A method and system for compensating for transient non-linear errors of an energy metering box transformer

Abstract

The application discloses a kind of electric energy metering box mutual inductor transient nonlinear error compensation method and system.The method comprises the following steps: S1 baseline modeling, the nonlinear error characteristic baseline model covering the transient error trend under different switching paths is established for each steady-state load point ratio difference angle difference;S2 synchronous sampling, voltage and current channel are synchronously sampled and steady-state and transient switching state are identified in real time;S3 cross-checking, cross-consistency checking is carried out using the physical conservation constraint between voltage and current, and the deviation vector containing ratio difference deviation component and angle difference deviation component is output;S4 dynamic compensation, the dynamic compensation coefficient is calculated in real time by combining the deviation vector and baseline model, and the sampling data is compensated;S5 electric energy accumulation, compensated instantaneous power is integrated and electric energy pulse is output.Three feedback loops are established between each step, a multi-scale closed-loop structure is formed, real-time dynamic compensation of mutual inductor transient nonlinear error is realized, and dependence on high-precision mutual inductor hardware is reduced.

Description

Technical Field

[0001] This invention relates to the field of electricity metering technology, specifically to a method and system for compensating for transient nonlinear errors in current transformers of electricity metering boxes. Background Technology

[0002] Existing electricity metering boxes convert primary-side electrical signals into secondary-side measurable signals through current transformers and voltage transformers, and then the metering chip in the smart meter performs power calculations and energy accumulation.

[0003] To ensure measurement accuracy, the industry generally adopts the method of selecting high-precision instrument transformer hardware and performing steady-state calibration at several rated load points. The compensation coefficient obtained from the calibration is then solidified into the metering chip and remains unchanged throughout the entire service life.

[0004] However, in real-world power applications, loads frequently switch transiently between light and overload conditions. The nonlinear response of the transformer core flux during this transition causes the phase angle error and ratio error to exhibit transient deviations significantly different from the steady-state calibration values ​​within a time window of milliseconds to seconds. Existing fixed compensation coefficients and linear interpolation methods cannot capture and correct these transient nonlinear errors. Furthermore, the independent processing of voltage and current sampling channels does not utilize the inherent physical conservation constraints between the two channels for cross-verification. In addition, the unidirectional irreversible nature of the energy accumulation process and the chain structure without feedback throughout the entire process mean that once transient errors occur, they cannot be traced and corrected, and they continue to accumulate unidirectionally with repeated load switching events. Current industry solutions to this problem are limited to using higher-precision transformer hardware or increasing the number of calibration points. Summary of the Invention

[0005] To achieve the above objectives, a method for compensating for transient nonlinear errors in current transformers within an energy metering box includes the following steps: S1 Baseline Modeling: Nonlinear error characteristic baseline modeling is performed on the current and voltage transformers within the energy metering box to establish a nonlinear error characteristic baseline model covering the ratio and angle differences at each steady-state load point, as well as the transient transition error variation trend under different load switching directions and rates; S2 Synchronous Sampling: Synchronous sampling is performed on the voltage and current sampling channels, and real-time load state analysis is conducted on the sampled data stream to identify steady-state operating states and transient switching states; S3 Cross-validation The following steps are implemented: S4 Dynamic Compensation: Based on the physical conservation constraint relationship between voltage and current, cross-consistency verification is performed on the multi-channel synchronous sampling data. Combined with the load impedance continuity constraint and the power factor physical feasible region constraint, the verification result is output in the form of a deviation vector containing the ratio deviation component and the angle deviation component; S5 Dynamic Compensation: Based on the deviation vector and the nonlinear error characteristic baseline model, the dynamic compensation coefficient of the current sampling period is calculated in real time. After performing dynamic error compensation on the sampling data, the instantaneous power is calculated; S6 Energy Accumulation: The compensated instantaneous power value is integrated over time to accumulate electrical energy and output an energy pulse.

[0006] Preferably, the steps form a multi-scale closed-loop structure, including: the dynamic compensation step evaluates the compensation correction amount for each sampling period, and when the compensation correction amount exceeds a preset level, it sends an instruction to the synchronous sampling step to increase the sampling density; when the compensation correction amount for multiple consecutive sampling periods is lower than the preset level, it sends an instruction to the synchronous sampling step to restore the normal sampling density; the power accumulation step maintains a power increment trend monitoring mechanism within a sliding time window, detects the systematic cumulative deviation trend associated with a specific load switching mode, and feeds back the cumulative deviation trend to the cross-validation step for continuous correction of the validation benchmark parameters; the cross-validation step periodically feeds back the statistical characteristics of power conservation deviation under different load states and switching paths accumulated during long-term operation to the baseline modeling step, and the baseline modeling step dynamically updates the nonlinear error characteristic baseline model based on the fed-back long-term validation statistical characteristics.

[0007] Preferably, the baseline modeling step includes: during the initial installation phase, using an automated load scanning method, gradually changing the load level within a load range of 1% to 400% of the basic current at a preset load step interval, and collecting the corresponding ratio difference and angle difference data as steady-state error feature points after each load level stabilizes; performing a fast switching operation between adjacent load levels, collecting transient error response data during the switching transition process according to the rising path from light load to heavy load and the falling path from heavy load to light load, respectively, and recording the transient offset direction and recovery time characteristics of the transformer ratio difference and angle difference at different switching rates; and comprehensively constructing the nonlinear error characteristic baseline model by combining the steady-state error feature points and the transient error response data. The nonlinear error characteristic baseline model uses the load level and load switching path as dual index dimensions, and a two-dimensional piecewise linear interpolation lookup table as the basic structure. Each grid node stores the steady-state ratio difference, steady-state angle difference, transient peak deviation, and recovery time constant, and the nonlinear error characteristic baseline model is stored in the updatable storage area of ​​the metering chip.

[0008] Preferably, the synchronous sampling step adds a transient event tag to the sampled data when the load enters a transient switching process. The transient event tag includes the switching start time, switching direction, current change rate, and steady-state load level before switching. The cross-validation step adaptively adjusts the cross-validation judgment threshold according to the transient event tag.

[0009] Preferably, the real-time load status analysis of the sampled data stream includes: continuously acquiring data from the current sampling channel at a preset reference sampling frequency, calculating the gradient of the change in the root mean square value of the current between adjacent sampling points; comparing the gradient with a preset transient switching identification threshold, and determining that the current load has entered a transient switching process when the gradient continuously exceeds the transient switching identification threshold within a series of sampling points; determining the switching direction based on the positive or negative direction of the current amplitude change, determining the current change rate based on the absolute value of the gradient, and determining the steady-state load level before the switching based on the average current value within the most recent steady-state time window before the switching occurs.

[0010] Preferably, the sampling frequency of the synchronous sampling step is dynamically adjusted according to the sampling density adjustment command fed back from the dynamic compensation step, including: when a command to increase the sampling density is received, the current sampling frequency is increased to a preset encrypted sampling frequency, and the encrypted sampling frequency is maintained within a preset encryption duration window; when a command to restore the normal sampling density is received, the current sampling frequency is restored to the reference sampling frequency; wherein, the encrypted sampling frequency is higher than the reference sampling frequency, and the length of the encryption duration window is adaptively determined according to the current change rate of the most recent transient switching event, and the greater the current change rate, the longer the encryption duration window.

[0011] Preferably, the cross-validation step includes: calculating a load impedance amplitude reference value by dividing the root mean square value of voltage by the root mean square value of current, and calculating an impedance phase angle reference value by subtracting the fundamental phase angle of current from the fundamental phase angle of voltage, based on the sampling data of the voltage sampling channel and the current sampling channel in the most recent consecutive power frequency cycles in steady state; using the load impedance amplitude reference value and the impedance phase angle reference value as continuity constraint benchmarks; within each sampling cycle, calculating the equivalent load impedance amplitude and equivalent impedance phase angle of the current cycle using the sampling data of the current sampling channel and the sampling data of the voltage sampling channel in the same way; obtaining the impedance amplitude deviation rate by dividing the current equivalent impedance amplitude by the load impedance amplitude reference value and then subtracting one; obtaining the impedance phase angle deviation by subtracting the impedance phase angle reference value from the current equivalent impedance phase angle; and verifying whether the impedance amplitude deviation rate and the impedance phase angle deviation exceed the adaptive judgment threshold; when the load is in a steady-state operating state, the adaptive judgment threshold is taken as the smaller value. To maintain high verification sensitivity; when the load is in a transient switching process, the adaptive judgment threshold is relaxed and adjusted according to the current change rate and switching direction to adapt to the reasonable variation range of load parameters during the transient transition; the output of the verification result in the form of a deviation vector includes: since the grid voltage is relatively stable during the transient switching and the ratio difference of the current transformer causes the equivalent impedance amplitude to deviate in the opposite direction, the impedance amplitude deviation rate is negatively extracted as the ratio difference deviation component; the impedance phase angle deviation is negatively extracted as the angle difference deviation component; before extracting the deviation vector, the measured equivalent impedance deviation direction is consistent with the expected transformer error deviation direction of the corresponding switching path in the nonlinear error characteristic baseline model. When the two directions are consistent, the ratio difference deviation component and the angle difference deviation component are combined into the deviation vector and continuously output in each sampling period during the transient switching process. When the two directions are inconsistent, the deviation vector is set to zero to avoid false compensation.

[0012] Preferably, the dynamic compensation step includes: using the error characteristics in the nonlinear error characteristic baseline model corresponding to the current load level and the current switching path as a basic compensation reference; superimposing the ratio deviation component in the deviation vector onto the ratio compensation factor in the basic compensation reference, and superimposing the angle deviation component in the deviation vector onto the angle compensation factor in the basic compensation reference to obtain the dynamic compensation coefficient for the current sampling period; during steady-state operation of the load, the amplitude of the deviation vector approaches zero, and the dynamic compensation coefficient remains consistent with the basic compensation reference; during transient switching transition, the dynamic compensation coefficient is adjusted in real time for each sampling period driven by the deviation vector; the evaluation method for the compensation correction amount is: applying the dynamic compensation coefficient and the basic compensation reference to the current sampling period respectively. After each set of sampled data undergoes error compensation, the instantaneous power is calculated. The absolute value of the difference between two instantaneous power values ​​is the compensation correction amount for the current sampling period. The energy increment trend monitoring mechanism within the sliding time window includes: continuously recording the energy increment value for each sampling period within a preset length sliding time window, and associating each energy increment value with the corresponding load switching event type; statistically analyzing the energy increment values ​​with the same load switching event type within the sliding time window to detect whether there is a continuous unidirectional offset trend; when a systematic offset trend associated with a specific load switching mode is detected, the direction and amplitude characteristics of the offset trend are fed back to the cross-validation step as the cumulative deviation trend to correct the judgment benchmark in the verification benchmark parameters corresponding to the load switching mode.

[0013] Preferably, the baseline modeling step dynamically updates the nonlinear error characteristic baseline model based on the returned long-term verification statistical characteristics, including: the cross-verification step continuously records the deviation vector statistical distribution under different load levels, different switching paths, and different switching rates during operation; when the accumulated running time reaches the preset model update cycle, the trend component reflecting the gradual change of the transformer error characteristics in the deviation vector statistical distribution is extracted and returned to the baseline modeling step as the model update increment; the baseline modeling step progressively corrects the ratio difference, angle difference, and transient response parameters of each grid node in the lookup table of the nonlinear error characteristic baseline model based on the model update increment, wherein the parameter change of any grid node in a single update does not exceed the preset upper limit of the initial calibration value of that node, and the timestamp of each update, the parameter difference before and after the update, and the statistical basis for triggering the update are all written to an immutable log storage area.

[0014] A transient nonlinear error compensation system for current transformers in an energy metering box, applicable to the aforementioned transient nonlinear error compensation method for current transformers in an energy metering box, including: The baseline modeling unit is used to perform baseline modeling of nonlinear error characteristics of current transformers and voltage transformers in the power metering box with a wide load range, establish a nonlinear error characteristic baseline model and store it in the updatable storage area of ​​the metering chip, and receive long-term verification statistical features returned from the cross-verification unit to dynamically update the nonlinear error characteristic baseline model. The adaptive sampling unit is used to perform synchronous sampling of the voltage sampling channel and the current sampling channel, perform real-time load state analysis on the sampled data stream and attach transient event tags when a transient switching process is detected, and dynamically adjust the sampling frequency according to the sampling density adjustment command fed back from the dynamic compensation unit. The cross-validation unit is used to perform cross-consistency verification on multi-channel synchronous sampling data with transient event labels using the physical conservation constraint relationship between voltage and current. It adaptively adjusts the verification judgment threshold according to the transient event labels, continuously corrects the verification benchmark parameters according to the long-term power drift trend information fed back from the power accumulation unit, outputs the verification results to the dynamic compensation unit in the form of a deviation vector, and sends the long-term verification statistical characteristics back to the baseline modeling unit. The dynamic compensation unit is used to calculate the dynamic compensation coefficient in real time based on the direction and magnitude of the deviation vector and the baseline model of nonlinear error characteristics, perform dynamic error compensation on the sampled data and calculate the instantaneous power, evaluate the compensation correction amount and generate a sampling density adjustment command based on this and feed it back to the adaptive sampling unit. The power accumulation unit is used to integrate the compensated instantaneous power value over time to accumulate power energy. It monitors the trend of power increment change within the sliding time window and feeds back the accumulated deviation trend to the cross-validation unit. The final confirmed power energy value is converted into power pulse output according to the pulse constant.

[0015] Compared with the prior art, the beneficial effects of the present invention are: (1) By establishing a cross-coupled closed-loop structure containing three feedback loops of short period, medium period and long period among the five process steps of baseline modeling, adaptive sampling, cross-consistency verification, dynamic error compensation and power accumulation, the inherent physical conservation constraint relationship between voltage sampling channel and current sampling channel is used to realize the real-time identification and dynamic compensation of transient nonlinear error of current transformer. This overcomes the long-standing industry belief that solving the problem of transient nonlinear error accumulation of current transformer angle difference ratio during light load and overload switching is highly dependent on high-precision current transformer and voltage transformer. It allows high-precision metering to be achieved even when using current transformers with relatively ordinary accuracy levels.

[0016] (2) The nested closed-loop adaptive structure of multiple time scales enables the system to continuously learn and self-optimize throughout the entire working cycle. It can automatically track the gradual change in error characteristics of the current transformer caused by factors such as aging and temperature changes, which significantly reduces the hardware cost and life-cycle maintenance cost of the power metering box. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall process flow of the method of the present invention; Figure 2 This is a schematic diagram of the overall system architecture of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1, please refer to Figure 1 This invention provides a method for compensating for transient nonlinear errors in current transformers of an energy metering box, comprising: S1, Baseline Modeling of Wide-Range Nonlinear Error Characteristics of Current Transformers: For current transformers and voltage transformers in the power metering box, error response data are collected across the entire load range from light load to overload and under various load switching paths. A baseline model of nonlinear error characteristics is established, covering the ratio and angle difference values ​​of each steady-state load point and the transient transition error variation trend under different load switching directions and switching rates. This model is stored in the updatable storage area of ​​the metering chip. The model is dynamically updated based on the long-term verification statistical characteristics returned from the cross-consistency verification step.

[0020] S2, Adaptive multi-channel synchronous sampling with load state awareness: Synchronous sampling is performed on the voltage sampling channel and the current sampling channel, and real-time load state analysis is performed on the sampled data stream. By monitoring the gradient of the current sample value sequence, it is determined whether the current load is in a steady-state operation state or a transient switching state. When the load enters a transient switching process, a transient event label is attached to the sampled data. This label includes the switching start time, switching direction, current change rate, and steady-state load level before switching. The sampling frequency is dynamically adjusted according to the sampling density adjustment command fed back from the dynamic error compensation step.

[0021] S3, Power Conservation Cross-Verification between Voltage and Current Channels: Utilizing the physical conservation constraints between voltage and current, cross-consistency verification is performed on multi-channel synchronous sampling data with transient event labels. Combining the continuity constraints of load impedance and the physical feasible region constraints of power factor, it verifies whether the instantaneous power value jointly calculated by the voltage sampling channel and the current sampling channel falls within a physically reasonable range. The cross-verification decision threshold is adaptively adjusted according to the transient event labels. The verification benchmark parameters are continuously corrected based on the long-term energy drift trend information fed back from the energy accumulation step. The verification results are output in the form of a deviation vector containing the ratio deviation component and the angle deviation component. The statistical characteristics accumulated during long-term operation are periodically fed back to the baseline modeling step to update the nonlinear error characteristic baseline model.

[0022] S4, Transient sensing dynamic error compensation based on cross-validation deviation: Based on the direction and magnitude of the deviation vector, and combined with the nonlinear error characteristic baseline model, the dynamic compensation coefficient of the current sampling period is calculated in real time. After performing dynamic error compensation on the voltage sampling data and current sampling data, the instantaneous power is calculated, and the compensation correction amount of each sampling period is evaluated. When the compensation correction amount exceeds the preset level, an instruction to increase the sampling density is issued to the synchronous sampling step. When the compensation correction amount of multiple consecutive sampling periods is lower than the preset level, an instruction to restore the normal sampling density is issued.

[0023] S5, Energy Accumulation and Output with Trend Monitoring: The compensated instantaneous power value is integrated over time to accumulate energy. An energy increment trend monitoring mechanism is maintained within a sliding time window to detect the systematic cumulative deviation trend associated with a specific load switching mode. This trend is fed back to the cross-consistency verification step to continuously correct the verification reference parameters. The finally confirmed energy value is converted into an energy pulse output according to the pulse constant.

[0024] It should be noted that the above five steps form a forward data chain with S1→S2→S3→S4→S5 as the main processing path. Three feedback loops are superimposed on this chain: the short-cycle feedback loop (S4→S2) allows for rapid adjustment of the sampling density based on the compensation results; the medium-cycle feedback loop (S5→S3) allows for correction of the cross-validation benchmark based on the energy accumulation trend; and the long-cycle feedback loop (S3→S1) allows for updating the transformer error baseline model based on the verification statistics. These three feedback loops correspond to adaptive adjustment time scales at the sampling period level, minute-to-hour level, and day-to-week level, respectively, forming a multi-scale nested closed-loop structure. The computational complexity of all the above steps is compatible with the processing capabilities of existing mature metering chips and can be executed in real-time on metering chips with integrated digital signal processing modules, without requiring additional dedicated processing hardware.

[0025] In an optional embodiment, baseline modeling of the nonlinear error characteristics over a wide load range further includes: during the installation and commissioning phase, a portable programmable precision load source and a reference standard energy meter of at least 0.02 grade are used to complete the process. The portable precision load source outputs each steady-state load in stages within the range of 1% to 400% of the base current according to a pre-programmed scan sequence, and the reference standard energy meter provides the true error value under each operating condition.

[0026] After each load level stabilizes, the corresponding ratio and angle difference data are collected as steady-state error feature points. Between adjacent load levels, rapid switching is performed according to the ascending path from light load to heavy load and the descending path from heavy load to light load, respectively, to collect transient error response data and record the transient offset direction and recovery time characteristics of the transformer ratio and angle difference at different switching rates. The steady-state error feature points and transient error response data are combined to construct a nonlinear error characteristic baseline model with load level and load switching path as dual index dimensions. The entire scanning and modeling process is automatically completed in one go during the installation and commissioning phase.

[0027] It should be noted that the nonlinear error characteristic baseline model uses a two-dimensional piecewise linear interpolation lookup table as its basic structure. The first dimension is the discretized grid points of the load current level, and the second dimension is the load switching path type. The corresponding steady-state ratio difference, steady-state angle difference, peak offset of transient error response, and recovery time constant are stored on each grid node.

[0028] In the extremely low load range of 1% to 5% of the basic current, the uncertainty of the acquired data is high due to the deviation of the transformer operating point from the linear region. The baseline model assigns a lower confidence level to the grid nodes in this range. Subsequent S4 dynamic compensation appropriately constrains the compensation adjustment range when referencing these nodes to reduce the risk of overcompensation. This structure enables the baseline model to not only cover steady-state conditions but also to account for the transient error differences caused by the transformer hysteresis effect under different switching directions, providing a refined reference benchmark related to the switching path for subsequent S3 cross-checking and S4 dynamic compensation.

[0029] In an optional embodiment, real-time load status analysis further includes: continuously acquiring current sampling channel data at a preset reference sampling frequency; calculating the gradient of the current root mean square (RMS) value change in each sampling period, i.e., the difference between the current RMS value in the current sampling period and the current RMS value in the previous sampling period divided by the sampling period; and determining that the load has entered a transient switching process when the gradient continuously exceeds the transient switching identification threshold of 10% per millisecond of the rated current for three or more consecutive sampling points. The switching direction is determined based on the positive or negative direction of the current amplitude change, the current change rate is determined based on the absolute value of the gradient, and the steady-state load level before switching is determined based on the average current value within the steady-state time window before switching.

[0030] It should be noted that the transient event labels generated by the load state analysis serve as context information for the forward transmission from S2 to S3, enabling the cross-validation step to adaptively adjust the validation sensitivity based on the specific characteristics of the transient event, thus demonstrating the positive coupling relationship between the sampling step and the validation step.

[0031] In an optional embodiment, the dynamic adjustment of the sampling frequency further includes: when a command to increase the sampling density is received, increasing the sampling frequency from the reference sampling frequency to an encrypted sampling frequency that is 3 times the reference sampling frequency and maintaining it within the encryption duration window; and when a command to restore the normal sampling density is received, restoring it to the reference sampling frequency.

[0032] The length of the encryption duration window is adaptively determined based on the current change rate of the most recent transient switching event, with a recommended range of 50 milliseconds to 500 milliseconds. The greater the current change rate, the longer the window.

[0033] It should be noted that the dynamic adjustment of the sampling frequency is the direct execution link of the S4→S2 short-cycle reverse coupling loop, which enables the actual needs of the downstream compensation step to drive the resource allocation strategy of the upstream sampling step in reverse, replacing the traditional approach of relying on high-precision current transformer hardware to suppress the impact of insufficient sampling.

[0034] In an optional embodiment, cross-consistency verification further includes: Based on the sampling data of the voltage sampling channel in the most recent eight or more consecutive power frequency cycles in a steady state, the root mean square value of voltage and the root mean square value of current are calculated in each power frequency cycle. The load impedance amplitude reference value is obtained by dividing the root mean square value of voltage by the root mean square value of current, and the impedance phase angle reference value is obtained by subtracting the fundamental phase angle of current from the fundamental phase angle of voltage. The above two reference values ​​are used as the continuity constraint benchmark.

[0035] Within each sampling period, the equivalent load impedance magnitude and equivalent impedance phase angle of the current period are calculated from the current sampling channel data and the voltage sampling channel data using the same method. Then, the impedance magnitude deviation rate (i.e., the current equivalent impedance magnitude divided by the reference impedance magnitude and then subtracted by 1) and the phase angle deviation (i.e., the current equivalent impedance phase angle minus the reference impedance phase angle) are calculated.

[0036] When the load is in steady state, the threshold for judging the impedance amplitude deviation rate is set to 2%, and the threshold for judging the phase angle deviation is set to 1 degree. When the load is switching transiently, the threshold for impedance amplitude deviation rate is relaxed to 15% and the threshold for phase angle deviation is relaxed to 5 degrees according to the current change rate, so as to adapt to the reasonable variation range of load parameters during the transient transition.

[0037] It should be noted that the threshold relaxation during transient switching is to accommodate the normal changes in the load impedance itself. However, the relaxation of the threshold does not mean the loss of the ability to detect transformer errors. The reason is that the equivalent impedance trajectory caused by normal load changes follows a physically achievable load characteristic curve, while the transient ratio and angle errors introduced by the hysteresis effect of the transformer core superimpose a set of systematic offset components on the equivalent impedance plane that are independent of the load and depend only on the magnetization state of the current transformer.

[0038] The specific separation method of cross-validation is as follows: based on the switching direction and current change rate in the transient event label transmitted by S2, the expected offset of transient error stored in the grid node corresponding to the switching path is retrieved in the lookup table of the baseline model. Then, the expected offset is subtracted from the total deviation of the measured equivalent impedance. The residual part after subtraction is attributed to normal load changes and is not included in the compensation. The subtracted expected offset component is extracted as a deviation vector after the consistency with the measured deviation direction is confirmed.

[0039] When the measured deviation direction is inconsistent with the expected direction of the baseline model, it indicates that the current deviation is mainly caused by atypical load changes rather than transient error of the transformer. At this time, the deviation vector output is zero to avoid false compensation, so that the cross-verification can simultaneously receive two feedback information from the transient event tag from S2 and the long-term drift trend from S5. The transient error of the transformer is transformed from an unobservable measure to an observable measure by utilizing the physical conservation constraint between the voltage channel and the current channel.

[0040] In an optional embodiment, the output of the deviation vector further includes: since the grid voltage is relatively stable during the transient switching period, the ratio difference of the current transformer causes the measured current amplitude to deviate from the true value, causing the equivalent impedance amplitude to deviate in the opposite direction relative to the true load impedance. Therefore, the impedance amplitude deviation rate is taken as a negative value and extracted as the ratio difference deviation component to reflect the amplitude transformation error introduced by the current transformer in the current sampling period.

[0041] Similarly, the phase difference of the current transformer causes the measured current phase to deviate from the true value, resulting in a reverse deviation of the equivalent impedance phase angle. Therefore, the negative value of the phase angle deviation is extracted as the phase difference deviation component to reflect the phase transformation error introduced by the current transformer in the current sampling period.

[0042] The ratio error and angle error components are combined into an error vector, which is continuously output periodically during transient switching to track the dynamic process of the transient error of the transformer.

[0043] It should be noted that the dual-component structure of the deviation vector enables the S4 dynamic compensation step to apply precise corrections to the two independent dimensions of ratio difference and angle difference, while the continuous output per sampling period ensures the real-time tracking capability of the rapid changes in transient error.

[0044] In an optional embodiment, the calculation of the dynamic compensation coefficient further includes: The error characteristics in the nonlinear error characteristic baseline model corresponding to the current load level and the current switching path are used as the basic compensation reference. This basic compensation reference includes the basic ratio difference compensation factor and the basic angle difference compensation factor.

[0045] The ratio deviation component in the deviation vector is added to the basic ratio deviation compensation factor to obtain the dynamic ratio deviation compensation factor, and the angle deviation component in the deviation vector is added to the basic angle deviation compensation factor to obtain the dynamic angle deviation compensation factor. Together, they constitute the dynamic compensation coefficient of the current sampling period.

[0046] The current sampling data is corrected for amplitude by applying a dynamic ratio difference compensation factor and for phase by applying a dynamic angle difference compensation factor. Then, the power is multiplied with the voltage sampling data to obtain the compensated instantaneous power value. During steady-state operation of the load, the deviation vector amplitude approaches zero, ensuring the dynamic compensation coefficient remains consistent with the basic compensation reference. During transient switching, the dynamic compensation coefficient is adjusted in real-time, driven by the deviation vector, on a sampling cycle basis.

[0047] The evaluation method for the compensation correction amount of a single sampling period is as follows: After performing error compensation on the same set of sampled data using the dynamic compensation coefficient and the basic compensation reference respectively, the instantaneous power is calculated. The absolute value of the difference between the two power values ​​is the compensation correction amount for that period. When the correction amount exceeds one-thousandth of the rated apparent power, an instruction to increase the sampling density is issued to the synchronous sampling step. When the correction amount of 5 or more consecutive sampling periods is lower than the threshold, an instruction to restore the normal sampling density is issued.

[0048] It should be noted that the calculation of the dynamic compensation coefficient depends on both the baseline model provided by S1 and the real-time deviation vector provided by S3. The baseline model provides a priori error estimate associated with the load condition, and the deviation vector provides the real-time correction amount for the current sampling period. The two work together to form a collaborative working relationship between the three steps S1, S3 and S4.

[0049] In an optional embodiment, the power increment trend monitoring further includes: continuously recording the power increment value of each sampling period within a 60-minute sliding time window and associating it with the corresponding load switching event type; performing statistical analysis on the power increment values ​​with the same load switching event type to detect a persistent unidirectional offset trend; and when a systematic offset trend associated with a specific load switching mode is detected, feeding back the direction and magnitude characteristics of the offset trend as a cumulative deviation trend to the cross-consistency verification step to correct the judgment benchmark in the verification benchmark parameters corresponding to the load switching mode.

[0050] It should be noted that the monitoring of incremental electricity trends constitutes the information source of the periodic feedback loop in S5→S3. It uses the long-term statistical regularity contained in the cumulative electricity data at the end to refine the cross-verification criteria of the intermediate links, thereby realizing the continuous optimization of micro-verification decisions based on macro-accumulated information.

[0051] In an optional embodiment, the dynamic update of the nonlinear error characteristic baseline model further includes: the cross-consistency verification step continuously records the statistical distribution of deviation vectors under different load levels, different switching paths, and different switching rates during operation. When the cumulative running time reaches a model update cycle of 168 hours, the trend component reflecting the gradual change of the transformer error characteristics in the statistical distribution of deviation vectors is extracted as the model update increment and sent back to the baseline modeling step. The baseline modeling step performs progressive correction on the ratio difference, angle difference, and transient response parameters of each grid node in the nonlinear error characteristic baseline model lookup table according to the model update increment, so that the model continuously tracks the evolution of the error characteristics of the transformer caused by aging and environmental changes.

[0052] It should be noted that the dynamic update of the baseline model constitutes the closed loop of the long-cycle feedback loop from S3 to S1, so that the transformer error model is no longer a one-time static calibration at the factory, but a dynamic model that is continuously optimized in actual operation.

[0053] To comply with the requirements of current electricity metering management regulations regarding the sealing management of metering parameters, each dynamic update of the baseline model is subject to a maximum allowable adjustment range limit. In a single update, the parameter change of any grid node must not exceed the preset upper limit of the initial calibration value of that node. Furthermore, the timestamp of each update, the parameter differences before and after the update, and the statistical basis for triggering the update are all written to an immutable log storage area for verification by the metering management department.

[0054] Example 2, please refer to Figure 2 This invention provides a technical solution: a transient nonlinear error compensation system for an electrical energy metering box transformer, applicable to the aforementioned transient nonlinear error compensation method for an electrical energy metering box transformer, comprising: Baseline modeling unit 1 is used to perform nonlinear error characteristic baseline modeling for current transformers and voltage transformers in the power metering box with a wide load range. It establishes a nonlinear error characteristic baseline model with a two-dimensional piecewise linear interpolation lookup table as the basic structure and stores it in the updatable storage area of ​​the metering chip. It receives long-term verification statistical features from cross-verification unit 3 to dynamically update the nonlinear error characteristic baseline model.

[0055] The adaptive sampling unit 2 is used to perform synchronous sampling of the voltage sampling channel and the current sampling channel, perform real-time load state analysis on the sampled data stream and attach transient event tags when a transient switching process is detected, and dynamically adjust the sampling frequency according to the sampling density adjustment command fed back from the dynamic compensation unit 4.

[0056] The cross-validation unit 3 is used to perform cross-consistency verification on multi-channel synchronous sampling data with transient event labels using the physical conservation constraint relationship between voltage and current. It adaptively adjusts the verification judgment threshold according to the transient event labels, continuously corrects the verification benchmark parameters according to the long-term power drift trend information fed back from the power accumulation unit 5, outputs the verification results to the dynamic compensation unit 4 in the form of a deviation vector, and sends the long-term verification statistical characteristics back to the baseline modeling unit 1.

[0057] The dynamic compensation unit 4 is used to calculate the dynamic compensation coefficient in real time based on the direction and magnitude of the deviation vector and the baseline model of nonlinear error characteristics, perform dynamic error compensation on the sampled data and calculate the instantaneous power, evaluate the compensation correction amount and generate a sampling density adjustment command based on this and feed it back to the adaptive sampling unit 2.

[0058] The power accumulation unit 5 is used to integrate the compensated instantaneous power value over time to accumulate power energy. It monitors the trend of power increment change within the sliding time window and feeds back the accumulated deviation trend to the cross-validation unit 3. It converts the finally confirmed power energy value into power pulse output according to the pulse constant.

[0059] The present invention designs the following experiment to verify the technical effect of the above technical solution.

[0060] I. Experimental Objective The effectiveness of the current transformer metering error dynamic compensation method described in this invention in both transient load disturbance suppression and long-term cumulative error control is verified. The performance improvement is quantified by comparing it with traditional calibration schemes and uncompensated schemes, and the independent contribution of each functional module is quantified by ablation experiments.

[0061] II. Experimental Conditions 2.1 The object under test Three 0.5-class current transformers of the same batch and model were selected, with a rated transformation ratio of 100A / 5A and a rated basic current Ib=100A, and were numbered CT-01, CT-02, and CT-03. All experimental data were taken as the arithmetic mean of the three tested transformers.

[0062] 2.2 Test Equipment (1) Standard current transformer: 0.02 class, rated ratio 100A / 5A, working range covers 1%Ib~400%Ib (1A~400A).

[0063] (2) Programmable load impact source: Output range 1%Ib~400%Ib (1A~400A), supports arbitrary load curve programming, step switching time ≤50ms, output accuracy ±0.5%.

[0064] (3) High-precision data acquisition system: sampling rate ≥10kS / s, A / D resolution 24bit, supporting synchronous acquisition of ratio difference and angle difference.

[0065] (4) Standard energy meter: 0.02 grade active energy meter, used as the basis for energy accumulation measurement.

[0066] (5) Temperature control environment chamber: temperature control range -25℃~+55℃, temperature control accuracy ±0.5℃.

[0067] 2.3 Test Environment The ambient temperature is set to 23±2℃, the relative humidity to 45%~65%RH, and the power frequency to 50±0.2Hz.

[0068] III. Experimental Design Five experimental schemes were set up, covering comparative experiments and ablation analysis.

[0069] Option 1 (Traditional calibration scheme, control group A): The traditional factory static multi-point calibration method is adopted. After completing the one-time error correction under rated operating conditions, it is put into operation. No dynamic compensation or baseline update operation is performed during operation.

[0070] Option 2 (Uncompensated Option, Control Group B): The current transformer under test is not loaded with any error compensation algorithm, and the original measurement output is used directly for measurement as the basic performance reference benchmark.

[0071] Scheme 3 (Complete Scheme of the Invention, Experimental Group): Fully implement the method described in this invention. Stage 1 (Initial Baseline Modeling, corresponding to step S1 of claim): Multi-point calibration covering the full load range from 1% Ib to 400% Ib to establish an initial error compensation model; Stage 2 (Real-time Transient Compensation, corresponding to the forward processing chain of steps S2 to S4): Perform dynamic error correction based on the load mutation detection results; Stage 3 (Medium-to-Long Cycle Dual Feedback Adaptive Update, corresponding to the three feedback loops of steps S4→S2, S5→S3, and S3→S1): Perform periodic baseline adaptive updates to continuously optimize compensation accuracy.

[0072] Option 4 (Disable transient compensation, ablation group A): Based on Option 3, disable the Phase 2 real-time transient compensation module, and retain only the Phase 1 baseline modeling and Phase 3 dual-feedback adaptive update to quantify the independent contribution of the Phase 2 module.

[0073] Option 5 (Disable Adaptive Update, Ablation Group B): Based on Option 3, disable the long-term dual-feedback adaptive update module in Phase 3, and retain only the baseline modeling in Phase 1 and the real-time transient compensation in Phase 2 to quantify the independent contribution of the Phase 3 module.

[0074] IV. Experimental Data and Analysis 4.1 72-hour transient response test Based on steady-state operation at 100% Ib, a step load impact from 100% Ib to 350% Ib was applied every 2 hours. Each impact lasted 10 seconds before returning to steady state, and this process was repeated for 72 hours. The peak transient ratio difference at each impact moment (the maximum value among all impact events) and the cumulative energy error δE3 over 72 hours were recorded. The specific results are shown in the table below:

[0075] Table 1: Results of Transient Ratio Peak Value and Cumulative Energy Error Scheme 3 exhibits a peak transient ratio difference of only 0.47% under load shock, a 74.3% reduction compared to Scheme 1's 1.83%; the 72-hour cumulative energy error decreases from +0.273% to +0.025%, a reduction of 90.8%. These data demonstrate that the Phase 2 real-time transient compensation module can quickly correct ratio difference offsets when load changes occur, effectively suppressing the cumulative transmission of transient errors to energy metering results.

[0076] 4.2 30-day extended stability test The system operates cyclically for 30 days based on a typical daily load curve, with the daily load curve set as follows: 8 hours of 30% Ib during the valley period, 8 hours of 70% Ib during the flat period, and 8 hours of 120% Ib during the peak period. The cumulative daily energy error (daily error) is recorded daily, and the total cumulative energy error δE over 30 days is calculated. 30 The specific results are shown in the table below:

[0077] Table 2: Daily Error Results of Cumulative Electricity Consumption Scheme 1 exhibits a slow daily error drifting between +0.089% and +0.110%, accumulating to +2.97% over 30 days, reflecting the inherent deficiency of traditional static calibration in lacking adaptive correction capabilities during long-term operation. Scheme 2, without any compensation, shows a daily error that continuously increases from +0.217% to +0.312%, accumulating to a high +7.84% over 30 days, reflecting the natural error drift level of a Class 0.5 current transformer without compensation. Scheme 3, through continuous optimization of the Phase 3 adaptive update mechanism, shows a significant convergence trend, with the daily error gradually converging from +0.011% to +0.002%, accumulating to only +0.14% over 30 days, a reduction of 95.3% compared to Scheme 1 and 98.2% compared to Scheme 2.

[0078] 4.3 Ablation Experiment Schemes 3, 4, and 5 were operated synchronously under the same daily load curve conditions as in Section 4.2, and the daily error and 30-day cumulative energy error for each scheme were recorded. The peak transient ratio difference was obtained by applying the step impact test described in Section 4.1 separately during the daily load curve operation interval, which does not affect the daily error statistics. The results are shown in the table below:

[0079] Table 3: Comparison of Transient Ratio Error Peak Value and Cumulative Energy Error Analysis of Ablation Group A (Scheme 4, Phase 2 Transient Compensation Disabled): After disabling the Phase 2 module, the peak transient ratio difference deteriorated from 0.47% to 0.83%, a degradation of 76.6%, indicating that the Phase 2 module is the core mechanism for suppressing transient load impacts. However, since the Phase 3 adaptive update is still running, the long-term daily error continues to converge (from +0.015% to +0.003%), and the 30-day cumulative error of +0.18% is only 28.6% higher than that of Scheme 3, indicating that the long-term cumulative performance is mainly guaranteed by the Phase 3 module.

[0080] Analysis of Ablation Group B (Scheme 5, Stage 3 Adaptive Update Disabled): After disabling the Stage 3 module, the peak transient ratio difference remained unchanged at 0.47%, indicating that the transient response performance was entirely provided by the Stage 2 module. However, the daily error lost its ability to converge continuously. After briefly dropping to +0.008% on day 7, it gradually rebounded with baseline drift, reaching +0.015% on day 14 and further increasing to +0.019% on day 30, showing a divergent trend. The 30-day cumulative error of +0.42% was 200.0% worse than that of Scheme 3, indicating that the Stage 3 module is the key mechanism for maintaining the continuous convergence of long-term compensation accuracy.

[0081] 4.4 Verification of the Baseline Adaptive Update Process A detailed record of the baseline update process of the Phase 3 module during the 30-day operation of Scheme 3 is provided. In the long-term dual feedback mechanism of Phase 3, the medium-term feedback submodule begins to make continuous minor adjustments to the compensation parameters after 24 hours of operation, and the long-term feedback submodule triggers the first overall baseline update after 72 hours of operation. The following is a record of the parameter changes during the first long-term update at the 72nd hour.

[0082] At the 100% Ib load point, the ratio difference compensation coefficient was corrected from +0.38% to +0.3989%, a change of +0.0189 percentage points, representing 4.97% of the initial value, close to but not exceeding the single update ratio limit of 5%. The angle difference compensation at the same load point was corrected from 8.2′ to 8.5′, a change of 0.3′, representing 3.66% of the initial value. The update process was completed smoothly within a single metering cycle, with transient ratio difference fluctuations not exceeding 0.02% during the switch, and no observable additional metering error was introduced.

[0083] V. Experimental Conclusions (1) The complete solution of the present invention (Solution 3) has a transient ratio difference peak value of only 0.47% under load impact conditions, which is 74.3% lower than the 1.83% of the traditional calibration solution, and the cumulative power error reduction over 72 hours reaches 90.8%. The compensated transient measurement error is still controlled within the rated ratio difference limit of 0.5 (±0.5%), ensuring that the tested transformer maintains its rated accuracy level without degradation under transient load disturbance conditions.

[0084] (2) In the 30-day extended test, the cumulative power error of the complete scheme of the present invention was only +0.14%, which was 95.3% lower than the +2.97% of the traditional calibration scheme and 98.2% lower than the +7.84% of the uncompensated scheme. Moreover, the daily error showed a continuous convergence trend from +0.011% to +0.002%, which verified the effectiveness of the long-term dual feedback adaptive update mechanism in the third stage in suppressing long-term error drift and driving continuous optimization of compensation accuracy.

[0085] (3) Ablation experiments show that the Phase 2 real-time transient compensation module mainly contributes to the rapid suppression of transient load impacts (the transient peak value deteriorated by 76.6% after shutdown, and only increased by 28.6% cumulatively over 30 days), while the Phase 3 medium-to-long-term dual-feedback module mainly contributes to the continuous convergence control of long-term cumulative errors (the cumulative error deteriorated by 200.0% over 30 days after shutdown, but the transient peak value remained unaffected). The two modules act on different time scales, complement each other, and together constitute a complete multi-time scale dynamic compensation system.

[0086] (4) The baseline adaptive update process is smooth and controllable. The single update amplitude of 4.97% is close to but does not exceed the safety threshold of 5%, and the angle difference update amplitude is 3.66%. No additional measurement error is introduced during the switching period, which verifies the engineering stability and safety of the progressive baseline update strategy of the present invention.

[0087] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A method for compensating for transient nonlinear errors in current transformers of an energy metering box, characterized in that, Includes the following steps: S1 Baseline Modeling Steps: Perform nonlinear error characteristic baseline modeling on the current transformers and voltage transformers in the power metering box to establish a nonlinear error characteristic baseline model covering the ratio difference and angle difference of each steady-state load point, as well as the transient transition error variation trend under different load switching directions and switching rates. S2 Synchronous Sampling Steps: Perform synchronous sampling on the voltage sampling channel and the current sampling channel, and perform real-time load state analysis on the sampled data stream to identify steady-state operating state and transient switching state; S3 cross-validation steps: Perform cross-consistency verification on multi-channel synchronous sampling data using the physical conservation constraint relationship between voltage and current. Combine the load impedance continuity constraint and the power factor physical feasible region constraint, and output the verification result in the form of a deviation vector containing the ratio deviation component and the angle deviation component. S4 Dynamic Compensation Step: Calculate the dynamic compensation coefficient for the current sampling period in real time based on the deviation vector and the nonlinear error characteristic baseline model, and calculate the instantaneous power after performing dynamic error compensation on the sampled data; S5 power accumulation step: The compensated instantaneous power value is integrated over time to accumulate power energy and output a power pulse.

2. The method according to claim 1, characterized in that, The dynamic compensation step evaluates the compensation correction amount for each sampling period. When the compensation correction amount exceeds a preset level, it sends an instruction to the synchronous sampling step to increase the sampling density. When the compensation correction amount for multiple consecutive sampling periods is lower than the preset level, it sends an instruction to the synchronous sampling step to restore the normal sampling density. The power accumulation step maintains a power increment trend monitoring mechanism within a sliding time window, detects the systematic cumulative deviation trend associated with a specific load switching mode, and feeds the cumulative deviation trend back to the cross-validation step for continuous correction of the validation benchmark parameters. The cross-validation step periodically transmits the statistical characteristics of power conservation deviation under different load states and switching paths accumulated during long-term operation back to the baseline modeling step. The baseline modeling step dynamically updates the nonlinear error characteristic baseline model based on the transmitted long-term validation statistical characteristics.

3. The method according to claim 1, characterized in that, The baseline modeling steps include: During the initial installation phase, an automated load scanning method is used to gradually change the load level within the load range of the basic current at a preset load step interval. After each load level stabilizes, the corresponding ratio difference and angle difference data are collected as steady-state error feature points. A fast switching operation is performed between adjacent load levels. Transient error response data during the switching transition process are collected according to the rising path from light load to heavy load and the falling path from heavy load to light load, respectively. The transient offset direction and recovery time characteristics of the transformer ratio difference and angle difference are recorded at different switching rates. The steady-state error feature points and the transient error response data are combined to construct the nonlinear error characteristic baseline model. The nonlinear error characteristic baseline model uses load level and load switching path as dual index dimensions and a two-dimensional piecewise linear interpolation lookup table as its basic structure. Each grid node stores the steady-state ratio difference, steady-state angle difference, transient peak deviation, and recovery time constant. The nonlinear error characteristic baseline model is stored in the updatable storage area of ​​the metering chip.

4. The method according to claim 1, characterized in that, When the synchronous sampling step detects that the load has entered a transient switching process, it adds a transient event tag to the sampled data. The transient event tag includes the switching start time, switching direction, current change rate, and steady-state load level before switching. The cross-validation step adaptively adjusts the cross-validation judgment threshold according to the transient event tag.

5. The method according to claim 4, characterized in that, The real-time load status analysis of the sampled data stream includes: The data of the current sampling channel is continuously collected at a preset reference sampling frequency, and the gradient of the change of the root mean square value of the current between adjacent sampling points is calculated. The changing gradient is compared with a preset transient switching identification threshold. When the changing gradient continuously exceeds the transient switching identification threshold within a series of consecutive sampling points, it is determined that the current load has entered a transient switching process. The switching direction is determined based on the positive or negative direction of the current amplitude change, the current change rate is determined based on the absolute value of the change gradient, and the steady-state load level before the switching is determined based on the average current value within the most recent steady-state time window before the switching occurs.

6. The method according to claim 5, characterized in that, The sampling frequency of the synchronous sampling step is dynamically adjusted according to the sampling density adjustment command fed back from the dynamic compensation step, including: When a command to increase sampling density is received, the current sampling frequency is increased to a preset encrypted sampling frequency, and the encrypted sampling frequency is maintained within a preset encryption duration window; When a command to restore normal sampling density is received, the current sampling frequency is restored to the reference sampling frequency; The encryption sampling frequency is higher than the reference sampling frequency, and the length of the encryption duration window is adaptively determined based on the current change rate of the most recent transient switching event. The greater the current change rate, the longer the encryption duration window.

7. The method according to claim 1, characterized in that, The cross-validation step includes: Based on the sampling data of the voltage sampling channel and the current sampling channel in the most recent consecutive power frequency cycles in a steady state, the load impedance amplitude reference value is calculated by dividing the voltage root mean square value by the current root mean square value, and the impedance phase angle reference value is calculated by subtracting the current fundamental phase angle from the voltage fundamental phase angle. The load impedance amplitude reference value and the impedance phase angle reference value are used as the continuity constraint benchmark. Within each sampling period, the equivalent load impedance magnitude and equivalent impedance phase angle of the current period are calculated using the sampling data from the current sampling channel and the sampling data from the voltage sampling channel in the same way. The impedance magnitude deviation rate is obtained by dividing the current equivalent impedance magnitude by the reference value of the load impedance magnitude and then subtracting one. The impedance phase angle deviation amount is obtained by subtracting the reference value of the impedance phase angle from the current equivalent impedance phase angle. The impedance magnitude deviation rate and the impedance phase angle deviation amount are then checked to see if they exceed the adaptive judgment threshold. When the load is in a steady-state operation, the adaptive judgment threshold is set to a smaller value to maintain a higher verification sensitivity; when the load is in a transient switching process, the adaptive judgment threshold is relaxed and adjusted according to the current change rate and switching direction to adapt to the reasonable range of load parameter changes during the transient transition. The step of outputting the verification result in the form of a deviation vector includes: Since the grid voltage is relatively stable during transient switching, the ratio difference of the current transformer causes the equivalent impedance amplitude to deviate in the opposite direction. The negative value of the impedance amplitude deviation rate is extracted as the ratio difference deviation component. The negative value of the impedance phase angle deviation is extracted as the angle difference deviation component; Before extracting the deviation vector, the measured equivalent impedance deviation direction is confirmed to be consistent with the expected transformer error deviation direction of the corresponding switching path in the nonlinear error characteristic baseline model. When the two directions are consistent, the ratio difference deviation component and the angle difference deviation component are combined into the deviation vector and continuously output in each sampling period during the transient switching process. When the two directions are inconsistent, the deviation vector is set to zero to avoid false compensation.

8. The method according to claim 2, characterized in that, The dynamic compensation step includes: The error characteristics in the nonlinear error characteristic baseline model corresponding to the current load level and the current switching path are used as the basic compensation reference; The ratio deviation component in the deviation vector is superimposed on the ratio compensation factor in the basic compensation reference, and the angle deviation component in the deviation vector is superimposed on the angle compensation factor in the basic compensation reference to obtain the dynamic compensation coefficient for the current sampling period. During steady-state operation under load, the magnitude of the deviation vector approaches zero, and the dynamic compensation coefficient remains consistent with the basic compensation reference. During transient switching transitions, the dynamic compensation coefficient is adjusted in real time on a sampling cycle driven by the deviation vector. The evaluation method for the compensation correction amount is as follows: after performing error compensation on the same set of sampled data in the current sampling period using the dynamic compensation coefficient and the basic compensation reference respectively, the instantaneous power is calculated, and the absolute value of the difference between the two instantaneous power values ​​is the compensation correction amount for the current sampling period. The mechanism for maintaining the monitoring of the incremental trend of electrical energy within the sliding time window includes: Within a preset sliding time window, the power increment value of each sampling period is continuously recorded, and each power increment value is associated with the corresponding load switching event type. Statistical analysis is performed on the incremental power values ​​with the same load switching event type within the sliding time window to detect whether there is a continuous unidirectional offset trend. When a systematic offset trend associated with a specific load switching mode is detected, the direction and magnitude characteristics of the offset trend are fed back to the cross-validation step as the cumulative deviation trend, in order to correct the judgment benchmark in the validation benchmark parameters corresponding to the load switching mode.

9. The method according to claim 2, characterized in that, The baseline modeling step dynamically updates the nonlinear error characteristic baseline model based on the returned long-term verification statistical characteristics, including: The cross-validation step continuously records the statistical distribution of the deviation vector under different load levels, different switching paths, and different switching rates during the operation. When the accumulated running time reaches the preset model update cycle, the trend component reflecting the gradual change of the transformer error characteristics in the statistical distribution of the deviation vector is extracted and used as the model update increment to be fed back to the baseline modeling step. The baseline modeling step progressively corrects the ratio difference, angle difference, and transient response parameters of each grid node in the lookup table of the nonlinear error characteristic baseline model according to the model update increment. In a single update, the parameter change of any grid node does not exceed the preset upper limit of the initial calibration value of that node, and the timestamp of each update, the parameter difference before and after the update, and the statistical basis for triggering the update are all written to an immutable log storage area.

10. A transient nonlinear error compensation system for an electrical energy metering box transformer, applicable to the method described in any one of claims 1 to 9, characterized in that, include: The baseline modeling unit (1) is used to establish and maintain the baseline model of the nonlinear error characteristics, and is stored in the updatable storage area of ​​the metering chip with a two-dimensional piecewise linear interpolation lookup table as the basic structure. The adaptive sampling unit (2) is used to perform synchronous sampling on the voltage sampling channel and the current sampling channel, and to perform real-time load state analysis on the sampled data stream to identify steady-state operation and transient switching states. The cross-validation unit (3) is used to perform cross-consistency verification on multi-channel synchronous sampling data using the physical conservation constraint relationship between voltage and current, and outputs the verification result to the dynamic compensation unit (4) in the form of a deviation vector. The dynamic compensation unit (4) is used to calculate the dynamic compensation coefficient in real time based on the deviation vector and the nonlinear error characteristic baseline model, and to calculate the instantaneous power after performing dynamic error compensation on the sampled data; The power accumulation unit (5) is used to perform time integration on the compensated instantaneous power value to accumulate power energy and output power pulses.

Images