Current transformer with temperature compensation function

Abstract

The invention relates to the technical field of transformers, in particular to a current transformer with a temperature compensation function. The current transformer comprises a magnetic circuit structure used for sensing a primary side current and outputting a secondary side current signal; the signal processing unit is integrated in the current transformer and is used for executing the current measurement process; wherein the current measurement process comprises the following steps: injecting a test signal into the magnetic circuit structure; wherein the frequency of the test signal is higher than the power frequency and is not integer multiples of the power frequency; measuring amplitude response and phase response of the magnetic circuit structure to the test signal; calculating magnetic conductivity response data under the test signal based on the amplitude response and the phase response; according to the magnetic conductivity response data and a preset magnetic conductivity-temperature mapping relation, determining a temperature variation; generating a current compensation coefficient based on the temperature variation; and correcting the original current signal by using the current compensation coefficient to obtain a compensated current signal. According to the invention, the measurement reliability of the mutual inductor under various temperature conditions is improved.

Description

Technical Field

[0001] This application relates to the field of current transformer technology, and in particular to a current transformer with temperature compensation function. Background Technology

[0002] Low-voltage current transformers, as key measuring devices in power systems, are widely used in energy metering, relay protection, and condition monitoring. Their basic working principle involves inducing primary current through a magnetic circuit structure and generating a proportionally reduced current signal on the secondary side, enabling safe measurement of large currents. With the rapid development of smart grids and the power Internet of Things, the requirements for the measurement accuracy of current transformers are constantly increasing. The IEC 61850-9-2 standard has set 0.2S class (ratio error ±0.2%, phase error ±10') as a basic requirement for high-precision metering. However, the measurement accuracy of current transformers is affected by various factors, among which the drift of magnetic circuit characteristics caused by temperature changes is one of the most significant technical challenges.

[0003] In traditional technologies, there are two main solutions to the temperature drift problem: one is to use special magnetic materials with extremely low temperature coefficients (such as microcrystalline alloys) to reduce the impact of temperature through the stability of the material itself; the other is to obtain the ambient temperature through an external temperature sensor and then compensate according to a preset temperature-error curve.

[0004] However, while the first approach reduces temperature sensitivity, its high material cost and low permeability affect the transformer's sensitivity. The second approach is less expensive, but the external temperature sensor cannot accurately reflect the actual operating temperature of the magnetic circuit. This is because the magnetic circuit temperature is significantly affected by various factors, including current heating, ambient temperature, and heat dissipation conditions, resulting in a substantial difference from the ambient temperature. For example, at an ambient temperature of 70°C, when the primary current reaches 120% of its rated value, the temperature in the core area of ​​the magnetic circuit may be 15-20°C higher than the ambient temperature, leading to significant errors in ambient temperature-based compensation. Therefore, neither of these two approaches can solve the technical problem of decreased measurement accuracy in traditional transformers due to temperature variations. Summary of the Invention

[0005] To address the technical problem of decreased measurement accuracy caused by temperature changes in traditional current transformers, this application provides a current transformer with temperature compensation function.

[0006] A current transformer with temperature compensation function, the current transformer comprising: A magnetic circuit structure is used to sense the primary side current and output the secondary side current signal; The signal processing unit, integrated inside the current transformer, is used to execute the current measurement process; The current measurement process includes: A test signal is injected into the magnetic circuit structure; wherein the frequency of the test signal is higher than the power frequency and is not an integer multiple of the power frequency; Measure the amplitude and phase response of the magnetic circuit structure to the test signal; Based on the amplitude response and phase response, calculate the permeability response data under the test signal; The temperature change is determined based on the permeability response data and the preset permeability-temperature mapping relationship; Based on the temperature change, a current compensation coefficient is generated. The original current signal is corrected using a current compensation coefficient to obtain the compensated current signal.

[0007] The aforementioned current transformer with temperature compensation innovatively achieves non-invasive measurement of magnetic circuit characteristics under normal operating conditions by injecting a test signal with a frequency higher than the power frequency but not an integer multiple of the power frequency into the magnetic circuit structure. This design avoids the limitations of traditional technologies: since the test signal frequency is not an integer multiple of the power frequency and its harmonics, additional characteristic information of the magnetic circuit can be independently obtained without affecting the main measurement function. When the test signal passes through the magnetic circuit, its transmission characteristics are directly affected by the current permeability of the magnetic circuit, and permeability, as an intrinsic property of magnetic materials, has a definite physical relationship with temperature.

[0008] This application achieves direct measurement of permeability by measuring the amplitude and phase responses of the magnetic circuit to test signals. Unlike traditional methods that rely on external temperature sensors for indirect inference, this method directly obtains the core parameters reflecting the actual operating state of the magnetic circuit. The amplitude response characterizes the loss characteristics of the magnetic circuit, while the phase response reflects its dynamic response characteristics; together, they constitute a complete characterization of permeability. This direct measurement method based on electromagnetic theory avoids the compensation errors caused by the difference between ambient temperature and the actual temperature of the magnetic circuit in traditional methods.

[0009] Furthermore, this application establishes a precise mapping from the measured signal to the magnetic permeability by calculating the permeability response data based on the amplitude and phase responses. This process utilizes the fundamental electromagnetic equations of the magnetic circuit to correlate the transmission characteristics of the test signal with the magnetic permeability, achieving an accurate characterization of the magnetic circuit state. Since the functional relationship between magnetic permeability and temperature is an inherent property of the material and is not affected by external environmental factors, this method can truly reflect the operating temperature of the magnetic circuit.

[0010] Finally, this application achieves precise compensation for temperature drift by determining the temperature change based on the permeability response data and generating a corresponding current compensation coefficient. This compensation mechanism is directly based on the actual temperature change of the magnetic circuit, rather than the ambient temperature or a hypothetical model, fundamentally solving the problems of hysteresis and inaccuracy in traditional compensation methods. When the temperature changes, the permeability changes accordingly; through real-time measurement and compensation, the current transformer can maintain stable measurement characteristics under various temperature conditions. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the accompanying drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below.

[0012] Figure 1 This is a system structure of a current transformer with temperature compensation function in one embodiment; Figure 2 This is a flowchart illustrating the current measurement process executed by the signal processing unit in one embodiment; Explanation of reference numerals in the attached figures: 10. Magnetic circuit structure; 20. Signal processing unit. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of this application clearer, this application will be described in further detail below. The described embodiments should not be regarded as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0014] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0015] See Figure 1 This embodiment provides a system structure for a current transformer with temperature compensation function. In this embodiment, the transformer includes a magnetic circuit structure 10 and a signal processing unit 20, which work together to achieve high-precision current measurement based on the permeability-temperature characteristics.

[0016] In some embodiments, the magnetic circuit structure 10 is formed by stacking high-permeability silicon steel sheets to create a closed magnetic circuit for sensing primary-side current and outputting secondary-side current signals. The design of the magnetic circuit structure 10 takes into account the operating frequency range (50Hz-1kHz) and current range (0-100A) to ensure good linearity and low-loss characteristics across the entire range. The material selection and geometry of the magnetic circuit structure 10 are optimized to provide predictable permeability-temperature characteristics within the target operating temperature range (-40°C to +85°C), laying the foundation for subsequent temperature compensation.

[0017] In some embodiments, the signal processing unit 20 is integrated inside the current transformer and implemented using a microcontroller, responsible for executing the complete measurement process. The signal processing unit 20 acquires signals through a high-precision ADC (24-bit resolution), generates test signals through a DAC, and possesses sufficient computing power to execute complex signal processing algorithms. The software algorithm of the signal processing unit 20 is the core of the entire system's intelligence, enabling the current transformer to autonomously complete temperature measurement and compensation without external intervention.

[0018] See Figure 2 The following will combine Figure 2 The steps shown illustrate the current measurement process performed by the signal processing unit provided in the embodiments of this application.

[0019] In step S101, a test signal is injected into the magnetic circuit structure.

[0020] The frequency of the test signal is higher than the power frequency but not an integer multiple of the power frequency.

[0021] In this embodiment, the injection of the test signal is the starting point of the entire temperature measurement process. Traditional current transformers only measure power frequency current and cannot obtain information about the temperature characteristics of the magnetic circuit. This embodiment innovatively introduces a non-power frequency test signal. By analyzing the response characteristics of the magnetic circuit to this signal, the change in permeability is indirectly measured, thereby deriving the temperature change. The selection of the test signal frequency is crucial: if the frequency is too low (close to the power frequency), it will be affected by power grid interference; if the frequency is too high, the skin effect of the magnetic circuit will significantly affect the measurement accuracy; if the frequency is an integer multiple of the power frequency, it will be impossible to distinguish between the test signal and power grid harmonics.

[0022] In some embodiments, the test signal is a single-frequency sine wave with a frequency set to 70Hz (higher than the 50Hz power frequency and not an integer multiple thereof). The choice of 70Hz is based on the following considerations: First, this frequency is far from the 50Hz power frequency and its harmonics (100Hz, 150Hz, etc.), avoiding power grid harmonic interference; second, 70Hz is still within the effective operating frequency band of the magnetic circuit, and the skin effect is negligible; third, the 70Hz signal is easy to generate and detect, with moderate hardware requirements. The amplitude of the test signal is controlled within the linear operating range of the magnetic circuit (typically 5%-10% of the rated current) to avoid magnetic circuit saturation.

[0023] In other embodiments, the test signal uses a multi-frequency swept signal (e.g., five discrete frequencies within the 60-80Hz range). By analyzing the response characteristics at different frequencies, a dispersion characteristic curve is constructed, improving the robustness of temperature measurement. Although multi-frequency testing increases measurement time, it can effectively distinguish temperature effects from other interference factors, making it particularly suitable for industrial sites with complex electromagnetic environments.

[0024] The timing of the test signal injection is also carefully designed. In some embodiments, the test signal is injected near the zero-crossing point of the power frequency current, when the primary current is at its minimum, minimizing its impact on system operation, and simultaneously maximizing the signal-to-noise ratio of the test signal. The signal processing unit 20 precisely controls the timing of the test signal injection by monitoring the zero-crossing point of the power frequency current in real time, ensuring measurement accuracy and system stability.

[0025] In step S102, the amplitude response and phase response of the magnetic circuit structure to the test signal are measured.

[0026] In this embodiment, the measurement of amplitude response and phase response is a crucial step in obtaining the characteristics of the magnetic circuit. Magnetic permeability, as a core parameter of the magnetic circuit, directly affects the turns ratio and phase characteristics of the transformer, and is closely related to temperature. By accurately measuring the transmission characteristics of the test signal in the magnetic circuit, the change in magnetic permeability can be indirectly deduced, thereby determining the temperature change.

[0027] In some embodiments, the amplitude response is obtained by sampling the secondary-side signal using a high-precision ADC and calculating its effective value. Specifically, the signal processing unit 20 performs 128-point synchronous sampling of the secondary-side signal, removes power frequency interference using digital filtering, and then calculates the amplitude of the test signal component. To improve measurement accuracy, the system employs lock-in amplification technology, multiplying the sampled signal with a reference signal that is in phase and frequency with the test signal and then integrating the result, effectively suppressing broadband noise and interference from unrelated frequency components.

[0028] In some embodiments, the phase response is obtained by measuring the time difference between the primary-side test signal and the secondary-side response signal. Specifically, the zero-crossing times t1 and t2 of the primary-side test signal and the secondary-side response signal are recorded, with a phase difference Δφ = 360° × (t2 - t1) / T (where T is the test signal period). To eliminate the influence of system delay, the system undergoes zero-point calibration before leaving the factory, recording the inherent phase offset and compensating for it in subsequent measurements.

[0029] In other embodiments, the amplitude response and phase response are obtained simultaneously using spectral analysis techniques. The signal processing unit 20 performs FFT analysis on the primary and secondary signals to extract the complex response at the test signal frequency points; the magnitude of this response is the amplitude response, and the phase angle is the phase response. This method can obtain complete frequency domain information in a single measurement, making it particularly suitable for multi-frequency testing scenarios.

[0030] It is worth noting that noise and interference during the measurement process can significantly affect the accuracy of the results. In some embodiments, the system uses an averaging method based on multiple measurements to reduce the impact of random noise; in other embodiments, the system monitors the signal-to-noise ratio in real time and automatically retryes the measurement when the signal-to-noise ratio falls below a threshold to ensure data reliability. These measures enable the measurement accuracy of the amplitude response and phase response to reach 0.1% and 0.05°, respectively, laying the foundation for subsequent permeability calculations.

[0031] In step S103, the permeability response data under the test signal is calculated based on the amplitude response and phase response.

[0032] In this embodiment, the calculation of permeability response data serves as a bridge connecting the measurement signal and temperature change. Magnetic permeability, as a fundamental property of magnetic materials, has a definite functional relationship with temperature. By analyzing the transmission characteristics of the test signal in the magnetic circuit, the current permeability value can be deduced, thereby determining the temperature change.

[0033] In some embodiments, the permeability response data is calculated as follows: First, the complex impedance Z = R + jX of the magnetic circuit is calculated based on the amplitude response A and the phase response φ; then, the complex impedance is converted into complex permeability μ = μ' - jμ'' based on the magnetic circuit geometry parameters and the number of coil turns; finally, the real part μ' of the complex permeability is extracted as the permeability response data. This calculation process is based on the fundamental equations of the magnetic circuit and takes into account the effects of core losses and winding resistance to ensure the accuracy of the results.

[0034] In other embodiments, the system employs a calibration method to directly establish the relationship between amplitude response and permeability. Before leaving the factory, the current transformer is tested at standard temperature points (e.g., -25°C, 25°C, 70°C), and the amplitude response values ​​at each temperature are recorded to construct an amplitude response-permeability mapping table. In actual operation, the system calculates the current permeability value based on the measured amplitude response through table lookup and interpolation. This method avoids complex theoretical calculations and is more suitable for resource-constrained embedded systems.

[0035] In some embodiments, the system also considers the effect of DC bias on permeability measurement. When a DC component exists on the primary side, the operating point of the magnetic circuit shifts, and the permeability characteristics change. The signal processing unit 20 corrects the permeability calculation results by monitoring the DC component of the primary side current, ensuring accurate temperature measurement even when a DC bias exists.

[0036] In step S104, the temperature change is determined based on the permeability response data and the preset permeability-temperature mapping relationship.

[0037] In this embodiment, the permeability-temperature mapping relationship is the core of temperature measurement. The permeability of magnetic materials exhibits a specific pattern with temperature: below the Curie temperature, the permeability generally decreases with increasing temperature; approaching the Curie temperature, the permeability drops sharply. By establishing a precise permeability-temperature mapping relationship, the measured permeability value can be converted into a temperature value.

[0038] In some embodiments, the permeability-temperature mapping relationship is represented by a polynomial fitting: T = a0 + a1μ + a2μ² + ... + a n μⁿ, where T is temperature, μ is permeability, and a0...a n These are the fitting coefficients. The polynomial has a fitting error of less than 0.5℃ within a temperature range of -40℃ to +85℃, meeting the accuracy requirements of the current transformer. The fitting coefficients were obtained through laboratory calibration, taking into account individual differences in the magnetic circuit structure 10 and its materials.

[0039] In other embodiments, the permeability-temperature mapping relationship is implemented using a piecewise linear function. The temperature range is divided into multiple intervals (e.g., -40℃ to 0℃, 0℃ to 40℃, 40℃ to 85℃), and a linear function is used to approximate each interval. The piecewise linear method is computationally simple and resource-efficient, making it particularly suitable for embedded systems. Although the accuracy is slightly lower than that of polynomial fitting, the error is controlled within ±1℃ in practical applications, fully meeting the compensation requirements of current transformers.

[0040] In some embodiments, the system also implements an online update function for the permeability-temperature mapping relationship. When the temperature sensor data is reliable, the system pairs the measured temperature with the permeability data to correct the mapping relationship and compensate for characteristic drift caused by long-term aging. This self-learning capability enables the transformer to maintain high accuracy throughout its service life.

[0041] In step S105, a current compensation coefficient is generated based on the temperature change.

[0042] In this embodiment, the generation of the current compensation coefficient is a crucial step in temperature compensation. Temperature changes lead to changes in permeability, which in turn affects the turns ratio and phase characteristics of the transformer. By using a pre-established temperature-error characteristic model, the measurement error at the current temperature can be calculated, and the corresponding compensation coefficient can be generated.

[0043] In some embodiments, the current compensation coefficient is obtained by looking up a pre-stored temperature-error mapping table. This mapping table is obtained through full-temperature range testing before leaving the factory and records the ratio error and phase angle error of the current transformer at different temperatures. The signal processing unit 20 looks up the corresponding error value from the mapping table based on the determined temperature change and generates the compensation coefficient. The lookup process uses linear interpolation to ensure a smooth transition between temperature points.

[0044] In other embodiments, the system employs an analytical model to calculate the compensation coefficient. Based on the equivalent circuit model and magnetic circuit characteristics of the current transformer, the mathematical relationship between temperature and measurement error is derived: ΔI / I = f(ΔT), where ΔI / I is the relative error and ΔT is the temperature change. This model considers the influence of multiple factors such as permeability and winding resistance, achieving high calculation accuracy, but requiring more computational resources.

[0045] In some embodiments, the system also considers the influence of load conditions on the compensation coefficient. The temperature characteristics of the current transformer vary slightly with load current, and the signal processing unit 20 fine-tunes the compensation coefficient based on the current load current magnitude to achieve more accurate compensation. This dynamic adjustment mechanism enables the current transformer to maintain high accuracy under various operating conditions.

[0046] In step S106, the original current signal is corrected using the current compensation coefficient to obtain the compensated current signal.

[0047] In this embodiment, the correction of the current signal is the final step in the entire temperature compensation process and directly affects the output accuracy. The original current signal contains errors caused by temperature; by applying a compensation coefficient, these errors can be effectively eliminated or significantly reduced, thereby improving measurement accuracy.

[0048] In some embodiments, current compensation is achieved through multiplication: I_compensated = I_original × (1 + K), where I_compensated is the compensated current, I_original is the original current, and K is the compensation coefficient. This method is simple and efficient, and suitable for ratio error compensation. For phase error, the system employs digital phase-shifting technology, adjusting the signal phase through an FIR filter to control the phase error within ±2'.

[0049] In other embodiments, the system employs a more complex compensation algorithm to simultaneously address amplitude and phase errors. The signal processing unit 20 represents the original current signal in complex form I = I_m∠θ, and then applies complex compensation coefficients C = C_m∠φ to obtain the compensated signal I' = I × C. This method can simultaneously correct amplitude and phase errors, making it particularly suitable for high-precision metrology applications.

[0050] In some embodiments, the system also implements a smooth transition in the compensation process. When the temperature changes rapidly, the compensation coefficient also changes rapidly, which may cause abrupt changes in the output signal. To avoid this, the system performs low-pass filtering or uses a ramp change on the compensation coefficient to ensure the continuity of the output signal and meet the power system's requirements for signal stability.

[0051] In the actual operating environment of power systems, low-voltage current transformers face complex and variable electromagnetic interference. A single measurement may be affected by transient interference, leading to distortion of the permeability response data and consequently, temperature measurement errors. For example, transient interference caused by switching operations may cause abrupt changes in permeability measurements, which may be misinterpreted as rapid temperature changes, triggering unnecessary current compensation and ultimately reducing measurement accuracy. The multi-measurement process reliability verification mechanism proposed in this embodiment effectively solves this problem, enabling the transformer to maintain high-precision measurements even in complex electromagnetic environments.

[0052] In this embodiment, when the signal processing unit 20 determines the temperature change, it first acquires the permeability response data of the most recent N measurement processes, where N is a positive integer ≥3 and includes the current measurement process. The selection of the N value is based on statistical principles and engineering experience: N=3 can provide basic trend judgment, but has limited anti-interference capability; N=5 achieves a good balance between computational complexity and reliability; N>7 results in diminishing marginal benefits and increases the delay of the signal processing unit 20. In some embodiments, the signal processing unit 20 dynamically adjusts the N value according to the level of environmental interference: N=3 in a stable environment and N=7 in a strong interference environment, achieving an optimal balance between reliability and real-time performance.

[0053] Determining the trend of permeability variation is a core aspect of this embodiment. In some embodiments, the signal processing unit 20 uses linear regression analysis to analyze the most recent N measured permeability values, calculating the slope k and correlation coefficient r. The slope k reflects the rate of change of permeability, and the correlation coefficient r reflects the regularity of the change. For example, when k = -0.2 and r = 0.95, it indicates that the permeability is steadily decreasing, consistent with the expected pattern of a slow temperature increase; while when k = 1.5 and r = 0.3, it indicates that the permeability changes abruptly and irregularly, possibly due to transient interference.

[0054] The expected pattern of temperature change serves as a benchmark for judging measurement reliability. In some embodiments, the signal processing unit 20 establishes a temperature change model based on thermodynamic principles: under conditions of no sudden external heat source changes, the rate of temperature change of the current transformer typically does not exceed 0.5℃ / min, corresponding to a permeability change rate not exceeding 0.3% / min. This expected pattern considers the current transformer's heat capacity, heat dissipation conditions, and typical operating environment, and is verified through extensive experimental data. When the permeability change trend is consistent with this expected pattern (e.g., the slope k is between -0.4 and 0.1 and the correlation coefficient r > 0.8), the signal processing unit 20 determines that the current measurement process is reliable; otherwise, it determines it is unreliable.

[0055] If the current measurement process is deemed reliable, the signal processing unit 20 directly determines the temperature change based on the permeability response data of the current measurement process using a preset permeability-temperature mapping relationship. This processing method fully utilizes the latest measurement data, ensuring the timeliness of temperature compensation. In some embodiments, the signal processing unit 20 also performs weighted processing on the current measurement data, giving higher weight to the most recent measurement points to better reflect the temperature change trend.

[0056] When the current measurement process is determined to be unreliable, the signal processing unit 20 does not use the current measurement data, but instead determines the temperature change based on the permeability response data corresponding to reliable historical measurement processes. In some embodiments, the signal processing unit 20 automatically backtracks the most recent M confirmed reliable measurement data (M≥2) and calculates their average value as the current temperature change. For example, if the 1st, 3rd, and 4th measurements out of the last 5 measurements are determined to be reliable, then the average permeability of these 3 measurements is used to determine the temperature. This processing method effectively avoids measurement errors caused by transient interference while maintaining the continuity of temperature measurement.

[0057] In some embodiments, the signal processing unit 20 also implements a self-learning capability for reliability determination. By recording the degree of agreement between historical determination results and actual temperature changes, the signal processing unit 20 dynamically adjusts the determination threshold (the threshold of the rank correlation coefficient r) to adapt to transformer aging and environmental changes. For example, as the usage time increases, the signal processing unit 20 may gradually adjust the correlation coefficient threshold from 0.8 to 0.75 to adapt to the slow changes in magnetic circuit characteristics.

[0058] In this embodiment, by analyzing the trends of multiple measurement process data and comparing them with expected patterns, the system intelligently identifies and eliminates interfered measurement data, ensuring the reliability of the temperature change determination.

[0059] In the aforementioned S101~S106, we used a single-frequency test signal (e.g., 70Hz) to measure the permeability-temperature characteristics. However, in actual power system environments, single-frequency testing has significant limitations: when the test frequency happens to be at a strong interference frequency, the measurement results may be severely distorted; furthermore, a single frequency cannot fully reflect the frequency characteristics of the magnetic circuit and makes it difficult to distinguish temperature effects from other interference factors. This embodiment innovatively introduces a multi-frequency testing mechanism, which analyzes the response characteristics of the magnetic circuit at multiple frequency points and constructs a dispersion characteristic curve, significantly improving the robustness and accuracy of temperature measurement.

[0060] In this embodiment, the test signal consists of multiple discrete frequency test sub-signals, rather than a single frequency. The frequency selection of these test sub-signals is based on the following principles: First, all frequencies are higher than the power frequency (50Hz) and not integer multiples of the power frequency to avoid interference with power grid harmonics; second, the frequency range covers the effective operating frequency band of the magnetic circuit (60-100Hz), which can reflect the frequency characteristics of permeability while avoiding the influence of the skin effect; third, the frequency points are evenly distributed to ensure the smoothness of the dispersion characteristic curve. In some embodiments, the signal processing unit 20 uses five discrete frequencies (65Hz, 72Hz, 80Hz, 88Hz, and 95Hz). These frequency points are optimized to effectively characterize the dispersion characteristics of the magnetic circuit, while controlling the test time within 200ms to meet real-time requirements.

[0061] The permeability response sub-data consists of the measurement results corresponding to each test sub-signal. In some embodiments, the signal processing unit 20 independently injects, measures the response, and calculates the permeability for each test sub-signal to obtain the permeability value at that frequency point. Unlike the single permeability value in claim 1, this embodiment obtains a permeability-frequency dataset, which more comprehensively reflects the frequency characteristics of the magnetic circuit. It is worth noting that, since the test sub-signals are injected sequentially, the signal processing unit 20 also considers the small temperature changes during the test process, and corrects all measurement data to the same temperature reference point through timestamp recording and linear interpolation to ensure the accuracy of the dispersion characteristic curve.

[0062] The construction of the dispersion characteristic curve is the core step in this embodiment. In some embodiments, the signal processing unit 20 fits the permeability response sub-data at each test sub-signal frequency point to generate a smooth permeability-frequency curve. The fitting method uses cubic spline interpolation, which ensures both the smoothness of the curve and accurate reflection of the measurement data. In other embodiments, the signal processing unit 20 directly uses discrete measurement points to construct the dispersion characteristic curve, avoiding errors introduced by fitting, which is particularly suitable for scenarios with high measurement accuracy and dense frequency points.

[0063] A pre-stored library of standard dispersion characteristic curves serves as a benchmark for temperature identification. In some embodiments, this library is established through precise calibration before the current transformer leaves the factory: the current transformer is placed in a constant temperature chamber, and temperature points are set at 5°C intervals from -40°C to +85°C. At each temperature point, multi-frequency responses are measured, and dispersion characteristic curves are constructed. These curves reflect the true frequency characteristics of the magnetic circuit at different temperatures, forming a temperature-dispersion characteristic mapping library. In other embodiments, the curve library uses a parametric representation, storing only key feature points and fitting parameters, significantly reducing storage space requirements, which is particularly suitable for resource-constrained embedded signal processing units 20.

[0064] Comparison of dispersion characteristic curves is a crucial step in confirming the reliability of temperature change measurements. In some embodiments, the signal processing unit 20 employs a Dynamic Time Warping (DTW) algorithm to calculate the similarity between the current dispersion characteristic curve and curves in a standard curve library. The DTW algorithm effectively handles local deformations of curves and is particularly suitable for comparing dispersion characteristic curves. In other embodiments, the signal processing unit 20 uses a method combining Euclidean distance and weighting coefficients to assign different weights to different frequency regions (e.g., higher weights for low-frequency regions due to their greater sensitivity to temperature), thereby improving the accuracy of the comparison.

[0065] A similarity threshold is used to determine the reliability of temperature readings. In some embodiments, the matching threshold is set to 0.92 (normalized similarity, where 1.0 represents a perfect match). This threshold is determined based on extensive experimental data: when the similarity is ≥0.92, the temperature measurement error is typically less than ±0.5℃; when the similarity is <0.92, the error may exceed ±2℃, indicating that the measurement is affected by interference. The signal processing unit 20 also implements a dynamic threshold adjustment mechanism: in a stable environment, the threshold is increased to 0.95 to pursue higher accuracy, and in a strong interference environment, the threshold is decreased to 0.88 to ensure availability.

[0066] After confirming the reliability of the temperature change, the signal processing unit 20 generates a current compensation coefficient. In some embodiments, the signal processing unit 20 not only uses the temperature value corresponding to the matched standard curve, but also employs a weighted average of adjacent temperature points to improve temperature resolution. For example, if the current curve has a similarity of 0.95 with the 25℃ curve and a similarity of 0.85 with the 30℃ curve, the final temperature calculation is (25×0.95+30×0.85) / (0.95+0.85)=27.3℃. This processing method improves the temperature resolution from 5℃ to 0.5℃, significantly enhancing the compensation accuracy.

[0067] In this embodiment, a dispersion characteristic curve is constructed through multi-frequency testing, extending temperature measurement from a single frequency point to the frequency domain, effectively distinguishing between temperature effects and interference factors. This method achieves a significant improvement in temperature measurement reliability without increasing hardware costs; it only requires optimizing the test signal and signal processing algorithm.

[0068] In the aforementioned implementation process, the signal processing unit 20 established a temperature measurement mechanism based on the permeability response. However, during long-term use, the magnetic properties of the magnetic circuit material may drift due to factors such as aging and stress changes, causing the original permeability-temperature mapping relationship to become inaccurate. For example, silicon steel sheets may experience magnetic aging under long-term alternating magnetic fields, causing changes in the permeability value at the same temperature. Traditional current transformers require periodic factory calibration, while in this embodiment, the signal processing unit 20 innovatively introduces an in-situ calibration and verification mechanism, enabling the current transformer to autonomously detect and correct mapping relationship drift, significantly extending the maintenance-free period.

[0069] In this embodiment, the temperature sensor is integrated inside and thermally coupled to the magnetic circuit structure 10, which is the basis for achieving accurate temperature comparison. In some embodiments, the temperature sensor used by the signal processing unit 20 is a high-precision digital temperature sensor with a temperature measurement range of -55°C to +125°C and an accuracy of ±0.5°C, meeting the operating temperature range requirements of the transformer. The sensor is embedded in the core region of the magnetic circuit and is in close contact with the surface of the magnetic core to ensure good thermal coupling. In other embodiments, the temperature sensor used by the signal processing unit 20 is a thermistor, which is lower in cost and has a faster response, but requires additional signal conditioning circuitry. Regardless of the sensor used, the signal processing unit 20 ensures a strict thermal coupling design: the sensor-magnetic circuit contact surface is coated with thermally conductive silicone grease and fixed with insulating adhesive to ensure a thermal resistance of less than 0.5°C / W, enabling the sensor to accurately reflect the actual temperature of the magnetic circuit.

[0070] Acquiring the reference temperature is the starting point for temperature calculation. In some embodiments, the signal processing unit 20 uses the ambient temperature as the initial calibration temperature upon initial power-on, measuring it with a high-precision external thermometer and inputting it into the system. In other embodiments, the signal processing unit 20 automatically records the last calibration temperature during stable operation, using it as a reference for subsequent measurements. The signal processing unit 20 stores the reference temperature in non-volatile memory, ensuring it is not lost even after power failure and allowing temperature tracking to continue after a system restart. In some embodiments, the signal processing unit 20 also implements a multiple backup mechanism for the reference temperature, storing the three most recent valid reference temperatures in different storage areas to prevent data loss due to a single point of failure.

[0071] Determining the calculated temperature is a crucial step in connecting the permeability measurement with the actual temperature. In some embodiments, the signal processing unit 20 calculates the current calculated temperature based on a reference temperature T0 and the aforementioned determined temperature change ΔT using the formula T_calc = T0 + ΔT. This calculation process by the signal processing unit 20 considers the continuity of temperature change, avoiding the cumulative error of absolute temperature measurement. In other embodiments, the signal processing unit 20 employs a recursive algorithm: T_calc(n) = T_calc(n-1) + ΔT(n), where n represents the current measurement period. This algorithm is less dependent on the reference temperature and is more suitable for long-term operation. It is worth noting that the calculated temperature calculated by the signal processing unit 20 reflects a temperature value extrapolated from the change in permeability, and its accuracy depends on the precision of the permeability-temperature mapping relationship.

[0072] Temperature deviation comparison is the trigger condition for calibration verification. In some embodiments, the signal processing unit 20 calculates the difference ΔT_dev = T_calc - T_meas between the calculated temperature T_calc and the temperature measured by the temperature sensor in real time, and monitors the trend of this deviation. In other embodiments, the signal processing unit 20 uses a sliding window mechanism to calculate the standard deviation of the deviation over the most recent 5 minutes to distinguish between transient interference and long-term drift. The deviation threshold set by the signal processing unit 20 is based on a large amount of experimental data: when |ΔT_dev| > 2.5℃ and lasts for more than 10 minutes, it indicates that the mapping relationship may have drifted significantly, requiring calibration verification. This threshold takes into account sensor accuracy (±0.5℃) and measurement noise (±0.3℃) to ensure that calibration is not frequently triggered by normal fluctuations.

[0073] The execution conditions of the calibration and verification process are carefully designed by the signal processing unit 20. In some embodiments, the signal processing unit 20 performs calibration only when the measured rate of temperature change is below a stable threshold (e.g., 0.1℃ / min), ensuring that the calibration is performed under stable temperature conditions and avoiding measurement errors during dynamic processes. The rate of temperature change is obtained by continuously measuring temperature values ​​and calculating the derivative by the signal processing unit 20, and a stable state is determined when |dT_meas / dt| < 0.1℃ / min. In other embodiments, the signal processing unit 20 also adds environmental condition judgment: calibration is performed only when the transformer load is stable (current change rate < 0.5A / s) and there is no strong electromagnetic interference, further improving the calibration quality.

[0074] The calibration and verification process comprises three key steps, all performed by the signal processing unit 20. First, the signal processing unit 20 records the stable value T_stable of the measured temperature, obtained through a 10-minute moving average to eliminate measurement noise. In some embodiments, the signal processing unit 20 also performs outlier removal, excluding data points with temperature fluctuations exceeding ±0.2℃ to ensure the reliability of T_stable. Second, the signal processing unit 20 measures the permeability response μ_stable of the magnetic circuit structure 10 to T_stable. This measurement employs the aforementioned multi-frequency testing method to ensure accurate results. In other embodiments, the signal processing unit 20 performs multiple measurements and averages them to further improve the accuracy of permeability measurement. Finally, the signal processing unit 20 updates the permeability-temperature mapping relationship based on T_stable and μ_stable. This update uses incremental correction: only mapping points near T_stable are adjusted to avoid instability caused by global recalibration. In some embodiments, the signal processing unit 20 also implements a weighted update mechanism, where the weight of historical data decays over time, allowing the mapping relationship to adapt to a slow aging process.

[0075] In some embodiments, the signal processing unit 20 also implements a self-learning capability for calibration verification. By analyzing historical calibration data, the signal processing unit 20 dynamically adjusts the deviation threshold and the stability threshold: when frequent triggering of invalid calibration is detected, the deviation threshold is appropriately increased; when drift is found to be not detected in time, the deviation threshold is appropriately decreased. This adaptive mechanism enables the signal processing unit 20 to maintain optimal calibration performance under different operating environments.

[0076] In this embodiment, the intelligent comparison between the calculated temperature and the measured temperature performed by the signal processing unit 20 verifies and corrects the permeability-temperature mapping relationship in place, thus solving the long-term accuracy drift problem caused by magnetic circuit aging.

[0077] In the aforementioned implementation process, the signal processing unit 20 has been able to accurately measure temperature changes and perform basic compensation. However, in actual power systems, the operating state of the instrument transformer is affected by multiple factors: DC bias can cause the magnetic circuit operating point to shift, changing the permeability characteristics; the magnitude of the load current affects the degree of magnetic circuit saturation; and temperature changes alter the basic properties of the material. Single-dimensional temperature compensation cannot adapt to this complex operating condition with multiple coupled factors. In this embodiment, the signal processing unit 20 innovatively introduces a multi-dimensional operating state determination mechanism, achieving precise compensation through a three-dimensional lookup table, significantly improving the measurement accuracy of the instrument transformer under various operating conditions.

[0078] Obtaining the DC bias parameter is a fundamental step in this embodiment. In some embodiments, the signal processing unit 20 detects the DC component of the primary current using a high-precision ADC and calculates the DC bias value using a moving average filtering algorithm. Specifically, the signal processing unit 20 samples the primary current signal at 1024 points and calculates its arithmetic mean as the DC bias parameter I_dc. To improve measurement accuracy, the signal processing unit 20 also implements a temperature compensation mechanism: when the ambient temperature changes, it automatically corrects the zero-point drift of the ADC to ensure that the DC bias measurement error is less than ±0.05%. In other embodiments, the signal processing unit 20 uses hardware circuit preprocessing, separating the DC component through a DC blocking capacitor and a precision rectifier circuit, and then digitizing it by the ADC. This method has a faster response but requires additional hardware support.

[0079] Measuring the effective value of the primary-side AC current is another key parameter for determining the operating state. In some embodiments, the signal processing unit 20 employs a true RMS calculation method: the primary-side current is rectified by full wave, squared, then low-pass filtered to obtain the average value, and finally the square root is taken to obtain the effective value I_rms. This method is applicable to arbitrary waveforms, including distorted currents containing harmonics. In other embodiments, the signal processing unit 20 employs a simplified algorithm: for the current signal under the sinusoidal assumption, the peak value is directly calculated and then divided by √2. This method has low computational complexity and is suitable for resource-constrained embedded systems. Regardless of the method used, the signal processing unit 20 ensures the accuracy of the RMS measurement: the measurement error does not exceed ±0.1% within the range of 5% to 120% of the rated current.

[0080] The determination of the working state is the core innovation of this embodiment. In some embodiments, the signal processing unit 20 uses the temperature change ΔT, the DC bias parameter I_dc, and the effective current value I_rms as three-dimensional coordinates to construct a working state space. The signal processing unit 20 first normalizes each parameter: converts ΔT into a percentage change relative to the reference temperature, I_dc into a percentage relative to the rated current, and I_rms into a multiple of the rated current. Then, the signal processing unit 20 substitutes the normalized parameters into the preset partitioning rules to determine the region to which the current working state belongs. For example, when -10% < ΔT < 5%, |I_dc| < 2%, and 0.2 < I_rms / I_rated < 0.8, it is determined as the "low temperature and low load" state; when 5% < ΔT < 20%, |I_dc| > 5%, and 0.8 < I_rms / I_rated < 1.2, it is determined as the "high temperature and high DC bias" state. This partitioning method enables the signal processing unit 20 to accurately identify the working conditions of the current transformer, providing a basis for subsequent compensation.

[0081] The construction and query of the temperature-bias-current three-dimensional look-up table are the keys to achieving precise compensation. In some embodiments, this look-up table is established through full-condition testing before the current transformer leaves the factory: in an environmental test chamber, systematically change the temperature (-40°C to +85°C, step size 5°C), apply different DC biases (-10% to +10% of the rated current, step size 2%), input different load currents (20% to 120% of the rated current, step size 10%), measure the ratio error and phase error under each combination, and form a complete three-dimensional data set. In some other embodiments, the look-up table is represented parametrically, only storing the key node data, and calculating the intermediate points through trilinear interpolation, significantly reducing the storage requirements. When the signal processing unit 20 determines the working state, it first locates the coordinate position of this state in the look-up table, and then uses the trilinear interpolation algorithm to calculate the precise compensation coefficient. For example, if the current state is located at (ΔT = 8°C, I_dc = 3.5%, I_rms = 0.75I_rated), the signal processing unit 20 will find the adjacent 8 nodes and calculate the weighted average as the final compensation coefficient.

[0082] In some embodiments, the signal processing unit 20 also implements a dynamic optimization mechanism for the look-up table. By analyzing the relationship between historical measurement data and actual errors, the signal processing unit 20 automatically adjusts the compensation coefficients in certain regions of the look-up table to compensate for the characteristic drift caused by long-term aging. For example, when it is found that the compensation effect in the "high temperature and high load" region is continuously poor, the signal processing unit 20 will fine-tune the compensation coefficients in this region to keep the error within the allowable range. This adaptive ability enables the current transformer to maintain high precision throughout its service life.

[0083] In this embodiment, the signal processing unit 20 realizes precise compensation under multi-factor coupling working conditions by establishing a three-dimensional working state model of temperature-bias-current.

[0084] In the above implementation process, the signal processing unit 20 can already obtain accurate current compensation coefficients according to the working state. However, in actual operation, the state of the mutual inductor may fluctuate due to factors such as environmental changes and component aging, resulting in changes in the reliability of the compensation coefficients. For example, when the temperature changes rapidly, the permeability-temperature mapping relationship may temporarily fail; when electromagnetic interference increases, the measurement accuracy of the test signal may decrease. If the current compensation coefficient is directly applied at this time, additional errors will be introduced instead. In this embodiment, the signal processing unit 20 innovatively introduces a system health index evaluation mechanism, dynamically adjusts the compensation strategy according to the current state of the mutual inductor, and ensures the stability of measurement while guaranteeing the accuracy.

[0085] The acquisition of the system health index is the basic link of this embodiment. In some embodiments, the signal processing unit 20 calculates the system health index SHI (System Health Index) through multi-source data fusion. Its value range is [0,1], where 1 represents the best state and 0 represents a serious fault. The signal processing unit 20 first collects multiple health-related parameters: the signal-to-noise ratio SNR of the test signal, the measurement stability σ_μ of the permeability, the temperature change rate dT / dt, and the total harmonic distortion THD of the current waveform, and then calculates SHI through weighted summation: SHI = w1·f(SNR)+w2·f(σ_μ)+w3·f(dT / dt)+w4·f(THD) where wi is the weight coefficient and f(·) is the normalization function. In some other embodiments, the signal processing unit 20 uses fuzzy logic to evaluate the health state: divides each parameter into three levels of "good", "general", and "poor", and obtains the health index through fuzzy rule reasoning. For example, when SNR > 30dB and dT / dt < 0.2℃ / min, it is judged as "good"; when 15dB < SNR < 30dB and 0.2℃ / min < dT / dt < 0.5℃ / min, it is judged as "general". This fuzzy evaluation method can better reflect the uncertainty in actual operation.

[0086] The setting of the first health threshold and the second health threshold is a key parameter in this embodiment. In some embodiments, the signal processing unit 20 sets the first health threshold to 0.85 and the second health threshold to 0.65. This setting is based on a large amount of experimental data: when SHI > 0.85, the error of the current compensation coefficient is usually less than ±0.05%; when 0.65 < SHI < 0.85, there may be a moderate degree of uncertainty in the current compensation coefficient; when SHI < 0.65, the system may be in a severely abnormal state and requires special handling. In other embodiments, the signal processing unit 20 implements a dynamic adjustment mechanism for the threshold: automatically adjusts the threshold according to the service life and working environment of the current transformer. For example, for a current transformer that has been used for more than 3 years, the first threshold is reduced from 0.85 to 0.80 to adapt to the performance changes caused by component aging.

[0087] When the system health index is higher than the first health threshold, the signal processing unit 20 uses the current compensation coefficient corresponding to the current working state to correct the original current signal. In some embodiments, the signal processing unit 20 directly applies the compensation coefficient because the system state is stable and the current measurement data is reliable at this time. For example, when SHI = 0.92, the signal processing unit 20 determines that the permeability measurement accuracy is high, the temperature change is stable, and the current compensation coefficient can accurately reflect the state of the current transformer. In other embodiments, the signal processing unit 20 also makes minor adjustments: applies a low-pass filter to the compensation coefficient to eliminate the influence of high-frequency noise and make the output signal smoother. This processing method is particularly effective when SHI is close to the threshold and can prevent frequent switching of the compensation strategy due to small fluctuations in the health index.

[0088] When the system health index is between the first health threshold and the second health threshold, the signal processing unit 20 performs a weighted average of the current compensation coefficient corresponding to the current working state and the historical compensation coefficient to obtain an effective compensation coefficient. In some embodiments, the signal processing unit 20 uses a linear weighting method: K_eff = α·K_current+(1 - α)·K_history where K_eff is the effective compensation coefficient, K_current is the current compensation coefficient, K_history is the historical compensation coefficient, and α is the weighting coefficient. The value of α is linearly related to the health index: when SHI = 0.85, α = 1.0; when SHI = 0.65, α = 0.3. This design ensures that when the health state is good, it mainly relies on the current data, and when the health state deteriorates, it gradually increases the weight of the historical data. In other embodiments, the signal processing unit 20 uses non-linear weighting: α=(SHI - 0.65) / 0.2, but sets a minimum value of 0.3 and a maximum value of 1.0 to avoid instability in extreme cases.

[0089] The dynamic adjustment mechanism of historical compensation coefficient weights is an innovation of this embodiment. In some embodiments, the signal processing unit 20 adjusts the weights of historical data according to the changing trend of the health index: if the health index continues to decline, indicating that the system state is deteriorating, the weight of historical data is increased to maintain the stability of the output; if the health index tends to stabilize, the weight of historical data is reduced so that the compensation can adapt to the current state more quickly. Specifically, the signal processing unit 20 calculates the short-term rate of change of the health index d(SHI) / dt. When d(SHI) / dt < -0.01 / min, the historical weight is increased by 10%; when d(SHI) / dt > 0.01 / min, the historical weight is reduced by 5%. In other embodiments, the signal processing unit 20 considers the quality of historical data: it preferentially selects historical data with higher health indices for weighting, for example, only using historical compensation coefficients with SHI > 0.7 to avoid introducing low-quality data.

[0090] In some embodiments, the signal processing unit 20 also implements a self-learning capability for the health index. By analyzing the relationship between historical health indices and actual measurement errors, the signal processing unit 20 dynamically adjusts the parameters of the health assessment model. For example, when it is found that the correlation between the health index and the actual error decreases under certain operating conditions, the weighting coefficients of each health parameter are automatically adjusted to make the health assessment more accurate. This adaptive mechanism enables the signal processing unit 20 to maintain optimal performance in different operating environments.

[0091] In this embodiment, the signal processing unit 20 assesses the current transformer status through the system health index and dynamically adjusts the compensation strategy accordingly. When the status is good, it makes full use of the latest data to ensure accuracy, and when the status changes, it integrates historical data to ensure stability.

[0092] In the aforementioned implementation process, the signal processing unit 20 constructs the dispersion characteristic curve using multi-frequency test signals with a fixed frequency distribution. However, in actual power system environments, the frequency characteristics of electromagnetic interference vary with operating conditions: industrial sites may experience significant 5-15kHz inverter interference, while residential areas may be affected by 40-80Hz grid harmonics. The fixed-frequency distribution test scheme cannot adapt to these variations, potentially leading to severe measurement distortion at certain frequency points. In this embodiment, the signal processing unit 20 innovatively introduces an interference sensing mechanism, dynamically optimizing the test signal frequency distribution based on the real-time electromagnetic environment, significantly improving the reliability and efficiency of dispersion characteristic curve measurement.

[0093] Determining the interference level is a fundamental step in this embodiment. In some embodiments, the signal processing unit 20 determines the interference level through a dual-channel evaluation: first, it calculates the signal-to-noise ratio (SNR) of the test signal, which is the ratio of the energy at the test signal frequency point to the noise energy in the adjacent frequency band; second, it acquires the environmental interference spectrum through an external electromagnetic interference detection module. The signal processing unit 20 fuses these two types of information and calculates the interference level index IL using a weighted scoring method. IL = 0.6 × (1 - SNR_norm) + 0.4 × (interference spectrum complexity) Where SNR_norm is the normalized signal-to-noise ratio (0 represents extremely low, 1 represents extremely high), and the interference spectrum complexity is obtained by calculating the peak-to-valley ratio and dispersion of the spectrum. In other embodiments, the signal processing unit 20 uses fuzzy logic classification: the signal-to-noise ratio is divided into three levels: "high," "medium," and "low," and the interference spectrum is divided into three levels: "simple," "general," and "complex." The final interference level is determined by preset rules (e.g., "high SNR + simple spectrum" → "stable environment," "low SNR + complex spectrum" → "strong interference environment"). This fuzzy classification method is more adaptable to the uncertainty of the actual electromagnetic environment.

[0094] The key innovation of this embodiment is the adjustment of the high-frequency test sub-signal density under strong interference environments. In some embodiments, the signal processing unit 20 divides the frequency range into a low-frequency band (60-80Hz) and a high-frequency band (80-100Hz). When the interference level index IL > 0.7 (indicating a strong interference environment), the density of the test sub-signals in the high-frequency band is increased. Specifically, the signal processing unit 20 increases the number of test points in the high-frequency band from 3 to 6, and decreases the point spacing from 10Hz to 3.3Hz, while keeping the number of test points in the low-frequency band unchanged. This adjustment is based on the characteristics of electromagnetic interference: interference in industrial environments is usually concentrated in the low-frequency region (<80Hz), while the high-frequency region is relatively clean. Increasing the density of high-frequency points can increase the number of reliable measurement points. In other embodiments, the signal processing unit 20 adopts adaptive density adjustment: according to the specific distribution of the interference spectrum, the frequency region where the density needs to be increased is dynamically determined. For example, if the interference spectrum shows strong interference in the 70-75Hz range, the test point density is increased in the 75-100Hz range.

[0095] Another innovation point of this embodiment is the optimization of the distribution of test sub-signals in the low-frequency band under a stable environment. In some embodiments, when the interference level index IL < 0.3 (indicating a stable environment), the signal processing unit 20 optimizes the distribution of test sub-signals in the low-frequency band, using logarithmic intervals instead of linear intervals. For example, five test points in the range of 60 - 80 Hz are set to 60 Hz, 64 Hz, 70 Hz, 76 Hz, and 80 Hz. This distribution better conforms to the natural change law of the permeability-frequency characteristics and provides higher resolution in the critical region (close to the power frequency). In other embodiments, the signal processing unit 20 intelligently adjusts the test point distribution according to the curvature change of the historical dispersion characteristic curve: increasing the point density in the region with large curvature (temperature-sensitive region) and decreasing the point density in the region with small curvature, so as to make the best use of the limited test point resources.

[0096] The frequency range definition between the high-frequency band and the low-frequency band is an important parameter of this embodiment. In some embodiments, the signal processing unit 20 uses 80 Hz as the demarcation point between the high and low frequency bands. This choice is based on a large amount of experimental data: below 80 Hz, the power grid harmonic interference is significant; above 80 Hz, the interference intensity usually drops by more than 40%, and the magnetic circuit can still maintain good linear characteristics. In other embodiments, the signal processing unit 20 realizes the dynamic adjustment of the demarcation point: automatically determining the optimal demarcation frequency according to the main frequency distribution of the environmental interference. For example, when strong interference is detected near 75 Hz, the demarcation point is moved up to 85 Hz to ensure the measurement reliability of the high-frequency band.

[0097] In some embodiments, the signal processing unit 20 also realizes a smooth transition mechanism for frequency adjustment. When the interference level fluctuates near the threshold, to avoid frequent switching of the frequency distribution of the test signal, a progressive adjustment is adopted: setting a transition region (0.3 < IL < 0.7), and within this region, the frequency distribution strategies of the stable environment and the strong interference environment are mixed in proportion. For example, when IL = 0.5, the number of test points in the high-frequency band is 4.5 (rounded to 4 or 5). Through this smooth transition, the continuity of the output signal of the mutual inductor is ensured, and the measurement jump caused by the sudden change of the test strategy is avoided.

[0098] In this embodiment, the signal processing unit 20 dynamically optimizes the frequency distribution strategy of the test signal by real-time sensing the electromagnetic environment, focusing on the high-frequency band to avoid interference in a strong interference environment, and optimizing the low-frequency band to improve the resolution in a stable environment.

[0099] In the aforementioned implementation process, the signal processing unit 20 has been able to measure permeability and perform temperature compensation using test signals. However, in actual power systems, test signals are often subject to two main types of interference: transient interference caused by switching operations and periodic interference caused by grid harmonics. These interferences can lead to distortion in permeability measurements, thereby affecting the accuracy of temperature calculations and current compensation. For example, transient interference generated by circuit breaker operations may cause abrupt changes in amplitude response, which may be misinterpreted as rapid temperature changes; the fifth harmonic of the grid may cause phase shifts, resulting in distortion of the dispersion characteristic curve. In this embodiment, the signal processing unit 20 innovatively introduces an interference identification and elimination mechanism. By processing the amplitude response and phase response separately, it accurately identifies and eliminates different types of interference, significantly improving the reliability of permeability measurements.

[0100] Time-domain differentiation of the amplitude response is crucial for identifying switching interference. In some embodiments, the signal processing unit 20 performs first-order time-domain differentiation on the amplitude response A(t) of the test signal to calculate the amplitude change rate dA / dt. Specifically, the signal processing unit 20 samples the amplitude response at 10μs intervals and calculates the differential value using the center difference algorithm. (dA / dt)_i=[A(t_{i+1})-A(t_{i-1})] / (2Δt) This algorithm effectively suppresses high-frequency noise while maintaining computational accuracy. In other embodiments, the signal processing unit 20 uses a Savitzky-Golay filter for differentiation, smoothing the signal while calculating the derivative, which is particularly suitable for scenarios with low signal-to-noise ratios. The signal processing unit 20 sets a first distortion threshold of 0.8% / ms, which is based on a large amount of experimental data: when the amplitude change rate exceeds 0.8% / ms, more than 95% of the cases are caused by switching operations; below this value, it is mainly caused by measurement noise. When the signal processing unit 20 detects that the amplitude change rate exceeds the first distortion threshold, it determines that there is a sudden amplitude change caused by a switching operation and uses this amplitude change rate as a transient feature. It is worth noting that the signal processing unit 20 defines the transient feature as a quantized value of the amplitude change rate, rather than a new feature, thus avoiding the problem of feature amplification.

[0101] Spectral analysis of the phase response is crucial for identifying harmonic interference. In some embodiments, the signal processing unit 20 performs FFT analysis on the phase response φ(t) of the test signal, focusing on the spectral characteristics at power frequency harmonic frequencies (50Hz, 100Hz, 150Hz, etc.). The signal processing unit 20 calculates the energy proportion at each harmonic frequency. Energy percentage = (harmonic frequency amplitude²) / (total energy of the entire analysis frequency band) The calculation takes into account the effect of spectral leakage and uses a Hanning window for preprocessing. In some embodiments, the signal processing unit 20 uses wavelet transform instead of FFT, which is particularly suitable for analyzing harmonic components in non-stationary signals. The signal processing unit 20 sets a second distortion threshold of 8%, which is based on the IEC 61000-4-7 standard: when the amplitude proportion at the power frequency harmonic exceeds 8%, the phase shift usually exceeds 0.5°, which is sufficient to affect the accuracy of temperature measurement. When the signal processing unit 20 detects that the amplitude proportion at the power frequency harmonic exceeds the second distortion threshold, it determines that there is a phase shift caused by the power frequency harmonic and uses the phase shift at that harmonic frequency as a harmonic feature. The phase shift is obtained by calculating the difference between the theoretical phase and the measured phase, in degrees (°).

[0102] The comprehensive determination of interference distortion is an innovation of this embodiment. In some embodiments, the signal processing unit 20 uses a logical "OR" to determine whether interference distortion exists: if transient characteristics or harmonic characteristics exist, interference distortion is determined to be present. This determination method ensures sensitivity to any type of interference. In other embodiments, the signal processing unit 20 uses a weighted determination: different weights are assigned according to the degree of influence of the interference type on the measurement accuracy. For example, the influence of switching transients on the amplitude response has a weight of 0.7, and the influence of harmonic interference on the phase response has a weight of 0.3. When the weighted sum exceeds a threshold of 0.5, interference distortion is determined to exist. The signal processing unit 20 also implements interference intensity assessment: the severity of interference is quantified by the amplitude change rate of transient characteristics and the phase shift of harmonic characteristics, providing a basis for subsequent adjustment of filter parameters.

[0103] When interference distortion is present, the signal processing unit 20 performs targeted filtering. In some embodiments, the signal processing unit 20 performs transient filtering on the amplitude response based on transient characteristics: when the amplitude change rate exceeds a threshold, an adaptive median filter is applied, and the window size is dynamically adjusted according to the change rate (the larger the change rate, the larger the window). For example, when the amplitude change rate is 1.2% / ms, a 7-point window is used; when the change rate is 2.0% / ms, an 11-point window is used. This dynamic adjustment ensures that abrupt changes are effectively eliminated without over-smoothing the valid signal. In other embodiments, the signal processing unit 20 performs harmonic filtering on the phase response based on harmonic characteristics: when a specific harmonic interference is detected, a notch filter is applied with a center frequency consistent with the interference frequency, and the bandwidth is dynamically adjusted according to the phase offset. For example, when the phase offset caused by the 5th harmonic is 0.8°, a notch filter with a bandwidth of 2Hz is used; when the offset is 1.5°, a sine wave filter with a bandwidth of 4Hz is used. The signal processing unit 20 ensures that the filtering process only applies to the interfered portion, preserving the integrity of the valid signal.

[0104] When there is no interference distortion, the signal processing unit 20 directly calculates the permeability response data based on the original amplitude and phase responses. In some embodiments, the signal processing unit 20 uses the least squares method to fit the permeability-frequency relationship to obtain a smoother dispersion characteristic curve. In other embodiments, the signal processing unit 20 performs simple linear interpolation to calculate the permeability value at a specified frequency point. This direct calculation method avoids unnecessary signal processing, reduces computational latency, and is particularly suitable for scenarios with high real-time requirements.

[0105] In some embodiments, the signal processing unit 20 also implements a self-verification mechanism for the filtering effect. By comparing the residual energy before and after filtering, the signal processing unit 20 evaluates the filtering effect: if the residual energy decreases by more than 60%, the filtering is confirmed to be effective; if the decrease is less than 30%, the filtering parameters are readjusted. This closed-loop verification ensures the quality of interference cancellation and avoids the problems of over-filtering or under-filtering.

[0106] In this embodiment, the signal processing unit 20 accurately identifies and eliminates different types of interference by separating and processing the amplitude response and phase response, thus avoiding the "one-size-fits-all" filtering process in traditional methods.

[0107] In the aforementioned implementation process, the signal processing unit 20 can measure temperature independently using either the permeability response or a temperature sensor. However, in actual power systems, a single measurement method has significant limitations: permeability-based measurements are easily interfered with during sudden current changes, while temperature sensor-based measurements have a delayed response and cannot reflect the actual operating temperature of the magnetic circuit. In this embodiment, the signal processing unit 20 innovatively introduces a dual-model fusion mechanism, intelligently switching the weights of the primary and secondary models according to the operating conditions, enabling the transformer to obtain high-precision temperature measurements under various operating conditions.

[0108] The main mapping model and the auxiliary mapping model are the core of this embodiment. In some embodiments, the main mapping model of the signal processing unit 20 calculates the main temperature change based on the aforementioned permeability-temperature mapping relationship using measured permeability response data. Specifically, the signal processing unit 20 substitutes the permeability response data into a pre-stored cubic polynomial model T_main=a0+a1μ+a2μ²+a3μ³, where μ is the permeability and a0-a3 are fitting coefficients. This model has a fitting error of less than 0.4℃ in the range of -40℃ to +85℃, accurately reflecting the relationship between permeability and temperature. In other embodiments, the main mapping model of the signal processing unit 20 uses a piecewise linear function, dividing the temperature range into 5 intervals. Each interval is approximated using a linear function, resulting in less computation and making it more suitable for resource-constrained embedded systems.

[0109] The auxiliary mapping model of the signal processing unit 20 calculates the auxiliary temperature change based on the aforementioned temperature sensor data and current change rate. In some embodiments, the auxiliary mapping model employs a heat conduction model: T_aux = T_meas + k·(dI / dt) Where T_meas is the temperature sensor measurement, dI / dt is the rate of change of current, and k is the thermo-electric coupling coefficient. This model considers the Joule heating effect caused by sudden current changes and can predict transient temperature changes in the magnetic circuit. In other embodiments, the auxiliary mapping model of the signal processing unit 20 is implemented using a neural network, which obtains the nonlinear relationship between the rate of change of current and the temperature response through training on historical data, making it particularly suitable for complex operating conditions. It is worth noting that the signal processing unit 20 ensures that the auxiliary mapping model only uses the measured temperature and rate of change of current defined in the model, avoiding the feature expansion problem.

[0110] The operating condition determination mechanism is a key innovation of this embodiment. In some embodiments, the signal processing unit 20 performs time-domain differentiation on the primary current waveform to calculate the current change rate dI / dt: (dI / dt)_i=[I(t_{i+1})-I(t_{i-1})] / (2Δt) Where Δt is the sampling interval (typically 10 μs). The signal processing unit 20 sets a stable operating condition threshold of 0.8 A / ms and a sudden change operating condition threshold of 2.5 A / ms. These thresholds are based on typical power system operating conditions: when |dI / dt| < 0.8 A / ms, the current changes smoothly and the magnetic circuit operates stably; when |dI / dt| > 2.5 A / ms, it is usually caused by switching operations, and the magnetic circuit is in a transient process. Simultaneously, the signal processing unit 20 monitors the signal-to-noise ratio (SNR) of the test signal, setting a reliable measurement threshold of 25 dB and an interference threshold of 18 dB. When SNR > 25 dB, the permeability measurement accuracy is high; when SNR < 18 dB, the measurement is severely affected by interference. The signal processing unit 20 uses logical judgment rules: If |dI / dt| < 0.8 A / ms and SNR > 25 dB → steady condition; If |dI / dt|>2.5A / ms or SNR<18dB → mutation condition; Other situations → Transitional working conditions (determined proportionally); The weight allocation logic is the core of achieving complementary advantages between the two models. In some embodiments, the signal processing unit 20 sets the main weight w_main and the auxiliary weight w_aux based on the operating condition determination result: Under steady-state conditions: w_main=0.85, w_aux=0.15 Under sudden change conditions: w_main=0.35, w_aux=0.65; During transitional operating conditions: weights are calculated using linear interpolation; This weighting setting is based on the fundamental technical principle: under stable operating conditions, permeability measurement is more accurate, and the main model should dominate; under abrupt operating conditions, temperature sensor response is more reliable, and the auxiliary model should dominate. In other embodiments, the signal processing unit 20 implements continuous adjustment of the weights: w_main=0.7+0.15×(1-|dI / dt| / 2.5)+0.15×(SNR-18) / 7, so that the weights change smoothly with the operating conditions, avoiding abrupt changes in the compensation coefficients.

[0111] The calculation of the temperature change is the final step in this embodiment. In some embodiments, the signal processing unit 20 calculates the final temperature change by weighted summation based on the primary weight, secondary weight, primary temperature change, and secondary temperature change: ΔT_final=w_main·ΔT_main+w_aux·ΔT_aux Where ΔT_main is the temperature change calculated by the primary mapping model, and ΔT_aux is the temperature change calculated by the auxiliary mapping model. In some embodiments, the signal processing unit 20 employs nonlinear weighting: when the difference between ΔT_main and ΔT_aux is large (2°C), the model result with smaller change is preferred to avoid the influence of outliers. The signal processing unit 20 also implements smoothing processing of temperature changes: a first-order low-pass filter is applied to ΔT_final, and the time constant is dynamically adjusted according to the operating conditions to ensure the continuity of the output signal.

[0112] In some embodiments, the signal processing unit 20 also implements an adaptive optimization mechanism for dual-model fusion. By analyzing the relationship between historical data and actual temperature errors, the signal processing unit 20 dynamically adjusts the operating condition judgment threshold and weight allocation strategy. For example, when the main model is found to perform better than expected under abrupt operating conditions, the main weight under abrupt operating conditions is appropriately increased; when the temperature sensor response slows down, the weight of the current change rate in the auxiliary model is increased. This self-learning capability enables the signal processing unit 20 to adapt to the aging of the transformer and environmental changes.

[0113] In this embodiment, the signal processing unit 20 intelligently determines the operating condition based on the current change rate and signal-to-noise ratio, and dynamically adjusts the weights of the primary and secondary models accordingly, achieving comprehensive performance that cannot be achieved by a single measurement method. This mechanism utilizes both the high accuracy of the primary model under stable operating conditions and the reliability of the secondary model under abrupt changes, enabling the transformer to obtain accurate temperature measurements under various operating conditions. This provides a solid guarantee for the precise measurement of the power system.

[0114] In the aforementioned implementation process, the signal processing unit 20 can measure permeability and perform temperature compensation through test signals. However, the timing of test signal injection has a decisive impact on measurement quality. Under grid voltage fluctuations or harmonic interference, the zero-crossing point of the power frequency current will drift. If the test signal is still injected according to a fixed timing sequence at this time, it may be in a region with a large current amplitude, causing the test signal to be submerged in strong interference. For example, when there is a 5th harmonic in the grid, the zero-crossing phase deviation can reach 1.5°, corresponding to a time deviation of 167μs. At a 50Hz power frequency, the current amplitude has reached 8.7% of the rated value, significantly reducing the signal-to-noise ratio of the test signal. In this embodiment, the signal processing unit 20 innovatively introduces a zero-crossing phase deviation monitoring mechanism, triggering test signal injection only when the zero-crossing phase deviation is small, ensuring that the measurement is performed in the region with the smallest current amplitude and the highest signal-to-noise ratio.

[0115] Zero-crossing phase deviation detection is a fundamental step in this embodiment. In some embodiments, the signal processing unit 20 obtains the actual zero-crossing time t_actual of the power frequency current through a high-precision zero-crossing detection circuit. Specifically, the signal processing unit 20 passes the primary current signal through a zero-crossing detector. When the current value crosses zero, a trigger pulse is generated. The phase difference between this pulse and the system clock is then measured by a time-to-digital converter (TDC), with an accuracy of ±50ns. In other embodiments, the signal processing unit 20 uses a software algorithm to detect the zero-crossing point: the current signal is sampled at 100kHz, and the precise zero-crossing position is determined using linear interpolation. This method requires no additional hardware but involves slightly more computation. The signal processing unit 20 calculates the theoretical zero-crossing time t_theoretical based on the system clock and the known power frequency period (50Hz corresponds to a 20ms period), and then calculates the zero-crossing phase deviation. Phase deviation = 360° × (t_actual - t_theoretical) / T Where T is the power frequency period (20ms). The signal processing unit 20 ensures that the phase deviation calculation takes into account the current direction (rising edge / falling edge), but takes the absolute value for subsequent determination.

[0116] The setting and dynamic adjustment of the tolerance threshold is a key innovation of this embodiment. In some embodiments, the signal processing unit 20 sets the tolerance threshold to 0.8°, a setting based on extensive experimental data: when the phase deviation is <0.8°, the current amplitude within a 100μs window near the zero crossing is less than 1% of the rated value, and the signal-to-noise ratio (SNR) of the test signal can be improved by more than 15dB; when the phase deviation is >0.8°, the SNR improvement effect significantly decreases. In other embodiments, the signal processing unit 20 implements a dynamic adjustment mechanism for the tolerance threshold: the tolerance threshold is automatically optimized based on historically measured SNR data. For example, when the system detects an increase in grid harmonic content, the tolerance threshold is appropriately increased (e.g., from 0.8° to 1.2°) to ensure sufficient testing opportunities; when the grid quality improves, the tolerance threshold is decreased (e.g., reduced to 0.6°) to pursue a higher SNR. The signal processing unit 20 also considers the influence of the transformer's operating temperature: in low-temperature environments (<0°C), the tolerance threshold is slightly relaxed by 0.1° to compensate for the temperature drift of electronic components.

[0117] The triggering mechanism for test signal injection is the core of this embodiment. In some embodiments, the signal processing unit 20 monitors the absolute value of the zero-crossing phase deviation in real time, and immediately triggers test signal injection when it falls below the tolerance threshold. Specifically, after detecting the actual zero-crossing, the signal processing unit 20 initiates a 500μs monitoring window, continuously comparing the phase deviation with the tolerance threshold within this window, and injecting a test signal once the condition is met. In other embodiments, the signal processing unit 20 employs predictive injection: based on the phase deviation change trend of the last 5 cycles, it predicts the zero-crossing position of the next cycle and adjusts the test signal injection timing in advance, ensuring that the injection time is precisely aligned with the region of minimum current. The signal processing unit 20 also implements a fine-tuning mechanism for the injection timing: based on the signal-to-noise ratio of the previous measurement, it dynamically adjusts the injection delay time to ensure that the test signal is always at the optimal measurement position.

[0118] In some embodiments, the signal processing unit 20 also implements a multi-cycle verification mechanism. When a zero-crossing phase deviation of a single cycle is detected to meet the conditions, a test signal is not immediately injected; instead, three cycles are continuously verified to ensure the stability of the power grid. This mechanism effectively avoids false triggering caused by transient interference and is particularly suitable for industrial environments with large power grid fluctuations. The signal processing unit 20 also sets a maximum waiting time (typically five power frequency cycles) to prevent the inability to inject a test signal for an extended period under continuous high interference conditions, thus ensuring the basic function of the transformer.

[0119] In this embodiment, the signal processing unit 20 monitors the zero-crossing phase deviation in real time and injects the test signal only during the window period when the power grid is stable and the current amplitude is minimal, thereby maximizing the signal-to-noise ratio of the test signal.

[0120] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0121] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A current transformer with temperature compensation function, characterized in that, The current transformer includes: A magnetic circuit structure is used to sense the primary side current and output the secondary side current signal; The signal processing unit, integrated inside the current transformer, is used to execute the current measurement process; The current measurement process includes: A test signal is injected into the magnetic circuit structure; wherein the frequency of the test signal is higher than the power frequency but not an integer multiple of the power frequency; Measure the amplitude and phase response of the magnetic circuit structure to the test signal; Based on the amplitude response and phase response, calculate the permeability response data under the test signal; The temperature change is determined based on the permeability response data and the preset permeability-temperature mapping relationship; Based on the temperature change, a current compensation coefficient is generated; The original current signal is corrected using the current compensation coefficient to obtain the compensated current signal.

2. The current transformer according to claim 1, characterized in that, When the signal processing unit performs the determination of the temperature change, it is specifically configured as follows: Obtain the permeability response data of the most recent N measurement processes; where N is a positive integer ≥3 and includes the current measurement process; Based on the permeability response data from the N measurement processes, the permeability variation trend is determined. When the trend of the change in magnetic permeability is consistent with the expected pattern of temperature change, the current measurement process is determined to be reliable, and the amount of temperature change is determined based on the magnetic permeability response data of the current measurement process. When the trend of magnetic permeability change is inconsistent with the expected pattern of temperature change, the current measurement process is determined to be unreliable, and the amount of temperature change is determined based on the magnetic permeability response data corresponding to reliable historical measurement processes.

3. The current transformer according to claim 1, characterized in that, The test signal includes multiple discrete frequency test sub-signals; the permeability response data includes permeability response sub-data for each test sub-signal; When the signal processing unit executes the generation of the current compensation coefficient, it is specifically configured as follows: Based on the permeability response sub-data, the current dispersion characteristic curve is constructed; The current dispersion characteristic curve is compared with the pre-stored standard dispersion characteristic curve library; When the similarity between the current dispersion characteristic curve and any standard dispersion characteristic curve in the standard dispersion characteristic curve library reaches the matching threshold, the reliability of the temperature change is confirmed, and a current compensation coefficient is generated.

4. The current transformer according to claim 1, characterized in that, The current transformer also includes a temperature sensor, which is integrated inside the magnetic circuit structure and thermally coupled to the magnetic circuit structure. The signal processing unit is further configured to: Obtain a reference temperature; wherein the reference temperature is the initial calibration temperature or the last calibration temperature; The calculation temperature is determined based on the reference temperature and the temperature change. The calculated temperature is compared with the measured temperature output by the temperature sensor to obtain the temperature deviation; When the temperature deviation exceeds the deviation threshold and the rate of change of the measured temperature is lower than the stability threshold, the calibration verification process is executed. The calibration and verification process includes: Record the stable value of the measured temperature; Measure the permeability response of the magnetic circuit structure to the stable value; The permeability-temperature mapping relationship is updated based on the stable value and the corresponding permeability response.

5. The current transformer according to claim 1 or 3, characterized in that, When the signal processing unit executes the generation of the current compensation coefficient, it is specifically configured as follows: Obtain the DC bias parameters and the effective value of the primary side AC current of the magnetic circuit structure; The operating status is determined based on the temperature change, the DC bias parameter, and the effective value. Query the temperature-bias-current three-dimensional lookup table to obtain the current compensation coefficient corresponding to the operating state.

6. The current transformer according to claim 5, characterized in that, When the signal processing unit executes the correction of the original current signal, it is specifically configured as follows: Obtain the system health index; When the system health index is higher than the first health threshold, the original current signal is corrected using the current compensation coefficient corresponding to the current working state. When the system health index is between the first health threshold and the second health threshold, the current compensation coefficient corresponding to the current working state and the historical compensation coefficient are weighted and averaged to obtain the effective compensation coefficient, and the original current signal is corrected using the effective compensation coefficient. The weight of the historical compensation coefficient is dynamically adjusted according to the system health index.

7. The current transformer according to claim 3, characterized in that, When the signal processing unit executes the injected test signal, it is specifically configured as follows: The interference level is determined based on the signal-to-noise ratio of the test signal and the external electromagnetic interference detection results. When the current interference level is a strong interference environment, increase the density of test sub-signals in the high-frequency band; When the current interference level is a stable environment, optimize the distribution of test sub-signals in the low-frequency band; The frequency range of the high-frequency band is higher than that of the low-frequency band.

8. The current transformer according to claim 1 or 3, characterized in that, When the signal processing unit executes the calculation of the permeability response data, it is specifically configured as follows: The amplitude response of the test signal is subjected to time-domain differentiation to obtain the amplitude change rate; when the amplitude change rate exceeds the first distortion threshold, it is determined that there is an amplitude mutation caused by a switching operation, and the amplitude change rate is used as a transient feature; Spectral analysis is performed on the phase response of the test signal. When the amplitude ratio at the power frequency harmonic point exceeds the second distortion threshold, it is determined that there is a phase shift caused by the power frequency harmonic, and the phase shift at the harmonic frequency point is taken as a harmonic feature. Based on the transient characteristics and the harmonic characteristics, determine whether interference distortion is included; If so, perform transient filtering on the amplitude response based on transient characteristics to eliminate amplitude abrupt changes, and perform harmonic filtering on the phase response based on harmonic characteristics to eliminate phase shifts, and calculate permeability response data based on the eliminated amplitude response and phase response; If not, then the permeability response data is calculated based on the amplitude response and the phase response.

9. The current transformer according to claim 4, characterized in that, The preset permeability-temperature mapping relationship includes a main mapping model and an auxiliary mapping model; When the signal processing unit performs the determination of the temperature change, it is specifically configured as follows: Perform time-domain differentiation on the primary current waveform to calculate the rate of change of current. When the rate of change of current is lower than the stable operating condition threshold and the signal-to-noise ratio of the test signal is higher than the reliable measurement threshold, it is determined to be a stable operating condition. When the rate of change of current is higher than the threshold for sudden change of operating condition or the signal-to-noise ratio is lower than the threshold for interference, it is determined to be a sudden change of operating condition. When the stable operating condition is determined, the primary weight of the primary mapping model is set to be higher than the secondary weight of the auxiliary mapping model. When the sudden change condition is determined, the auxiliary weight is set to be higher than the main weight; Based on the magnetic permeability response data, the main temperature change is calculated using the main mapping model. Based on the measured temperature and the rate of change of current, the auxiliary temperature change is calculated using the auxiliary mapping model. The temperature change is calculated by weighted summation based on the primary weight, the secondary weight, the primary temperature change, and the secondary temperature change.

10. The current transformer according to claim 1, characterized in that, The signal processing unit is specifically configured to control the injection of test signals as follows: When the absolute value of the zero-crossing phase deviation of the power frequency current is detected to be lower than the tolerance threshold, the test signal injection is triggered.

Images