A method and system for monitoring power quality in an electricity metering box
By reconstructing the virtual synchronous sampling sequence and frequency correction, the deviation problem of harmonic monitoring in power metering boxes under asynchronous sampling is solved, realizing high-precision power quality monitoring under low sampling rate and limited computing power, which is suitable for end-point power metering boxes in industrial parks.
Patent Information
- Application Number
- CN202611601003.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-16
- Publication Date
- 2026-11-17
AI Technical Summary
Existing power metering boxes cannot accurately distinguish between interharmonics and integer harmonics under asynchronous sampling conditions, resulting in deviations in harmonic amplitude and total harmonic distortion rate of voltage. Furthermore, due to limitations in hardware cost and power consumption, the sampling rate or computing power cannot be increased to improve monitoring accuracy.
By obtaining the change time of the circuit breaker auxiliary contacts, determining the monitoring window by combining the voltage zero-crossing sequence, reconstructing the virtual synchronous sampling sequence, performing frequency domain transformation and spectral peak analysis, and combining the sidelobe leakage degree for frequency correction, an iterative projection stripping strategy is adopted to reduce the risk of spectral leakage and misclassification caused by asynchronous sampling.
Under conditions of low sampling rate and limited computing power, it improves the accuracy and reliability of power quality monitoring, reduces communication bandwidth pressure, and adapts to the hardware constraints of low-power metering chips.
Smart Images

Figure CN123115907A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of power quality monitoring technology, and in particular to a method and system for power quality monitoring of a power metering box. Background Technology
[0002] With the rapid development of industrial parks, a large number of nonlinear loads such as frequency converters and DC charging piles have been connected to the end-point power grid. These loads generate a large number of interharmonics (such as 25 Hz, 75 Hz, etc.) and integer harmonics during operation, leading to a deterioration in power quality. Therefore, high-precision power quality monitoring of power metering boxes is particularly important. However, existing end-point power metering boxes in industrial parks typically embed low-power metering chips. Limited by hardware cost and power consumption, their sampling rate is fixed and low (e.g., 1.6 kHz), and they use a fixed 128-point rectangular window for spectrum analysis. They only have limited computing power and narrowband communication interfaces (such as RS-485), making it impossible to upload raw waveform data for extended periods, and it is also impossible to improve monitoring accuracy by significantly increasing the hardware sampling rate or adopting high-resolution spectrum analysis algorithms.
[0003] In actual operation, the power grid frequency typically fluctuates between 49.5 Hz and 50.5 Hz, causing the voltage zero-crossing moment to be asynchronous with the fixed sampling moment. When using a fixed-length window for Fast Fourier Transform (FFT), this asynchronous truncation can cause severe spectral leakage. In particular, interharmonics such as 25 Hz and 75 Hz are non-orthogonal to the 50 Hz fundamental frequency and integer harmonics in the frequency domain, and the sidelobes of the rectangular window attenuate slowly, resulting in overlap between the spectral leakage components of the interharmonics and the spectral components of the integer harmonics.
[0004] When dealing with multi-source aliasing, existing conventional FFT methods have difficulty reliably distinguishing adjacent components under limited frequency resolution. They are prone to misclassifying interharmonic sidebands as integer harmonics such as the 2nd or 3rd order, which may lead to deviations in the output harmonic amplitude and total harmonic distortion (THD) of the voltage, and thus fail to provide accurate data support for power quality assessment. Summary of the Invention
[0005] To address the technical problem of severe harmonic amplitude distortion caused by interharmonic leakage under asynchronous sampling, and the inability to rely on increasing the hardware sampling rate, this application provides a power quality monitoring method and system for power metering boxes.
[0006] In a first aspect, this application provides a method for monitoring the power quality of an electricity metering box, including: acquiring the time of change of the auxiliary contact of a circuit breaker and the incoming current; and determining the event anchor point time within a preset search window after the change time based on the short-time average difference between the absolute values of the instantaneous values of the incoming current before and after the change of the candidate current sampling point. The measured fundamental period is determined based on the voltage zero-crossing sequence, the monitoring window is determined based on the event anchor point time and the voltage zero-crossing sequence, and the virtual synchronous sampling sequence is obtained by interpolation reconstruction. The virtual synchronous sampling sequence is transformed in the frequency domain, and candidate frequencies are determined by combining the spectral peak and the preset a priori frequency band. The sidelobe leakage degree is calculated based on the main lobe and left and right sidelobe energies corresponding to the candidate frequencies. The degree of sidelobe leakage is used to determine whether to enter the frequency correction evaluation. The candidate frequencies that meet the correction conditions are corrected a finite number of times. Based on the integer multiple relationship between the corrected candidate frequencies and the measured fundamental frequency and the effectiveness of the candidate components, the integer harmonic candidate components and interharmonic candidate components are determined. Using the virtual synchronous sampling sequence as the current time domain remaining sequence, the fundamental component is stripped first, and then the integer harmonic candidate components and interharmonic candidate components that meet the validity conditions are sorted and iteratively projected and stripped to obtain the integer harmonic amplitude set and the interharmonic frequency and amplitude set respectively. The total harmonic distortion rate of the voltage is calculated based on the integer harmonic amplitude set.
[0007] Optionally, the method for determining the event anchor point time includes: Within the preset search window after the auxiliary contact of the circuit breaker changes position, calculate the short-time average of the absolute values of the instantaneous values of the incoming current at preset points before and after the candidate current sampling point, and determine the per-unit value of the current step based on the ratio of the absolute value of the difference between the two short-time averages to the preset rated current reference. When the current step per unit value is detected to be greater than the preset trigger threshold for the first time, a valid switching event is confirmed, and the corresponding sampling point time is determined as the event anchor point time; when there is no sampling point in the search window that meets the trigger condition, no valid switching event anchor point is generated; for candidate sampling points for which complete short-term average calculation data cannot be obtained, the corresponding current step per unit value determination is not performed.
[0008] Optionally, the acquisition of the virtual synchronization sampling sequence includes: The current data processing time is obtained, and the event reference time stamp and event timeliness confidence are determined based on the time difference between the current data processing time and the event anchor time corresponding to the most recent valid switching event. The event reference time stamp is used for event association positioning in the monitoring window, and the voltage zero-crossing sequence is used to determine the phase reference time of virtual synchronous sampling. The phase reference time of virtual synchronous sampling is determined based on the voltage zero-crossing sequence. When there is no valid switching event or the time difference reaches the preset event failure duration, the historical event anchor time is not used as the event reference time stamp of the current monitoring window. Based on the event reference time stamp, the current data processing time, and the coverage area of the original sampling data, a monitoring window containing a preset number of complete fundamental frequency cycles is determined. When a valid event reference time stamp exists and the original sampling data covers a preset number of consecutive complete fundamental frequency cycles after the corresponding event, a preset number of consecutive complete fundamental frequency cycles starting from the first effective voltage positive zero-crossing moment after the event are selected. Otherwise, the monitoring window ends at the last effective voltage positive zero-crossing moment within the coverage area of the current original sampling data, and a preset number of consecutive complete fundamental frequency cycles are selected backward. Based on the voltage zero-crossing moment and the measured period duration of each fundamental frequency cycle, the same number of phase-uniform virtual sampling points are constructed within each fundamental frequency cycle, and a virtual synchronous sampling sequence is obtained through linear interpolation.
[0009] Optionally, the determination of the candidate frequency includes: The amplitude spectrum is obtained by performing a discrete Fourier transform on the virtual synchronous sampling sequence. Frequency points that satisfy the local peak condition and the preset peak threshold are selected, and frequency points to be detected within the preset prior frequency band are selected. At the same time, frequency points to be detected corresponding to preset integer harmonics are set. The above frequency points are merged and deduplicated to form an initial candidate frequency set. The background noise amplitude is estimated based on the initial candidate frequency set, and a candidate frequency set is formed by screening according to the effective amplitude detection condition.
[0010] Optionally, the calculation of the sidelobe leakage degree includes: Based on the frequency resolution of the amplitude spectrum, the main lobe region and the left and right side lobe regions are determined with the candidate frequency as the center. The frequency point sets corresponding to the regions do not overlap. The squares of the frequency point amplitudes in each region are accumulated to obtain the main lobe energy, the left side lobe energy and the right side lobe energy. When the main lobe energy is greater than the preset lower energy limit, the ratio of the sum of the left and right side lobe energies to the main lobe energy is used as the side lobe leakage degree; when the main lobe energy is not greater than the lower energy limit, the corresponding candidate component is marked as an invalid candidate component.
[0011] Optionally, the frequency correction includes: Determine whether the left and right sidelobe regions of the current candidate frequency have common frequency points with the main lobe regions of other reference components; the other reference components include the measured fundamental wave components other than the current candidate components, and the components to be detected that meet the effective amplitude detection conditions and whose amplitude estimates are greater than the current candidate components among the preset integer harmonic detection frequencies. When the common frequency point exists, the current candidate frequency is retained and frequency correction based on the energy difference between the left and right sidelobes is not performed; when the common frequency point does not exist and the sidelobe leakage degree is greater than the preset leakage degree judgment threshold, the trial correction direction is determined according to the energy difference between the left and right sidelobes, the trial frequency is generated with a preset correction step size, and the same amplitude estimation and normalization rules are used to calculate the relevant amplitudes corresponding to the candidate frequency and the trial frequency before correction. When the test frequency is within the preset maximum frequency offset range, does not enter the main lobe region of other identified spectral peaks, and the correlation amplitude corresponding to the test frequency is greater than the correlation amplitude corresponding to the candidate frequency before correction, the correction is accepted. After each correction, the main lobe and left and right side lobe regions are redefined with the updated candidate frequency, the common frequency point judgment is re-executed, and the side lobe leakage degree and the energy difference between the left and right side lobes are recalculated. When the preset maximum number of corrections is reached, the frequency correction condition is no longer met, or the preset correction acceptance condition is not met, the frequency correction ends and the final candidate frequency is retained.
[0012] Optionally, the classification of the candidate components includes: Based on the frequency deviation between the corrected candidate frequency and the measured fundamental frequency, as well as the preset frequency matching tolerance, effective amplitude detection conditions and frequency estimation reliability conditions, the integer harmonic candidate components are determined. Candidate components that do not meet the integer multiple frequency matching condition but meet the effective amplitude detection and frequency estimation reliability conditions are identified as interharmonic candidate components. Candidate components that do not meet the effective amplitude detection condition or the frequency estimation reliability condition are marked as low-confidence candidate components.
[0013] Optionally, the iterative projection stripping includes: First, the fundamental component is projected and stripped. Then, based on the main lobe energy in the spectrum corresponding to the current time domain remaining sequence, the remaining candidate components that meet the validity conditions are sorted. Sine reference sequences and cosine reference sequences are constructed according to the frequencies of the components to be stripped. In-phase components and quadrature components are obtained by correlation summation and binary coefficient solution, and the amplitude and phase are calculated accordingly. Based on the amplitude and phase, the projected components are constructed. When the preset stripping validity condition is met, the current remaining time-domain sequence is updated, and the unprocessed candidate components are reordered.
[0014] Optionally, obtaining the set of integer harmonic amplitudes, the set of interharmonic frequencies and amplitudes, and the total harmonic distortion rate of the voltage includes: Integer harmonic components that pass the validity determination are recorded in the integer harmonic amplitude set, and the frequency and amplitude of interharmonic components that pass the validity determination are recorded in the interharmonic frequency and amplitude set; candidate components that fail the validity determination are marked with low confidence and their initial amplitudes are not directly used as valid harmonic amplitudes. The estimated total harmonic distortion rate of voltage is calculated based on the effective integer harmonic voltage amplitude, and the detection status and corresponding detection limit of the integer harmonics not detected within the preset maximum analysis harmonic order range are recorded.
[0015] Secondly, this application provides an energy quality monitoring system for an energy metering box, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned energy quality monitoring method for the energy metering box is implemented.
[0016] This application has the following technical advantages: 1. This application combines the switching event timescale and the voltage zero-crossing sequence to determine the monitoring window and reconstruct the virtual synchronous sampling sequence. It uses current step to determine the event reference timescale and the voltage zero-crossing sequence to determine the phase reference. Furthermore, it uses the measured fundamental period of the voltage zero-crossing sequence to perform multi-cycle recursion, constructing a virtual sampling timescale corresponding to the measured fundamental phase. This mechanism reduces the impact of spectral leakage caused by asynchronous truncation, provides a stable phase reference for subsequent frequency domain analysis, and requires no increase in the hardware sampling rate throughout, effectively adapting to the hardware constraints of low-power metering chips.
[0017] 2. This application analyzes the main lobe and left and right side lobes corresponding to candidate frequencies, calculates the degree of side lobe leakage, and checks whether the side lobe region of the candidate frequency shares a common frequency with the main lobe region of other reference components. When a common frequency exists, the current candidate frequency is retained, and directional correction is not performed using the energy difference between the left and right side lobes, reducing the risk of incorrect correction when other components interfere with the side lobe information. When the side lobe information meets the correction conditions, the candidate frequency is evaluated through correlation amplitude comparison and a finite number of corrections. Combining spectral boundary processing, the integer multiple relationship between the final candidate frequency and the measured fundamental frequency, and the validity of the candidate components, classification is performed to reduce the risk of misclassification between integer harmonics and interharmonics. The calculation process mainly involves frequency domain transformation, frequency point set comparison, energy accumulation, correlation summation, and a finite number of iterative processes, without the need for large-scale joint optimization calculations, making it suitable for the limited computing power of power metering boxes.
[0018] 3. This application employs an iterative orthogonal projection stripping strategy, sorting components by main lobe energy from largest to smallest. This strategy successively eliminates the projections of components to be stripped in the time domain, reducing the impact of already stripped, stronger components on the estimation of subsequent candidate components and mitigating the risk of weak interharmonics being misclassified as integer harmonics. Simultaneously, it clearly defines the output destinations for integer harmonics and interharmonics. THD is calculated from the effective integer harmonic amplitudes, with interharmonic components recorded separately, improving the consistency between THD calculation results and the definition of integer harmonics. Multi-cycle projection processing helps reduce the interference of non-integer multiple frequency components on the estimation of other components, improving the accuracy and reliability of power quality monitoring results under complex operating conditions.
[0019] 4. Under the stringent constraints of a low sampling rate of 1.6 kHz, limited computing power, and narrowband RS-485 communication, the method of this application achieves power quality feature extraction suitable for low sampling rate and limited computing power conditions. It only requires uploading the corrected set of integer harmonic amplitudes, the set of interharmonic frequencies and amplitudes, and THD and other feature quantities, without having to upload the original waveform for a long time, thus reducing the communication bandwidth pressure and adapting to the constraints of narrowband RS-485 communication. Attached Figure Description
[0020] Figure 1 This is a flowchart of a method for monitoring the power quality of an electricity metering box according to an embodiment of this application. Detailed Implementation
[0021] This application discloses a method for monitoring the power quality of an electricity metering box, referring to... Figure 1 This includes steps S1-S3.
[0022] S1: Determine the monitoring window based on the switching event time stamp and the voltage zero-crossing sequence, and reconstruct the virtual synchronous sampling sequence to establish a fundamental phase reference and reduce the spectral leakage caused by asynchronous sampling.
[0023] This step focuses on the electrical performance testing scenario of end-point electricity metering boxes in industrial parks. The box embeds a low-power metering chip, employs a fixed 128-point rectangular analysis window and a raw sampling rate of 1.6 kHz, and simultaneously connects to frequency converters, DC charging piles, and lighting loads, potentially introducing interharmonics such as 25 Hz and 75 Hz. The metering box is only equipped with RS-485 narrowband communication and limited computing power, making it unable to continuously upload raw waveforms. The grid frequency fluctuates between 49.5 Hz and 50.5 Hz, and the voltage zero-crossing is asynchronous with the sampling time. The fixed window truncation causes harmonic and interharmonic spectral leakage and superposition. The harmonic amplitudes and interharmonic frequencies output by conventional Fast Fourier Transform (FFT) show significant deviations under light loads and multi-source aliasing. Therefore, the core electrical performance testing problem to be solved in this step is: the large deviation in harmonic amplitude caused by interharmonic leakage under asynchronous sampling, and the inability to rely on increasing the hardware sampling rate.
[0024] Event triggering itself does not increase the sampling rate, but only provides a time stamp. If the event detection jitter is large, it will affect the reliability of the timing of the switching event reference, but the voltage fundamental phase is still determined by the voltage zero-crossing sequence. Therefore, this step combines the circuit breaker auxiliary contact status, incoming current mutation, and voltage zero-crossing sequence to locate the event and determine the phase reference, and makes explicit handling of rated current, temperature application, and event failure fallback, so that subsequent frequency domain stripping is based on a stable phase reference. The incoming current step generated by load switching is used to determine the switching event reference time stamp, which is not directly used as the phase zero point of the voltage fundamental; the voltage zero-crossing sequence is used to determine the voltage fundamental phase reference time and the measured duration of each fundamental period, and the two are used together to determine the monitoring window and the virtual synchronous sampling time.
[0025] This embodiment uses power quality monitoring of one phase voltage as an example; other phase voltages can be processed in the same way. The incoming line current is used for switching event detection and is not included in the calculation of the total harmonic distortion (THD). Specifically, the input is the instantaneous sampling sequence of the three-phase voltage. , , Instantaneous sampling sequence of incoming line current Circuit breaker auxiliary contact state sequence and the temperature inside the chamber First, taking the change of auxiliary contact of the circuit breaker as a candidate event, select candidate current sampling points near the time of the change, and calculate the current before and after the candidate sampling points respectively. The short-time average of the instantaneous absolute value of the incoming line current, in this embodiment. Sixteen points are selected, corresponding to 10 ms. The specific values can be calibrated based on the mechanical action delay characteristics of the circuit breaker connected to the metering box and the incoming current noise level. If the values are too small, the mean difference is easily affected by measurement noise; if the values are too large, they may cross the natural load fluctuation cycle, introducing misjudgments. In this embodiment, the actual action delay of the connected circuit breaker can be used to determine the mean difference. Calibration is performed; when the original sampling rate is 1.6 kHz, 16 points correspond to 10 ms, which is about half a cycle of the 50 Hz fundamental wave. This helps to reduce the impact of the cancellation of the positive and negative half cycles of the current, but it is still necessary to combine the auxiliary contact displacement information and the current waveform change to determine whether there is a valid switching event.
[0026] The per-unit value of the current step is calculated as follows: before taking candidate sampling points The short-time average of the instantaneous absolute value of the incoming line current is After taking candidate sampling points The short-time average of the instantaneous absolute value of the incoming line current is Then the per-unit value of the current step is: In the formula, The reference current is a positive rated current, in amperes (A), sourced from standard initialization configuration or historical statistics. It is used to convert the short-time average difference of the incoming current into a dimensionless current step per unit value, thus setting the trigger threshold. It achieves cross-device versatility regardless of the rated current rating of a specific metering box. Taking the absolute value of the instantaneous current reduces the impact of positive and negative half-cycle cancellation of AC current. and The absolute value of the difference is used to simultaneously detect current changes caused by load switching on and off.
[0027] Preset search window after the change of the circuit breaker auxiliary contact Within the timeframe, the per-unit value of the current step is calculated point by point according to the original sampling time sequence. When the per-unit value of the current step exceeds the preset trigger threshold for the first time... When selecting candidate sampling points, confirm valid switching events and determine the time corresponding to the sampling point as the event anchor time. Instead of directly using the mechanical displacement time of the circuit breaker auxiliary contacts, a valid switching event anchor point is not generated when no sampling point that meets the triggering conditions exists within the search window. For candidate sampling points for which a complete short-term average window cannot be obtained, no corresponding judgment is performed. In this embodiment... The value is set to 0.05. The specific value can be calibrated based on the load fluctuation characteristics and incoming current noise level of the node where the metering box is located. A value that is too small may cause false triggers due to normal load fluctuations, while a value that is too large may miss light load switching events. For the end load of an industrial park, fluctuations below 5% of the rated current are generally considered normal load fluctuations or measurement noise. In this embodiment... The time interval is set to 10 ms. The specific value can be calibrated based on the mechanical action delay characteristics of the circuit breaker connected to the metering box: a value that is too small may miss current steps with large delays, while a value that is too large may introduce subsequent irrelevant fluctuations. In this embodiment, the calibrated action delay range of the circuit breaker is 2 to 8 ms, so setting it to 10 ms covers this range and leaves a margin. A preset search window is used to limit the search range of candidate current sampling points. The sampling data required for calculating the short-time average before and after a candidate sampling point can be located outside the search window; the current step per-unit value determination is only performed when the corresponding before and after sampling data is complete. The event start sampling sequence number is output. It is used to locate event anchor points in the original sampling sequence for data backtracking, event logging, and debugging.
[0028] Regarding the timeliness determination of event anchors: obtain the current data processing time. (Refers to the absolute moment when data processing of this virtual synchronous sampling window begins, sourced from a real-time clock). If a valid switching event exists, and the current data processing time is the same as the time of the most recent valid switching event... The time difference is less than the preset backtracking window Then adopt this Used as an event reference time marker, and the event timeliness confidence level is marked as normal; if the time difference is greater than or equal to the preset backtracking window And less than the preset event expiration time In this case, the method will still be used. As a reference time marker for the event, the event's timeliness confidence is marked as decreased; if there is no valid event to be switched, or the time difference is greater than or equal to the preset event expiration time. In this case, the historical switching event time will no longer be used as the event reference time stamp for the current monitoring window. Instead, the phase reference time will be determined based on the voltage zero-crossing sequence. The phase reference time will not be redefined as the event anchor point time. In this embodiment Take 5 s as the reference value. The specific value can be calibrated according to the frequency fluctuation characteristics of the power grid and the accuracy requirements of the phase reference. If the value is too small, the most recent effective switching event may be eliminated prematurely. If the value is too large, historical switching events with a long time interval may continue to participate in the event association judgment of the current monitoring window. For a normally operating power grid, when the frequency change rate is less than 0.02 Hz / s, the change of the fundamental period within 5 s is less than 0.04 ms, which can be ignored relative to the 20 ms fundamental period. A timescale of 60 seconds is used, with the specific value determined based on the load switching frequency and event correlation timeliness requirements of the industrial park. A value that is too small will frequently degenerate into a voltage zero-crossing anchor point, while a value that is too large may reduce the timeliness of event correlation. For end-point loads in industrial parks, the switching interval of nonlinear loads is typically on the order of several seconds to tens of seconds; a timescale of 60 seconds ensures effective event anchor points for most periods. The event reference timescale is used to record the switching event and its temporal correlation with the monitoring data, and is not directly used as the phase zero point of the virtual synchronous sampling sequence. The phase reference of the virtual synchronous sampling sequence is determined by the positive voltage zero-crossing moment. When a valid event reference timescale exists and the original sampling data covers the continuous period following the event... For a complete fundamental frequency cycle, the monitoring window starts at the moment when the first positive zero-crossing of the effective voltage after the event occurs, and selects the nearest continuous period after the event. The monitoring window is defined as a complete fundamental cycle. When there is no valid event reference timescale or the original sampled data does not meet the above conditions, the monitoring window ends at the last positive zero-crossing of the effective voltage within the current original sampled data coverage area, and the window is then selected backwards. A continuous complete fundamental frequency cycle is used as the monitoring window. The event timeliness confidence level marks the timeliness of the event reference timescale; when the event reference timescale has not expired and the original sampling data meets the window selection conditions, the event reference timescale can still be used for event correlation and monitoring window positioning. The actual monitoring window is always based on the complete and valid fundamental frequency cycle that can be obtained within the coverage area of the original sampling data. In this embodiment, the original sampling data is stored in a circular buffer, and the buffer length covers at least a continuous fundamental frequency cycle. The sampling data required for a complete fundamental frequency cycle and corresponding zero-crossing detection; when the complete fundamental frequency cycle after an event is no longer within the buffer coverage, that event will no longer be used as the location basis for the current monitoring window. When the coverage of the original sampling data is insufficient to obtain... If a continuous, valid, and complete fundamental frequency cycle is not obtained, or if adjacent original sampling data required for the virtual sampling time cannot be obtained, the current virtual synchronous sampling sequence will not be constructed, the current monitoring window will be marked as invalid, and the current S2 and S3 processes will not be executed; the monitoring window will be re-determined after subsequent sampling data meet the above conditions.
[0029] The positive zero-crossing moment when the voltage changes from negative to non-negative is selected as the fundamental phase reference moment, so that the virtual sampling points of each fundamental period have corresponding phase positions. Specifically, the interval between adjacent sampling values changing from negative to non-negative in the original voltage sampling sequence is detected, and linear interpolation is performed using adjacent sampling values and their sampling times to determine the positive zero-crossing moment of the voltage; the measured fundamental period is determined based on the time difference between adjacent valid positive zero-crossing moments. In this embodiment, the validity of the time interval between adjacent positive zero-crossing moments is further determined based on a preset fundamental frequency range. The preset fundamental frequency range is 49.5 Hz to 50.5 Hz, and the corresponding effective fundamental period range is approximately 19.802 ms to 20.202 ms. Zero-crossing intervals outside this range are not considered as effective fundamental periods. For detection results where the effective voltage value of the data segment to be detected is less than 10% of the rated effective voltage value, or the measured period exceeds the above effective fundamental period range, it is not used to construct the virtual synchronous sampling moment; the effective voltage value of the data segment to be detected is calculated based on the corresponding original voltage sampling data. The voltage zero-crossing comparison threshold is set as follows. ,in This is the current effective voltage value. The rated voltage is used (220 V phase voltage for three-phase four-wire metering boxes, and 380 V line voltage for three-phase three-wire metering boxes) to avoid false zero-crossing judgments caused by low-voltage noise. The voltage zero-crossing comparison threshold is used to determine the noise immunity effectiveness of zero-crossing events. It is not directly used as the interpolation level for the zero-voltage crossing moment. The voltage zero-crossing comparison threshold is a non-negative voltage amplitude threshold, and its negative and positive values are used as the basis for determining the validity before and after the zero-crossing interval, respectively. The valid positive zero-crossing moment is still determined by linear interpolation based on the interval where adjacent sampled values cross the zero level. In this embodiment, for the detected positive zero-crossing interval, the voltage sampled data before and after the interval are checked. When there is a sampled value lower than the negative value of the voltage zero-crossing comparison threshold before the zero-crossing interval, and a sampled value higher than the positive value of the voltage zero-crossing comparison threshold after the zero-crossing interval, the zero-crossing event is considered a candidate for validity, and then validity is determined in conjunction with the adjacent zero-crossing interval.
[0030] Internal temperature value Used to mark sampling clock drift confidence: when Exceeding the preset temperature deviation threshold At that time, the confidence level of this virtual synchronization sampling sequence is marked as decreased. Where, The value represents the internal temperature of the chamber, in °C; 25 represents the default ambient temperature of 25 °C, in °C. The temperature deviation threshold is expressed in °C, and in this embodiment, it is set to 10 °C. The specific value can be determined based on the temperature and frequency characteristics of the actual metering chip clock device and the allowable timescale error: if the value is too small, normal temperature fluctuations will trigger degradation, leading to frequent changes in the confidence level; if the value is too large, the crystal oscillator frequency deviation will significantly affect the timescale accuracy without degradation, resulting in the phase error not being marked. In this embodiment, when the temperature inside the chamber deviates from the 25 °C reference temperature by more than 10 °C, the sampling timescale confidence is reduced. The confidence level marker generated in S1 due to temperature exceeding the limit or timescale degradation is a sequence-level confidence level marker, which is transmitted to S2 and S3 along with the virtual synchronous sampling sequence. This allows S3 to add a low confidence level prompt when outputting the final result, but does not skip the orthogonal projection stripping process. If the temperature is not configured, 25 °C is used as the default ambient temperature. The sampling timescale confidence is used to evaluate the reliability of the sampling time, and the event timescale confidence is used to evaluate the timeliness of the event reference timescale. Both are recorded separately and transmitted to subsequent processing steps along with the virtual synchronous sampling sequence.
[0031] The effective voltage positive zero-crossing time at the beginning of the current monitoring window is taken as the phase reference time, denoted as . ,structure The virtual synchronization sampling time for each consecutive fundamental frequency period. The total sequence number of the virtual synchronization sampling is denoted as... , The calculation formula is: In the formula, For the first One virtual synchronous sampling moment; The effective voltage at the beginning of the current monitoring window is the moment of positive zero crossing, in seconds. For the first The measured duration of each fundamental frequency period, nominal initial value The time is 20 ms, determined by the time difference between adjacent positive zero-crossing moments of effective voltage; The fundamental period numbers involved in the summation; For the first The measured duration of each fundamental frequency cycle; The number of virtual sampling points per fundamental frequency period is 32 in this embodiment. The specific value can be calibrated according to the highest harmonic order to be monitored and the computing power of the low-power chip: if the value is too small, the Nyquist frequency will be insufficient and high-order harmonics will aliasing; if the value is too large, the DFT calculation will increase. At the nominal fundamental frequency of 50 Hz, 32 points per period corresponds to a nominal equivalent sampling rate of 1.6 kHz and a Nyquist frequency of 800 Hz. The 15th harmonic corresponds to 750 Hz, which is lower than the Nyquist frequency. However, its amplitude estimation accuracy is still affected by the original sampling rate, interpolation error and sampling time scale error, while taking into account the computing burden of the low-power chip. The fundamental frequency period for continuous projection is 4 in this embodiment. The specific value can be determined based on the required interharmonic frequency intervals and DFT frequency resolution requirements: if the value is too small, the DFT resolution will be insufficient to resolve 25 Hz interharmonics, and single-cycle integration leakage will be severe for non-integer multiple frequencies; if the value is too large, the data window will span multiple fundamental frequency periods, increasing the impact of frequency fluctuations. In this embodiment… A value of 4 corresponds to a frequency interval of 12.5 Hz when the measured average fundamental frequency is 50 Hz. When the fundamental frequency changes, the physical frequency interval is calculated based on the measured average fundamental frequency. This parameter configuration is used to provide a candidate frequency analysis grid and reduce the impact of asynchronous truncation, but it cannot guarantee accurate resolution of the 25 Hz and 75 Hz components under all mixed signal conditions. This is the sequence number of the current fundamental frequency period. , ; This is the virtual sampling sequence number within the current fundamental frequency period. When the power grid frequency changes slowly, it can also be approximated. ,according to Unified recursion.
[0032] The values of the virtual synchronization sampling sequence are obtained by linear interpolation of the original sampling sequence: for the first... virtual sampling time Find the matching sequence in the original sampling sequence Two adjacent raw sampling times ,according to Calculate the virtual sampled values. The truncation error of the linear interpolation is... ,in s represents the original sampling interval; the actual interpolation error is also related to the frequency, amplitude, and waveform variations of the signal being analyzed. Significant amplitude deviations may occur near higher harmonic frequencies, therefore these are not uniformly considered negligible. The output is... A virtual synchronous sampling sequence of consecutive fundamental frequency periods (total) The virtual sampling points and their corresponding time-stamped sets are used for subsequent short-window spectrum estimation and iterative orthogonal projection. It should be noted that the virtual synchronous sampling sequence is uniformly arranged according to the fundamental wave phase, and has the same number of virtual sampling points within each fundamental wave period. However, the actual physical time interval between adjacent virtual sampling points can vary with the measured fundamental frequency period. When performing a discrete Fourier transform on this sequence, the resulting spectrum first corresponds to the fundamental frequency order coordinates, and then is converted to the physical frequency based on the measured average fundamental frequency within the monitoring window; when the fundamental frequency period changes within the monitoring window, this conversion is an approximation in the sense of average frequency. Subsequent steps S2 and S3 use the same frequency coordinates and reference sequence as this virtual synchronous sampling sequence.
[0033] Without the virtual synchronization sampling process described in this step, the fixed analysis window may be affected by the asynchrony between the original sampling clock and the fundamental period. Spectral leakage caused by asynchronous truncation may still exist, affecting the distinction between adjacent harmonics and interharmonic components. The virtual synchronization sampling process uses the switching event reference timescale to determine the event correlation and uses the voltage zero-crossing sequence to establish a fundamental phase reference. This allows subsequent spectrum estimation to be performed on a virtual sampling sequence with uniform phase, thereby reducing the impact of spectral leakage caused by asynchronous truncation and improving the reliability of main lobe and side lobe energy evaluation. Therefore, the virtual synchronization sampling sequence and its timescale set output in this step are necessary intermediate quantities for all subsequent electrical performance determinations.
[0034] S2: Evaluate the energy of the main lobe and side lobes, check the overlap between the side lobe region and the main lobe region of other reference components, perform a finite number of corrections on candidate frequencies that do not overlap and meet the conditions, and determine the candidate harmonic categories.
[0035] The objective electrical performance testing constraints addressed in this step are: slow sidelobe attenuation of the rectangular window, and the non-orthogonality of the 25 Hz and 75 Hz interharmonics with the 50 Hz fundamental and integer harmonics in the frequency domain. If the integer harmonic category is determined solely based on the frequency of the spectral peak, the interharmonics and their spectral leakage components may be incorrectly classified as integer harmonics, leading to deviations in the integer harmonic amplitude and total harmonic distortion rate of the voltage.
[0036] Specifically, the input is S1 and the output is... The virtual synchronous sampling sequence, the corresponding virtual sampling timescale, and the measured average fundamental frequency within the monitoring window are used. Short-window spectrum estimation is performed in a known manner: for A rectangular window is applied to the virtual synchronous sampling sequence, and a discrete Fourier transform is performed to obtain the complex results at each frequency point. The amplitude spectrum is obtained after taking the modulus value. , ,in For frequency point index, For the first The magnitude of the DFT complex spectrum at each frequency point. The virtual synchronous sampling sequences are arranged at equal intervals in phase coordinates, with the physical frequency interval denoted as . According to the measured average fundamental frequency and the number of consecutive fundamental periods within the monitoring window The ratio is determined when the measured average fundamental frequency is 50 Hz. , hour, Hz, corresponding to the nominal equivalent sampling frequency Take 1600 Hz. For non-DC, non-Nyquist frequency points, under the condition that the single-frequency component is aligned with the DFT frequency point, the single-sided peak amplitude can be obtained from... Calculation. For candidate components not aligned with the DFT frequency point, Primarily used for candidate detection and main / sidelobe energy evaluation, it is not directly used as the final effective voltage amplitude. The nominal equivalent sampling frequency of the virtual synchronous sampling sequence is denoted as... ,satisfy ;when For 12.5 Hz, 4. When it is 32, The frequency is 1600 Hz. The complete DFT results are used to store the spectral data; candidate frequency detection and background noise estimation only use frequencies within the preset analysis frequency range that are greater than zero and less than... For single-sided spectral frequencies, conjugate images corresponding to negative frequencies are not considered as independent candidates. Local peak comparisons are performed only for frequencies with valid left and right adjacent frequencies. Main lobe and side lobe energy evaluation uses valid frequencies in the single-sided spectrum; when the region exceeds the range of the single-sided spectrum, it is handled according to the boundary candidate rules described later.
[0037] exist Point DFT amplitude spectrum In the middle, take the one that satisfies and The local maximum frequency point, and Greater than the preset peak threshold In this embodiment Take 0.1 times the maximum value of the amplitude spectrum. The specific value can be calibrated according to the background noise level and the sensitivity requirements for candidate frequency detection to filter out spurious peaks caused by background noise. For integer harmonic frequencies within the preset maximum analysis harmonic order range, set additional frequency points to be detected; when they do not reach the general candidate peak threshold... In this embodiment, the absence of the corresponding integer harmonic is not directly assumed. Instead, the formation of a valid candidate component is determined based on subsequent effective amplitude detection conditions. In this embodiment, the effective amplitude detection conditions are determined based on the estimated voltage amplitude corresponding to the candidate frequency and a preset amplitude detection threshold. The voltage amplitude estimate and the amplitude detection threshold use the same amplitude normalization caliber. When the estimated voltage amplitude is not less than the amplitude detection threshold, the effective amplitude detection conditions are satisfied. In this embodiment, an initial candidate frequency set is first formed based on local spectral peaks, preset integer harmonic detection frequencies, and preset a priori frequency bands. After excluding the main lobe frequencies corresponding to the initial candidate frequencies, the background noise amplitude is estimated. Then, the preset amplitude detection threshold is determined, and candidate components that meet the effective amplitude detection conditions are selected accordingly. After excluding the fundamental frequency and the main lobe frequencies corresponding to the candidate components, the median of the normalized amplitude of the remaining frequency points is used as the background noise amplitude, and three times this is used as the preset amplitude detection threshold. When the estimated amplitude of the candidate component is not less than this threshold, it is determined that it meets the effective amplitude detection conditions. The 3 times value is a preferred example value and can be adjusted according to the requirements of measurement noise and false detection. If the value is too small, false detection is likely to occur, and if the value is too large, weak harmonics may be missed. When the estimated value of the background noise amplitude is zero, the zero amplitude is not directly used as the basis for detecting the effective component. The candidate component amplitude must still be greater than zero and the subsequent validity judgment conditions must be met.
[0038] The initial candidate frequency set is obtained by merging and deduplicating the frequency points corresponding to the peak values of the short-window spectrum, the preset integer harmonic detection frequencies, and the detection frequencies in the preset a priori frequency band. After screening based on effective amplitude detection conditions, a candidate frequency set for subsequent frequency evaluation is formed. When there are no remaining frequency points that can be used for background noise estimation, the amplitude detection threshold judgment is marked as invalid, and candidate components that depend on this threshold are not directly recognized as valid components. Each candidate frequency in the candidate frequency set... For a corresponding candidate component, the candidate component is... It is the frequency center and has the main lobe energy. Total energy of side lobes and the degree of side lobe leakage Attributes such as... If the DFT frequency point within the prior frequency band is already covered by the short window spectral peak, it will not be added to the candidate frequency set again to avoid the same frequency being processed repeatedly. The prior frequency bands are 25Hz±12.5Hz and 75Hz±12.5Hz, covering the two frequency bands [12.5,37.5] Hz and [62.5,87.5] Hz, based on the following: On a Hz DFT grid, interharmonic frequency shifts may cause it to fall on adjacent frequency points. A bandwidth of ±12.5 Hz can cover this shift range. On a 12.5 Hz frequency grid corresponding to the 50 Hz nominal fundamental frequency, the above-mentioned a priori frequency band can cover multiple DFT frequency points near the corresponding frequency to expand the candidate search range, but it does not mean that interharmonic components necessarily exist in the frequency band.
[0039] For each candidate frequency The main lobe region and the left and right side lobe regions are delineated. When initially delineating the regions, candidate frequencies are used. Centered on; after each iteration and correction, based on the updated... The center was redefined. Centered on, Divide into main lobe region, and Divide it into the left lateral lobe region, and The right sidelobe region is designated as the main lobe region. The main lobe region is a set of frequency points with the center frequency as the core, and one frequency resolution interval on each side. When the candidate frequency center is aligned with the DFT frequency point, the main lobe region contains 3 frequency points; when the candidate frequency center is not aligned with the DFT frequency point, the main lobe region usually contains 2 frequency points, depending on the actual number of DFT frequency points falling within the main lobe region. The energy of each region is obtained by summing the squares of the frequency point amplitudes within the region, and is denoted as the main lobe energy. energy of the left lateral lobe and right side lobe energy (unit Total energy of the side lobes The common boundary frequency points of the main lobe region and the left and right side lobe regions are assigned according to the above interval opening and closing relationship, and the energy of different regions is not counted repeatedly.
[0040] Before determining the frequency correction direction based on the energy difference between the left and right sidelobes, it is checked whether the left and right sidelobe regions of the current candidate frequency share common frequencies with the main lobe regions of other reference components. Other reference components include the measured fundamental wave component other than the current candidate component, and the components to be detected among the preset integer harmonic detection frequencies that meet the effective amplitude detection conditions and whose amplitude estimates are greater than the current candidate component; each amplitude estimate uses the same normalization caliber. The center frequency of the measured fundamental wave component is the measured average fundamental wave frequency output by S1, and the center frequency of the integer harmonic detection components is the frequency of the corresponding integer harmonic detection frequency. The main lobe region of each reference component is determined according to the aforementioned main lobe region delineation rules. When common frequencies exist, the sidelobe overlap mark is recorded, the current candidate frequency is retained, and frequency correction based on the energy difference between the left and right sidelobes is not performed. This process is used to avoid the main lobe energy of other reference components affecting the correction direction of the current candidate frequency; the main lobe energy, left and right sidelobe energy, and sidelobe leakage degree are still calculated and recorded according to the original frequency set, and the overall energy of a certain sidelobe is not set to zero.
[0041] Calculate the degree of sidelobe leakage: In the formula, The sidelobe leakage degree of the candidate component is dimensionless. In the calculation... First, determine the energy of the main lobe. Is it greater than the preset lower limit of main lobe energy? If the energy level is not greater than the preset lower limit of the main lobe energy, the candidate component is marked as an invalid candidate component, and the above division and frequency correction are not performed. In this embodiment, the preset lower limit of the main lobe energy is the square of the DFT amplitude detection threshold at the same scale, wherein the DFT amplitude detection threshold at the same scale is half the product of the preset amplitude detection threshold and the total number of virtual sampling points. This lower limit is used to avoid noise interference in the calculation of the side lobe leakage degree when the main lobe energy is too low. It is used to evaluate the spectral energy distribution in the vicinity of candidate frequencies, and is not used as a separate basis for classifying integer harmonics or interharmonics.
[0042] When candidate frequency satisfy or When the corresponding sidelobe region exceeds the single-side spectrum range, the sidelobe information is incomplete. Therefore, the energy of that sidelobe is set to zero, and the candidate component is marked as a boundary candidate. For components marked as boundary candidates, regardless of... Is it greater than None of them enter the frequency center correction loop based on the energy difference between the left and right sidelobes; boundary candidates are not directly classified into the interharmonic frequency and amplitude set, but are classified according to the relationship between their candidate frequency and integer multiples of the measured fundamental frequency, as well as the effective amplitude detection condition and the frequency estimation reliability condition. For boundary candidates that cannot meet the effective classification conditions, their candidate frequency, initial amplitude estimate, and low confidence mark are recorded separately. The basis is that the sidelobe information at the spectral boundary is incomplete, the directional correction based on the energy of one sidelobe is unreliable, and forced correction will cause the center to continuously move outward from the boundary and fail to converge. For boundary candidates, calculation is still performed. And recorded, but the This is for reference only for subsequent statistics and alarms, and should not be used as a criterion for whether to enter the correction cycle.
[0043] For candidate components that are not marked as boundary candidates and do not have sidelobe overlap markers, when Greater than the leakage level determination threshold When, proceed to candidate frequency correction evaluation; when Not greater than At that time, the current candidate frequency is retained. For boundary candidates or candidate components with sidelobe overlap markers, regardless of Is it greater than All candidate frequencies are retained and do not enter the frequency correction loop based on the energy difference between the left and right sidelobes. All candidate components are further classified and their validity is determined; they are not directly identified as valid integer harmonics or interharmonics simply because candidate frequencies are retained. In this embodiment... The threshold is set to 0.12. The specific value can be calibrated based on the interharmonic density and frequency center correction sensitivity requirements at the site. This threshold is only used to determine whether to proceed with correction evaluation when the sidelobe information meets the correction requirements; it is not a fixed threshold for distinguishing between integer harmonics and interharmonics.
[0044] Calculate the candidate frequency center correction: In the formula, This is the sign function. The physical meaning of this formula is: when the energy of the right sidelobe... Energy greater than that of the left side lobe At that time, the correction amount A positive value indicates a correction towards higher frequencies; when... Less than hour, A negative value indicates a correction towards lower frequencies; when the two are equal, the value is zero. This correction generates a trial correction direction and a trial frequency for candidate frequencies based on the energy distribution of the left and right sidelobes. The trial correction direction does not directly represent the direction of the true frequency deviation; whether to accept this correction requires further amplitude evaluation.
[0045] The frequency of the probe is: To evaluate the frequency of probing Is it better than the candidate frequency before correction? The correlation amplitude of the current virtual synchronous sampling sequence is calculated on reference sequences corresponding to two frequencies, and compared using a consistent amplitude estimation and normalization method. The correlation amplitude is obtained by summing the correlation between the virtual synchronous sampling sequence and the corresponding sine and cosine reference sequences, and solving for the binary coefficients. The specific calculation method is the same as that for the projection coefficients in S3. The correlation amplitude is used to evaluate the amplitude response at a specified candidate frequency and is not directly used as the basis for determining the harmonic category. The correlation amplitude evaluation in S2 uses the same reference phase construction method as in S3 to maintain consistency between the frequency coordinates used in the frequency correction evaluation and the subsequent projection stripping.
[0046] Generate a test frequency within a preset maximum frequency offset range. And based on the candidate frequency before correction With the frequency of probing The corresponding amplitude evaluation has been revised in this instance. If the frequency shift does not exceed the preset maximum frequency offset range, does not enter the main lobe region of other identified spectral peaks, and the correlation amplitude corresponding to the test frequency is greater than the correlation amplitude corresponding to the candidate frequency before correction, then this correction will be accepted and... Assign to Otherwise, retain the version before modification. Stop the correction in the current direction. After each correction, the main lobe and left and right side lobe regions are redefined with the updated candidate frequencies. The spectral boundaries and common frequencies with the main lobe regions of other reference components are rechecked, and the side lobe leakage and the energy difference between the left and right side lobes are recalculated. If the updated candidate component is not a boundary candidate, there is no side lobe overlap marker, and the side lobe leakage is still greater than the leakage threshold, the next trial correction direction is determined based on the recalculated energy difference between the left and right side lobes, and the next trial frequency is generated with a preset step size; otherwise, the correction is stopped. The maximum number of corrections is reached. If the next test frequency exceeds the preset maximum frequency offset range or does not meet the preset acceptance conditions, the correction ends and the candidate frequency is retained. After the correction is completed, the continuous frequency estimates of the final candidate frequency are retained. The most recent DFT frequency point is only used as a spectrum index and the final candidate frequency is not forcibly snapped to the DFT frequency point. The maximum number of frequency center corrections is a preset hyperparameter. In this embodiment, the step size is taken three times. The specific value can be calibrated according to the frequency center offset range and the convergence requirement of the correction: if the value is too small, the candidate frequency offset may not be fully evaluated; if the value is too large, the computational load will increase and over-correction may be introduced. The step size of each trial correction is... ,exist Hz corresponding The cumulative offset limit for three consecutive attempts to correct in the same direction is 18.75 Hz. In this embodiment, the preset maximum frequency offset range is 18.75 Hz above and below the initial candidate frequency. The actual correction must not exceed the preset analysis frequency range and must not enter the main lobe region of other identified spectral peaks. The above values are examples of the correction range and do not guarantee coverage of all true frequency deviations. When the maximum number of corrections is reached... If, after that, a candidate frequency that meets the preset reliability conditions still cannot be obtained, the corresponding candidate component is marked as a low-confidence candidate component, and is not directly based on... Whether it exceeds the limit determines its harmonic category.
[0047] For candidate components that have completed frequency correction evaluation, they are classified according to the integer multiple relationship between the final candidate frequency and the measured average fundamental frequency. The measured average fundamental frequency within the monitoring window is denoted as... When there are integer harmonic orders within the preset analysis range , so that the final candidate frequency is When the frequency deviation between the two frequencies is not greater than the preset frequency matching tolerance, and the candidate component meets the effective amplitude detection condition and the frequency estimation reliability condition, it is marked as the corresponding integer harmonic candidate component. In this embodiment, the preset frequency matching tolerance is preferably one-quarter of the frequency resolution, that is, 3.125 Hz when the frequency resolution is 12.5 Hz. This tolerance is used to allow for limited frequency estimation deviation; if the value is too small, integer harmonics are easily missed, and if the value is too large, adjacent interharmonics are easily misclassified. When the candidate frequency does not meet the integer multiple frequency matching condition, but meets the effective amplitude detection and frequency estimation reliability conditions, it is marked as an interharmonic candidate component. For candidate components that do not meet the validity condition, a low confidence mark is retained, and they are not directly confirmed as effective integer harmonics or interharmonics. In this embodiment, the frequency estimation reliability condition is that the candidate frequency is within the preset analysis frequency range, and the frequency correction does not exceed the preset maximum frequency offset range and does not enter the main lobe region of other identified spectral peaks; the effective amplitude detection condition is determined separately according to the aforementioned preset amplitude detection threshold. Candidate components that do not meet the corresponding conditions are marked as low-confidence candidate components.
[0048] The output is the corrected candidate frequency set and its corresponding component-level confidence level labels. The candidate frequency set includes integer harmonic candidate frequencies, interharmonic candidate frequencies, and low-confidence candidate frequencies that meet the validity criteria; each candidate component also records its main lobe energy. Initial spectral amplitude Candidate category labeling. For candidate components whose left and right sidelobe regions share common frequencies with the main lobe regions of other reference components, a sidelobe overlap label is recorded. This label only indicates that the current sidelobe information is not used for directional frequency correction and does not replace the effective amplitude detection condition and frequency estimation reliability condition; candidate components with retained frequencies still need to pass the corresponding validity judgment before entering S3 as effective candidate components for projection estimation. Among them, integer harmonic candidate components and interharmonic candidate components that meet the validity condition enter S3 for sorting and projection stripping; low-confidence candidate components are recorded separately and do not directly participate in the projection stripping of effective components. The components recorded in S2 Used for candidate frequency evaluation, it is not directly used as the final effective voltage amplitude; the final integer harmonic amplitude and interharmonic amplitude are determined by the projection estimation results in S3. The fundamental component undergoes validity checking and projection processing first in S3, and its effective voltage amplitude is denoted as... , is used as the denominator in the calculation of total harmonic distortion of voltage.
[0049] The main lobe and side lobe energy evaluation is used to determine whether candidate frequencies need to be included in the correction evaluation. For candidate components whose side lobe regions share common frequencies with the main lobe regions of other reference components, the current candidate frequencies are retained to avoid using the energy difference between the left and right side lobes affected by other components for directional correction. For candidate components whose side lobe information meets the correction conditions, a finite number of corrections are performed according to relevant amplitude comparisons and correction boundaries. Subsequently, the candidate category is determined by combining the integer multiple relationship between the final candidate frequency and the measured fundamental frequency and the validity conditions. S1 provides a phase-uniform virtual sampling sequence and corresponding time scales. S2 performs candidate frequency evaluation and classification on this sequence, providing candidate frequency and validity information for the iterative projection stripping in S3.
[0050] S3: Prioritize projection stripping of the fundamental component, then sort and iteratively project strip the remaining effective candidate components according to the current main lobe energy to obtain the set of integer harmonic amplitudes and the set of interharmonic frequencies and amplitudes, and calculate the total harmonic distortion rate of the voltage.
[0051] It should be noted that even after frequency center correction, interharmonics and integer harmonics may still be non-orthogonal in the frequency domain, and sidelobes of higher-energy components may remain at the frequencies of weaker components. If iterative stripping is performed, strong harmonic sidelobes will remain at the frequencies of weak interharmonics, causing misclassification. However, by stripping the remaining spectrum residuals according to the main lobe energy from largest to smallest, the effects of non-orthogonal aliasing can be gradually reduced under low computational power conditions. The technical problem to be solved in this step is that fixed-window FFT causes spectral leakage leading to misclassification of harmonic amplitudes when there is a high density of asynchronous interharmonics. The solution is not to increase the sampling rate, but to use the candidate frequency set and validity information determined in the previous step to prioritize the fundamental component, and then perform iterative projection stripping on the integer and interharmonic candidate components that meet the validity conditions, so as to reduce the impact of stronger components on subsequent amplitude estimation and reduce the risk of interharmonics being misclassified as integer harmonics.
[0052] Specifically, the inputs are the virtual synchronization sampling sequence and corresponding time stamp output by S1, and the corrected candidate frequency set and main lobe energy output by S2. Candidate component categories and validity marking. First, it is determined whether the fundamental component meets the preset fundamental validity conditions. In this embodiment, the fundamental validity conditions are: the measured fundamental frequency is within the range of 49.5 Hz to 50.5 Hz, the fundamental period used to construct the monitoring window passes the validity check, and the effective value of the monitoring window voltage is not less than 10% of the rated effective value. If these conditions are not met, the monitoring window is marked as invalid. When the conditions are met, the fundamental component is used as the first component to be stripped for projection stripping, and its effective voltage amplitude is recorded as... This is used as the denominator in the calculation of the total harmonic distortion (THD) of the voltage. After the fundamental wave projection is completed, the validity of the fundamental wave projection coefficients and whether the obtained effective value of the fundamental wave voltage is not less than 10% of the effective value of the rated voltage are checked. If the fundamental wave projection coefficients are invalid or the effective value of the fundamental wave voltage is lower than this threshold, the THD of this voltage is marked as invalid, and the initial spectral amplitude of the fundamental wave is not used to replace the projection result. After the fundamental wave processing is completed, the integer harmonic candidate components and interharmonic candidate components other than the fundamental wave component output by S2 that meet the effective amplitude detection condition and the frequency estimation reliability condition are taken as the remaining components to be stripped, and the main lobe energy in the corresponding spectrum of the current time domain residual sequence is used as the basis for further processing. Sort the components from largest to smallest and perform projection stripping sequentially. Low-confidence candidate components that do not meet the validity criteria are temporarily excluded from the projection stripping. The valid amplitude detection results in S2 are used to determine the candidates to be processed, but are not directly used as the basis for confirming the final valid components. After the fundamental component is stripped, the remaining candidate components are re-projected onto the current time-domain residual sequence for amplitude estimation, and are verified according to the amplitude detection and projection coefficient validity criteria in this step; candidate components with sidelobe overlap markers but meeting the validity criteria in S2 are also processed according to the same rules.
[0053] An orthogonal projection is performed on the current time-domain residual sequence at the final candidate frequency of the component to be stripped. The first round's current time-domain residual sequence is the virtual synchronous sampling sequence output by S1. After each amplitude and phase estimation of the current component to be stripped, a corresponding time-domain projection component is constructed based on the estimation results. Only when the component meets the preset validity conditions and the projection coefficients are calculated effectively is the projection component subtracted from the current time-domain residual sequence to form the current time-domain residual sequence for the next round. Therefore, orthogonal projection is always performed on the current time-domain residual sequence of the current round.
[0054] Orthographic projection This is performed on a virtual synchronous sampling sequence corresponding to a consecutive fundamental frequency period. Specifically, it is based on the frequency of the component to be stripped. Construct cosine and sine reference sequences; when the component to be stripped is the fundamental component... Take the measured average fundamental frequency When the component to be stripped is another valid candidate component, Take the final candidate frequency of the candidate component after processing by S2. To maintain consistency with the phase-uniform sampling sequence of S1, the first... The reference phase corresponding to each virtual sampling point is denoted as . And construct a cosine reference sequence. Sine reference sequence The aforementioned reference phase is calculated based on the measured average fundamental frequency within the monitoring window, and is suitable for situations where the fundamental period varies little within the window. When the fundamental periods vary significantly, the conversion error between the physical frequency and the phase order may affect the estimation of interharmonic frequencies and amplitudes. The current time-domain residual sequence is then used. The components are multiplied point-by-point with each of the two reference sequences and summed. Combined with the autocorrelation and cross-correlation sums of the two reference sequences, the in-phase and quadrature components are obtained through a binary coefficient solution. For interharmonic components that are not integer multiples of the fundamental frequency, the sine and cosine reference sequences are not necessarily strictly orthogonal. Therefore, a binary coefficient solution method is used to reduce the amplitude estimation deviation caused by the non-orthogonality of the reference sequences. In this embodiment, when all effective fundamental periods within the monitoring window are within the aforementioned effective fundamental period range, the component estimation is performed using the reference phase based on the measured average fundamental frequency; the corresponding frequency and amplitude are the estimation results within that monitoring window.
[0055] Within the projection interval, the current time-domain residual sequence can be approximated as: ,in The residual. When the sine and cosine reference sequences satisfy the approximate orthogonality condition, the in-phase components... and orthogonal components The following normalized correlation summation formula can be used for approximate calculation: In the formula, The current time-domain residual sequence is at the th The value of each virtual sampling point, in V; The amplitude of the component to be stripped, in V; The frequency of the component to be stripped, in Hz; For the first The reference phase angle corresponding to each virtual sampling point, in rad, is calculated according to... calculate; The measured average fundamental frequency within the monitoring window, in Hz; The phase of the component to be stripped, in rad; The residual is expressed in V. The number of consecutive fundamental frequency periods is 4 in this embodiment; The number of virtual sampling points per fundamental frequency period is 32 in this embodiment; This is the total virtual sampling sequence number. ; These are in-phase components, in units of V; These are orthogonal components, in units of V; This is a correlation summation normalization factor under approximately orthogonal conditions, used to restore the correlation summation result to the amplitude scale, so that... , In this embodiment, to avoid increasing the threshold for judging the approximate orthogonality condition, the actual amplitude estimation uniformly adopts a binary coefficient solution method. Specifically, based on the autocorrelation and cross-correlation sums of the cosine and sine reference sequences, and the correlation sums between the current time-domain residual sequence and the two reference sequences, the in-phase and orthogonal components are solved according to the following binary normal equation: In the formula, the meanings of each symbol are consistent with the aforementioned definitions; in the bivariate normal equation, the summation range is from the total virtual sampling index 0 to the total number of virtual sampling points minus 1; each element in the coefficient matrix is the autocorrelation sum of the cosine reference sequence, the cross-correlation sum of the sine and cosine reference sequences, and the autocorrelation sum of the sine reference sequence, respectively; the right side of the equation is the correlation sum between the current time-domain residual sequence and the corresponding reference sequence. When the above coefficient matrix is not invertible or the solution result is not a finite value, the corresponding candidate component is marked as invalid, and this projection stripping is not performed.
[0056] The above-mentioned solution for binary coefficients mainly involves multiplication, accumulation, and a small number of division operations, and is suitable for metering chips with limited computing power. (Based on in-phase components) and orthogonal components Synthesize the peak amplitude of the component to be stripped. and in accordance with Calculate the phase, where This is the arctangent function for the four quadrants, returning the phase angle in rad.
[0057] Complete the in-phase component for the current component to be stripped. Orthogonal components Amplitude and phase After estimation, it is determined whether the projection coefficients meet the validity conditions. In this embodiment, the validity conditions for the projection coefficients are: the in-phase component and the quadrature component obtained on the current time-domain residual sequence are both finite values, the amplitude synthesized by the two meets the preset amplitude detection threshold, and the sum of squares of the residual sequence after stripping is not greater than the sum of squares of the residual sequence before stripping. The amplitude and the detection threshold use the same normalization caliber, and the initial amplitude recorded in S2 is not used to replace the projection amplitude of this round for judgment; if the above conditions are not met, the residual sequence is not updated; when the projection coefficients are valid and the corresponding candidate components have met the valid detection and frequency reliability conditions of S2, the current projection stripping is accepted, and the projection component is subtracted from the current time-domain residual sequence to form the next round of time-domain residual sequence; otherwise, the current time-domain residual sequence before stripping is retained. The subtraction formula is: In the formula, The current time-domain residual sequence is at the th The value of each virtual sampling point, in V; For the updated time-domain residual sequence at the th The value of each virtual sampling point, in V; For candidate components that fail the validity determination of projection coefficients, they are marked as low-confidence candidate components, and their candidate frequencies, initial amplitude estimates, and failure flags are recorded. They are then removed from the current sorted set of components to be stripped, and the projection stripping of the same candidate component is not repeated.
[0058] After each successful projection stripping of the component to be stripped, a rectangular window is reapplied to the updated temporal residual sequence and then... Point DFT is performed to obtain the updated residual spectrum; the main lobe energies of the unprocessed valid candidate components are recalculated and sorted, and the candidate component at the top of the current sort is selected for the next projection stripping. Candidate components that have already been stripped or marked as low confidence are not included in the sorting set of the current round. Each iteration uses the same number of DFT points, rectangular window, and frequency resolution as in S2, only the input sequence is different.
[0059] S2 performs an initial DFT on the virtual synchronous sampling sequence. S3 re-executes the DFT each time a successful stripping occurs and other valid candidate components still exist. Each iteration uses the same number of DFT points as S2. Rectangular window and frequency interval The only difference is the input sequence. Since the actual number of components to be stripped varies with the validity determination result, the number of DFT executions is determined by the actual number of valid candidate components processed within the current monitoring window, and the computational complexity is limited by the number of candidate components and the number of iterations. Repeat this stripping process until all components to be stripped have been effectively stripped or processed to low confidence.
[0060] The output includes the set of integer harmonic amplitudes, the set of interharmonic frequencies and amplitudes, and the total harmonic distortion (THD) of the voltage, all after validity checks. Low-confidence candidate components are individually marked. The THD is calculated according to the conventional definition. In the formula, The total harmonic distortion (THD) of voltage is expressed in % (%). For integer harmonic orders, Corresponding to the fundamental frequency, Corresponding to each integer harmonic; interharmonic frequencies that do not satisfy the integer multiple relationship are not included. The range is separately retained in the set of interharmonic frequencies and amplitudes; For the first time after validity determination The effective value of the second integer harmonic voltage, in V; The effective value of the fundamental voltage after validity determination, in V; In this embodiment, the fundamental and integer harmonic voltage amplitudes adopt a uniform effective value. When the peak amplitude is calculated by projection... When, divide it by Obtain the corresponding effective voltage value; when the effective value of the fundamental voltage obtained by projection estimation is less than 10% of the effective value of the rated voltage, the total harmonic distortion rate of the voltage is not calculated and the corresponding invalid flag is output.
[0061] The highest integer harmonic order included in the calculation is used as a preset hyperparameter, and in this embodiment, it is set to 15. The specific value can be calibrated according to the highest harmonic order of interest in the monitoring task and the aliasing protection band requirements: if the value is too high, the higher harmonics will be unreliable due to aliasing; if the value is too low, the harmonic components of interest will be missed. At a nominal sampling rate of 1.6 kHz, the Nyquist frequency is 800 Hz. At the nominal fundamental frequency of 50 Hz, the highest integer harmonic frequency is 750 Hz, which is lower than the Nyquist frequency; however, this frequency relationship cannot guarantee that the linear interpolation and amplitude estimation errors meet the requirements. Therefore, the actual effective analysis range still needs to be determined in combination with the original sampling conditions and high-frequency amplitude errors.
[0062] The set of integer harmonic amplitudes includes the fundamental frequency and the voltage amplitudes of each integer harmonic, estimated by S3 projection and passing the validity criterion. The corresponding fundamental frequency, used as the denominator in THD calculations. The numerators used for THD calculations correspond to each effective integer harmonic. The interharmonic frequency and amplitude set includes interharmonic frequencies and voltage amplitudes classified as interharmonic candidates in S2 and projected and validated in S3. Interharmonic amplitudes are not included in THD calculations. Whether a candidate component is included in the effective integer harmonic amplitude set or the effective interharmonic frequency and amplitude set is based on the projection estimation and validity determination results on the current time-domain residual sequence, not directly determined by the initial amplitude or candidate category label in S2. For components with sidelobe overlap labels and retained candidate frequencies, their effective amplitudes are recorded only after passing the same projection estimation and validity determination; if they fail, they are recorded as low-confidence candidate components and not included in the effective amplitude set. For low-confidence candidates and candidate components that fail the projection coefficient validity determination, their candidate frequencies, initial amplitude estimates, and validity labels are recorded separately and not directly included in the effective integer harmonic amplitude set or the effective interharmonic frequency and amplitude set. For the preset highest analysis harmonic order... For integer harmonics that do not meet the effective detection criteria within the range, their undetected state and corresponding detection limit are recorded; the undetected state does not mean that the true amplitude of the harmonic is zero. THD is calculated based on the effective integer harmonic amplitudes obtained within the monitoring window. During the calculation, only the effective values of the integer harmonic voltages that pass the validity judgment are included in the summation. For undetected integer harmonics, their unknown true amplitudes are not directly considered as zero. The calculation result is used as the estimated value of the total harmonic distortion rate of the voltage corresponding to the effective detected component, and the undetected state, detection limit, and effective analysis range are recorded at the same time.
[0063] The corrected set of integer harmonic amplitudes, the set of interharmonic frequencies and amplitudes, and the total harmonic distortion (THD) of the voltage constitute the power quality monitoring results. These results are uploaded to the local monitoring equipment via an RS-485 communication interface, and alarm information is generated based on the alarm conditions configured on the local monitoring equipment. The monitoring results include the effective set of integer harmonic amplitudes, the effective set of interharmonic frequencies and amplitudes, THD, and corresponding validity markers; long-term uploading of the original voltage waveform is not required.
[0064] This step, together with the aforementioned virtual synchronous sampling window reconstruction, main lobe and side lobe energy evaluation, and candidate frequency classification, forms a data flow closed loop. The acquisition node outputs the instantaneous sampling sequences of three-phase voltage, incoming current, circuit breaker auxiliary contact status, and box temperature. The event detection node determines the reference time scale of the valid switching event based on the circuit breaker auxiliary contact displacement and incoming current step, and outputs the event's initial sampling sequence number. The virtual synchronous sampling node determines the phase reference time and the measured duration of each fundamental period based on the positive zero-crossing voltage sequence, constructing... The system generates a virtual synchronous sampling sequence and its corresponding timescale set. The candidate frequency processing node performs short-window spectral analysis on the virtual synchronous sampling sequence, performs a finite number of candidate frequency corrections based on the energy distribution of the main lobe and side lobes, and outputs integer harmonic candidate components, interharmonic candidate components, and low-confidence candidate components by combining the integer multiple relationship between the candidate frequencies and the measured average fundamental frequency, as well as validity conditions. The projection stripping node prioritizes the fundamental component, then iteratively projects and strips the remaining candidate components that meet the validity conditions, forming effective integer harmonic amplitude sets and effective interharmonic frequency and amplitude sets, respectively. The THD is calculated based on the effective integer harmonic amplitudes. The output node uploads the monitoring results and corresponding validity markers via an RS-485 communication interface, without continuously uploading the original waveform.
[0065] This application also discloses an energy quality monitoring system for an energy metering box, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, an energy quality monitoring method for an energy metering box according to this application is implemented.
[0066] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.
[0067] In this application, the aforementioned memory can be any tangible medium that contains or stores a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as resistive random access memory (DRAM), dynamic random access memory (DRAM), static random access memory (SRAM), etc., or any other medium that can be used to store desired information and can be accessed by an application program, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device.
[0068] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.
Claims
1. A method for monitoring power quality in an electricity metering box, characterized by include: Obtain the time of change of auxiliary contact of circuit breaker and incoming current. Within a preset search window after the time of change, determine the time of event anchor point based on the short-time average difference between the absolute values of the instantaneous values of the incoming current before and after the change of candidate current sampling points. The measured fundamental period is determined based on the voltage zero-crossing sequence, the monitoring window is determined based on the event anchor point time and the voltage zero-crossing sequence, and the virtual synchronous sampling sequence is obtained by interpolation reconstruction. The virtual synchronous sampling sequence is transformed in the frequency domain, and candidate frequencies are determined by combining the spectral peak and the preset a priori frequency band. The sidelobe leakage degree is calculated based on the main lobe and left and right sidelobe energies corresponding to the candidate frequencies. The degree of sidelobe leakage is used to determine whether to enter the frequency correction evaluation. The candidate frequencies that meet the correction conditions are corrected a finite number of times. Based on the integer multiple relationship between the corrected candidate frequencies and the measured fundamental frequency and the effectiveness of the candidate components, the integer harmonic candidate components and interharmonic candidate components are determined. Using the virtual synchronous sampling sequence as the current time domain remaining sequence, the fundamental component is stripped first, and then the integer harmonic candidate components and interharmonic candidate components that meet the validity conditions are sorted and iteratively projected and stripped to obtain the integer harmonic amplitude set and the interharmonic frequency and amplitude set respectively. The total harmonic distortion rate of the voltage is calculated based on the integer harmonic amplitude set.
2. The power quality monitoring method of the electric energy metering box according to claim 1, characterized in that, The methods for determining the event anchor point time include: Within the preset search window after the auxiliary contact of the circuit breaker changes position, calculate the short-time average of the absolute values of the instantaneous values of the incoming current at preset points before and after the candidate current sampling point, and determine the per-unit value of the current step based on the ratio of the absolute value of the difference between the two short-time averages to the preset rated current reference. When the current step per unit value is detected to be greater than the preset trigger threshold for the first time, a valid switching event is confirmed, and the corresponding sampling point time is determined as the event anchor point time; when there is no sampling point in the search window that meets the trigger condition, no valid switching event anchor point is generated; for candidate sampling points for which complete short-term average calculation data cannot be obtained, the corresponding current step per unit value determination is not performed.
3. The electric energy quality monitoring method of an electric energy metering box according to claim 2, characterized in that, The acquisition of the virtual synchronization sampling sequence includes: The current data processing time is obtained, and the event reference time stamp and event timeliness confidence are determined based on the time difference between the current data processing time and the event anchor time corresponding to the most recent valid switching event. The event reference time stamp is used for event association positioning in the monitoring window, and the voltage zero-crossing sequence is used to determine the phase reference time of virtual synchronous sampling. The phase reference time of virtual synchronous sampling is determined based on the voltage zero-crossing sequence. When there is no valid switching event or the time difference reaches the preset event failure duration, the historical event anchor time is not used as the event reference time stamp of the current monitoring window. Based on the event reference time stamp, the current data processing time, and the coverage area of the original sampling data, a monitoring window containing a preset number of complete fundamental frequency cycles is determined. When a valid event reference time stamp exists and the original sampling data covers a preset number of consecutive complete fundamental frequency cycles after the corresponding event, a preset number of consecutive complete fundamental frequency cycles starting from the first effective voltage positive zero-crossing moment after the event are selected. Otherwise, the monitoring window ends at the last effective voltage positive zero-crossing moment within the coverage area of the current original sampling data, and a preset number of consecutive complete fundamental frequency cycles are selected backward. Based on the voltage zero-crossing moment and the measured period duration of each fundamental frequency cycle, the same number of phase-uniform virtual sampling points are constructed within each fundamental frequency cycle, and a virtual synchronous sampling sequence is obtained through linear interpolation.
4. The electric energy quality monitoring method of an electric energy metering box according to claim 1, characterized in that, The determination of the candidate frequency includes: The amplitude spectrum is obtained by performing a discrete Fourier transform on the virtual synchronous sampling sequence. Frequency points that satisfy the local peak condition and the preset peak threshold are selected, and frequency points to be detected within the preset prior frequency band are selected. At the same time, frequency points to be detected corresponding to preset integer harmonics are set. The above frequency points are merged and deduplicated to form an initial candidate frequency set. The background noise amplitude is estimated based on the initial candidate frequency set, and a candidate frequency set is formed by screening according to the effective amplitude detection condition.
5. The method for monitoring power quality of an electric energy metering box according to claim 4, characterized in that, The calculation of the sidelobe leakage degree includes: Based on the frequency resolution of the amplitude spectrum, the main lobe region and the left and right side lobe regions are determined with the candidate frequency as the center. The frequency point sets corresponding to the regions do not overlap. The squares of the frequency point amplitudes in each region are accumulated to obtain the main lobe energy, the left side lobe energy and the right side lobe energy. When the main lobe energy is greater than the preset lower energy limit, the ratio of the sum of the left and right side lobe energies to the main lobe energy is used as the side lobe leakage degree; when the main lobe energy is not greater than the lower energy limit, the corresponding candidate component is marked as an invalid candidate component.
6. The electric energy quality monitoring method of an electric energy metering box according to claim 5, characterized in that, The frequency correction includes: Determine whether the left and right sidelobe regions of the current candidate frequency have common frequency points with the main lobe regions of other reference components; the other reference components include the measured fundamental wave components other than the current candidate components, and the components to be detected that meet the effective amplitude detection conditions and whose amplitude estimates are greater than the current candidate components among the preset integer harmonic detection frequencies. When the common frequency point exists, the current candidate frequency is retained and frequency correction based on the energy difference between the left and right sidelobes is not performed; when the common frequency point does not exist and the sidelobe leakage degree is greater than the preset leakage degree judgment threshold, the trial correction direction is determined according to the energy difference between the left and right sidelobes, the trial frequency is generated with a preset correction step size, and the same amplitude estimation and normalization rules are used to calculate the relevant amplitudes corresponding to the candidate frequency and the trial frequency before correction. When the test frequency is within the preset maximum frequency offset range, does not enter the main lobe region of other identified spectral peaks, and the correlation amplitude corresponding to the test frequency is greater than the correlation amplitude corresponding to the candidate frequency before correction, the correction is accepted. After each correction, the main lobe and left and right side lobe regions are redefined with the updated candidate frequency, the common frequency point judgment is re-executed, and the side lobe leakage degree and the energy difference between the left and right side lobes are recalculated. When the preset maximum number of corrections is reached, the frequency correction condition is no longer met, or the preset correction acceptance condition is not met, the frequency correction ends and the final candidate frequency is retained.
7. The method for monitoring power quality of an electric energy metering box according to claim 1, characterized in that, The classification of the candidate components includes: Based on the frequency deviation between the corrected candidate frequency and the measured fundamental frequency, as well as the preset frequency matching tolerance, effective amplitude detection conditions and frequency estimation reliability conditions, the integer harmonic candidate components are determined. Candidate components that do not meet the integer multiple frequency matching condition but meet the effective amplitude detection and frequency estimation reliability conditions are identified as interharmonic candidate components. Candidate components that do not meet the effective amplitude detection condition or the frequency estimation reliability condition are marked as low-confidence candidate components.
8. The method for monitoring power quality of an electric energy metering box according to claim 1, characterized in that, The iterative projection stripping includes: First, the fundamental component is projected and stripped. Then, based on the main lobe energy in the spectrum corresponding to the current time domain remaining sequence, the remaining candidate components that meet the validity conditions are sorted. Sine reference sequences and cosine reference sequences are constructed according to the frequencies of the components to be stripped. In-phase components and quadrature components are obtained by correlation summation and binary coefficient solution, and the amplitude and phase are calculated accordingly. Based on the amplitude and phase, the projected components are constructed. When the preset stripping validity condition is met, the current remaining time-domain sequence is updated, and the unprocessed candidate components are reordered.
9. A method for monitoring power quality in an electricity metering box according to claim 8, characterized in that, The acquisition of the set of integer harmonic amplitudes, the set of interharmonic frequencies and amplitudes, and the total harmonic distortion rate of voltage includes: Integer harmonic components that pass the validity determination are recorded in the integer harmonic amplitude set, and the frequency and amplitude of interharmonic components that pass the validity determination are recorded in the interharmonic frequency and amplitude set; candidate components that fail the validity determination are marked with low confidence and their initial amplitudes are not directly used as valid harmonic amplitudes. The estimated total harmonic distortion rate of voltage is calculated based on the effective integer harmonic voltage amplitude, and the detection status and corresponding detection limit of the integer harmonics not detected within the preset maximum analysis harmonic order range are recorded.
10. A power quality monitoring system for an electricity metering box, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the power quality monitoring method for an electricity metering box according to any one of claims 1-9 is implemented.
