High-voltage pulse voltage measurement method and system based on a capacitive voltage divider

Abstract

The application discloses a high-voltage pulse voltage measurement method and system based on a capacitor voltage divider, and belongs to the technical field of high-voltage pulse voltage measurement. The method specifically comprises the following steps: obtaining an output voltage division signal of a low-voltage arm of the capacitor voltage divider under the action of a to-be-measured high-voltage pulse, and synchronously collecting a dielectric loss angle change amount of a high-voltage arm of the capacitor voltage divider to form an actual voltage division ratio drift sequence; performing inversion compensation processing on the output voltage division signal according to the actual voltage division ratio drift sequence to generate a preliminary recovered voltage waveform; extracting an oscillation attenuation feature from the preliminary recovered voltage waveform, and constructing a high-frequency attenuation correction factor according to the oscillation attenuation feature; and performing time-domain reconstruction on the preliminary recovered voltage waveform based on the high-frequency attenuation correction factor to obtain a target high-voltage pulse voltage waveform corresponding to the to-be-measured high-voltage pulse. The application eliminates the voltage division ratio drift error through inversion compensation, offsets high-frequency attenuation and retains real high-frequency components through time-domain reconstruction, and improves the restoration accuracy of the high-voltage pulse voltage waveform.

Description

Technical Field

[0001] This invention relates to the field of high-voltage pulse voltage measurement technology, and in particular to a high-voltage pulse voltage measurement method and system based on a capacitive voltage divider. Background Technology

[0002] When measuring high-voltage pulse voltage using a capacitive voltage divider, the output signal of the low-voltage arm is typically multiplied by a nominal fixed voltage division ratio to estimate the high-voltage side voltage waveform. This estimation relies on the assumption that the capacitance ratio between the high-voltage and low-voltage arms of the divider remains constant during measurement. However, under actual high-voltage pulse conditions, the dielectric loss angle of the high-voltage arm capacitor dynamically changes with factors such as electric field strength and temperature, causing the actual voltage division ratio to drift with the pulse duration and no longer maintain a constant value. If a fixed voltage division ratio is continued for calculation, time-varying errors will be introduced into the voltage waveform's amplitude and rise time, reducing the accuracy of the measurement results. While some solutions attempt to correct the voltage division ratio through post-calibration or temperature compensation, these optimization methods cannot reflect the rapid real-time changes in the voltage division ratio during a single pulse, and significant dynamic measurement biases still exist.

[0003] On the other hand, due to the stray inductance and distributed capacitance inherent in capacitive voltage dividers, their high-frequency response is insufficient when measuring steep-leading high-voltage pulses. The output waveform often exhibits high-frequency decaying oscillations caused by parasitic parameters. These oscillations manifest as multiple reciprocating fluctuations after the voltage crosses zero, with the amplitude gradually decreasing, causing the measured waveform to deviate significantly from the true pulse shape at the trailing edge. Conventional processing methods often employ low-pass filtering or smoothing algorithms to suppress oscillations. However, while these suppression methods filter out the oscillation components, they also weaken the high-frequency components of the true signal, resulting in a slower pulse leading edge, lower peak value, and an inability to recover the fine characteristics of the original pulse.

[0004] To obtain a voltage waveform that truly reflects the full picture of a high-voltage pulse, two measurement challenges need to be addressed simultaneously: first, how to compensate for the real-time drift of the voltage divider ratio caused by the change in dielectric loss angle during the pulse, thereby reducing the resulting dynamic amplitude error and time axis distortion; and second, how to utilize the zero-crossing oscillation characteristics inherent in the waveform itself to specifically correct the high-frequency attenuation effect of the voltage divider without damaging the true high-frequency signal components. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing measurement methods, which make it difficult to accurately reconstruct the high-voltage pulse voltage waveform from the low-voltage arm output signal due to the real-time drift of the voltage divider ratio and the influence of high-frequency attenuation oscillations. This invention provides a high-voltage pulse voltage measurement method and system based on a capacitive voltage divider. By synchronously acquiring the change in the dielectric loss angle of the high-voltage arm to form an actual voltage divider ratio drift sequence, the output voltage divider signal is inverted and compensated time-by-time to reduce the dynamic amplitude error and time axis distortion caused by the voltage divider ratio drift during the pulse. Simultaneously, by extracting oscillation attenuation characteristics from the initially recovered voltage waveform to construct a high-frequency attenuation correction factor, and performing time-domain reconstruction based on this correction factor, the high-frequency attenuation effect of the voltage divider is offset while fully preserving the true high-frequency signal components, thus improving the accuracy of the reconstructed high-voltage pulse voltage waveform.

[0006] The objective of this invention is achieved through the following technical solution: High-voltage pulse voltage measurement methods based on capacitive voltage dividers include: The output voltage division signal of the low-voltage arm of the capacitor voltage divider under the action of the high-voltage pulse under test is obtained, and the change of dielectric loss angle of the high-voltage arm of the capacitor voltage divider is collected simultaneously to form the actual voltage division ratio drift sequence. The output voltage divider signal is inverted and compensated based on the actual voltage divider ratio drift sequence to generate a preliminary recovered voltage waveform. Oscillation attenuation characteristics are extracted from the initial recovered voltage waveform, and a high-frequency attenuation correction factor is constructed based on these characteristics. The initial recovered voltage waveform is reconstructed in the time domain based on the high-frequency attenuation correction factor to obtain the target high-voltage pulse voltage waveform corresponding to the high-voltage pulse under test.

[0007] Furthermore, the synchronous acquisition of the dielectric loss angle change of the high-voltage arm of the capacitor divider to form the actual voltage division ratio drift sequence includes: During the period of the high voltage pulse under test, the change in dielectric loss angle of the high voltage arm of the capacitor voltage divider is monitored in real time at a preset sampling frequency. Based on the corresponding change in dielectric loss angle, the real-time voltage division ratio correction coefficient for each sampling moment is calculated in conjunction with the rated voltage division ratio of the capacitive voltage divider. The real-time voltage division ratio correction coefficients at each sampling time are arranged in chronological order of sampling time to form the actual voltage division ratio drift sequence.

[0008] Furthermore, the step of performing inversion compensation processing on the output voltage divider signal based on the actual voltage divider ratio drift sequence to generate a preliminary recovered voltage waveform includes: Establish a mapping relationship between each real-time voltage division ratio correction coefficient in the actual voltage division ratio drift sequence and the sampling point in the output voltage division signal at the same sampling time; Based on the mapping relationship, amplitude compensation calculations are performed on each sampling point of the output voltage divider signal using the corresponding voltage divider ratio correction coefficient to obtain a discrete compensation amplitude sequence. Interpolation is performed on the discrete compensation amplitude sequence to obtain the preliminary recovery voltage waveform in the continuous time domain.

[0009] Furthermore, the step of performing inversion compensation processing on the output voltage divider signal based on the actual voltage divider ratio drift sequence to generate a preliminary recovered voltage waveform includes: The output voltage divider signal is divided into several voltage divider signal segments according to the sampling time. Each voltage divider signal segment is centered on a single sampling time and contains a main sampling point and several adjacent sampling points before and after it. Match the real-time voltage division ratio correction coefficient corresponding to the center sampling time of each voltage division signal segment from the actual voltage division ratio drift sequence, and align the real-time voltage division ratio correction coefficient with the voltage division signal segment in time. Based on the corresponding real-time voltage division ratio correction coefficient after timing alignment, each voltage division signal segment is subjected to amplitude scaling and time axis stretching to obtain the compensated waveform segment. The compensated waveform segments are arranged sequentially according to the sampling time, and adjacent compensated waveform segments are merged by weighted averaging of overlapping regions to obtain the preliminary recovered voltage waveform in the continuous time domain.

[0010] Furthermore, the step of extracting oscillation attenuation features from the initial recovered voltage waveform and constructing a high-frequency attenuation correction factor based on the oscillation attenuation features includes: Zero-crossing detection is performed on the preliminary recovered voltage waveform, and all zero-crossing points in the preliminary recovered voltage waveform where the voltage amplitude crosses the zero level are extracted; The length of the corresponding oscillation period is determined based on the time interval between each adjacent zero crossing point, and the main oscillation frequency value is determined based on the length of each oscillation period. Extract the apparent amplitude of each oscillation period and calculate the ratio of the attenuation amplitude between adjacent oscillation periods; The high-frequency attenuation correction factor is calculated based on the length of each oscillation period, the main oscillation frequency, and the ratio of attenuation amplitude between adjacent oscillation periods.

[0011] Furthermore, the step of extracting oscillation attenuation features from the initial recovered voltage waveform and constructing a high-frequency attenuation correction factor based on the oscillation attenuation features includes: Extract the trailing edge oscillation interval from the initial recovered voltage waveform, where the voltage amplitude gradually decays to zero after the voltage peak. Multiple consecutive oscillation peak amplitudes and corresponding oscillation trough amplitudes are extracted within the trailing edge oscillation interval, and adjacent oscillation peaks and troughs are combined into an oscillation period. Calculate the attenuation rate for each oscillation cycle, and use the arithmetic mean of the attenuation rates for all oscillation cycles as the high-frequency attenuation correction factor.

[0012] Furthermore, the step of extracting oscillation attenuation features from the initial recovered voltage waveform and constructing a high-frequency attenuation correction factor based on the oscillation attenuation features includes: Local maxima detection is performed on the preliminary recovered voltage waveform, and the sampling points that meet the local maxima conditions are determined as the peak points of the oscillation envelope; Calculate the amplitude attenuation difference between two adjacent oscillation envelope peaks, and combine the time interval between two adjacent oscillation envelope peaks and the amplitude of the previous oscillation envelope peak to generate an attenuation rate descriptor for each oscillation envelope peak. The high-frequency attenuation correction factor is calculated based on the attenuation rate descriptor corresponding to the peak point of each oscillation envelope.

[0013] Furthermore, the step of reconstructing the initially recovered voltage waveform in the time domain based on the high-frequency attenuation correction factor to obtain the target high-voltage pulse voltage waveform corresponding to the high-voltage pulse under test includes: Multiply the amplitude of each sampling point in the initially recovered voltage waveform by the high-frequency attenuation correction factor to obtain the amplitude-corrected voltage sequence; The voltage sequence after amplitude correction is subjected to waveform correction processing, and the voltage sequence after waveform correction processing is subjected to difference compensation processing to obtain the target high-voltage pulse voltage waveform.

[0014] Furthermore, the step of reconstructing the initially recovered voltage waveform in the time domain based on the high-frequency attenuation correction factor to obtain the target high-voltage pulse voltage waveform corresponding to the high-voltage pulse under test includes: The high-frequency attenuation correction factor is decomposed into an amplitude attenuation component and a frequency offset component; The initial recovered voltage waveform is adjusted exponentially based on the amplitude attenuation component to obtain the amplitude-compensated waveform; The amplitude compensation waveform is frequency shifted based on the frequency offset component to obtain the frequency shift corrected waveform. The frequency shift correction waveform is subjected to zero-phase filtering to obtain the target high-voltage pulse voltage waveform.

[0015] A high-voltage pulse voltage measurement system based on a capacitive voltage divider, used to perform any of the above-mentioned high-voltage pulse voltage measurement methods, includes: The signal acquisition module is used to acquire the output voltage division signal of the low-voltage arm of the capacitor voltage divider under the action of the high-voltage pulse to be tested, and simultaneously acquire the change in dielectric loss angle of the high-voltage arm of the capacitor voltage divider. The drift sequence generation module is used to generate an actual voltage ratio drift sequence based on the change in dielectric loss angle. The inversion compensation module is used to perform inversion compensation processing on the output voltage divider signal according to the actual voltage divider ratio drift sequence, and generate a preliminary recovered voltage waveform. The correction factor construction module is used to extract oscillation attenuation characteristics from the initial recovered voltage waveform and construct a high-frequency attenuation correction factor based on the oscillation attenuation characteristics. The time-domain reconstruction module is used to reconstruct the initial recovered voltage waveform in the time domain based on the high-frequency attenuation correction factor, and obtain the target high-voltage pulse voltage waveform corresponding to the high-voltage pulse under test.

[0016] The beneficial effects of this invention are: By synchronously acquiring the change in dielectric loss angle of the high-voltage arm during pulse action, a time-varying actual voltage division ratio drift sequence is formed to reflect the rapid real-time change of the voltage division ratio during a single pulse. Inversion compensation processing is then performed based on the actual voltage division ratio drift sequence, using the actual voltage division ratio at each sampling moment for correction. This reduces the time-varying voltage division error introduced by the dynamic changes in dielectric loss angle with factors such as electric field strength and temperature, making the recovered preliminary voltage waveform closer to the actual situation on the high-voltage side in terms of amplitude ratio and pulse rise time parameters. Furthermore, oscillation attenuation characteristics that characterize the actual attenuation effect of the capacitive voltage divider's parasitic parameters on high-frequency components are directly extracted from the preliminary recovered voltage waveform to construct a high-frequency attenuation correction factor. This correction factor is then used to reconstruct the preliminary recovered voltage waveform in the time domain, specifically offsetting the amplitude drop and phase distortion of the voltage divider in the high-frequency band. This accurately restores the steep leading edge, overshoot peak, and trailing edge attenuation process of the pulse, fully preserving the true pulse details and high-frequency characteristics, and improving the accuracy of the reconstructed high-voltage pulse waveform. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of a process of the present invention; Figure 2 This is a schematic diagram of a high-voltage pulse voltage waveform recovery and correction process shown in Embodiment 5 of the present invention. Detailed Implementation

[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0019] Example 1: Refer to Figure 1 This embodiment provides a high-voltage pulse voltage measurement method based on a capacitive voltage divider, including: The output voltage division signal of the low-voltage arm of the capacitor voltage divider under the action of the high-voltage pulse under test is obtained, and the change of dielectric loss angle of the high-voltage arm of the capacitor voltage divider is collected simultaneously to form the actual voltage division ratio drift sequence. The output voltage divider signal is inverted and compensated based on the actual voltage divider ratio drift sequence to generate a preliminary recovered voltage waveform. Oscillation attenuation characteristics are extracted from the initial recovered voltage waveform, and a high-frequency attenuation correction factor is constructed based on these characteristics. The initial recovered voltage waveform is reconstructed in the time domain based on the high-frequency attenuation correction factor to obtain the target high-voltage pulse voltage waveform corresponding to the high-voltage pulse under test.

[0020] When acquiring high-voltage pulse voltage using a capacitive voltage divider, the output voltage division signal of the low-voltage arm of the capacitive voltage divider under the action of the high-voltage pulse to be measured is first acquired. Specifically, a sampling resistor is first connected in series between the high-voltage arm and the low-voltage arm of the capacitive voltage divider. The resistance value of the sampling resistor is selected such that the voltage drop across the sampling resistor is much smaller than the voltage across the low-voltage arm, thereby avoiding affecting the inherent voltage division characteristics of the capacitive voltage divider. Then, the voltage difference across the sampling resistor is acquired using a differential probe or differential amplifier, and this voltage difference is used as the output voltage division signal. Based on this, during the action of the high-voltage pulse to be measured, the output voltage value of the low-voltage arm of the capacitive voltage divider is continuously recorded at a fixed sampling frequency. The setting of the fixed sampling frequency satisfies the Nyquist sampling theorem, and the fixed sampling frequency is greater than twice the highest frequency component of the high-voltage pulse to be measured.

[0021] Considering that during the application of the high-voltage pulse, the dielectric loss angle of the high-voltage arm capacitor will change dynamically with factors such as electric field strength and temperature, the actual voltage division ratio will drift with the pulse application time. Moreover, this drift occurs in real time and changes every moment, and is not a fixed deviation that can be obtained by simple post-pulse measurement after the pulse ends.

[0022] Therefore, while continuously acquiring the output voltage value of the low-voltage arm of the capacitor voltage divider, the dielectric loss angle change of the high-voltage arm is monitored in real time by a dielectric loss test device connected to the high-voltage arm. The dynamic behavior of the voltage division ratio during the pulse action is quantified by the change in dielectric loss angle. Then, based on the change in dielectric loss angle, a time-by-time voltage division ratio correction coefficient is formed that is precisely aligned with the timing of the output voltage division signal. The actual voltage division ratio drift sequence after timing alignment is used to perform point-by-point inversion compensation on the output voltage division signal, reducing the dynamic amplitude error and time axis distortion caused by the real-time drift of the voltage division ratio.

[0023] The synchronous acquisition of the dielectric loss angle change of the high-voltage arm of the capacitor voltage divider to form the actual voltage division ratio drift sequence includes: During the period of the high voltage pulse under test, the change in dielectric loss angle of the high voltage arm of the capacitor voltage divider is monitored in real time at a preset sampling frequency. Based on the corresponding change in dielectric loss angle, the real-time voltage division ratio correction coefficient for each sampling moment is calculated in conjunction with the rated voltage division ratio of the capacitive voltage divider. The real-time voltage division ratio correction coefficients at each sampling time are arranged in chronological order of sampling time to form the actual voltage division ratio drift sequence.

[0024] The change in dielectric loss angle is obtained by real-time measurement of the phase difference between the current flowing through the high-voltage arm capacitor and the voltage across the high-voltage arm. The deviation of this phase difference from the initial phase difference calibrated under power frequency reference conditions is the change in dielectric loss angle. This change in dielectric loss angle reflects the change in the equivalent series resistance of the high-voltage arm capacitor caused by dielectric relaxation polarization under a pulsed electric field. The power frequency reference conditions refer to the initial calibration state of the capacitor divider under standard power frequency voltage and at room temperature.

[0025] The real-time voltage division ratio correction coefficient for each sampling moment is calculated based on the change in dielectric loss angle. In this embodiment, the real-time voltage division ratio correction coefficient is calculated by multiplying the rated voltage division ratio of the capacitive voltage divider by a correction function associated with the change in dielectric loss angle. The correction function can be expressed as follows: ,in, This represents the change in the dielectric loss angle at sampling time t. This change in the dielectric loss angle is the difference between the current dielectric loss angle and the initial dielectric loss angle under the power frequency reference condition.

[0026] Based on the change in dielectric loss angle at each sampling time, the corresponding real-time voltage division ratio correction coefficient is calculated according to the rated voltage division ratio of the capacitive voltage divider using a correction function. The real-time voltage division ratio correction coefficients calculated at each sampling time are arranged in chronological order to form an actual voltage division ratio drift sequence. This actual voltage division ratio drift sequence describes the degree of deviation of the actual voltage division ratio from the rated voltage division ratio at each moment during the entire high-voltage pulse process.

[0027] Since the output voltage divider signal reflects the voltage across the low-voltage arm, its relationship with the actual voltage on the high-voltage side is determined by the current voltage division ratio. However, when the voltage division ratio drifts with changes in the dielectric loss angle, the relationship between the output voltage divider signal and the high-voltage side voltage is no longer a simple fixed ratio, but rather a time-varying ratio. If this output voltage divider signal is directly multiplied by a fixed voltage division ratio to calculate the high-voltage side voltage, it's equivalent to using a constant ratio to calculate the output of a time-varying proportional system, and the result will inevitably deviate from the true value.

[0028] Therefore, after forming the actual voltage division ratio drift sequence, the output voltage division signal is compensated for amplitude and corrected for time axis by point-by-point or segment-by-segment method using the actual voltage division ratio drift sequence. This reduces the amplitude error and rise time distortion introduced by the real-time voltage division ratio drift, and obtains the true low-voltage side voltage waveform, that is, the waveform after eliminating the influence of voltage division ratio drift. This waveform is then used as the initial recovery voltage waveform.

[0029] The step of performing inversion compensation processing on the output voltage divider signal based on the actual voltage divider ratio drift sequence to generate a preliminary recovered voltage waveform includes: The output voltage divider signal is divided into several voltage divider signal segments according to the sampling time. Each voltage divider signal segment is centered on a single sampling time and contains a main sampling point and several adjacent sampling points before and after it. Match the real-time voltage division ratio correction coefficient corresponding to the center sampling time of each voltage division signal segment from the actual voltage division ratio drift sequence, and align the real-time voltage division ratio correction coefficient with the voltage division signal segment in time. Based on the corresponding real-time voltage division ratio correction coefficient after timing alignment, each voltage division signal segment is subjected to amplitude scaling and time axis stretching to obtain the compensated waveform segment. The compensated waveform segments are arranged sequentially according to the sampling time, and adjacent compensated waveform segments are merged by weighted averaging of overlapping regions to obtain the preliminary recovered voltage waveform in the continuous time domain.

[0030] When dividing the voltage divider signal into segments, the segment length can be selected based on the time resolution of the actual voltage divider drift sequence and the sampling rate of the output voltage divider signal, and can be set to cover the time scale of voltage divider changes.

[0031] After dividing the output voltage divider signal into multiple voltage divider signal segments according to the sampling time, the peak amplitude and rise time are extracted for each voltage divider signal segment. The peak amplitude is taken from the maximum absolute value of the voltage within the voltage divider signal segment, and the rise time is the time length from the moment when the voltage reaches 10% of the peak amplitude to the moment when the voltage reaches 90% of the peak amplitude.

[0032] The real-time voltage division ratio correction coefficients corresponding to the center sampling time of each voltage division signal segment are found from the actual voltage division ratio drift sequence. These real-time correction coefficients are then aligned one-to-one with the voltage division signal segments in terms of time index, ensuring that the timestamps of the real-time correction coefficients coincide with the center time of each segment. If the time resolution of the actual voltage division ratio drift sequence is inconsistent with the sampling rate of the output voltage division signal, the correction coefficients need to be resampled using interpolation to obtain correction coefficients that are accurately aligned with the center time of the segments.

[0033] Each voltage divider signal segment is subjected to amplitude scaling and time axis stretching using a real-time voltage divider ratio correction factor after timing alignment. The amplitude scaling process involves dividing the voltage values ​​of all sampling points within the voltage divider signal segment by the real-time voltage divider ratio correction factor to obtain the compensated peak amplitude and waveform shape. The time axis stretching process involves multiplying the time coordinates corresponding to each sampling point within the voltage divider signal segment by the real-time voltage divider ratio correction factor, thus proportionally expanding the time intervals within the voltage divider signal segment.

[0034] The voltage divider signal segments that have undergone amplitude scaling and time axis stretching are used as compensated waveform segments. All compensated waveform segments are arranged sequentially according to the sampling time. Adjacent compensated waveform segments are merged in the overlapping area by weighted averaging. The weighting coefficient changes linearly according to the distance from the center of the segment, that is, the closer to the center, the greater the weight, and the farther away, the smaller the weight. After splicing and merging, a preliminary recovery voltage waveform that is continuous in time is generated.

[0035] By using segments as processing units and merging multiple sampling points, it can balance high compensation accuracy and computational efficiency, making it suitable for measurement scenarios where hardware acquisition rates are limited or voltage divider ratio drift changes relatively smoothly.

[0036] Due to the inherent stray inductance and distributed capacitance of the capacitive voltage divider, although the initial recovered voltage waveform obtained under steep-front pulse excitation has eliminated the amplitude and time axis deviations caused by the voltage division ratio drift, its trailing edge still contains high-frequency decaying oscillations caused by parasitic parameters. These oscillations manifest as multiple reciprocating fluctuations after the voltage crosses zero, with a gradually decreasing amplitude, causing the trailing edge waveform to deviate from the true pulse shape. Therefore, characteristic parameters that can describe this oscillation decay law are extracted from the initial recovered voltage waveform to construct a high-frequency attenuation correction factor that characterizes the actual attenuation effect of the capacitive voltage divider's parasitic parameters on high-frequency components. This allows for directional correction of the initial recovered voltage waveform, ensuring its accuracy.

[0037] The step of extracting oscillation attenuation features from the initial recovered voltage waveform and constructing a high-frequency attenuation correction factor based on the oscillation attenuation features includes: Zero-crossing detection is performed on the preliminary recovered voltage waveform, and all zero-crossing points in the preliminary recovered voltage waveform where the voltage amplitude crosses the zero level are extracted; The length of the corresponding oscillation period is determined based on the time interval between each adjacent zero crossing point, and the main oscillation frequency value is determined based on the length of each oscillation period. Extract the apparent amplitude of each oscillation period and calculate the ratio of the attenuation amplitude between adjacent oscillation periods; The high-frequency attenuation correction factor is calculated based on the length of each oscillation period, the main oscillation frequency, and the ratio of attenuation amplitude between adjacent oscillation periods.

[0038] Zero-crossing detection is performed on the initially recovered voltage waveform to extract all zero-crossing points when the voltage amplitude crosses from a positive value to zero level and from a negative value to zero level. The position of the zero-crossing point is determined by linear interpolation between adjacent sampling points. Since the signal change between two adjacent sampling points can be approximated as linear, linear interpolation can effectively ensure the accuracy of the zero-crossing point position.

[0039] Calculate the time interval between two adjacent zero-crossing points with the same direction of symbol change. The same direction of symbol change means that both cross from positive to negative level (falling edge crossing zero) or both cross from negative to positive level (rising edge crossing zero). The time interval between two zero-crossing points in the same direction is taken as the length of an oscillation period.

[0040] By statistically analyzing the lengths of all oscillation periods, and taking the reciprocal of each period, a set of instantaneous frequency values ​​is obtained. This set of instantaneous frequency values ​​collectively constitutes the high-frequency oscillation frequency distribution of the preliminary recovered voltage waveform. From this high-frequency oscillation frequency distribution, the instantaneous frequency value with the largest amplitude spectral density is determined as the dominant oscillation frequency value. This dominant oscillation frequency value represents the most important and energy-concentrated frequency component in the high-frequency damped oscillation, typically corresponding to the resonant frequency dominated by the parasitic parameters of the capacitive voltage divider.

[0041] In the initial recovery voltage waveform, half of the sum of the absolute values ​​of the positive peak and the negative peak in each oscillation cycle is extracted as the apparent amplitude of that oscillation cycle. Since the positive peak and the negative peak are often not completely symmetrical in a single oscillation cycle, the average of the two can effectively offset the apparent amplitude deviation introduced by the waveform asymmetry.

[0042] The ratio of the apparent amplitude between two adjacent oscillation periods is calculated to obtain the attenuation amplitude ratio. The attenuation amplitude ratio reflects the degree of attenuation of the oscillation amplitude between adjacent periods. The closer the ratio is to 1, the slower the attenuation. The closer the ratio is to 0, the more severe the attenuation.

[0043] The high-frequency attenuation correction factor is calculated based on the length of each oscillation period, the main oscillation frequency, and the ratio of attenuation amplitude between adjacent oscillation periods. Its expression is as follows: ; in, For high-frequency attenuation correction factor, The number of complete oscillation periods extracted. For the first The apparent amplitude of each oscillation period, For the first The apparent amplitude of each oscillation period, For the first The length of each oscillation period The main oscillation frequency value. The number of oscillation periods. The value of is determined by the number of zero-crossing points detected in the initial recovered voltage waveform, and At least 3. Apparent amplitude and The main oscillation frequency value is obtained by searching for the extreme points of the initially recovered voltage waveform data within the corresponding oscillation period. It is the frequency value corresponding to the maximum amplitude density selected from the frequency distribution of high-frequency oscillation, and its value ranges from 0.1 times to 1 times the equivalent bandwidth of the rising edge of the high-voltage pulse under test.

[0044] The calculated high-frequency attenuation correction factor describes the attenuation law of the oscillation amplitude over time in the entire trailing edge oscillation interval, as well as the quantitative relationship between the attenuation and the oscillation frequency. Since the high-frequency attenuation correction factor originates from the high-frequency response characteristics of the initial recovery voltage waveform itself, it can characterize the actual attenuation effect of the parasitic parameters of the capacitor divider on the high-frequency components and can be directly used for directional correction of the initial recovery voltage waveform.

[0045] Since the high-frequency attenuation correction factor is a scalar, it is a set of parameters obtained by offline calculation rather than a waveform that can be directly output. Therefore, the amplitude compensation and phase correction are applied to the preliminary recovery voltage waveform according to the high-frequency attenuation correction factor to offset the high-frequency amplitude drop and phase distortion caused by stray inductance and distributed capacitance of the capacitor divider, so as to realize the time-domain reconstruction of the preliminary recovery voltage waveform and restore the high-frequency components of the preliminary recovery voltage waveform to a state close to the true value.

[0046] The step of reconstructing the initially recovered voltage waveform in the time domain based on the high-frequency attenuation correction factor to obtain the target high-voltage pulse voltage waveform corresponding to the high-voltage pulse under test includes: Multiply the amplitude of each sampling point in the initially recovered voltage waveform by the high-frequency attenuation correction factor to obtain the amplitude-corrected voltage sequence; The voltage sequence after amplitude correction is subjected to waveform correction processing, and the voltage sequence after waveform correction processing is subjected to difference compensation processing to obtain the target high-voltage pulse voltage waveform.

[0047] Since the high-frequency attenuation effect of the voltage divider can be approximated as a uniform amplitude attenuation ratio across the entire frequency range, when the high-frequency attenuation correction factor is a scalar value, it can uniformly affect all sampling points in the preliminary recovered voltage waveform. Based on this, the amplitude of each sampling point in the preliminary recovered voltage waveform is multiplied by the high-frequency attenuation correction factor to obtain the amplitude-corrected voltage sequence.

[0048] The voltage sequence after amplitude correction is then subjected to waveform correction processing, which includes phase compensation and amplitude difference suppression processing.

[0049] First, phase compensation processing is performed on the amplitude-corrected voltage sequence. The phase offset to be adjusted at each sampling point is calculated based on the main oscillation frequency, and its expression is as follows: ; in, For the first The phase offset applied at each sampling point, The main oscillation frequency value, The sampling time interval is the reciprocal of the fixed sampling frequency.

[0050] The amplitude-corrected voltage sequence is then subjected to a phase delay or phase lead corresponding to the phase offset of each sampling point by a digital all-pass filter, resulting in a phase-corrected voltage sequence. It should be noted that the amplitude-frequency response of the digital all-pass filter is constant; it does not change the magnitude of the signal, but only corrects the phase characteristics of the signal.

[0051] Finally, amplitude difference suppression processing is performed on the phase-corrected voltage sequence. For each current sampling point in the phase-corrected voltage sequence, the amplitude of the current sampling point and the amplitude of the previous sampling point are obtained, the difference between the amplitude of the current sampling point and the amplitude of the previous sampling point is calculated, and half of this difference is subtracted from the amplitude of the current sampling point. ,in, The first voltage in the phase-corrected voltage sequence The amplitude of each sampling point The first voltage in the phase-corrected voltage sequence The amplitude of -1 sampling point, The first voltage sequence after time-domain reconstruction The amplitude of each sampling point. Those skilled in the art should understand that the difference suppression processing is equivalent to applying first-order differential feedback to the signal, and its function is to suppress local jumps between sampling points while maintaining the overall contour of the signal.

[0052] The result after subtraction is used as the amplitude of the corresponding sampling point in the reconstructed voltage sequence in the time domain, and finally the target high-voltage pulse voltage waveform is obtained.

[0053] This embodiment generates a time-by-time actual voltage division ratio drift sequence by synchronously acquiring the change in the dielectric loss angle of the high-voltage arm. This sequence is then used to invert and compensate the output voltage division signal, overcoming the shortcomings of existing fixed voltage division ratio conversion or post-calibration methods that cannot reflect the rapid real-time changes in the voltage division ratio during a single pulse. Furthermore, by stretching the voltage division signal segments along the time axis, the distortion effect of the voltage division ratio drift on the pulse rise time is synchronously corrected, achieving accurate restoration of both amplitude ratio and time parameters. Then, through zero-crossing detection and apparent amplitude analysis, oscillation attenuation characteristics are extracted from the initially recovered voltage waveform to construct a high-frequency attenuation correction factor. This factor characterizes the actual attenuation effect of the capacitive voltage divider's parasitic parameters on high-frequency components, overcoming the shortcomings of conventional low-pass filtering or smoothing algorithms that indiscriminately suppress true high-frequency components, thus accurately restoring the steep leading edge, overshoot peak, and trailing edge attenuation process of the pulse. Simultaneously, by using segments as processing units, high compensation accuracy and computational efficiency are balanced. The sampling frequency is set based on the Nyquist sampling theorem, reducing the requirements for acquisition hardware and storage capacity while ensuring signal non-aliasing.

[0054] Example 2: This example provides a high-voltage pulse voltage measurement method based on a capacitive voltage divider. The main difference from Example 1 is that the inversion compensation processing step uses point-by-point mapping and interpolation instead of the segment segmentation and weighted fusion method in Example 1. This avoids compensation errors that may be introduced due to the sharing of the same correction coefficient within segments, and also avoids abrupt changes in the reconstructed waveform. This can improve the compensation accuracy while simplifying the processing logic, and is applicable to measurement scenarios with fast voltage divider drift rate and extremely steep pulse leading edge.

[0055] In practical implementation, a sampling resistor is first connected in series between the high-voltage and low-voltage arms of the capacitor voltage divider. A differential voltage probe connected to both ends of the sampling resistor acquires the differential voltage signal. This differential voltage signal is amplified and filtered by a signal conditioning circuit and recorded as the output voltage divider signal. In this embodiment, the resistance value of the sampling resistor is selected to be less than one-thousandth of the equivalent impedance of the low-voltage arm, so that the connection of the sampling resistor does not change the inherent low-frequency voltage division ratio of the capacitor voltage divider. During the application of the high-voltage pulse under test, the output voltage value of the low-voltage arm of the capacitor voltage divider is continuously acquired at a fixed sampling frequency. The value of the fixed sampling frequency is determined based on the steepness of the rising edge of the high-voltage pulse under test, and the set value of the fixed sampling frequency is greater than five times the equivalent highest frequency component of the rising edge of the high-voltage pulse under test. By setting a higher sampling frequency, the fine structure of the pulse rising edge is captured, reducing the loss of rising edge details due to insufficient sampling points.

[0056] Furthermore, the change in dielectric loss angle of the high-voltage arm is monitored synchronously. The change in dielectric loss angle is obtained by measuring the phase difference between the current flowing through the capacitor of the high-voltage arm and the voltage across the high-voltage arm in real time. The offset of the phase difference relative to the initial phase difference calibrated under the power frequency reference condition is the change in dielectric loss angle.

[0057] At each sampling moment, a real-time voltage division ratio correction coefficient is calculated based on the change in the dielectric loss angle. The expression for calculating the real-time voltage division ratio correction coefficient in this embodiment is as follows: ; in, This is the real-time voltage division ratio correction factor at sampling time t. This represents the rated voltage division ratio of the capacitor voltage divider under standard power frequency conditions. To be at the sampling time The change in the tangent of the dielectric loss angle. This is the initial dielectric loss tangent measured under standard power frequency conditions. Rated voltage divider ratio. The initial dielectric loss tangent is determined based on the ratio of the high-voltage arm capacitance to the low-voltage arm capacitance of the capacitor divider. After the capacitor voltage divider is initially installed, the voltage is measured and stored using the power frequency bridge method. The value is the measured tangent of the dielectric loss angle at the current sampling time and... The difference.

[0058] The real-time voltage division ratio correction coefficients calculated at each sampling time are arranged sequentially according to the sampling time order. After arrangement, the actual voltage division ratio drift sequence is formed. Then, the inversion compensation process is performed through point-by-point mapping and interpolation to generate the corresponding preliminary recovery voltage waveform.

[0059] The step of performing inversion compensation processing on the output voltage divider signal based on the actual voltage divider ratio drift sequence to generate a preliminary recovered voltage waveform includes: Establish a mapping relationship between each real-time voltage division ratio correction coefficient in the actual voltage division ratio drift sequence and the sampling point in the output voltage division signal at the same sampling time; Based on the mapping relationship, amplitude compensation calculations are performed on each sampling point of the output voltage divider signal using the corresponding voltage divider ratio correction coefficient to obtain a discrete compensation amplitude sequence. Interpolation is performed on the discrete compensation amplitude sequence to obtain the preliminary recovery voltage waveform in the continuous time domain.

[0060] The mapping relationship is established through timestamp alignment. For the sub-sampling point level time difference between the sampling time and the actual voltage division ratio drift sequence recording time in the output voltage divider signal due to the clock deviation of the acquisition device, the real-time voltage division ratio correction coefficient is resampled by linear interpolation so that the time index of the resampled real-time voltage division ratio correction coefficient is consistent with the sampling time index of the output voltage divider signal.

[0061] After establishing the mapping relationship, an amplitude compensation operation is performed on each sampling point of the output voltage divider signal. The amplitude compensation operation is to divide the voltage value of the sampling point by the real-time voltage divider ratio correction coefficient after resampling that has a mapping relationship with the sampling point, and the result is used as the compensation amplitude of the sampling point.

[0062] After completing the amplitude compensation calculation for all sampling points, a set of discrete compensation amplitude sequences is obtained. Cubic spline interpolation is then performed on the discrete compensation amplitude sequences. The interpolation nodes are selected from all sampling points in the compensation amplitude sequences. The boundary conditions are set as natural boundary conditions with zero second derivatives. The interpolation step size is set to one-fifth of the original sampling time interval. After the interpolation is completed, a function curve of the compensation amplitude as a function of time defined in the continuous time domain is obtained, which is the preliminary recovery of the voltage waveform.

[0063] Using the same method as in Example 1, the high-frequency attenuation correction factor is calculated by performing zero-crossing detection on the preliminary recovered voltage waveform, extracting the oscillation period length and main oscillation frequency value, calculating the apparent amplitude of each oscillation period and the ratio of attenuation amplitude between adjacent oscillation periods.

[0064] However, unlike Embodiment 1, the zero-crossing detection in this embodiment is achieved by finding two consecutive interpolated voltage values ​​with opposite signs, and applying linear interpolation to the adjacent time points to determine the precise time value of the zero-crossing moment when the voltage amplitude crosses the zero level.

[0065] After extracting all zero-crossing points from the preliminary recovered voltage waveform, zero-crossing points with the same sign change direction are selected sequentially. The time difference between two adjacent zero-crossing points with the same sign change direction is calculated, and this time difference is taken as the length of an oscillation period. Then, all extracted oscillation period lengths are analyzed, and the reciprocal of each oscillation period length is calculated to obtain the instantaneous frequency value corresponding to that oscillation period. The instantaneous frequency values ​​corresponding to all oscillation periods are statistically analyzed to obtain the high-frequency oscillation distribution of the preliminary recovered voltage waveform. From this high-frequency oscillation distribution, the frequency value with the highest frequency of instantaneous frequency occurrence is selected as the main oscillation frequency value.

[0066] Simultaneously, in the initial recovery of the voltage waveform, the positive and negative amplitude peak values ​​are extracted for each oscillation cycle. The positive amplitude peak value is the positive maximum of the voltage value within the oscillation cycle, and the negative amplitude peak value is the absolute value of the negative maximum of the voltage value within the oscillation cycle. The average of the positive and negative amplitude peak values ​​is used as the representative amplitude value of the oscillation cycle. The ratio of the representative amplitude values ​​of two adjacent oscillation cycles is calculated, and this ratio is used as the attenuation amplitude ratio between corresponding oscillation cycles.

[0067] A high-frequency attenuation correction factor is then constructed based on the main oscillation frequency value, the length of each oscillation period, and the ratio of attenuation amplitude corresponding to each oscillation period length. The high-frequency attenuation correction factor is constructed by taking the natural logarithm of the ratio of attenuation amplitude corresponding to each oscillation period length, dividing the value after taking the natural logarithm by the product of the oscillation period length and the main oscillation frequency value, and averaging over all oscillation periods.

[0068] After constructing the high-frequency attenuation correction factor, the amplitude of each sampling point in the preliminary recovered voltage waveform is multiplied by the high-frequency attenuation correction factor in the same way as in Example 1 to obtain the amplitude-corrected voltage sequence. Then, the amplitude-corrected voltage sequence is subjected to phase compensation processing and amplitude difference suppression processing in sequence to obtain the target high-voltage pulse voltage waveform.

[0069] In practice, the high-frequency attenuation correction factor is multiplied point-by-point with the amplitude of each sampling point in the preliminary recovered voltage waveform. The result of the multiplication operation constitutes the amplitude-corrected voltage sequence. The amplitude of each sampling point in the amplitude-corrected voltage sequence is equal to the voltage amplitude at the corresponding moment in the preliminary recovered voltage waveform multiplied by the high-frequency attenuation correction factor. Phase compensation processing is then performed on the amplitude-corrected voltage sequence. The phase compensation process involves calculating the phase offset to be adjusted for each sampling point based on the main oscillation frequency. A phase delay corresponding to the phase offset is applied to the amplitude-corrected voltage sequence through an all-pass digital filter, resulting in a phase-corrected voltage sequence. Each sampling point in the phase-corrected voltage sequence is used as the current processing sampling point. The amplitude of the current processing sampling point and the amplitude of the previous sampling point are obtained. The difference between the amplitude of the current processing sampling point and the amplitude of the previous sampling point is calculated. Half of this difference is subtracted from the amplitude of the current processing sampling point, and the result is used as the voltage amplitude at the corresponding moment in the time-domain reconstructed voltage sequence. The above operation is performed sequentially on all sampling points in the phase-corrected voltage sequence to obtain the final target high-voltage pulse voltage waveform.

[0070] Compared to Example 1, this example further refines the compensation granularity by using timestamp alignment and linear interpolation resampling to ensure that each sampling moment has a precisely corresponding correction coefficient. This overcomes the accuracy loss caused by sharing the same correction coefficient within a segment, making it particularly suitable for measurement scenarios with a high voltage divider ratio drift rate. Furthermore, by reconstructing the continuous waveform through cubic spline interpolation, the continuity of the first and second derivatives is guaranteed at the nodes, eliminating the need for weighted fusion to handle segment boundary issues. This simplifies the processing logic while improving compensation accuracy. Simultaneously, the sampling frequency is increased to five times the highest equivalent frequency component of the leading edge to fully capture the fine structure of the pulse leading edge, providing higher temporal resolution raw data for compensation.

[0071] Example 3: This example provides a high-voltage pulse voltage measurement method based on a capacitive voltage divider. The main difference from Example 2 is that a high-frequency attenuation correction factor is constructed using the trailing edge peak-valley attenuation rate method, and waveform correction in time-domain reconstruction is performed through sliding window filtering and difference compensation to reduce the computational load and the demand for data processing and storage resources. While effectively suppressing residual oscillations, it also prevents the main characteristics of the signal from being excessively smoothed. This method is suitable for measurement scenarios with high measurement speed requirements or limited computing resources and relatively regular trailing edge oscillation characteristics.

[0072] In practical implementation, a sampling resistor is connected in series between the high-voltage and low-voltage arms of the capacitor voltage divider. The resistance value of the sampling resistor is chosen to be one-thousandth of the equivalent impedance of the low-voltage arm, so that the voltage drop across the sampling resistor is much smaller than the voltage across the low-voltage arm, thereby reducing the influence of the inherent voltage division characteristics of the capacitor voltage divider. The voltage difference across the sampling resistor is then acquired using a differential voltage probe. The acquired voltage difference signal is amplified and low-pass filtered by a signal conditioning circuit and recorded as the output voltage divider signal. Furthermore, during the application of the high-voltage pulse under test, the output voltage value of the low-voltage arm is continuously sampled and recorded at a fixed sampling frequency. The value of the fixed sampling frequency is determined based on the equivalent bandwidth of the rising edge of the high-voltage pulse under test, and the fixed sampling frequency is set to five times the highest equivalent frequency component of the leading edge of the high-voltage pulse under test.

[0073] In this embodiment, the generation process of the actual voltage division ratio drift sequence is the same as in Embodiment 1. During continuous sampling, a dielectric loss test device connected to the high-voltage arm is used to monitor the change in the dielectric loss angle of the high-voltage arm in real time. This change in the dielectric loss angle is obtained by measuring the phase difference between the current flowing through the capacitor of the high-voltage arm and the voltage across the high-voltage arm, and subtracting the initial phase difference calibrated under power frequency reference conditions. At each sampling moment, a real-time voltage division ratio correction coefficient is calculated based on the change in the dielectric loss angle. This real-time voltage division ratio correction coefficient is obtained by multiplying the rated voltage division ratio of the capacitor voltage divider by a correction function associated with the change in the dielectric loss angle. The correction function is taken as... ,in Sampling time The tangent of the change in the dielectric loss angle. The real-time voltage division ratio correction coefficients calculated at each sampling time are arranged sequentially according to the sampling time to form the actual voltage division ratio drift sequence.

[0074] Using the same method as in Example 2, a mapping relationship is established between each real-time voltage division ratio correction coefficient in the actual voltage division ratio drift sequence and the sampling points in the output voltage division signal at the same sampling time. Amplitude compensation is performed on each sampling point of the output voltage division signal to obtain a discrete compensation amplitude sequence. Then, cubic spline interpolation is performed on the discrete compensation amplitude sequence to obtain a preliminary recovered voltage waveform in the continuous time domain. Based on the preliminary recovered voltage waveform, a high-frequency attenuation correction factor is constructed using the trailing edge peak-trough attenuation rate method.

[0075] Specifically, the step of extracting oscillation attenuation characteristics from the initial recovered voltage waveform and constructing a high-frequency attenuation correction factor based on the oscillation attenuation characteristics includes: Extract the trailing edge oscillation interval from the initial recovered voltage waveform, where the voltage amplitude gradually decays to zero after the voltage peak. Multiple consecutive oscillation peak amplitudes and corresponding oscillation trough amplitudes are extracted within the trailing edge oscillation interval, and adjacent oscillation peaks and troughs are combined into an oscillation period. Calculate the attenuation rate for each oscillation cycle, and use the arithmetic mean of the attenuation rates for all oscillation cycles as the high-frequency attenuation correction factor.

[0076] The method for extracting the trailing-edge oscillation interval is as follows: The moment of the global voltage peak is detected in the initially recovered voltage waveform. This global voltage peak moment is used as the starting point of the trailing-edge oscillation interval. The search continues forward until the voltage amplitude is consistently less than one-hundredth of the global voltage peak for several consecutive oscillation cycles. This moment is then taken as the ending point of the trailing-edge oscillation interval. Within the extracted trailing-edge oscillation interval, multiple consecutive oscillation peak amplitudes and corresponding oscillation trough amplitudes are extracted. The oscillation peak amplitude is obtained by searching for local maxima of the voltage value within the trailing-edge oscillation interval, and the oscillation trough amplitude is obtained by searching for the absolute values ​​of local minima of the voltage value within the trailing-edge oscillation interval. During extraction, each adjacent oscillation peak and trough is combined into one oscillation cycle. The extraction order of the oscillation peaks is sequential, starting from the starting point of the trailing-edge oscillation interval and proceeding forward, with the extraction order of the oscillation troughs corresponding accordingly.

[0077] Based on the extracted peak and trough amplitudes of the oscillations, the attenuation rate for each oscillation cycle is calculated, and its expression is as follows: ; in, For the first The decay rate of the oscillation period corresponding to each oscillation peak. Indicates the number of times the oscillation interval is detected within the trailing edge. The amplitude of each oscillation peak, Indicates the relationship with the first The oscillation peaks are adjacent and located at the first... The first oscillation peak after the second The amplitude of each oscillation trough, The value of is incremented sequentially from 1 to the total number of complete oscillation periods detected within the subsequent oscillation interval. The total number of complete oscillation periods detected within the subsequent oscillation interval. The number of paired oscillation peaks and troughs that appear within the subsequent oscillation interval is determined by statistical analysis. The value is an integer greater than or equal to 3. (Oscillating wave peak amplitude) The method of obtaining the amplitude of the oscillation wave trough is to detect local maxima one by one from the starting point to the back within the oscillation interval. The method for obtaining this value is to detect the local minimum value adjacent to the oscillation peak within the same oscillation period and take its absolute value. Attenuation rate The value ranges from 0 to 1, and the attenuation rate is... The value reflects the degree of attenuation of the voltage amplitude from one oscillation peak to the next oscillation trough within the subsequent oscillation interval.

[0078] The arithmetic mean of the attenuation rates for all calculated oscillation periods is taken, and this average attenuation rate is used as the high-frequency attenuation correction factor. The amplitude of each sampling point in the initially recovered voltage waveform is multiplied by the average attenuation rate. The result of the multiplication operation constitutes the attenuation-compensated set of sampling point amplitudes, which contains the same number of amplitude data points as the initial recovered voltage waveform, thus obtaining the amplitude-corrected voltage sequence.

[0079] The voltage sequence after amplitude correction is then subjected to waveform correction processing. In this embodiment, the waveform correction processing includes sliding window filtering and difference compensation processing.

[0080] In practice, a sliding window filter is applied to the amplitude set of attenuation-compensated sampling points. The length of the sliding window is set to cover the number of sampling points corresponding to three oscillation periods. The length of the oscillation period is the arithmetic mean of the lengths of all oscillation periods during the attenuation rate calculation within the subsequent oscillation interval. The Hanning window function is used as the window function for the sliding window. The sliding window is sequentially moved across the amplitude set of attenuation-compensated sampling points, moving one sampling point at a time. The weighted average of the amplitudes of all sampling points within the sliding window and the Hanning window coefficient is calculated, and this weighted average is assigned to the sampling point corresponding to the center of the sliding window.

[0081] After the sliding window filtering process is completed, a smoothed voltage amplitude sequence is obtained. For each amplitude in the smoothed voltage amplitude sequence, the time index of that amplitude in the smoothed voltage amplitude sequence is obtained. The time point with the same time index is located in the preliminary recovered voltage waveform, and the amplitude of the previous sampling moment of that time point is extracted. The difference compensation amount between the amplitude of that time index in the smoothed voltage amplitude sequence and the amplitude of the previous sampling moment in the preliminary recovered voltage waveform is calculated. The difference compensation amount is taken as half of the difference between the corresponding amplitude in the smoothed voltage amplitude sequence and the amplitude of the previous sampling moment in the preliminary recovered voltage waveform.

[0082] The target high-voltage pulse voltage waveform is generated by adding the calculated difference compensation amount to the amplitude of each sampling point in the smoothed voltage amplitude sequence.

[0083] Compared to Examples 1 and 2, this example only requires truncating the trailing edge oscillation interval and extracting the peak and trough amplitudes, using the arithmetic mean of the attenuation rate as a correction factor. This reduces the number of operational steps and computational load, effectively decreasing the demand for data processing and storage resources. Furthermore, the sliding window filter's window length is related to the oscillation period length, allowing the filtering operation to match the inherent frequency characteristics of the waveform. Additionally, the Hanning window's sidelobes decay rapidly, effectively suppressing residual oscillations while preventing excessive smoothing of the main signal characteristics. Moreover, it eliminates the need for a digital filter, making implementation simpler.

[0084] Example 4: This example provides a high-voltage pulse voltage measurement method based on a capacitive voltage divider. The main difference from Example 3 is that it uses the envelope peak point and attenuation rate descriptor method to construct a high-frequency attenuation correction factor. By detecting the local maxima of the oscillation envelope peak point and calculating the amplitude attenuation difference between adjacent envelope points one by one, an attenuation rate descriptor reflecting the attenuation speed of each envelope segment is generated to comprehensively reflect the local changes in the attenuation process, thereby adapting to the measurement needs of complex waveforms with non-uniform attenuation.

[0085] In practical implementation, a sampling resistor is connected in series between the high-voltage and low-voltage arms of the capacitor voltage divider. The resistance value of the sampling resistor is selected to be less than one-thousandth of the equivalent impedance of the low-voltage arm, so that the voltage drop across the sampling resistor is much smaller than the voltage drop across the low-voltage arm, reducing the impact of the sampling resistor on the inherent voltage division characteristics of the capacitor voltage divider. The voltage difference signal across the sampling resistor is acquired by a differential voltage probe connected to the sampling resistor. After the voltage difference signal is amplified and subjected to anti-aliasing low-pass filtering by a signal conditioning circuit, it is recorded as the output voltage divider signal.

[0086] During the period of the high voltage pulse under test, the output voltage value of the low voltage arm is continuously collected at a fixed sampling frequency. The fixed sampling frequency is set according to the equivalent bandwidth of the rising edge of the high voltage pulse under test, and the value of the fixed sampling frequency is set to five times the equivalent highest frequency component of the leading edge of the high voltage pulse under test.

[0087] While continuously acquiring the voltage value at the output terminal of the low-voltage arm, the change in the dielectric loss angle of the high-voltage arm is monitored in real time through a dielectric loss testing device connected to the high-voltage arm. The change in dielectric loss angle is obtained by measuring the phase difference between the current flowing through the capacitor of the high-voltage arm and the voltage across the high-voltage arm in real time, and subtracting the initial phase difference obtained under power frequency reference conditions from this real-time phase difference. The difference is the change in dielectric loss angle.

[0088] At each sampling moment, a real-time voltage divider ratio correction coefficient is calculated based on the change in dielectric loss angle. The calculation method for the real-time voltage divider ratio correction coefficient is to multiply the rated voltage divider ratio of the capacitive voltage divider by a correction function. In this embodiment, the form of the correction function is as follows: ,in Indicates the sampling time The tangent of the change in the dielectric loss angle. The real-time voltage division ratio correction coefficients calculated at each sampling time are arranged sequentially according to the sampling time to form the actual voltage division ratio drift sequence.

[0089] Using the same method as in Example 3, a mapping relationship is established between each real-time voltage division ratio correction coefficient in the actual voltage division ratio drift sequence and the sampling point in the output voltage division signal at the same sampling time. An amplitude compensation operation is performed on each sampling point of the output voltage division signal to obtain a discrete compensation amplitude sequence. In this example, the amplitude compensation operation is to divide the voltage value of the sampling point by the real-time voltage division ratio correction coefficient after resampling that has a mapping relationship with the sampling point. The result of the division operation is used as the compensation amplitude of the sampling point.

[0090] The discrete compensation amplitude sequence is then subjected to cubic spline interpolation to obtain the preliminary recovered voltage waveform in the continuous time domain. Based on the preliminary recovered voltage waveform, a high-frequency attenuation correction factor is constructed using the envelope peak point and attenuation rate descriptor method.

[0091] Specifically, the step of extracting oscillation attenuation characteristics from the initial recovered voltage waveform and constructing a high-frequency attenuation correction factor based on the oscillation attenuation characteristics includes: Local maxima detection is performed on the preliminary recovered voltage waveform, and the sampling points that meet the local maxima conditions are determined as the peak points of the oscillation envelope; Calculate the amplitude attenuation difference between two adjacent oscillation envelope peaks, and combine the time interval between two adjacent oscillation envelope peaks and the amplitude of the previous oscillation envelope peak to generate an attenuation rate descriptor for each oscillation envelope peak. The high-frequency attenuation correction factor is calculated based on the attenuation rate descriptor corresponding to the peak point of each oscillation envelope.

[0092] The local maximum detection (MMR) iterates through each interpolated sampling point in the initially recovered voltage waveform, determining whether the voltage value of the current sampling point is simultaneously greater than the voltage values ​​of the previous and next sampling points. Sampling points satisfying this condition are identified as oscillation envelope peak points. The MMR traversal range is from the moment the voltage peak appears in the initially recovered voltage waveform until the end of the waveform. All detected oscillation envelope peak points are arranged sequentially in chronological order, resulting in an ordered set of oscillation envelope peak points. The amplitude attenuation difference between two adjacent oscillation envelope peak points is calculated. The amplitude attenuation difference is calculated as follows: for the ... The voltage amplitude of each oscillation envelope peak point is used to determine its value. Subtract the first Voltage amplitude at each oscillation envelope peak point , obtained the The amplitude attenuation difference between the peak points of the oscillation envelope and the (u+1)th oscillation envelope peak point forming the u-th adjacent oscillation envelope peak point pair .

[0093] A decay rate descriptor is generated based on the amplitude attenuation difference, and the calculation expression for the decay rate descriptor is as follows: ; in, For the u-th adjacent oscillation envelope peak point pair, the corresponding decay rate descriptor is... This indicates the first detected value in the initial recovery voltage waveform. Voltage amplitude at each oscillation envelope peak point This indicates the first detected value in the initial recovery voltage waveform. Voltage amplitude at each oscillation envelope peak point For the first The peak point of the oscillation envelope and the first The amplitude attenuation difference between the peak points of the oscillation envelope, For the first The peak point of the oscillation envelope and the first The time interval between the peak points of the oscillation envelope The value starts from 1 and increases sequentially up to the total number of detected oscillation envelope peaks. , This represents the total number of oscillation envelope peak points detected from the initially recovered voltage waveform. The value of is determined by the local maximum detection process. The value is an integer greater than or equal to 3. The voltage amplitude at each oscillation envelope peak point. Read directly from the sampled data of the initially recovered voltage waveform, time interval It is obtained by multiplying the difference in sampling indices corresponding to the peak points of the two oscillation envelopes by the interpolated sampling time interval.

[0094] Based on the time interval and decay rate descriptor corresponding to each oscillation envelope peak point, a high-frequency attenuation correction factor is calculated. The value of the high-frequency attenuation correction factor is obtained by taking the arithmetic mean of the calculated decay rate descriptor for all adjacent oscillation envelope peak points, and then normalizing it by multiplying it by the product of the average time interval between each oscillation envelope peak point and the main oscillation frequency value. The main oscillation frequency value is obtained by performing zero-crossing detection on the preliminary recovery voltage waveform and extracting the reciprocal of the time interval between zero-crossing points and taking the statistical mean.

[0095] Next, the high-frequency attenuation correction factor is multiplied point-by-point by the amplitude of each sampling point in the preliminary recovered voltage waveform. This point-by-point multiplication is performed by sequentially traversing each sampling point in the preliminary recovered voltage waveform, multiplying the voltage value of that sampling point by the value of the high-frequency attenuation correction factor, and using the product as the voltage value of the corresponding sampling point in the amplitude-corrected voltage sequence. Phase compensation processing is then applied to the amplitude-corrected voltage sequence. This is performed by calculating the required phase offset for each sampling point based on the main oscillation frequency. A phase delay or phase lead corresponding to the phase offset is applied to the amplitude-corrected voltage sequence using an all-pass digital filter. The phase response curve of the all-pass digital filter exhibits linear characteristics within the required frequency range. After applying the phase offset, the phase-corrected voltage sequence is obtained.

[0096] The target high-voltage pulse voltage waveform is obtained by subtracting half of the amplitude difference between the previous sampling point and the amplitude of each sampling point in the phase-corrected voltage sequence.

[0097] Compared to Example 3, this example generates a decay rate descriptor reflecting the decay rate of each envelope segment by detecting local maxima at the peak points of the oscillation envelope and calculating the amplitude decay difference between adjacent envelope points. This descriptor integrates information from three dimensions: amplitude decay, decay time interval, and current absolute amplitude level, effectively overcoming the limitation that the global average decay rate cannot reflect the change in decay rate over time. Furthermore, for complex waveforms with non-uniform decay of the oscillation envelope, this example also captures the local differences in decay rate at each time interval by describing each segment and incorporates them into the calculation of the correction factor, achieving accurate correction for non-uniform decay oscillations.

[0098] Example 5: This example provides a high-voltage pulse voltage measurement method based on a capacitive voltage divider. (See attached document.) Figure 2 The main difference from Example 4 is that the high-frequency attenuation correction factor is calculated using the logarithmic formula method, and time-domain reconstruction is performed using exponential compensation, frequency shift correction, and zero-phase filtering. Considering that the amplitude envelope of typical high-frequency attenuated oscillations usually follows an exponential decay law, the exponential relationship is transformed into a linear relationship by taking the natural logarithm of the amplitude ratio, so that the attenuation rate between each oscillation period has better linear comparability. At the same time, in the time-domain reconstruction stage, the correction factor is decomposed into amplitude attenuation component and frequency shift component and processed separately. The amplitude compensation adopts exponential compensation, which is inverse to the logarithmic processing in the construction of the correction factor. The frequency correction realizes the overall shift of the signal in the frequency domain through complex single-sideband modulation, thereby correcting the main oscillation frequency shift. The forward and reverse filtering make the net phase response of the filter in the signal zero, ensuring that the output waveform does not shift in the time axis, thus improving the accuracy of amplitude correction. It is suitable for measurement scenarios with the highest requirements for waveform fidelity.

[0099] In practice, a sampling resistor is connected in series between the high-voltage and low-voltage arms of the capacitive voltage divider. The resistance value of the sampling resistor is selected to be less than one-thousandth of the equivalent impedance of the low-voltage arm. A differential voltage probe connected to the sampling resistor acquires the voltage difference signal across it. This voltage difference signal is amplified and low-pass filtered for anti-aliasing by a signal conditioning circuit before being recorded as the output voltage divider signal. During the high-voltage pulse under test, the output voltage value of the low-voltage arm is continuously acquired at a fixed sampling frequency. The fixed sampling frequency is determined based on the equivalent bandwidth of the rising edge of the high-voltage pulse under test, and is set to five times the highest equivalent frequency component of the leading edge of the high-voltage pulse. Simultaneously with the continuous acquisition of the low-voltage arm output voltage value, the change in the dielectric loss angle of the high-voltage arm is monitored in real time using a dielectric loss testing device connected to the high-voltage arm. At each sampling moment, a real-time voltage divider ratio correction coefficient is calculated based on the change in the dielectric loss angle. The real-time voltage divider ratio correction coefficient is calculated by multiplying the rated voltage divider ratio of the capacitive voltage divider by a correction function associated with the change in the dielectric loss angle. The real-time voltage division ratio correction coefficients calculated at each sampling time in the above manner are arranged sequentially according to the order of the sampling times to form the actual voltage division ratio drift sequence.

[0100] Using the same method as in Example 4, a mapping relationship is established between each real-time voltage division ratio correction coefficient in the actual voltage division ratio drift sequence and the sampling point in the output voltage division signal at the same sampling time. An amplitude compensation operation is performed on each sampling point of the output voltage division signal to obtain a discrete compensation amplitude sequence. In this example, the amplitude compensation operation is to divide the voltage value of the sampling point by the real-time voltage division ratio correction coefficient after resampling that has a mapping relationship with the sampling point. The result of the division operation is used as the compensation amplitude of the sampling point.

[0101] The discrete compensation amplitude sequence is then subjected to cubic spline interpolation to obtain the preliminary recovered voltage waveform in the continuous time domain. Based on the preliminary recovered voltage waveform, a high-frequency attenuation correction factor is constructed using the envelope peak point and attenuation rate descriptor method.

[0102] However, unlike Example 4, this example first performs local maximum detection on the preliminary recovered voltage waveform. Local maximum detection involves traversing each interpolated sampling point after the voltage peak occurrence time in the preliminary recovered voltage waveform, determining whether the voltage value of the current sampling point is simultaneously greater than the voltage values ​​of the previous and next sampling points. Sampling points satisfying this condition are identified as oscillation envelope peak points. All detected oscillation envelope peak points are arranged sequentially according to time, resulting in an ordered sequence of oscillation envelope peak points. The number of oscillation envelope peak points in the sequence is denoted as _____. , The value of is an integer greater than or equal to 3, which is automatically determined by the local maximum detection process.

[0103] Calculate the amplitude attenuation difference between two adjacent oscillation envelope peaks, for the th The voltage amplitude of each oscillation envelope peak point is used to determine its value. Subtract the first Voltage amplitude at each oscillation envelope peak point The amplitude attenuation difference between the two is obtained. A decay rate descriptor is generated based on this amplitude attenuation difference, and the decay rate descriptor corresponds to each pair of adjacent oscillation envelope peak points. It is obtained by dividing the amplitude attenuation difference by the product of the time interval between the peak points of the pair of oscillation envelopes and the voltage amplitude of the previous peak point of the oscillation envelope.

[0104] Based on the time interval and attenuation rate descriptor corresponding to each oscillation envelope peak point, the high-frequency attenuation correction factor is calculated. The calculation expression is as follows: ; in, For high-frequency attenuation correction factor, The total number of detected oscillation envelope peaks. Indicates the first The voltage amplitude at each peak of the oscillation envelope is directly read from the interpolated sampled data of the initially recovered voltage waveform. Indicates the first The voltage amplitude at each peak of the oscillation envelope is also directly read from the interpolated sampled data of the initially recovered voltage waveform. Indicates the first The peak point of the oscillation envelope and the first The time interval between the peak points of the oscillation envelope is obtained by multiplying the difference in time indices corresponding to two peak points of the oscillation envelope by the interpolated sampling time interval. The value of starts from 1 and increases sequentially to 1. .

[0105] Based on the high-frequency attenuation correction factor, exponential compensation, frequency shift correction and zero-phase filtering are used for time-domain reconstruction.

[0106] The step of reconstructing the initially recovered voltage waveform in the time domain based on the high-frequency attenuation correction factor to obtain the target high-voltage pulse voltage waveform corresponding to the high-voltage pulse under test includes: The high-frequency attenuation correction factor is decomposed into an amplitude attenuation component and a frequency offset component; The amplitude-compensated waveform is obtained by exponentially compensating the initial recovery voltage waveform based on the amplitude attenuation component. The amplitude compensation waveform is frequency shifted based on the frequency offset component to obtain the frequency shifted corrected waveform. The frequency shift correction waveform is subjected to zero-phase filtering to obtain the target high-voltage pulse voltage waveform.

[0107] The high-frequency attenuation correction factor is decomposed into an amplitude attenuation component and a frequency offset component. The value is the high-frequency attenuation correction factor. The value of the frequency offset component The acquisition method is as follows: zero-crossing detection is performed on the preliminary recovered voltage waveform, and the zero-crossing time points of all voltage amplitudes crossing the zero level are extracted. The reciprocal of the time interval between the zero-crossing time points of adjacent symbols with the same direction of change is calculated to obtain a set of instantaneous frequency values. The main oscillation frequency value is determined from this set of instantaneous frequency values. Then, the difference between the main oscillation frequency value and the reference oscillation frequency value of the capacitor voltage divider under the power frequency reference condition is taken. The difference is the frequency offset component. The reference oscillation frequency value is pre-determined and stored after the capacitor divider is installed using the low-voltage square wave injection method.

[0108] Using amplitude attenuation component To perform exponential compensation adjustment on the initial recovered voltage waveform, first, for each sampling point in the initial recovered voltage waveform, obtain the time value corresponding to the m-th sampling point. And calculate the exponential compensation coefficient for the m-th sampling point. The voltage amplitude at the m-th sampling point is multiplied by the exponential compensation coefficient, and the product is used as the voltage amplitude at the corresponding moment in the amplitude compensation waveform. After performing the above operation on all sampling points of the initially recovered voltage waveform, the complete amplitude compensation waveform is obtained.

[0109] Using frequency offset components Frequency shift correction is performed on each frequency component of the amplitude compensation waveform. This frequency shift correction is achieved through complex single-sideband modulation, constructing a modulation signal corresponding to each sampling time, which is expressed as follows: The voltage amplitude of the amplitude compensation waveform at each sampling moment is multiplied by the corresponding modulation signal value at that sampling moment, and the real part of the product is extracted as the voltage amplitude of the frequency shift correction waveform at that sampling moment. After traversing all sampling moments of the amplitude compensation waveform, the frequency shift correction waveform is obtained.

[0110] Zero-phase filtering is applied to the frequency shift correction waveform to eliminate any phase delay that may be introduced during the frequency shift correction process. Zero-phase filtering is performed using a forward and reverse filtering method. First, the frequency shift correction waveform is passed forward through a Butterworth low-pass digital filter to obtain a forward filtered sequence. Then, the forward filtered sequence is passed backward through the same Butterworth low-pass digital filter to obtain a reverse filtered sequence. Finally, the reverse filtered sequence is reversed again, and the output sequence is the target high-voltage pulse waveform. The Butterworth low-pass digital filter is set to order 4, and its cutoff frequency is set to 1.5 times the equivalent highest frequency component of the rising edge of the high-voltage pulse under test.

[0111] Compared to embodiments 1 to 4, this embodiment calculates the correction factor by taking the natural logarithm of the amplitude ratio of adjacent envelope peak points, forming a mathematical inverse relationship with exponential compensation adjustment. This fully utilizes the physical characteristic that the high-frequency decaying oscillation amplitude envelope follows an exponential decay law, making the amplitude correction more accurate. Furthermore, the correction factor is decomposed into an amplitude decay component and a frequency offset component. Complex single-sideband modulation is used to achieve a complete shift of the signal in the frequency domain, thereby correcting the main oscillation frequency offset and overcoming the limitation of phase compensation methods in compensating for frequency offset. Simultaneously, forward and reverse filtering achieves zero-phase-delay smoothing, ensuring that the output waveform does not produce any offset on the time axis, meeting the measurement requirements of precision measurement scenarios with the highest waveform fidelity requirements.

[0112] Example 6: This example provides a high-voltage pulse voltage measurement system based on a capacitive voltage divider, including: The signal acquisition module is used to acquire the output voltage division signal of the low-voltage arm of the capacitor voltage divider under the action of the high-voltage pulse to be tested, and simultaneously acquire the change in dielectric loss angle of the high-voltage arm of the capacitor voltage divider. The drift sequence generation module is used to generate an actual voltage ratio drift sequence based on the change in dielectric loss angle. The inversion compensation module is used to perform inversion compensation processing on the output voltage divider signal according to the actual voltage divider ratio drift sequence, and generate a preliminary recovered voltage waveform. The correction factor construction module is used to extract oscillation attenuation characteristics from the initial recovered voltage waveform and construct a high-frequency attenuation correction factor based on the oscillation attenuation characteristics. The time-domain reconstruction module is used to reconstruct the initial recovered voltage waveform in the time domain based on the high-frequency attenuation correction factor, and obtain the target high-voltage pulse voltage waveform corresponding to the high-voltage pulse under test.

[0113] The signal acquisition module includes a sampling resistor, a differential amplifier, a first analog-to-digital converter (ADC), a dielectric loss tester, and a second ADC. The sampling resistor is connected in series between the high-voltage and low-voltage arms of a capacitive voltage divider. The input of the differential amplifier is connected to both ends of the sampling resistor to acquire the voltage difference across it; this voltage difference represents the output voltage divider signal of the low-voltage arm. The output of the differential amplifier is connected to the input of the first ADC, which converts the analog output voltage divider signal into a digital output voltage divider signal at a fixed sampling frequency. The dielectric loss tester is connected to the high-voltage arm of the capacitive voltage divider to monitor the change in the dielectric loss angle of the high-voltage arm in real time. Its output is connected to the input of the second ADC, which converts the analog dielectric loss angle change into a digital dielectric loss angle change. The first and second ADCs are driven by the same clock source to achieve synchronous acquisition of the two signals.

[0114] The input of the drift sequence generation module is connected to the output of the signal acquisition module, the input of the inversion compensation module is connected to the output of the drift sequence generation module, the input of the correction factor construction module is connected to the output of the inversion compensation module, and the input of the time-domain reconstruction module is connected to the outputs of both the inversion compensation module and the correction factor construction module. All four modules—drift sequence generation, inversion compensation, correction factor construction, and time-domain reconstruction—are equipped with data processing and analysis components, such as microprocessors, and can reconstruct the high-voltage pulse waveform from the two acquired signals.

[0115] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any way. Other variations and modifications are possible without departing from the technical solutions described in the claims.

Claims

1. A high-voltage pulse voltage measurement method based on a capacitive voltage divider, characterized in that, include: The output voltage division signal of the low-voltage arm of the capacitor voltage divider under the action of the high-voltage pulse under test is obtained, and the change of dielectric loss angle of the high-voltage arm of the capacitor voltage divider is collected simultaneously to form the actual voltage division ratio drift sequence. The output voltage divider signal is inverted and compensated based on the actual voltage divider ratio drift sequence to generate a preliminary recovered voltage waveform. Oscillation attenuation characteristics are extracted from the initial recovered voltage waveform, and a high-frequency attenuation correction factor is constructed based on these characteristics. The initial recovered voltage waveform is reconstructed in the time domain based on the high-frequency attenuation correction factor to obtain the target high-voltage pulse voltage waveform corresponding to the high-voltage pulse under test.

2. The high-voltage pulse voltage measurement method based on a capacitive voltage divider according to claim 1, characterized in that, The synchronous acquisition of the dielectric loss angle change of the high-voltage arm of the capacitor divider to form the actual voltage division ratio drift sequence includes: During the period of the high voltage pulse under test, the change in dielectric loss angle of the high voltage arm of the capacitor voltage divider is monitored in real time at a preset sampling frequency. Based on the corresponding change in dielectric loss angle, the real-time voltage division ratio correction coefficient for each sampling moment is calculated in conjunction with the rated voltage division ratio of the capacitive voltage divider. The real-time voltage division ratio correction coefficients at each sampling time are arranged in chronological order of sampling time to form the actual voltage division ratio drift sequence.

3. The high-voltage pulse voltage measurement method based on a capacitive voltage divider according to claim 2, characterized in that, The step of performing inversion compensation processing on the output voltage divider signal based on the actual voltage divider ratio drift sequence to generate a preliminary recovered voltage waveform includes: Establish a mapping relationship between each real-time voltage division ratio correction coefficient in the actual voltage division ratio drift sequence and the sampling point in the output voltage division signal at the same sampling time; Based on the mapping relationship, amplitude compensation calculations are performed on each sampling point of the output voltage divider signal using the corresponding voltage divider ratio correction coefficient to obtain a discrete compensation amplitude sequence. Interpolation is performed on the discrete compensation amplitude sequence to obtain the preliminary recovery voltage waveform in the continuous time domain.

4. The high-voltage pulse voltage measurement method based on a capacitive voltage divider according to claim 2, characterized in that, The step of performing inversion compensation processing on the output voltage divider signal based on the actual voltage divider ratio drift sequence to generate a preliminary recovered voltage waveform includes: The output voltage divider signal is divided into several voltage divider signal segments according to the sampling time. Each voltage divider signal segment is centered on a single sampling time and contains a main sampling point and several adjacent sampling points before and after it. Match the real-time voltage division ratio correction coefficient corresponding to the center sampling time of each voltage division signal segment from the actual voltage division ratio drift sequence, and align the real-time voltage division ratio correction coefficient with the voltage division signal segment in time. Based on the corresponding real-time voltage division ratio correction coefficient after timing alignment, each voltage division signal segment is subjected to amplitude scaling and time axis stretching to obtain the compensated waveform segment. The compensated waveform segments are arranged sequentially according to the sampling time, and adjacent compensated waveform segments are merged by weighted averaging of overlapping regions to obtain the preliminary recovered voltage waveform in the continuous time domain.

5. The high-voltage pulse voltage measurement method based on a capacitive voltage divider according to claim 1, characterized in that, The step of extracting oscillation attenuation features from the initial recovered voltage waveform and constructing a high-frequency attenuation correction factor based on the oscillation attenuation features includes: Zero-crossing detection is performed on the preliminary recovered voltage waveform, and all zero-crossing points in the preliminary recovered voltage waveform where the voltage amplitude crosses the zero level are extracted; The length of the corresponding oscillation period is determined based on the time interval between each adjacent zero crossing point, and the main oscillation frequency value is determined based on the length of each oscillation period. Extract the apparent amplitude of each oscillation period and calculate the ratio of the attenuation amplitude between adjacent oscillation periods; The high-frequency attenuation correction factor is calculated based on the length of each oscillation period, the main oscillation frequency, and the ratio of attenuation amplitude between adjacent oscillation periods.

6. The high-voltage pulse voltage measurement method based on a capacitive voltage divider according to claim 1, characterized in that, The step of extracting oscillation attenuation features from the initial recovered voltage waveform and constructing a high-frequency attenuation correction factor based on the oscillation attenuation features includes: Extract the trailing edge oscillation interval from the initial recovered voltage waveform, where the voltage amplitude gradually decays to zero after the voltage peak. Multiple consecutive oscillation peak amplitudes and corresponding oscillation trough amplitudes are extracted within the trailing edge oscillation interval, and adjacent oscillation peaks and troughs are combined into an oscillation period. Calculate the attenuation rate for each oscillation cycle, and use the arithmetic mean of the attenuation rates for all oscillation cycles as the high-frequency attenuation correction factor.

7. The high-voltage pulse voltage measurement method based on a capacitive voltage divider according to claim 1, characterized in that, The step of extracting oscillation attenuation features from the initial recovered voltage waveform and constructing a high-frequency attenuation correction factor based on the oscillation attenuation features includes: Local maxima detection is performed on the preliminary recovered voltage waveform, and the sampling points that meet the local maxima conditions are determined as the peak points of the oscillation envelope; Calculate the amplitude attenuation difference between two adjacent oscillation envelope peaks, and combine the time interval between two adjacent oscillation envelope peaks and the amplitude of the previous oscillation envelope peak to generate an attenuation rate descriptor for each oscillation envelope peak. The high-frequency attenuation correction factor is calculated based on the attenuation rate descriptor corresponding to the peak point of each oscillation envelope.

8. The high-voltage pulse voltage measurement method based on a capacitive voltage divider according to claim 1, characterized in that, The step of reconstructing the initially recovered voltage waveform in the time domain based on the high-frequency attenuation correction factor to obtain the target high-voltage pulse voltage waveform corresponding to the high-voltage pulse under test includes: Multiply the amplitude of each sampling point in the initially recovered voltage waveform by the high-frequency attenuation correction factor to obtain the amplitude-corrected voltage sequence; The voltage sequence after amplitude correction is subjected to waveform correction processing, and the voltage sequence after waveform correction processing is subjected to difference compensation processing to obtain the target high-voltage pulse voltage waveform.

9. The high-voltage pulse voltage measurement method based on a capacitive voltage divider according to claim 1, characterized in that, The step of reconstructing the initially recovered voltage waveform in the time domain based on the high-frequency attenuation correction factor to obtain the target high-voltage pulse voltage waveform corresponding to the high-voltage pulse under test includes: The high-frequency attenuation correction factor is decomposed into an amplitude attenuation component and a frequency offset component; The initial recovered voltage waveform is adjusted exponentially based on the amplitude attenuation component to obtain the amplitude-compensated waveform; The amplitude compensation waveform is frequency shifted based on the frequency offset component to obtain the frequency shift corrected waveform. The frequency shift correction waveform is subjected to zero-phase filtering to obtain the target high-voltage pulse voltage waveform.

10. A high-voltage pulse voltage measurement system based on a capacitive voltage divider, used to perform the high-voltage pulse voltage measurement method according to any one of claims 1 to 9, characterized in that, include: The signal acquisition module is used to acquire the output voltage division signal of the low-voltage arm of the capacitor voltage divider under the action of the high-voltage pulse to be tested, and simultaneously acquire the change in dielectric loss angle of the high-voltage arm of the capacitor voltage divider. The drift sequence generation module is used to generate an actual voltage ratio drift sequence based on the change in dielectric loss angle. The inversion compensation module is used to perform inversion compensation processing on the output voltage divider signal according to the actual voltage divider ratio drift sequence, and generate a preliminary recovered voltage waveform. The correction factor construction module is used to extract oscillation attenuation characteristics from the initial recovered voltage waveform and construct a high-frequency attenuation correction factor based on the oscillation attenuation characteristics. The time-domain reconstruction module is used to reconstruct the initial recovered voltage waveform in the time domain based on the high-frequency attenuation correction factor, and obtain the target high-voltage pulse voltage waveform corresponding to the high-voltage pulse under test.

Images