SiC power semiconductor ultrahigh-voltage current detection method and system

Abstract

The invention provides a SiC power semiconductor ultrahigh-voltage current detection method and system. The method comprises the steps that differential sampling processing is carried out on magnetic field signals collected by an inner ring Hall sensor array and an outer ring Hall sensor array which are arranged in a preset range of a SiC power semiconductor; performing nonlinear characteristic curve modeling processing on the output signal of the inner ring sensor according to the output signal of the outer ring sensor and the ratio of the output signal of the inner ring sensor to the output signal of the outer ring sensor to obtain a sensor saturation characteristic mapping relation under different magnetic field intensities; performing dynamic weight distribution processing on the inner and outer ring sensor signal ratio according to the mapping relation; and finally, performing segmented reconstruction processing on a current detection result to obtain a complete current measurement value. According to the method, the inner and outer ring differential structure and the dynamic weight distribution strategy are utilized, the saturation problem of the magnetic sensor under the ultra-high current is effectively solved, and the accuracy of current detection is ensured.

Description

Technical Field

[0001] The present invention relates to the field of current detection technology, and in particular to a SiC power semiconductor ultra-high voltage current detection method and system. Background Art

[0002] In current high-voltage power systems, silicon carbide (SiC) power semiconductors are widely used in ultra-high voltage current control and conversion equipment due to their excellent electrical properties, high temperature stability and fast switching characteristics. These SiC power devices are usually integrated into complex power modules with compact structures and contain multiple high-frequency switching units for precise control and conversion of current. Existing ultra-high voltage current detection methods mainly rely on magnetic sensors, which measure the magnetic field generated by the induction current. They have the advantages of simple structure and non-contact measurement, and are widely used in industrial applications.

[0003] However, when dealing with ultra-high voltage current, due to the high current density characteristics of SiC power semiconductors, the magnetic sensor easily enters a magnetic saturation state, causing its magnetic response to no longer be linear, which in turn causes measurement errors and system instability. This nonlinear response is manifested as: when the current exceeds a certain threshold, the linear relationship between the output signal of the magnetic sensor and the actual current value deviates significantly, and the measurement result no longer accurately reflects the actual current size. This problem seriously affects the accuracy of current detection, limits the application of SiC power semiconductors under high load conditions, and becomes an important factor restricting its performance. Summary of the invention

[0004] The main purpose of the present invention is to solve the technical problem that in the existing ultra-high voltage current detection method, the magnetic sensor is easy to enter the magnetic saturation state when processing ultra-high current, resulting in its magnetic response no longer being linear, thereby causing measurement errors; A first aspect of the present invention provides a SiC power semiconductor ultra-high voltage current detection method, the SiC power semiconductor ultra-high voltage current detection method comprising: Perform differential sampling processing on magnetic field signals collected by an inner ring Hall sensor array and an outer ring Hall sensor array set within a preset range of the SiC power semiconductor to obtain an inner ring sensor output signal, an outer ring sensor output signal, and a ratio of the inner and outer ring sensor output signals; According to the output signal of the outer ring sensor and the ratio of the output signals of the inner and outer ring sensors, a nonlinear characteristic curve modeling process is performed on the output signal of the inner ring sensor to obtain a mapping relationship of the sensor saturation characteristics under different magnetic field strengths; According to the sensor saturation characteristic mapping relationship, the inner ring sensor output signal, and the outer ring sensor output signal, a dynamic weight distribution process is performed on the inner and outer ring sensor signal ratio to obtain a sensor weight coefficient within different magnetic field strength ranges; According to the sensor weight coefficient, the inner loop sensor output signal and the outer loop sensor output signal, the current detection result of the SiC power semiconductor is reconstructed in sections to obtain a complete current measurement value.

[0005] Optionally, in a first implementation of the first aspect of the present invention, the magnetic field signals collected by the inner ring Hall sensor array and the outer ring Hall sensor array set within a preset range of the SiC power semiconductor are differentially sampled to obtain the inner ring sensor output signal, the outer ring sensor output signal, and the ratio of the inner and outer ring sensor output signals, including: Performing weighted sum processing on sampling signals of an inner ring Hall sensor array arranged within a preset range of a SiC power semiconductor to obtain an inner ring sensor output signal; Performing weighted sum processing on sampling signals of an outer ring Hall sensor array arranged within a preset range of a SiC power semiconductor to obtain an outer ring sensor output signal; The inner loop sensor output signal and the outer loop sensor output signal are subjected to signal filtering processing, and the inner and outer loop sensor output signal ratio is calculated based on the inner and outer loop sensor output signal after signal filtering.

[0006] Optionally, in a second implementation of the first aspect of the present invention, the nonlinear characteristic curve modeling processing is performed on the output signal of the inner loop sensor according to the output signal of the outer loop sensor and the ratio of the output signals of the inner and outer loop sensors to obtain the sensor saturation characteristic mapping relationship under different magnetic field strengths, including: Calculate the theoretical magnetic field strength value at the setting position of the inner ring Hall sensor array according to the corresponding relationship between the output signal of the outer ring sensor and the set theoretical magnetic field attenuation coefficient; Establishing an initial mapping curve of sensor saturation characteristics according to the ratio of the output signals of the inner and outer ring sensors and the theoretical magnetic field strength value; Performing piecewise polynomial fitting processing on the initial mapping curve of the sensor saturation characteristic to obtain nonlinear correction coefficients in different magnetic field intensity intervals; The sensor saturation characteristic initial mapping curve is corrected according to the nonlinear correction coefficient to obtain a sensor saturation characteristic mapping relationship.

[0007] Optionally, in a third implementation of the first aspect of the present invention, establishing an initial mapping curve of a sensor saturation characteristic according to the ratio of the output signals of the inner and outer ring sensors and the theoretical magnetic field strength value includes: The theoretical magnetic field strength value is divided into intervals to obtain a plurality of magnetic field strength calculation intervals, and an initial data point sequence is established according to the average value of the output signal ratio of the inner and outer ring sensors in each magnetic field strength calculation interval; Performing least square curve fitting processing on the initial data point sequence to obtain a saturation characteristic fitting curve, and dividing the magnetic field intensity range into a linear region and a saturation region according to the inflection point position of the saturation characteristic fitting curve to obtain a piecewise function expression; The piecewise function expressions of the linear region and the saturation region are subjected to continuity constraint processing to obtain a continuous transition function, and the piecewise function expressions are corrected according to the continuous transition function to obtain an initial mapping curve of the sensor saturation characteristic.

[0008] Optionally, in a fourth implementation of the first aspect of the present invention, the dynamic weight allocation processing is performed on the ratio of the inner and outer loop sensor signals according to the sensor saturation characteristic mapping relationship, the inner loop sensor output signal, and the outer loop sensor output signal, and the sensor weight coefficients within different magnetic field strength ranges are obtained, including: Determine the saturation threshold of the inner ring Hall sensor array in different magnetic field strength intervals according to the sensor saturation characteristic mapping relationship; Calculating the sensor saturation of the inner ring Hall sensor array under the corresponding magnetic field strength according to the inner ring sensor output signal and the outer ring sensor output signal; Determining initial parameters for weight allocation according to a comparison result between the sensor saturation under the magnetic field strength and the saturation threshold; The initial parameters of the weight distribution are subjected to smooth transition processing to obtain sensor weight coefficients within different magnetic field strength ranges.

[0009] Optionally, in a fifth implementation of the first aspect of the present invention, calculating the sensor saturation of the inner-loop Hall sensor array under the corresponding magnetic field strength according to the inner-loop sensor output signal and the outer-loop sensor output signal includes: Performing magnetic field intensity conversion processing on the output signal of the outer ring sensor to obtain the magnetic field intensity value of the current detection point; Determine a theoretical output signal value according to a corresponding position of the magnetic field intensity value of the current detection point in the sensor saturation characteristic mapping relationship; Calculating the deviation between the output signal of the inner loop sensor and the theoretical output signal value to obtain an actual deviation value; According to the ratio of the actual deviation value to the preset standard deviation, the sensor saturation of the inner ring Hall sensor array under the corresponding magnetic field strength is obtained.

[0010] Optionally, in a sixth implementation of the first aspect of the present invention, the segmented reconstruction processing of the current detection result of the SiC power semiconductor according to the sensor weight coefficient, the inner loop sensor output signal and the outer loop sensor output signal to obtain a complete current measurement value includes: Divide the magnetic field strength into a first section below a first magnetic field strength threshold, a second section from the first magnetic field strength threshold to a second magnetic field strength threshold, and a third section above the second magnetic field strength threshold; The inner loop sensor output signal is used to calculate the current value of the first section, the weighted combination value of the inner loop sensor output signal and the outer loop sensor output signal is used to calculate the current value of the second section according to the sensor weight coefficient, and the outer loop sensor output signal is used to calculate the current value of the third section to obtain segment current detection data; Performing continuity analysis on current detection data at critical points between the first section and the second section, and between the second section and the third section, to obtain critical point transition parameters; The segmented current detection data is subjected to smooth transition processing according to the critical point transition parameter to obtain a complete current measurement value.

[0011] A second aspect of the present invention provides a SiC power semiconductor ultra-high voltage current detection system, the SiC power semiconductor ultra-high voltage current detection system comprising: The differential acquisition module is used to perform differential sampling processing on the magnetic field signals collected by the inner ring Hall sensor array and the outer ring Hall sensor array set within the preset range of the SiC power semiconductor to obtain the inner ring sensor output signal, the outer ring sensor output signal and the ratio of the inner and outer ring sensor output signals; A feature modeling module is used to perform nonlinear characteristic curve modeling processing on the output signal of the inner ring sensor according to the output signal of the outer ring sensor and the ratio of the output signals of the inner and outer ring sensors, so as to obtain a sensor saturation characteristic mapping relationship under different magnetic field strengths; A weight distribution module is used to dynamically distribute the weights of the inner and outer ring sensor signal ratios according to the sensor saturation characteristic mapping relationship, the inner ring sensor output signal, and the outer ring sensor output signal to obtain sensor weight coefficients within different magnetic field strength ranges; The data reconstruction module is used to perform segmented reconstruction processing on the current detection result of the SiC power semiconductor according to the sensor weight coefficient, the inner loop sensor output signal and the outer loop sensor output signal to obtain a complete current measurement value.

[0012] The above-mentioned SiC power semiconductor ultra-high voltage current detection method and system performs differential sampling processing on the magnetic field signals collected by the inner ring Hall sensor array and the outer ring Hall sensor array set within the preset range of the SiC power semiconductor; according to the outer ring sensor output signal and the ratio of the inner and outer ring sensor output signals, the inner ring sensor output signal is modeled with a nonlinear characteristic curve to obtain the sensor saturation characteristic mapping relationship under different magnetic field intensities; according to the mapping relationship, the inner and outer ring sensor signal ratio is dynamically weighted; finally, the current detection result is segmented and reconstructed to obtain a complete current measurement value. The present invention utilizes the inner and outer ring differential structure and the dynamic weight allocation strategy to effectively solve the saturation problem of the magnetic sensor under ultra-high current and ensure the accuracy of current detection.

[0013] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0014] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a schematic diagram of a first embodiment of a method for detecting ultra-high voltage current of a SiC power semiconductor according to an embodiment of the present invention; Figure 2 Schematic diagram of an embodiment of a SiC power semiconductor ultra-high voltage current detection system in an embodiment of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] The terms "including" and "having" and any variations thereof mentioned in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device end including a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or device ends.

[0018] To facilitate understanding of this embodiment, a SiC power semiconductor ultra-high voltage current detection method disclosed in an embodiment of the present invention is first introduced in detail. Figure 1 As shown, the method comprises the following steps: 101. Perform differential sampling processing on magnetic field signals collected by an inner ring Hall sensor array and an outer ring Hall sensor array arranged within a preset range of a SiC power semiconductor to obtain an inner ring sensor output signal, an outer ring sensor output signal, and a ratio of the inner and outer ring sensor output signals; In one embodiment of the present invention, the differential sampling processing of magnetic field signals collected by the inner-ring Hall sensor array and the outer-ring Hall sensor array set within the preset range of the SiC power semiconductor to obtain the inner-ring sensor output signal, the outer-ring sensor output signal and the ratio of the inner-ring sensor output signals includes: weighted summing processing of the sampling signals of the inner-ring Hall sensor array set within the preset range of the SiC power semiconductor to obtain the inner-ring sensor output signal; weighted summing processing of the sampling signals of the outer-ring Hall sensor array set within the preset range of the SiC power semiconductor to obtain the outer-ring sensor output signal; signal filtering processing of the inner-ring sensor output signal and the outer-ring sensor output signal, and calculating the ratio of the inner-ring sensor output signals based on the inner-ring sensor output signal and the outer-ring sensor output signal after signal filtering.

[0019] Specifically, the inner ring Hall sensor array consists of 8 Hall sensors evenly distributed in a circular area close to the current path. A 16-channel analog multi-way switch is used to select the output signal of each sensor, and the sampling frequency is 100kHz. The installation angle of each Hall sensor is precisely adjusted to ensure that its sensing surface is perpendicular to the magnetic induction line. The weight coefficient is assigned according to the position of each sensor in the array. The weight coefficient of the sensor close to the center of the current path is 0.8-1.0, and the weight coefficient of the sensor far from the center is 0.6-0.8. The signal acquisition uses a 24-bit analog-to-digital converter with a sampling resolution of 1μT. The collected digital signal is processed in real time by a field programmable gate array (FPGA). The sampled signal of each sensor is multiplied by the corresponding weight coefficient, and a pipeline structure is used to realize high-speed multiplication operation. Then, the summation operation of the 8-way signal is completed by a 32-bit fixed-point adder, and the output signal of the inner ring sensor is finally obtained. The data transmission in this process adopts a differential signal mode to suppress common-mode interference, and optical fiber isolation technology is used to ensure the anti-interference of signal acquisition.

[0020] Specifically, the outer ring Hall sensor array is also composed of 8 Hall sensors, which are evenly distributed in a circular area far away from the current path. The spacing between the sensors is determined by electromagnetic field simulation optimization. Considering the magnetic field attenuation effect, the weight coefficient of each sensor needs to be calculated according to its position and the attenuation characteristics of the magnetic field strength. Based on the Biot-Savart law, a magnetic field attenuation model is established, and the weight coefficient of the outer ring sensor is set between 0.4 and 0.6. The weight distribution adopts an adaptive algorithm and is adjusted in real time according to the magnetic field strength. The signal acquisition system uses the same hardware architecture as the inner ring sensor array, but a programmable gain amplifier is added to the signal conditioning circuit to compensate for the signal strength reduction caused by magnetic field attenuation. The sampled data is processed by FPGA to complete the weighted and summed operations to obtain the output signal of the outer ring sensor. A double buffer mechanism is used in the entire signal processing process to ensure the real-time performance of data processing.

[0021] Specifically, the output signals of the inner and outer ring sensors are filtered by a digital signal processor (DSP), and an eighth-order Butterworth low-pass filter with a cutoff frequency of 10kHz is used to filter out high-frequency noise and power frequency interference. The order and cutoff frequency of the filter are determined by analyzing the signal spectrum characteristics to ensure that interference is suppressed to the greatest extent while retaining the effective signal. The filtered signal is divided by 32-bit floating-point operations. The data is normalized before calculation to adjust the numerical range to a suitable interval for the operation. The division operation is implemented by an iterative algorithm, and the calculation accuracy and speed are improved by combining a lookup table and Newton iteration. In order to prevent the calculation overflow caused by the divisor being too small, a threshold is set for judgment. When the output signal of the outer ring sensor is lower than the threshold, the exception handling mechanism is activated. The final output signal ratio of the inner and outer ring sensors is processed by data smoothing to eliminate the influence of quantization noise. The calculation results are transmitted to the host computer through the optical fiber communication interface for data storage and display.

[0022] 102. Construct a feature vector according to the voltage fluctuation sampling data, the load current sampling data and the environmental parameter data, and input the feature vector into a pre-trained voltage fluctuation prediction model to obtain voltage fluctuation prediction data of the power supply voltage of each vehicle-standard MCU; In one embodiment of the present invention, the nonlinear characteristic curve modeling processing is performed on the inner loop sensor output signal according to the outer loop sensor output signal and the ratio of the inner and outer loop sensor output signals to obtain the sensor saturation characteristic mapping relationship under different magnetic field strengths, including: calculating the theoretical magnetic field intensity value at the setting position of the inner loop Hall sensor array according to the correspondence between the outer loop sensor output signal and the set theoretical magnetic field attenuation coefficient; establishing the sensor saturation characteristic initial mapping curve according to the ratio of the inner and outer loop sensor output signals and the theoretical magnetic field intensity value; performing piecewise polynomial fitting processing on the sensor saturation characteristic initial mapping curve to obtain nonlinear correction coefficients for different magnetic field strength intervals; and correcting the sensor saturation characteristic initial mapping curve according to the nonlinear correction coefficient to obtain the sensor saturation characteristic mapping relationship.

[0023] Specifically, the outer ring sensor array is set at a fixed distance from the current path, and the magnetic field strength at this position theoretically follows the attenuation law of the Biot-Savart law. When setting the theoretical magnetic field attenuation coefficient, the magnetic field distribution model around the current path is first established by finite element simulation software, and the theoretical values ​​of the magnetic field strength at the outer ring position and the inner ring position are calculated under different current values. These theoretical values ​​are compared with the actual calibration experimental data to obtain the corrected magnetic field attenuation coefficient. The influence of factors such as the actual installation error of the sensor and the uneven distribution of the surrounding magnetic permeability are considered in the correction process. In order to improve the calculation accuracy, a high-precision current source is used in the calibration experiment to generate a standard current in the range of 0-1000A, with a step of 10A, and 1000 data are collected at each current value to take the average value. At the same time, a three-axis magnetic field compensation coil is arranged around the outer ring sensor, and dynamic compensation of the background magnetic field is realized through feedback control, reducing the influence of the background magnetic field to less than 0.1μT. The output signal of the outer ring sensor is substituted into the attenuation model, and the theoretical magnetic field strength value at the inner ring position is calculated according to the known distance ratio from the outer ring to the inner ring. The piecewise linear interpolation method is used in the calculation to map the output signal of the outer ring sensor to the corresponding magnetic field strength value, and then the theoretical magnetic field strength value at the inner ring position is converted according to the attenuation coefficient. The determination of the distance ratio is guaranteed by precision machining and three-coordinate measurement system, with machining accuracy of ±0.01mm and measurement accuracy of ±0.005mm.

[0024] Specifically, the calculated theoretical magnetic field strength value is divided into 50 intervals according to the logarithmic scale. The range of each interval is 1.2 times that of the previous interval, covering the magnetic field strength range of 0.1T to 2T as a whole. In each interval, 100 sets of sampling data of the output signal ratio of the inner and outer ring sensors are collected. The sampling frequency is set to 100kHz, and the acquisition time of each set of data is 10ms. The collected data is first subjected to median filtering to remove burst interference, and then the outliers are removed using the 3σ criterion. The arithmetic mean of the remaining data is calculated as the characteristic data point of the interval. To ensure the reliability of the data, the ambient temperature, power supply voltage and other working parameters are recorded synchronously during the acquisition process of each interval. When these parameters exceed the set range, the corresponding data will be marked and re-collected. These characteristic data points are sorted according to the size of the magnetic field strength to obtain the initial discrete data sequence. The discrete data sequence is smoothed by the cubic spline interpolation method. The selection of interpolation nodes is adaptively determined based on the data distribution density. The node spacing automatically decreases where the data changes drastically and automatically increases where the data changes gently. During the interpolation process, the smoothness of the curve is controlled by adjusting the smoothing factor, and the generalized cross-validation criterion is used to select the smoothing factor. The residual analysis of the interpolation results is performed to ensure that the interpolation error is within ±0.1%. The smoothed curve is the initial mapping curve of the sensor saturation characteristic, which is described by 1000 points and stored as a floating-point array.

[0025] Specifically, the initial mapping curve of the sensor saturation characteristic is numerically analyzed, and the first-order derivative and second-order derivative of the curve in the entire magnetic field intensity range are calculated by the five-point method. Differential intervals of different scales are used in derivative calculation, and the calculation accuracy is improved by weighted averaging. The inflection point of the curve is determined according to the rate of change of the first-order derivative and the zero point of the second-order derivative, which marks the beginning of the sensor entering the saturation state. The inflection point is determined by an iterative method. First, the region where the derivative changes significantly is roughly located, and then the inflection point is accurately located by subdividing the interval within the region. The curve is divided into a linear region and a saturation region with the inflection point as the boundary. A first-order polynomial is used for fitting in the linear region, and the fitting coefficient is obtained by solving the normal equation group by the least squares method. In the saturation region, it is subdivided into multiple sub-intervals, each sub-interval is fitted by a fifth-order polynomial, and derivative continuity constraints are imposed at the boundary points of adjacent sub-intervals. The QR decomposition method is used to solve the overdetermined equation group in the fitting process, and the numerical stability of the solution is ensured by conditional number analysis. The fitting effect is evaluated by calculating the root mean square error of the fitting residual. When the error exceeds the set threshold, the sub-interval division method is automatically adjusted and refitted. The polynomial coefficients obtained by fitting are compared with the ideal linear response to obtain the nonlinear correction coefficients for each magnetic field intensity interval.

[0026] Specifically, in each magnetic field intensity interval, the nonlinear correction coefficient is applied to the corresponding segment of the initial mapping curve to correct the nonlinear response of the sensor. The correction adopts a segmented processing method. The correction process of each interval takes into account the influence of the temperature coefficient, which is obtained by calibrating at multiple temperature points in the range of -40℃ to 85℃. A transition region with a width of 5% of the theoretical magnetic field intensity is added at the interval boundary point. A cubic Hermite interpolation polynomial is used in the transition region to achieve smooth transition. The coefficients of the interpolation polynomial are obtained by analytical solution to ensure that the function value and the first-order derivative at the boundary point are continuous. The specific width of the transition region is determined by the particle swarm optimization algorithm, and the optimization goal is to minimize the root mean square error of the corrected curve. The corrected segmented curves are combined into a complete mapping relationship, and this mapping relationship is discretized and stored as a lookup table containing 10,000 points. The sampling interval of the lookup table adopts a non-uniform division strategy. In the area where the nonlinear characteristics change drastically, the sampling interval is set to 0.1 times the nominal value, and in the area where the characteristics change gently, the sampling interval is increased to 5 times the nominal value. Each data point in the lookup table contains three parameters: magnetic field strength value, correction coefficient and temperature correction factor. The data format is stored in 32-bit floating point. The final sensor saturation characteristic mapping relationship is used to correct the output signal of the inner loop sensor in real time through online search and linear interpolation.

[0027] Furthermore, the establishment of the initial mapping curve of the sensor saturation characteristic according to the ratio of the output signals of the inner and outer ring sensors and the theoretical magnetic field strength value includes: performing interval division processing on the theoretical magnetic field strength value to obtain multiple magnetic field strength calculation intervals, and establishing an initial data point sequence according to the mean value of the output signal ratio of the inner and outer ring sensors in each of the magnetic field strength calculation intervals; performing least squares curve fitting processing on the initial data point sequence to obtain a saturation characteristic fitting curve, and dividing the magnetic field intensity range into a linear region and a saturation region according to the inflection point position of the saturation characteristic fitting curve to obtain a piecewise function expression; performing continuity constraint processing on the piecewise function expressions of the linear region and the saturation region to obtain a continuous transition function, and correcting the piecewise function expression according to the continuous transition function to obtain the initial mapping curve of the sensor saturation characteristic.

[0028] Specifically, the entire theoretical magnetic field strength measurement range of 0.1T to 2T is divided into 50 intervals according to the logarithmic scale, and the range of each interval is 1.2 times that of the previous interval. When dividing the intervals, the precise installation positions of the inner and outer ring sensors are determined by three-coordinate measurement, and the position measurement accuracy is controlled within the range of ±0.005mm. For each divided magnetic field strength interval, a high-precision programmable current source is used to generate a standard excitation current at the corresponding magnetic field strength. The accuracy of the current source is 0.01%, and the long-term stability is better than 0.005%. 100 groups of data are sampled in each interval, each group of sampling lasts for 10ms, the sampling frequency is 100kHz, and 1000 sampling points are obtained. The collected data is first processed by a digital bandpass filter, and the effective frequency band of the signal is determined by spectrum analysis to filter out power frequency interference and high-frequency noise. Then the 3σ criterion is used to eliminate abnormal data points, and the arithmetic mean is calculated for the remaining data. During the data processing process, the working parameters such as ambient temperature and humidity are recorded synchronously. When these parameters exceed the set range, they are automatically marked and re-collected. The 50 interval means finally obtained constitute the initial data point sequence, and each data point contains the theoretical magnetic field intensity value and the corresponding ratio of the output signals of the inner and outer ring sensors.

[0029] Specifically, the least squares curve fitting is performed on the initial data point sequence, and the quintic polynomial is used as the fitting basis function. The QR decomposition method is used to solve the normal equations in the fitting process, and the condition number analysis is used to ensure the numerical stability of the solution process. The polynomial coefficients obtained by fitting are rounded off to retain six significant figures. The fitting curve is numerically analyzed, and the first-order derivative and second-order derivative are calculated using the five-point difference formula. The difference step size is optimized and determined by the Richardson extrapolation method. The inflection point position of the fitting curve is determined by finding the zero point of the second-order derivative. The inflection point search adopts the dichotomy method. The initial search interval is the entire magnetic field intensity range, and the search accuracy is set to 0.001T. According to the inflection point position, the magnetic field intensity range is divided into a linear region and a saturation region. In the linear region, a linear function is used to describe the sensor response characteristics, and in the saturation region, a cubic polynomial is used to describe the nonlinear characteristics. The coefficients of the piecewise function are obtained by performing least squares fitting on the data points in the two regions respectively, and the root mean square error of the fitting residual is controlled within 0.1%.

[0030] Specifically, a transition interval with a width of 0.1T is added at the boundary between the linear region and the saturation region, and a continuous transition function is constructed in the transition interval using a cubic Hermite interpolation polynomial. The construction of the transition function must meet the continuity constraints of the function value and derivative value at the boundary point. The Lagrange multiplier method is used to handle the constraints, and the constraints are converted into an equivalent optimization problem. The optimization process is solved using the Trust-Region algorithm, and the iteration termination condition is that the residual is less than 0.0001 or the maximum number of iterations is 100. The obtained transition function coefficients are analyzed for numerical stability to ensure that there will be no numerical oscillation in the entire transition interval. The transition function is combined with the original piecewise function, and the smooth transition of the three regions is achieved by adjusting the function value and derivative value at both ends of the transition interval. Finally, the combined piecewise function is corrected as a whole. The correction process takes into account the influence of temperature, and the temperature coefficient in the range of -40℃ to 85℃ is obtained by experimental calibration. The corrected function expression is discretized and stored as a lookup table containing 10,000 points. The sampling interval of the lookup table is dense in the nonlinear region and sparse in the linear region, and finally the initial mapping curve of the sensor saturation characteristic is obtained.

[0031] 103. According to the mapping relationship of the sensor saturation characteristics, the output signal of the inner loop sensor, and the output signal of the outer loop sensor, a dynamic weight distribution process is performed on the ratio of the inner and outer loop sensor signals to obtain the sensor weight coefficient within different magnetic field strength ranges; In one embodiment of the present invention, the dynamic weight allocation processing is performed on the ratio of inner and outer loop sensor signals according to the sensor saturation characteristic mapping relationship, the inner loop sensor output signal, and the outer loop sensor output signal to obtain the sensor weight coefficient within different magnetic field strength ranges, including: determining the saturation threshold of the inner loop Hall sensor array in different magnetic field strength intervals according to the sensor saturation characteristic mapping relationship; calculating the sensor saturation of the inner loop Hall sensor array under the corresponding magnetic field strength according to the inner loop sensor output signal and the outer loop sensor output signal; determining the initial weight allocation parameters according to the comparison result of the sensor saturation under the magnetic field strength and the saturation threshold; and performing smooth transition processing on the initial weight allocation parameters to obtain the sensor weight coefficient within different magnetic field strength ranges.

[0032] Specifically, according to the mapping relationship of the sensor saturation characteristics, the magnetic field intensity range of 0.1T to 2T is divided into three main intervals. First, the first-order derivative of the mapping curve in the entire range is calculated by numerical analysis method, using a five-point differential formula with a differential step of 0.001T. The change rate of the first-order derivative is analyzed, and three key saturation thresholds are set: when the derivative change rate is less than 5%, it is set to a mild saturation threshold of 0.2, the derivative change rate is between 5% and 15% and the medium saturation threshold is 0.5, and the derivative change rate is greater than 15% and the deep saturation threshold is 0.8. In the threshold determination process, a high-precision current source is used to perform multi-point calibration in the temperature range of -40℃ to 85℃, with a calibration point interval of 5℃, 1000 sets of data are collected at each temperature point, and the sampling duration of each set of data is 10ms and the sampling frequency is 100kHz. The temperature compensation coefficient is obtained by least squares fitting, and the compensation coefficient is described by a cubic polynomial. The polynomial coefficient is determined by the QR decomposition method. The setting of the threshold also takes into account the temperature drift characteristics of the sensor. A ±3% hysteresis interval is set near each threshold point to prevent frequent state switching due to temperature fluctuations. The width of the hysteresis interval is determined by analyzing the impact of temperature on the sensor output. During the analysis process, a thermal imager is used to monitor the temperature distribution of the sensor in real time, with a temperature resolution of 0.1°C. A temperature-threshold relationship lookup table is established based on experimental data. The lookup table contains 1000 discrete points, each of which contains three parameters: temperature value, threshold and compensation coefficient. Cubic spline interpolation is used to achieve continuous adjustment of the threshold, and the selection of the interpolation node is adaptively determined based on the temperature change rate. The threshold data is finally stored in the Block RAM of the FPGA, with a storage depth of 4096, and each data occupies 32 bits.

[0033] Specifically, the output signal of the outer ring sensor is used as a reference benchmark. The signal is sampled by a 24-bit Σ-Δ analog-to-digital converter with a sampling frequency of 100kHz and an integration time of 10μs. The signal acquisition circuit adopts a four-layer PCB design, the signal layer and the power layer adopt a complete ground plane design, and the analog part and the digital part are strictly separated. The current magnetic field strength value is calculated through the pre-calibrated magnetic field strength conversion relationship. The conversion relationship is implemented by a piecewise linear interpolation method. The interpolation coefficient is stored in the lookup table with a table depth of 8192. The temperature coefficient of the sensor is considered in the calculation process. The temperature information is provided by a high-precision PT100 temperature sensor with a temperature measurement accuracy of 0.1℃. The temperature measurement circuit adopts a four-wire connection method to eliminate the influence of lead resistance. The temperature signal is sampled by a 24-bit Σ-Δ ADC with a sampling frequency of 10Hz, and the data is smoothed by an IIR low-pass filter. The calculated magnetic field strength value is substituted into the sensor saturation characteristic mapping relationship to obtain the theoretical output value of the inner ring sensor under the magnetic field strength. The mapping relationship is accessed using a binary search algorithm. The search process is implemented in the FPGA, and the average search time is less than 1μs. At the same time, the actual output signal of the inner ring sensor is sampled and digitally filtered. The filtering adopts a tenth-order Butterworth low-pass filter with a cutoff frequency of 10kHz. The coefficient of the filter is optimized and calculated by the MATLAB tool. The difference between the actual output signal and the theoretical output value is divided by the theoretical value to obtain the relative deviation. The division operation is implemented using the CORDIC algorithm with a calculation accuracy of 32-bit floating point. The relative deviation is then divided by the preset standard deviation of 0.05 to obtain the saturation value of the inner ring Hall sensor array under the current magnetic field strength. All calculation processes are implemented in the FPGA using a pipeline structure, which is divided into 8 pipelines, and the delay of each stage does not exceed 100ns.

[0034] Specifically, a four-level weight distribution structure is established to compare the saturation value with the light saturation threshold of 0.2, the medium saturation threshold of 0.5, and the deep saturation threshold of 0.8. The comparison circuit is implemented with a high-speed comparator, and the comparison result is processed by a state machine with a working clock frequency of 100MHz. Eight states are designed in the state machine: the initial state, three comparison states, three weight distribution states, and one output state. The processing time of each state does not exceed 10ns, ensuring that the processing delay of the entire state machine is within 100ns. When the saturation is less than 0.2, the weight of the inner loop sensor is set to 1.0 and the weight of the outer loop sensor is set to 0. At this time, the sensor is considered to be working in a completely linear region; when the saturation is between 0.2 and 0.5, the weight of the inner loop sensor is reduced from 1.0 to 0.6 according to linear interpolation, and the weight of the outer loop sensor is increased from 0 to 0.4. The interpolation uses 1000 points to achieve a smooth transition; when the saturation is between 0.5 and 0.8, the weight of the inner loop sensor continues to decrease linearly to 0.2, and the weight of the outer loop sensor is increased to 0.8, and 1000 points are also used for interpolation; when the saturation is greater than 0.8, the weight of the inner loop sensor is set to 0 and the weight of the outer loop sensor is set to 1.0, indicating that the inner loop sensor is fully saturated. The weight calculation uses a 32-bit floating point unit with a calculation accuracy better than 0.001. Each weight value is equipped with a 64-bit timestamp to record the timing information of the weight change. A 100ns delay is set during the threshold comparison process, which is implemented by a programmable delay unit with a delay accuracy of 1ns. The comparison results are processed through a three-stage pipeline, and each stage of the pipeline is equipped with a data verification mechanism.

[0035] Specifically, a digital low-pass filter is used to smooth the initial parameters of weight allocation. The filter is a fifth-order Butterworth type with a cutoff frequency of 1kHz. The coefficients of the filter are calculated by MATLAB tools and converted into 32-bit fixed-point format. A total of 25 filter coefficients are designed. A pipeline FIR filter is implemented in FPGA, which contains 25 multipliers and 24 adders, and the processing delay is controlled within 1μs. The input data and coefficients of the filter are both in 32-bit fixed-point format, and the decimal point position is fixed at the 16th bit. In order to prevent data overflow, an overflow detection circuit is set at the output of each adder. When the weight coefficient changes in a step, a linear interpolation algorithm is used to achieve a smooth transition. The interpolation interval is 100 sampling points, and the time interval of each sampling point is 10μs. At the same time, a fast interpolation circuit based on a lookup table is implemented. The lookup table contains 1000 pre-calculated interpolation coefficients, which are stored in a dual-port RAM and support simultaneous read and write operations. The weight coefficients after smoothing are normalized to ensure that the sum of the weights of the inner and outer loop sensors is always 1. The normalization calculation is implemented using an iterative divider, and the required accuracy can be achieved after 4 iterations. The delay of the filter is compensated by a shift register with a depth of 32 and a clock frequency of 100MHz. The final sensor weight coefficients are stored in a double buffer with a buffer size of 4KB and transmitted to the digital signal processor through the SPI interface at a rate of 10Mbps for subsequent calculations.

[0036] Furthermore, the calculation of the sensor saturation of the inner-loop Hall sensor array under the corresponding magnetic field strength according to the inner-loop sensor output signal and the outer-loop sensor output signal includes: performing magnetic field strength conversion processing on the outer-loop sensor output signal to obtain the magnetic field strength value of the current detection point; determining the theoretical output signal value according to the corresponding position of the magnetic field strength value of the current detection point in the sensor saturation characteristic mapping relationship; performing deviation calculation between the inner-loop sensor output signal and the theoretical output signal value to obtain an actual deviation value; and obtaining the sensor saturation of the inner-loop Hall sensor array under the corresponding magnetic field strength according to the ratio of the actual deviation value to a preset standard deviation.

[0037] Specifically, the magnetic field intensity conversion of the output signal of the outer ring sensor first establishes a relationship model between magnetic field intensity and current based on the Biot-Savart law. Since the outer ring sensor is far away from the current path, the magnetic field intensity at its location is proportional to the current and inversely proportional to the distance. When converting the output signal of the outer ring sensor, first obtain the actual distance value from the outer ring sensor to the center of the current path, which is obtained by three-dimensional coordinate measurement. Then, according to the sensitivity coefficient of the outer ring sensor (the coefficient is obtained by experimental calibration and represents the output voltage value of the sensor under unit magnetic field intensity), the sensor output signal is divided by the sensitivity coefficient. Then, the calculated result is multiplied by the ratio of the distance value from the sensor to the current path to the calibration distance to obtain the spatial distance correction coefficient. At the same time, considering the influence of ambient temperature on sensor sensitivity, a temperature correction factor is introduced, which is obtained by looking up the table according to the current temperature value. Finally, the corrected result is substituted into the magnetic field intensity calculation formula to obtain the actual magnetic field intensity value of the current detection point, which is expressed in Tesla (T). The entire conversion process takes into account multiple influencing factors such as spatial position, temperature influence and sensor intrinsic characteristics to ensure the accuracy of the conversion result.

[0038] Specifically, when searching for the theoretical output signal value corresponding to the magnetic field strength value of the current detection point in the sensor saturation characteristic mapping relationship, first substitute the current magnetic field strength value into the piecewise function of the mapping relationship. The mapping relationship is divided into two parts: the linear region and the saturation region. In the linear region, the theoretical output signal value is equal to the magnetic field strength value multiplied by the sensitivity coefficient of the sensor in the linear region; in the saturation region, the theoretical output signal value is calculated by a cubic polynomial, and the coefficient of the polynomial has been determined in the early modeling. For the magnetic field strength value near the boundary between the two regions, the theoretical output value needs to be calculated by the transition function. The transition function uses a cubic Hermite interpolation polynomial to ensure the continuity of the function value and derivative at the boundary point. Considering the influence of temperature on the sensor characteristics, the theoretical output value also needs to be corrected according to the current temperature, and the correction coefficient is obtained through the temperature lookup table. The final theoretical output signal value reflects the ideal output characteristics that the inner ring sensor should have under the current magnetic field strength.

[0039] Specifically, when calculating the deviation between the actual output signal of the inner loop sensor and the theoretical output signal value, the two signals are first converted into the same physical quantity unit. Since the actual output signal has been processed by the previous conditioning circuit, it is necessary to first remove the gain and bias effects introduced by the signal conditioning circuit. The deviation calculation adopts the form of relative error, dividing the difference between the actual output signal and the theoretical output signal by the theoretical output signal value. The calculated relative error is the actual deviation value, which reflects the degree of deviation between the actual output of the inner loop sensor and the ideal linear output. During the calculation process, it is necessary to judge whether the theoretical output signal value is zero. When the theoretical value is close to zero, the preset minimum threshold is used instead to avoid division by zero errors.

[0040] Specifically, the actual deviation value is divided by the preset standard deviation (the standard deviation value is 0.05) to obtain the saturation value of the inner loop sensor. The determination of the standard deviation value is based on a large amount of statistical analysis of experimental data, which reflects the maximum deviation range allowed by the sensor under normal working conditions. When the calculated saturation value is less than 1, it means that the sensor is working within the normal range; when the saturation value is greater than 1, it means that the sensor has begun to show saturation characteristics, and the larger the saturation value, the deeper the saturation. The influence of measurement noise must also be considered in the calculation process, and the saturation value jitter caused by noise can be avoided by setting a dead zone. The final output saturation value will serve as an important basis for determining the weight distribution of the inner and outer loop sensors.

[0041] 104. According to the sensor weight coefficient, the inner loop sensor output signal and the outer loop sensor output signal, the current detection result of the SiC power semiconductor is reconstructed in sections to obtain a complete current measurement value.

[0042] In one embodiment of the present invention, the current detection result of the SiC power semiconductor is reconstructed in sections according to the sensor weight coefficient, the inner-loop sensor output signal and the outer-loop sensor output signal to obtain a complete current measurement value, including: dividing the magnetic field strength into a first section below a first magnetic field strength threshold, a second section from the first magnetic field strength threshold to a second magnetic field strength threshold, and a third section above the second magnetic field strength threshold; the inner-loop sensor output signal is used to calculate the current value of the first section, the weighted combination value of the inner-loop sensor output signal and the outer-loop sensor output signal is used to calculate the current value of the second section according to the sensor weight coefficient, and the outer-loop sensor output signal is used to calculate the current value of the third section to obtain section current detection data; the current detection data at the critical points between the first section and the second section and between the second section and the third section are analyzed for continuity to obtain critical point transition parameters; the section current detection data is subjected to smooth transition processing according to the critical point transition parameters to obtain a complete current measurement value.

[0043] Specifically, two key magnetic field strength threshold points are determined according to the saturation characteristics of the inner-loop Hall sensor. The first magnetic field strength threshold is set at 0.5 Tesla, which corresponds to the position where the inner-loop sensor begins to have a slight nonlinear response. Below this threshold, the inner-loop sensor maintains good linearity. The second magnetic field strength threshold is set at 1.2 Tesla, which corresponds to the position where the inner-loop sensor enters a deep saturation state, at which point the output response of the sensor deviates seriously from the linear characteristics. The entire measurement range is divided into three sections through these two threshold points: the first section covers the range of 0 to 0.5 Tesla, at which point the inner-loop sensor operates in the linear region; the second section covers the range of 0.5 to 1.2 Tesla, at which point the inner-loop sensor begins to have nonlinear characteristics but is not yet fully saturated; the third section covers the range above 1.2 Tesla, at which point the inner-loop sensor is already in a deep saturation state. This segmentation method takes into account the physical characteristics and actual application requirements of the sensor, and the selection of the threshold is based on statistical analysis of a large amount of experimental data.

[0044] Specifically, in the first section, due to the low magnetic field strength, the inner loop sensor works in the linear region, and the output signal of the inner loop sensor is directly divided by its calibrated sensitivity coefficient to obtain the current value. The calibrated sensitivity coefficient is determined by the slope of the input-output characteristic curve of the sensor in this section. In the second section, the inner loop sensor begins to show nonlinear characteristics, and it is necessary to use the information of the inner and outer loop sensors at the same time. The output signal of the inner loop sensor is multiplied by its corresponding weight coefficient, and the output signal of the outer loop sensor is multiplied by its corresponding weight coefficient. The two are added to obtain a weighted combination value, and then the current value is calculated using the comprehensive sensitivity coefficient in this interval. In the third section, the inner loop sensor is already severely saturated, and the current value is mainly calculated by the output signal of the outer loop sensor. The current value can be obtained by dividing the output signal of the outer loop sensor by its sensitivity coefficient at this magnetic field strength. The current calculation of all sections needs to take temperature correction into account, and the temperature correction coefficient is obtained by looking up the temperature characteristic curve table.

[0045] Specifically, the current detection data near the two magnetic field intensity threshold points need to be analyzed for continuity. First, at the first threshold point of 0.5 Tesla, the current values ​​of 50 sampling points before and after the threshold point are extracted to analyze the changing trends of these data. The data before and after the threshold point are fitted by the least squares method to obtain the slope and intercept of the two segments of data. The width of the transition interval is determined based on the slope difference, and the width needs to ensure that the first-order derivative of the current value is continuous in the transition interval. The same method is applied to the data analysis at the second threshold point of 1.2 Tesla. For each threshold point, the transition parameters required to ensure data continuity are calculated based on the changing characteristics of the data before and after, including the transition interval width, the slope adjustment coefficient, and the bias adjustment value.

[0046] Specifically, the critical point transition parameters are used to smoothly connect the current detection data of the three sections. In the transition interval of the first threshold point, a cubic polynomial function is used to achieve smooth transition. The coefficients of the polynomial are determined by the current values ​​and their derivative values ​​at both ends of the transition interval. The same method is used for smoothing in the transition interval of the second threshold point. The data after transition processing ensures the continuity of the current value and the rate of change in the entire measurement range. Finally, the smoothed data is normalized to ensure the dimensional consistency of the measurement results and obtain the complete current measurement value in amperes.

[0047] In this embodiment, differential sampling is performed on the magnetic field signals collected by the inner ring Hall sensor array and the outer ring Hall sensor array set within the preset range of the SiC power semiconductor; the inner ring sensor output signal is modeled and processed according to the nonlinear characteristic curve according to the outer ring sensor output signal and the ratio of the inner and outer ring sensor output signals, and the sensor saturation characteristic mapping relationship under different magnetic field strengths is obtained; the inner and outer ring sensor signal ratio is dynamically weighted according to the mapping relationship; finally, the current detection result is reconstructed in sections to obtain the complete current measurement value. The present invention utilizes the inner and outer ring differential structure and the dynamic weight allocation strategy to effectively solve the saturation problem of the magnetic sensor under ultra-high current and ensure the accuracy of current detection.

[0048] The above describes the SiC power semiconductor ultra-high voltage current detection method in the embodiment of the present invention. The following describes the SiC power semiconductor ultra-high voltage current detection system in the embodiment of the present invention. Figure 2 , an embodiment of the SiC power semiconductor ultra-high voltage current detection system in the embodiment of the present invention includes: The differential acquisition module 201 is used to perform differential sampling processing on the magnetic field signals collected by the inner ring Hall sensor array and the outer ring Hall sensor array set within the preset range of the SiC power semiconductor to obtain the inner ring sensor output signal, the outer ring sensor output signal and the ratio of the inner and outer ring sensor output signals; A feature modeling module 202 is used to perform nonlinear characteristic curve modeling processing on the output signal of the inner loop sensor according to the output signal of the outer loop sensor and the ratio of the output signals of the inner and outer loop sensors, so as to obtain a sensor saturation characteristic mapping relationship under different magnetic field strengths; The weight distribution module 203 is used to perform dynamic weight distribution processing on the ratio of inner and outer loop sensor signals according to the sensor saturation characteristic mapping relationship, the inner loop sensor output signal, and the outer loop sensor output signal to obtain sensor weight coefficients within different magnetic field strength ranges; The data reconstruction module 204 is used to perform segment-by-segment reconstruction processing on the current detection result of the SiC power semiconductor according to the sensor weight coefficient, the inner loop sensor output signal and the outer loop sensor output signal to obtain a complete current measurement value.

[0049] In an embodiment of the present invention, the SiC power semiconductor ultra-high voltage current detection system runs the above-mentioned SiC power semiconductor ultra-high voltage current detection method, and the SiC power semiconductor ultra-high voltage current detection system performs differential sampling processing on the magnetic field signals collected by the inner ring Hall sensor array and the outer ring Hall sensor array set within the preset range of the SiC power semiconductor; according to the outer ring sensor output signal and the ratio of the inner and outer ring sensor output signals, the inner ring sensor output signal is modeled with a nonlinear characteristic curve to obtain the sensor saturation characteristic mapping relationship under different magnetic field intensities; according to the mapping relationship, the inner and outer ring sensor signal ratio is dynamically weighted; finally, the current detection result is segmented and reconstructed to obtain a complete current measurement value. The present invention utilizes the inner and outer ring differential structure and the dynamic weight allocation strategy to effectively solve the saturation problem of the magnetic sensor under ultra-high current and ensure the accuracy of current detection.

[0050] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device, or unit can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0051] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0052] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting ultra-high voltage current of SiC power semiconductor, characterized in that: The SiC power semiconductor ultra-high voltage current detection method comprises: Perform differential sampling processing on magnetic field signals collected by an inner ring Hall sensor array and an outer ring Hall sensor array set within a preset range of the SiC power semiconductor to obtain an inner ring sensor output signal, an outer ring sensor output signal, and a ratio of the inner and outer ring sensor output signals; According to the output signal of the outer ring sensor and the ratio of the output signals of the inner and outer ring sensors, a nonlinear characteristic curve modeling process is performed on the output signal of the inner ring sensor to obtain a mapping relationship of the sensor saturation characteristics under different magnetic field strengths; According to the sensor saturation characteristic mapping relationship, the inner ring sensor output signal, and the outer ring sensor output signal, a dynamic weight distribution process is performed on the inner and outer ring sensor signal ratio to obtain a sensor weight coefficient within different magnetic field strength ranges; According to the sensor weight coefficient, the inner loop sensor output signal and the outer loop sensor output signal, the current detection result of the SiC power semiconductor is reconstructed in sections to obtain a complete current measurement value.

2. The SiC power semiconductor ultra-high voltage current detection method according to claim 1, characterized in that: The differential sampling process of the magnetic field signals collected by the inner ring Hall sensor array and the outer ring Hall sensor array arranged within the preset range of the SiC power semiconductor to obtain the inner ring sensor output signal, the outer ring sensor output signal and the ratio of the inner and outer ring sensor output signals includes: Performing weighted sum processing on sampling signals of an inner ring Hall sensor array arranged within a preset range of a SiC power semiconductor to obtain an inner ring sensor output signal; Performing weighted sum processing on sampling signals of an outer ring Hall sensor array arranged within a preset range of a SiC power semiconductor to obtain an outer ring sensor output signal; The inner loop sensor output signal and the outer loop sensor output signal are subjected to signal filtering processing, and the inner and outer loop sensor output signal ratio is calculated based on the inner and outer loop sensor output signal after signal filtering.

3. The SiC power semiconductor ultra-high voltage current detection method according to claim 1, characterized in that: According to the output signal of the outer ring sensor and the ratio of the output signals of the inner and outer ring sensors, the nonlinear characteristic curve modeling process is performed on the output signal of the inner ring sensor to obtain the sensor saturation characteristic mapping relationship under different magnetic field strengths, which includes: Calculate the theoretical magnetic field strength value at the setting position of the inner ring Hall sensor array according to the corresponding relationship between the output signal of the outer ring sensor and the set theoretical magnetic field attenuation coefficient; Establishing an initial mapping curve of sensor saturation characteristics according to the ratio of the output signals of the inner and outer ring sensors and the theoretical magnetic field strength value; Performing piecewise polynomial fitting processing on the initial mapping curve of the sensor saturation characteristic to obtain nonlinear correction coefficients in different magnetic field intensity intervals; The sensor saturation characteristic initial mapping curve is corrected according to the nonlinear correction coefficient to obtain a sensor saturation characteristic mapping relationship.

4. The SiC power semiconductor ultra-high voltage current detection method according to claim 3, characterized in that: The establishing of the initial mapping curve of the sensor saturation characteristic according to the ratio of the output signals of the inner and outer ring sensors and the theoretical magnetic field strength value comprises: The theoretical magnetic field strength value is divided into intervals to obtain a plurality of magnetic field strength calculation intervals, and an initial data point sequence is established according to the average value of the output signal ratio of the inner and outer ring sensors in each magnetic field strength calculation interval; Performing least square curve fitting processing on the initial data point sequence to obtain a saturation characteristic fitting curve, and dividing the magnetic field intensity range into a linear region and a saturation region according to the inflection point position of the saturation characteristic fitting curve to obtain a piecewise function expression; The piecewise function expressions of the linear region and the saturation region are subjected to continuity constraint processing to obtain a continuous transition function, and the piecewise function expressions are corrected according to the continuous transition function to obtain an initial mapping curve of the sensor saturation characteristic.

5. The SiC power semiconductor ultra-high voltage current detection method according to claim 1, characterized in that: According to the sensor saturation characteristic mapping relationship, the inner ring sensor output signal, and the outer ring sensor output signal, the dynamic weight distribution processing is performed on the inner and outer ring sensor signal ratio to obtain the sensor weight coefficient within different magnetic field strength ranges, including: Determine the saturation threshold of the inner ring Hall sensor array in different magnetic field strength intervals according to the sensor saturation characteristic mapping relationship; Calculating the sensor saturation of the inner ring Hall sensor array under the corresponding magnetic field strength according to the inner ring sensor output signal and the outer ring sensor output signal; Determining initial parameters for weight allocation according to a comparison result between the sensor saturation under the magnetic field strength and the saturation threshold; The initial parameters of the weight distribution are subjected to smooth transition processing to obtain sensor weight coefficients within different magnetic field strength ranges.

6. The SiC power semiconductor ultra-high voltage current detection method according to claim 5, characterized in that: The calculating the sensor saturation of the inner ring Hall sensor array under the corresponding magnetic field strength according to the inner ring sensor output signal and the outer ring sensor output signal comprises: Performing magnetic field intensity conversion processing on the output signal of the outer ring sensor to obtain the magnetic field intensity value of the current detection point; Determine a theoretical output signal value according to a corresponding position of the magnetic field intensity value of the current detection point in the sensor saturation characteristic mapping relationship; Calculating the deviation between the output signal of the inner loop sensor and the theoretical output signal value to obtain an actual deviation value; According to the ratio of the actual deviation value to the preset standard deviation, the sensor saturation of the inner ring Hall sensor array under the corresponding magnetic field strength is obtained.

7. The SiC power semiconductor ultra-high voltage current detection method according to claim 1, characterized in that: The step of performing segmented reconstruction processing on the current detection result of the SiC power semiconductor according to the sensor weight coefficient, the inner loop sensor output signal and the outer loop sensor output signal to obtain a complete current measurement value includes: Divide the magnetic field strength into a first section below a first magnetic field strength threshold, a second section from the first magnetic field strength threshold to a second magnetic field strength threshold, and a third section above the second magnetic field strength threshold; The inner loop sensor output signal is used to calculate the current value of the first section, the weighted combination value of the inner loop sensor output signal and the outer loop sensor output signal is used to calculate the current value of the second section according to the sensor weight coefficient, and the outer loop sensor output signal is used to calculate the current value of the third section to obtain segment current detection data; Performing continuity analysis on current detection data at critical points between the first section and the second section, and between the second section and the third section, to obtain critical point transition parameters; The segmented current detection data is subjected to smooth transition processing according to the critical point transition parameter to obtain a complete current measurement value.

8. A SiC power semiconductor ultra-high voltage current detection system, characterized in that: The SiC power semiconductor ultra-high voltage current detection system comprises: The differential acquisition module is used to perform differential sampling processing on the magnetic field signals collected by the inner ring Hall sensor array and the outer ring Hall sensor array set within the preset range of the SiC power semiconductor to obtain the inner ring sensor output signal, the outer ring sensor output signal and the ratio of the inner and outer ring sensor output signals; A feature modeling module is used to perform nonlinear characteristic curve modeling processing on the output signal of the inner ring sensor according to the output signal of the outer ring sensor and the ratio of the output signals of the inner and outer ring sensors, so as to obtain a sensor saturation characteristic mapping relationship under different magnetic field strengths; A weight distribution module is used to dynamically distribute the weights of the inner and outer ring sensor signal ratios according to the sensor saturation characteristic mapping relationship, the inner ring sensor output signal, and the outer ring sensor output signal to obtain sensor weight coefficients within different magnetic field strength ranges; The data reconstruction module is used to perform segmented reconstruction processing on the current detection result of the SiC power semiconductor according to the sensor weight coefficient, the inner loop sensor output signal and the outer loop sensor output signal to obtain a complete current measurement value.

Images