency. The time alignment unit performs timestamp correction, outlier removal, and window averaging on the above data according to the same sampling window.

[0028] In one implementation, for data with different sampling frequencies, the time alignment unit uses... Using the analysis window as a baseline, the effective or mean values ​​of electrical characteristics within the window are taken, the mean value of temperature characteristics is taken, and the amplitude and phase of the dominant frequency are taken for vibration characteristics. A real-time operating feature set corresponding to the same analysis window is then generated. The advantage of this design is that it avoids subsequent shifts in loss component inversion caused by inconsistencies in the acquisition times of electrical, temperature, and vibration data.

[0029] The phase-locked stress disturbance module is used to determine the phase offset of the core vibration main frequency with the fundamental phase of the excitation current as a reference, and to generate the core stress disturbance factor by combining the harmonic content, clamp vibration amplitude, and local temperature rise gradient. The phase-locked stress disturbance module includes a phase reference establishment unit, a vibration phase calculation unit, a temperature rise gradient calculation unit, and a disturbance factor generation unit.

[0030] The phase reference establishment unit is used to extract the fundamental phase from the excitation current waveform. The vibration phase calculation unit is used to extract the dominant frequency phase of the iron core vibration from the iron core vibration spectrum. And calculate the phase shift of the core vibration dominant frequency phase relative to the fundamental phase of the excitation current. The formula for its calculation is:

[0031] in, This represents the phase shift of the core vibration frequency, in units of... ; The phase of the dominant frequency of the iron core vibration, in units of ; The fundamental phase of the excitation current, in units of ; This indicates that the phase difference is converted to... Within the range. The reason for using the phase offset rather than the vibration amplitude alone is that external mechanical vibration transmission may only lead to an increase in vibration amplitude, while changes in the micro-gap in the core overlap area, attenuation of clamping force, or release of residual stress usually cause a synchronous shift in the vibration response relative to the excitation phase. Therefore, this parameter can more accurately distinguish between external vibration and internal stress disturbances in the core.

[0032] The temperature rise gradient calculation unit is used to calculate the temperature rise gradient of the lap joint area relative to the average core temperature, and the temperature rise gradient of the clamping area relative to the average core temperature. Specifically, the lap joint temperature rise gradient is determined based on the difference between the lap joint temperature and the average core temperature, and the clamping area temperature rise gradient is determined based on the difference between the clamping area temperature and the average core temperature; both are measured in units of [missing unit]. The difference in Celsius is numerically the same as the difference in Kelvin.

[0033] When generating the core stress disturbance factor, the disturbance factor generation unit first converts the phase offset, total harmonic distortion, clamp vibration amplitude, temperature rise gradient in the lap area, and temperature rise gradient in the clamping area into dimensionless deviations, and then normalizes them according to the historical stable operating range of the same amorphous core. In this embodiment, to reduce simple weighted summation, the root mean square form is used to generate the core stress disturbance factor. The calculation formula is:

[0034] in, is the core stress disturbance factor, which is dimensionless; This refers to the dimensionless deviation of the phase offset. The dimensionless deviation of the total harmonic distortion rate; This refers to the dimensionless deviation of the clamp vibration amplitude. This refers to the dimensionless deviation of the temperature rise gradient in the overlap area. This refers to the dimensionless deviation of the temperature rise gradient in the clamping zone. Each dimensionless deviation is determined by dividing the difference between the current value and the historical stable median by the stable fluctuation scale. The stable fluctuation scale uses... ,in This represents the median absolute deviation of historical stable samples. This processing method ensures that phase, harmonics, vibration, and temperature gradients are on the same dimensionless scale, avoiding the direct addition of data from different units.

[0035] Among them, the historical stable samples were taken from the period after the amorphous transformer was put into operation without triggering any early warnings and when the load rate was within a certain range. to Total harmonic distortion rate not exceeding The DC bias current does not exceed the rated excitation current. The operating window; when a single device has insufficient stable samples. When the timing is right, at least the same capacity, the same batch of strip material, and the same core structure should be used. The initial stable range was established using stable samples from the equipment, and the stable operating data of the equipment reached a certain level. After a period of time, the system switches to a stable range for each individual device. The advantage of this approach is that the threshold source is clear, and it can take into account both the insufficient sample size in the initial stage of operation and the differences between individual devices later on.

[0036] When the phase offset deviates from the historical stable range, and the temperature rise gradient in the overlap area or the clamping area increases synchronously, the disturbance factor generation unit marks this state as an enhanced stress disturbance state. When the vibration amplitude of the clamp increases but the temperature rise gradients in the overlap area and the clamping area do not increase synchronously, the disturbance factor generation unit marks this state as an external vibration transmission state. The advantage of this design is that the system can distinguish the loss risks caused by changes in clamping force, micro-gap in the overlap joint, residual stress release, and local magnetic flux distortion from the vibration increase caused by external mechanical vibration transmission.

[0037] The standard no-load conversion module is used to convert the current operating state to a virtual standard no-load condition, obtaining the virtual standard no-load loss value. The module includes a voltage-frequency conversion unit, a temperature conversion unit, an operating disturbance deduction unit, and a standard loss generation unit. The voltage-frequency conversion unit converts the current voltage and frequency to preset standard voltage and frequency. The temperature conversion unit converts the core temperature, winding temperature, oil temperature, and ambient temperature to preset standard temperatures. The operating disturbance deduction unit deducts the effects of load rate, total harmonic distortion, DC bias, and ambient temperature changes on the estimated no-load loss value. The standard loss generation unit outputs the virtual standard no-load loss value after eliminating operating condition disturbances.

[0038] Let the first The estimated operating losses within each analysis window are: Virtual standard no-load loss value Determined according to the following formula:

[0039] in, For the first The virtual standard no-load loss value for each analysis window, in units of ; This is an estimated loss value under the current operating condition, in units of... ; This is the voltage deviation correction amount, in units of ; This is the frequency deviation correction amount, in units of ; This is the load thermal coupling correction amount, in units of ; This is the harmonic distortion correction amount, in units of ; This is the DC bias correction amount, in units of ; This is the ambient temperature correction amount, in units of... The above corrections are all obtained by fitting test data or historical stable operation data of the same type of transformer. For example, the voltage deviation correction is determined by the calibration curve between voltage deviation and no-load loss variation, and the harmonic distortion correction is determined by the calibration curve between total harmonic distortion rate and the content of each harmonic current and additional loss. All corrections have been converted before being used in this formula. Therefore, their dimensions are consistent.

[0040] Specifically, when the operating voltage is higher than the standard voltage, the standard no-load conversion module deducts the change in magnetic flux density caused by the voltage increase from the current loss estimate; when the operating frequency deviates from the standard frequency, the standard no-load conversion module corrects for the changes in hysteresis loss and eddy current loss caused by the frequency deviation; when the load rate is high, the standard no-load conversion module deducts the thermal coupling effects caused by winding heating, oil temperature rise, and core temperature changes; when the total harmonic distortion rate or DC bias increases, the standard no-load conversion module deducts the additional loss effects caused by non-sinusoidal excitation and bias excitation. The advantage of this design is that it makes the virtual standard no-load loss value closer to the core's actual state, reducing the interference of operating condition fluctuations on the core loss judgment.

[0041] The loss component inversion module is used to determine the loss increment based on the difference between the virtual standard no-load loss value and the initial loss baseline, and to perform component inversion on the loss increment. The loss component inversion module includes an increment determination unit, a first subtraction unit, a second subtraction unit, a body degradation determination unit, and a component reliability verification unit.

[0042] The increment determination unit calculates the difference between the virtual standard no-load loss value and the initial loss baseline to obtain the loss increment. The first deduction unit determines the operating disturbance loss component based on voltage deviation, frequency deviation, and load rate, and deducts the operating disturbance loss component from the loss increment. The second deduction unit deducts the harmonic additional loss component, DC bias loss component, and thermal coupling loss component sequentially based on the total harmonic distortion rate, harmonic current content, DC bias, and temperature change. The core degradation determination unit determines the remaining loss as the core degradation loss component after further deducting the stress anomaly loss component.

[0043] In this embodiment, the core body degradation loss component Determined according to the following formula:

[0044] in, The component representing the degradation loss of the iron core body, in units of ; This is the virtual standard no-load loss value, in units of ; This is the initial loss baseline, in units of ; The operating disturbance loss component, in units of ; Added loss component to harmonics, unit: ; This represents the DC bias loss component, in units of... ; This represents the thermal coupling loss component, in units of... ; This represents the stress anomaly loss component, in units of... .in, Determined by the calibration curves corresponding to voltage deviation, frequency deviation, and load rate; It is determined by the total harmonic distortion rate and the additional loss curves corresponding to the content of each harmonic current; It is determined by the ratio of DC bias to the rated excitation current; It is determined by a thermal coupling model between core temperature, oil temperature, ambient temperature, and winding temperature; It is determined by the relationship between the core stress disturbance factor and the stress-added loss calibration. (Using...) The purpose is to avoid interpreting negative residual values ​​after operating condition disturbance correction as physical degradation losses, thus preventing negative degradation amounts that do not have a physical meaning.

[0045] The operating disturbance loss component is used to characterize the loss changes caused by voltage deviation, frequency deviation, and load rate changes; the harmonic additional loss component is used to characterize the additional core loss caused by harmonic distortion; the DC bias loss component is used to characterize the loss changes caused by the deviation of the magnetization state due to DC bias; the thermal coupling loss component is used to characterize the impact of changes in winding temperature, oil temperature, core temperature, and ambient temperature on the loss; the stress anomaly loss component is used to characterize the additional loss corresponding to the increase of the core stress disturbance factor; and the core body deterioration loss component is used to characterize the incremental core body loss that still exists after deducting the above components.

[0046] The component reliability verification unit is used to calculate the correlation between the operating disturbance loss component and the voltage frequency deviation, the correlation between the harmonic additional loss component and the total harmonic distortion rate, the correlation between the DC bias loss component and the DC bias amount, the correlation between the thermal coupling loss component and the temperature change, and the correlation between the stress anomaly loss component and the core stress disturbance factor. When any correlation is lower than the corresponding component reliability threshold, the component reliability verification unit triggers the re-inversion of that loss component and prohibits the direct use of the remaining loss before the re-inversion as the core body degradation loss component. The advantage of this design is that it can avoid the incorrect attribution of remaining loss to core body degradation due to the unreliability of a certain component inversion, thereby improving the accuracy of loss source judgment.

[0047] The prediction and early warning output module outputs the predicted core loss value, the source of loss anomalies, and the early warning level. The module includes a loss prediction unit, a source discrimination unit, and a graded early warning unit. The loss prediction unit outputs the current predicted core loss value based on the initial loss baseline, the core degradation loss component, and the core stress disturbance factor. It also outputs the trend of change within the prediction period based on the changes in the core degradation loss component over multiple consecutive sampling windows. The prediction period can be... , or .

[0048] The source identification unit determines the source of loss anomalies based on the proportion of each loss component in the loss increment, and outputs corresponding source tags for voltage frequency disturbance, harmonic distortion disturbance, DC bias disturbance, thermal coupling disturbance, stress disturbance, and core degradation. If the operating disturbance loss component has the highest proportion, a voltage frequency disturbance source tag is output; if the harmonic additional loss component has the highest proportion, a harmonic distortion disturbance source tag is output; if the DC bias loss component has the highest proportion, a DC bias disturbance source tag is output; if the thermal coupling loss component has the highest proportion, a thermal coupling disturbance source tag is output; if the stress anomaly loss component has the highest proportion, a stress disturbance source tag is output; if the core degradation loss component continues to increase and its proportion exceeds a set ratio, a core degradation source tag is output.

[0049] In this embodiment, the core degradation early warning threshold and stress disturbance threshold are not based on fixed empirical values, but are determined by historical stable operating data. Any monitored quantity stability threshold Determined according to the following formula:

[0050] in, For monitoring quantity The stability threshold; Monitoring quantity in historical stable samples The percentile; Monitoring quantity in historical stable samples the median; Monitoring quantity in historical stable samples The median absolute deviation. For the core body degradation loss component, Pick The unit is Regarding the core stress disturbance factor, Pick , dimensionless; for phase offset, Pick The unit is The reason for using this threshold determination method is that the first... Percentiles can reflect the upper limit of stable operation, while the median plus three times the robust fluctuation scale can suppress the impact of occasional outliers on the threshold. Taking the larger value of the two can reduce the probability of false alarms.

[0051] The graded early warning unit is used to output an abnormal core loss warning when the core body deterioration loss component continuously exceeds the core body deterioration threshold, and the abnormal stress loss component and the phase offset of the core vibration dominant frequency meet the same-direction enhancement condition. The same-direction enhancement condition means that within a continuous observation period, the abnormal stress loss component shows an upward trend, while the deviation of the core vibration dominant frequency phase offset from the historical stable range increases. In a specific operational judgment, when... continuous The number of analysis windows exceeds its stability threshold, and Exceeding its stability threshold, at the same time When the loss exceeds its stable threshold, the prediction and early warning output module outputs an abnormal warning for the iron core body loss. Since each analysis window is... ,continuous Each analysis window corresponds to This time length can filter out fluctuations in a single sampling while meeting the timely early warning requirements of transformer online monitoring. When the increase in loss is mainly dominated by harmonic additional loss components, DC bias loss components, or thermal coupling loss components, and the core body deterioration loss components do not continuously exceed the threshold, the system only outputs an operating condition disturbance warning and does not determine it as an abnormal core body loss.

[0052] The baseline update control module is used to control the updating, pausing, or freezing of the initial loss baseline based on the early warning results. The baseline update control module includes a stable state determination unit, a baseline slow update unit, and an abnormal freezing unit. The stable state determination unit determines that the amorphous transformer is in a stable operating state when voltage deviation, frequency deviation, total harmonic distortion rate, DC bias, load rate, core temperature change rate, core stress disturbance factor, and core degradation loss components are all within their corresponding stable threshold ranges. The baseline slow update unit updates the initial loss baseline by a single update increment less than one percent of the current initial loss baseline after the stable operating state has been maintained for a preset observation period. The abnormal freezing unit stops the initial loss baseline update when the prediction and early warning output module outputs an abnormal core loss warning, an enhanced stress disturbance state, or an operating condition disturbance indication, and uses the initial loss baseline before the update was stopped as the comparison benchmark for subsequent loss component inversion.

[0053] Regarding baseline updates, the baseline update control module only updates when the voltage deviation does not exceed [a certain threshold]. Frequency deviation does not exceed Total harmonic distortion rate not exceeding The DC bias current does not exceed the rated excitation current. Load rate is to The core temperature change rate does not exceed When the core stress disturbance factor and the core degradation loss component do not exceed their stability thresholds, the amorphous transformer is considered to be in a stable operating state. This stable operating state is maintained continuously. Only then can the baseline slow update unit allow updating of the initial loss baseline, and the magnitude of each update is less than that of the current initial loss baseline. When the system outputs an alert for abnormal core loss, increased stress disturbance, or operational disturbance, the abnormality freeze unit stops updating the initial loss baseline and uses the initial loss baseline before the update stops as the comparison benchmark for subsequent loss component inversion. The advantage of this design is that when the amorphous core is in a long-term stable operating state, the system can adapt to the normal aging process through small updates; when core abnormalities, increased stress disturbances, or operational disturbances occur, the system can prevent abnormal data from being written to the baseline, avoiding contamination of the initial loss baseline, thereby ensuring the stability of subsequent prediction and warning results.

[0054] Example 2: In another embodiment, the amorphous transformer core loss prediction system is deployed on a station-side server and communicates with the transformer online monitoring terminal. Field sensors collect data on three-phase voltage, three-phase current, frequency, load rate, total harmonic distortion (THD), DC bias, excitation current waveform, core temperature, winding temperature, oil temperature, ambient temperature, lap zone temperature, clamping zone temperature, clamp vibration amplitude, and core vibration spectrum. After data filtering, timestamping, and anomaly removal, the online monitoring terminal sends the data to the station-side server.

[0055] After receiving the real-time operating feature set, the station server first calls the batch process fingerprint modeling module to read the batch process fingerprint and initial loss baseline of the corresponding core number, and then calls the phase-locked stress disturbance module to generate the core stress disturbance factor. Subsequently, the standard no-load conversion module converts the current operating state to a virtual standard no-load condition, the loss component inversion module performs orderly deduction and reliability verification of the loss increment, the prediction and early warning output module outputs the predicted value of core body loss, the source of loss anomaly and the early warning level, and the baseline update control module controls the update status of the initial loss baseline according to the early warning results.

[0056] When the system detects an increase in the virtual standard no-load loss value, and a simultaneous increase in total harmonic distortion (THD), DC bias, or load rate, while the core stress disturbance factor remains within the historical stable range, the prediction and early warning output module outputs an operational condition disturbance warning, and the baseline update control module suspends the initial loss baseline update. When the system detects that the THD, DC bias, load rate, and temperature change are all within the stable threshold range, but the core body degradation loss component continuously exceeds the core degradation threshold, and the core vibration dominant frequency phase offset continuously deviates from the historical stable range, the prediction and early warning output module outputs an abnormal core body loss warning, and the baseline update control module freezes the initial loss baseline.

[0057] In a specific application scenario, a certain unit During the operation of the amorphous alloy distribution transformer, the system is in three consecutive... The analysis window identified that the virtual standard no-load loss value remained consistently higher than the initial loss baseline, while the total harmonic distortion remained at [a certain level]. to The DC bias is lower than the rated excitation current. The load rate is at to The core temperature change rate is lower than This indicates that the operating condition disturbance is not significant; however, the core stress disturbance factor continuously exceeds its stability threshold, and the phase offset of the core vibration main frequency continues to increase. At this point, after deducting the operating disturbance loss component, harmonic additional loss component, DC bias loss component, thermal coupling loss component, and stress anomaly loss component, the loss component inversion module still obtains the core body deterioration loss component that continuously exceeds the stability threshold. Based on this, the prediction and early warning output module outputs a core body loss anomaly warning. Through this processing method, the system can avoid misjudging ordinary harmonic disturbances or temperature rise disturbances as core body anomalies.

[0058] Conversely, when the virtual standard no-load loss value increases in another operating window, but the total harmonic distortion rate increases to In the above scenario, when the harmonic-induced additional loss component accounts for the largest proportion of the loss increment, and the core stress disturbance factor, the core vibration dominant frequency phase offset, and the core body degradation loss component all do not exceed their corresponding stability thresholds, the prediction and early warning output module only outputs a harmonic distortion disturbance warning and does not output an abnormal core body loss warning, and the baseline update control module suspends baseline updates. This prevents harmonic transient disturbances from being written into the initial loss baseline and avoids over-warning of the core body condition.

[0059] Example 3: In this embodiment, the amorphous transformer core loss prediction system is applied to a transformer with a rated capacity of The rated voltage on the high-voltage side is The rated voltage on the low-voltage side is Rated frequency is An amorphous alloy distribution transformer. This transformer is installed in the park's power distribution room, and its operating load rate is typically... to During this period, the system collects operating data through voltage and current acquisition devices, oil temperature sensors, iron core surface temperature sensors, lap joint temperature sensors, clamping area temperature sensors, and clamp vibration sensors.

[0060] In this embodiment, the batch process fingerprint modeling module reads the manufacturing data and factory test data of the amorphous iron core. The thickness of the amorphous strip is... The strip width is The winding tension is The overlap length is The number of overlapping layers is Layer, annealing temperature is The heat preservation time is Cooling rate is The clamping preload is The factory test data includes the no-load loss. for Factory open-circuit voltage for Factory frequency for Factory core temperature for The factory-set vibration frequency is .

[0061] The system will use standard voltage Set as , standard frequency Set as Standard temperature Set as Voltage correction index Made from the same type of amorphous iron core Double the rated voltage Double the rated voltage and The no-load loss was obtained by fitting the test under twice the rated voltage. In this embodiment, we take... Frequency correction index Depend on , and The no-load loss test fitting results were obtained in this embodiment. Temperature correction factor From the same batch of prototypes and The results of the no-load loss test were fitted to the values ​​obtained in this embodiment. .

[0062] Therefore, the batch process fingerprint modeling module determines the initial loss baseline of the single amorphous iron core according to the above initial loss baseline calculation formula, and obtains the result after substituting the parameters. Approximately This initial loss baseline is linked to the strip batch, winding tension, overlap structure, annealing process, and clamp preload of the core and stored as a comparison benchmark for subsequent loss component inversion. The advantage of this design is that it avoids ignoring manufacturing differences between different batches of the same type of core, improving the specificity of loss prediction for a single piece of equipment.

[0063] Run the feature synchronous acquisition module to This serves as the analysis window. The electrical feature sampling period is... The temperature feature sampling period is The vibration feature sampling period is The time alignment unit timestamps the data within each analysis window and removes data that exceeds the mean of three adjacent sampling points. The abrupt change points; for the removed voltage, current, load factor, and total harmonic distortion, the window mean is taken; for vibration data extraction... The vibration amplitude and phase corresponding to the dominant frequency are calculated by taking the window average of the temperature data and calculating the temperature rise gradient. The advantage of this design is that it can suppress the impact of transient interference on phase shift and loss components without reducing the timeliness of early warning.

[0064] During a certain runtime period, the system collected the following data in three consecutive analysis windows: the load rates were respectively , and The total harmonic distortion rates are respectively , and The DC bias is respectively the rated excitation current. , and The core temperature change rate is less than The above data are all within the stable operating range, indicating that there are no obvious harmonic disturbances, DC bias disturbances, or thermal coupling disturbances during this period.

[0065] Simultaneously, the phase-locked stress disturbance module extracts the fundamental phase of the excitation current from the excitation current waveform. Extracting from the vibration spectrum of the iron core Main frequency phase And calculate the phase shift of the dominant frequency of the iron core vibration. In three consecutive analysis windows, They are respectively , and The temperature rise gradients in the overlap area are respectively , and The temperature rise gradients in the clamping zone are respectively , and The disturbance factor generation unit converts the phase offset, total harmonic distortion rate, clamp vibration amplitude, temperature rise gradient in the lap area, and temperature rise gradient in the clamping area into dimensionless deviations, and then obtains the core stress disturbance factor in the root mean square form. The three analysis windows correspond to They are respectively , and .

[0066] In this embodiment, the historical stable samples are taken from before the transformer was put into operation. Operational data that has not triggered any alerts for the past day, and the filter criteria is that the load rate is within a certain range. to Total harmonic distortion rate not exceeding The DC bias current does not exceed the rated excitation current. The core temperature change rate does not exceed Based on the stability thresholds determined from this historical stable sample, the core stress disturbance factor threshold is: The phase offset threshold is The threshold value for the degradation loss component of the iron core body is The above threshold is determined by the first... The percentile and median are combined with three times the robust fluctuation scale, and the larger value is used to determine the value, rather than using fixed values ​​based on human experience.

[0067] The standard no-load conversion module obtains virtual standard no-load loss values ​​for three analysis windows based on the current operating status. , respectively , and Taking the second analysis window as an example, the loss component inversion module determines the operating disturbance loss components based on the calibration curve. for Harmonic additional loss components for DC bias loss component for Thermal coupling loss component for Stress abnormal loss component for According to the formula for calculating the degradation loss component of the iron core, after substituting the data from the second analysis window, we obtain... for Because this value exceeds the range determined by historical stable samples. If the core stress disturbance factor and phase offset both exceed the corresponding stability threshold within three consecutive analysis windows, the prediction and early warning output module determines that the amorphous transformer has a risk of abnormal core loss and outputs an early warning of abnormal core loss.

[0068] During this early warning process, the component reliability verification unit further verifies the correlation between each loss component and its corresponding operating characteristics. Since the total harmonic distortion rate does not exceed... The DC bias current does not exceed the rated excitation current. The load rate and temperature change rate are both within a stable range, while the core stress disturbance factor and phase offset continuously exceed the threshold. Therefore, the system labels the source of the anomaly as "stress disturbance accompanied by core degradation" rather than as harmonic distortion disturbance, DC bias disturbance, or thermal coupling disturbance. The advantage of this design is that it avoids directly judging core degradation due to increased operating losses, and it also avoids misattributing stress-related losses to grid harmonics or load temperature rise.

[0069] Upon receiving an abnormal core loss warning, the baseline update control module immediately stops the initial loss baseline update and resets the initial loss baseline from before the warning. This serves as a benchmark for subsequent loss component inversion. Simultaneously, the system outputs maintenance suggestions to the operation and maintenance platform, prompting priority checks on the core overlap area, clamp pre-tightening status, and core fixing structure. The advantage of this design is that it prevents abnormal operating data from being written into the initial loss baseline, avoiding the system mistakenly identifying continuously increasing abnormal losses as new normal conditions.

[0070] During subsequent maintenance, if the maintenance personnel retighten the clamp bolts and retest, the system will continue to function normally. The internally identified core stress disturbance factor is lower than Phase offset is lower than Furthermore, the degradation loss of the iron core itself is lower than that of the iron core. If the baseline update control module is unfrozen, it will still follow the baseline that is less than the current initial loss baseline. The update magnitude is updated slowly. Through this processing method, this embodiment can fully reflect the closed-loop process from data acquisition, phase locking, loss calculation, component inversion, anomaly warning to baseline freezing and recovery update, improving the feasibility of the amorphous transformer core loss prediction system in actual operation and maintenance.

[0071] Therefore, this invention achieves the differentiation and judgment of the sources of increased losses in amorphous transformer cores through the collaborative processing of establishing process fingerprints for batches of single cores, synchronously acquiring multi-source operational features, identifying phase-locked stress disturbances, converting virtual standard no-load loads, orderly inverting loss components, and updating and freezing baselines. Furthermore, by unifying parameter sources, threshold determination rules, and dimensions, the loss prediction process is made repeatable and feasible, thereby improving the accuracy of loss prediction and early warning for amorphous transformer cores.

[0072] The above embodiments provide a detailed description of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A system for predicting core losses in amorphous transformers, characterized in that, The system includes: The module includes a batch process fingerprint modeling module, a synchronous acquisition module for operational features, a phase-locked stress disturbance module, a standard no-load conversion module, a loss component inversion module, a prediction and early warning output module, and a baseline update control module. The batch process fingerprint modeling module is used to obtain the manufacturing end data and factory test data of a single amorphous transformer core, establish the batch process fingerprint according to the core number, and obtain the initial loss baseline by converting the factory no-load loss. The synchronous acquisition module for operational features is used to acquire electrical features, temperature features, and vibration features during operation, and to form a real-time operational feature set with a unified timestamp. The phase-locked stress disturbance module is used to determine the phase offset of the main frequency of core vibration with the fundamental phase of the excitation current as a reference, and to generate the core stress disturbance factor by combining the harmonic content, the vibration amplitude of the clamp and the local temperature rise gradient. The standard no-load conversion module is used to convert the current operating state to a virtual standard no-load condition to obtain a virtual standard no-load loss value. The loss component inversion module is used to determine the loss increment based on the difference between the virtual standard no-load loss value and the initial loss baseline, and decomposes the loss increment into operating disturbance loss component, harmonic additional loss component, DC bias loss component, thermal coupling loss component, stress anomaly loss component and core body deterioration loss component. The prediction and early warning output module is used to output the predicted value of the iron core body loss, the source of the loss anomaly and the early warning level; The baseline update control module is used to control the updating, pausing, or freezing of the initial loss baseline based on the early warning results.

2. The amorphous transformer core loss prediction system according to claim 1, characterized in that, The batch process fingerprint modeling module includes: Manufacturing parameter binding unit, factory status binding unit, and baseline conversion unit; The manufacturing parameter binding unit is used to bind the amorphous strip thickness, strip width, strip batch number, winding tension, number of winding layers, overlap length, number of overlap layers, annealing temperature, holding time, cooling rate, and clamping preload. The factory status binding unit is used to bind the factory no-load loss, factory excitation current, factory no-load voltage, factory frequency, factory core temperature and factory vibration frequency. The baseline conversion unit is used to convert the factory no-load loss to the reference state corresponding to the standard voltage, standard frequency and standard temperature, and the conversion result is used as the initial loss baseline of the single amorphous iron core.

3. The amorphous transformer core loss prediction system according to claim 1, characterized in that, The operational feature synchronous acquisition module includes: Electrical data acquisition unit, temperature data acquisition unit, vibration data acquisition unit, and time alignment unit; The electrical acquisition unit is used to acquire three-phase voltage, three-phase current, frequency, load rate, total harmonic distortion, harmonic current content, DC bias, and excitation current waveform. The temperature acquisition unit is used to acquire the core temperature, winding temperature, oil temperature, ambient temperature, lap area temperature, and clamping area temperature. The vibration acquisition unit is used to acquire the vibration amplitude of the clamp, the vibration spectrum of the iron core, and the main frequency of the iron core vibration. The time alignment unit is used to perform timestamp correction, outlier removal, and window averaging on the above data according to the same sampling window.

4. The amorphous transformer core loss prediction system according to claim 3, characterized in that, The phase-locked stress disturbance module includes: Phase reference establishment unit, vibration phase calculation unit, temperature rise gradient calculation unit, and disturbance factor generation unit; The phase reference establishment unit is used to extract the fundamental phase from the excitation current waveform; The vibration phase calculation unit is used to extract the main frequency phase of the core vibration from the core vibration spectrum and calculate its phase offset relative to the fundamental phase of the excitation current. The temperature rise gradient calculation unit is used to calculate the temperature rise gradient of the overlapping area relative to the average temperature of the iron core, and the temperature rise gradient of the clamping area relative to the average temperature of the iron core, respectively. The disturbance factor generation unit is used to generate core stress disturbance factors based on phase offset, total harmonic distortion, clamp vibration amplitude, temperature rise gradient of lap joint area and temperature rise gradient of clamping area.

5. The amorphous transformer core loss prediction system according to claim 4, characterized in that: When generating the core stress disturbance factor, the disturbance factor generation unit first converts the phase offset, total harmonic distortion rate, clamp vibration amplitude, temperature rise gradient of the lap area and temperature rise gradient of the clamping area into dimensionless deviations, and then normalizes them according to the historical stable operating range of the same amorphous core. When the phase offset deviates from the historical stable range, and the temperature rise gradient of the overlapping area or the temperature rise gradient of the clamping area increases synchronously, the disturbance factor generation unit marks this state as a stress disturbance enhancement state. When the vibration amplitude of the clamp increases but the temperature gradient does not increase synchronously, the disturbance factor generation unit marks this state as an external vibration transmission state.

6. The amorphous transformer core loss prediction system according to claim 1, characterized in that, The standard no-load conversion module includes: Voltage-frequency conversion unit, temperature conversion unit, operating disturbance deduction unit, and standard loss generation unit; The voltage-frequency conversion unit is used to convert the current voltage and current frequency to a preset standard voltage and a preset standard frequency. The temperature conversion unit is used to convert the core temperature, winding temperature, oil temperature and ambient temperature to a preset standard temperature. The operation disturbance deduction unit is used to deduct the effects of load rate, total harmonic distortion rate, DC bias and ambient temperature changes on the estimated no-load loss. The standard loss generation unit is used to output the virtual standard no-load loss value after excluding operating condition disturbances.

7. The amorphous transformer core loss prediction system according to claim 6, characterized in that, The loss component inversion module includes: The system includes an incremental determination unit, a first deduction unit, a second deduction unit, and a body degradation determination unit. The increment determination unit is used to calculate the difference between the virtual standard no-load loss value and the initial loss baseline to obtain the loss increment; The first deduction unit is used to determine the operating disturbance loss component based on the voltage deviation, frequency deviation and load rate, and deduct the operating disturbance loss component from the loss increment; The second subtraction unit is used to subtract the harmonic additional loss component, the DC bias loss component, and the thermal coupling loss component in sequence based on the total harmonic distortion rate, the content of each harmonic current, the DC bias amount, and the temperature change. The core degradation determination unit is used to determine the remaining loss as the core degradation loss component after further deducting the stress anomaly loss component.

8. The amorphous transformer core loss prediction system according to claim 7, characterized in that: The loss component inversion module also includes a component reliability verification unit; The component reliability verification unit is used to calculate the correlation between the operating disturbance loss component and the voltage frequency deviation, the correlation between the harmonic additional loss component and the total harmonic distortion rate, the correlation between the DC bias loss component and the DC bias quantity, the correlation between the thermal coupling loss component and the temperature change, and the correlation between the stress anomaly loss component and the core stress disturbance factor. when When any correlation is lower than the corresponding component confidence threshold, the component confidence verification unit triggers the re-inversion of the loss component and prohibits the use of the remaining loss before the re-inversion as the core body deterioration loss component.

9. The amorphous transformer core loss prediction system according to claim 8, characterized in that, The prediction and early warning output module includes: Loss prediction unit, source identification unit, and graded early warning unit; The loss prediction unit is used to output the current core body loss prediction value and the change trend within the prediction period based on the initial loss baseline, the core body deterioration loss component and the core stress disturbance factor. The source discrimination unit is used to determine the source of loss anomaly based on the proportion of each loss component in the loss increment, and output the corresponding source labels in voltage frequency disturbance, harmonic distortion disturbance, DC bias disturbance, thermal coupling disturbance, stress disturbance and core body degradation. The graded early warning unit is used to output an abnormal core body loss warning when the core body deterioration loss component continuously exceeds the core body deterioration threshold and the abnormal stress loss component and the phase offset of the core vibration main frequency meet the same direction enhancement condition. When the increase in loss is dominated by operating disturbance loss components, harmonic additional loss components, DC bias loss components, or thermal coupling loss components, an operating condition disturbance warning will be output.

10. The amorphous transformer core loss prediction system according to claim 9, characterized in that, The baseline update control module includes: Stable state determination unit, baseline slow update unit, and abnormal freeze unit; The stability determination unit is used to determine that the amorphous transformer is in a stable operating state when the voltage deviation, frequency deviation, total harmonic distortion rate, DC bias, load rate, core temperature change rate, core stress disturbance factor and core body deterioration loss component are all within the corresponding stability threshold range. The baseline slow update unit is used to update the initial loss baseline by a single update magnitude of less than one percent of the current initial loss baseline after the preset observation period has been reached in a stable operating state. The abnormal freezing unit is used to stop updating the initial loss baseline when the prediction and early warning output module outputs an abnormal loss warning of the iron core body, an enhanced stress disturbance state, or an operating condition disturbance prompt, and uses the initial loss baseline before stopping the update as the comparison benchmark for subsequent loss component inversion.

Images