Voltage sag detection method and device based on single-phase inverter amplitude fast tracking

Abstract

The application relates to the technical field of voltage sag detection, and discloses a voltage sag detection method and device based on single-phase inverter amplitude fast tracking, which comprises the following steps: acquiring and performing frequency domain analysis and preliminary filtering on the output voltage signal of a single-phase inverter; performing dynamic tracking processing on the filtered voltage amplitude sequence to determine a candidate voltage sag event; performing multi-scale signal decomposition on the candidate event, identifying and excluding transient interference matched with preset load mutation characteristics, thereby confirming a real voltage sag event; comparing the real event with a database, calculating the event duration and recovery time of the real event; and finally performing correlation logic calculation to generate a voltage sag event report. The application in-depth analyzes the transient characteristics of the event through a signal decomposition technology, effectively distinguishes real voltage sags from transient interferences such as load mutations, solves the problem of low detection reliability and precision in the prior art, and realizes accurate identification of slight voltage sags.

Description

Technical Field

[0001] This invention relates to the field of voltage sag detection technology, and in particular to a voltage sag detection method and apparatus based on rapid amplitude tracking of a single-phase inverter. Background Technology

[0002] Currently, voltage sag detection technology is a key research area for ensuring grid stability and equipment safety, its importance lying in its ability to effectively improve system reliability. This technology is widely used in various scenarios involving single-phase inverters, such as residential photovoltaic power generation systems, uninterruptible power supplies (UPS), new energy vehicle charging piles, and various precision electronic devices. In these applications, timely and accurate capture of voltage amplitude changes can effectively prevent cascading failures. The core technical challenges of voltage sag detection lie in the real-time nature of amplitude tracking and the accuracy of interference signal identification, especially in the complex electromagnetic environment of inverter operation, where sudden load changes or switching noise can easily interfere with signal purity.

[0003] In existing technologies, voltage sag detection methods often rely on a single signal processing path, determining the occurrence of a sag event by setting a fixed amplitude threshold. These methods perform preliminary processing of the acquired voltage signal and then directly compare the amplitudes. While simple to implement, they perform poorly in complex environments. The main drawback of existing technologies lies in their slow response to real-time dynamic signals and their inability to effectively distinguish between genuine voltage fluctuations and transient interference caused by sudden load changes such as motor startup. They are prone to misinterpreting external noise as valid events. For example, when a motor in a factory automation circuit generates high-frequency interference at startup and superimposes it onto the voltage signal, the detection system may amplify this transient noise as a false voltage sag alarm, causing unnecessary equipment downtime and production interruptions.

[0004] In summary, existing technologies suffer from limitations in reliability and accuracy that fail to meet practical requirements. Summary of the Invention

[0005] This invention provides a voltage sag detection method and device based on rapid amplitude tracking of a single-phase inverter, in order to solve the problem that the detection reliability and accuracy are difficult to meet practical needs.

[0006] In a first aspect, to solve the above-mentioned technical problems, the present invention provides a voltage sag detection method based on fast amplitude tracking of a single-phase inverter, comprising:

[0007] The output voltage signal of the single-phase inverter is acquired, and frequency domain analysis and preliminary filtering are performed to obtain the voltage amplitude sequence after preliminary filtering.

[0008] Based on the pre-filtered voltage amplitude sequence, dynamic tracking processing is performed and candidate voltage sag events are determined to obtain a candidate event list;

[0009] For each candidate event in the event candidate list, signal decomposition and event boundary detail fusion are performed to obtain the actual voltage sag event confirmation result;

[0010] Based on the confirmation results of the actual voltage sag events, a comparison is made with the pre-established voltage sag event database. If the similarity is higher than the preset similarity threshold, it is determined to be a recurring pattern event, and the event duration sequence is calculated and obtained.

[0011] Based on the event duration sequence, recovery boundary matching is performed. If the matching is successful, the recovery time is confirmed, and the sequence is fused to obtain the recovery time sequence.

[0012] Based on the event duration sequence and the recovery time sequence, correlation logic calculations are performed to generate a final voltage sag event report.

[0013] Preferably, the step of acquiring the output voltage signal of the single-phase inverter, performing frequency domain analysis and preliminary filtering to obtain the preliminarily filtered voltage amplitude sequence includes:

[0014] The output voltage signal is acquired from the single-phase inverter;

[0015] The output voltage signal is digitized to obtain a digitized voltage signal sequence;

[0016] Based on the digitized voltage signal sequence, frequency domain analysis is performed to obtain a frequency domain characteristic spectrum.

[0017] High-frequency noise components are extracted from the frequency domain feature spectrum. If the high-frequency noise components exceed a preset noise threshold, the high-frequency noise components are filtered out to obtain a pre-filtered voltage amplitude sequence.

[0018] Preferably, the step of performing dynamic tracking processing and determining candidate voltage sag events based on the pre-filtered voltage amplitude sequence to obtain an event candidate list includes:

[0019] Based on the pre-filtered voltage amplitude sequence, state estimation filtering is performed to obtain a smooth amplitude tracking curve;

[0020] Based on the smooth amplitude tracking curve, the voltage change is analyzed and calculated within a continuous time window to obtain the amplitude decrease magnitude of each time window.

[0021] If the magnitude of the voltage drop in each time window is greater than a preset percentage decrease, then the event corresponding to each time window is determined as a candidate voltage sag event, and a candidate event list is obtained.

[0022] Preferably, the step of performing signal decomposition and event boundary detail fusion on each candidate event in the event candidate list to obtain the actual voltage sag event confirmation result includes:

[0023] Based on each candidate event in the event candidate list, perform multi-scale decomposition to obtain the multi-scale components of each candidate event;

[0024] Based on the multi-scale components of each candidate event, transient mode features are extracted to obtain the transient mode features of each candidate event;

[0025] The transient mode features of each candidate event are matched with preset load mutation features. If they do not match, the amplitude drop information and event boundary details are fused to obtain the actual voltage sag event confirmation result.

[0026] Preferably, the step of comparing the confirmed results of the actual voltage sag events with a pre-established voltage sag event database, and determining the event as a recurring pattern event if the similarity is higher than a preset similarity threshold, and calculating and obtaining the event duration sequence, includes:

[0027] Based on the confirmation results of the actual voltage sag events, feature vectors are extracted and compared with a preset voltage sag event database to obtain a comparison similarity score.

[0028] If the comparison similarity score is higher than the preset similarity threshold, it is determined to be a repetitive pattern event, and the corresponding event boundary details are obtained.

[0029] Based on the event boundary details, timestamp correlation analysis and event duration calculation are performed to obtain the event duration sequence.

[0030] Preferably, the step of performing recovery boundary matching based on the event duration sequence, confirming the recovery time if the matching is successful, and performing sequence fusion to obtain the recovery time sequence includes:

[0031] Based on the event duration sequence, the event recovery phase is located, and recovery boundary features are extracted to obtain the recovery boundary sequence;

[0032] Based on the recovery boundary features, a comparison is made with a pre-established recovery boundary database. If the similarity is higher than a preset recovery threshold, the match is successful, and the event boundary detail sequence is obtained.

[0033] Based on the event boundary detail sequence, timestamp fusion and correlation calculations are performed to obtain the recovered time series.

[0034] Preferably, the step of performing correlation logic calculations based on the event duration sequence and the recovery time sequence to generate a final voltage sag event report includes:

[0035] Based on the event duration sequence and the recovery time sequence, the optimized detection result sequence is obtained by matching and verifying with the smooth amplitude tracking curve and performing attribute fusion.

[0036] Based on the optimized detection result sequence, multi-scale component decomposition is performed to obtain a refined version of the event boundary detail sequence;

[0037] Based on the refined event boundary detail sequence, the refined event duration and refined recovery time are calculated, and multi-scale analysis is performed to quantify the stationarity of the recovery process, thereby obtaining the system reliability assessment details.

[0038] The final voltage sag event report is generated based on the refined event duration, the refined recovery time, and the system reliability assessment details.

[0039] Secondly, the present invention provides a voltage sag detection device based on fast amplitude tracking of a single-phase inverter, comprising:

[0040] The signal acquisition and filtering module is used to acquire the output voltage signal of the single-phase inverter, perform frequency domain analysis and preliminary filtering, and obtain the voltage amplitude sequence after preliminary filtering.

[0041] The candidate event determination module is used to perform dynamic tracking processing and determine candidate voltage sag events based on the pre-filtered voltage amplitude sequence, and obtain an event candidate list.

[0042] The real event confirmation module is used to perform signal decomposition and event boundary detail fusion on each candidate event in the event candidate list to obtain the real voltage sag event confirmation result;

[0043] The duration calculation module is used to compare the actual voltage sag event confirmation result with a pre-established voltage sag event database. If the similarity is higher than a preset similarity threshold, it is determined to be a recurring pattern event, and the event duration sequence is calculated and obtained.

[0044] The recovery time calculation module is used to perform recovery boundary matching based on the event duration sequence. If the matching is successful, the recovery time is confirmed, and the sequence is fused to obtain the recovery time sequence.

[0045] The report generation module is used to perform correlation logic calculations based on the event duration sequence and the recovery time sequence to generate a final voltage sag event report.

[0046] Thirdly, the present invention also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the voltage sag detection method based on fast amplitude tracking of a single-phase inverter as described in any one of the above.

[0047] Fourthly, the present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the voltage sag detection method based on fast amplitude tracking of a single-phase inverter as described above.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] (1) This invention first performs frequency domain analysis on the acquired voltage signal to filter out high-frequency noise, and then performs multi-scale signal decomposition on candidate voltage sag events to extract their transient mode features and match them with preset load change features. This method not only focuses on the amplitude change of the signal, but also analyzes the structural features of the event in depth. By identifying and excluding transient interference events with load change features, this invention can effectively distinguish between real voltage sags and transient noise. Therefore, this method significantly reduces the problem of misjudgment and missed judgment caused by interference, and greatly improves the reliability and accuracy of voltage sag detection.

[0050] (2) This invention uses a state estimation filtering algorithm to dynamically track the voltage amplitude sequence after preliminary filtering, generating a smooth amplitude tracking curve that reflects voltage changes in real time. Compared to the traditional static threshold judgment method, this dynamic tracking curve can more sensitively capture instantaneous voltage fluctuations. By analyzing the drop amplitude of the smooth curve within a continuous time window to determine candidate events, this invention achieves rapid and continuous tracking of voltage amplitude. Therefore, this method improves the real-time response capability and sensitivity of the detection system, enabling timely detection of voltage sag events, including minor sags.

[0051] (3) This invention ensures the accuracy of the output results through multi-stage post-processing, including verification, boundary refinement, and final calculation of confirmed real events. The process first verifies the preliminary detection results against a smooth curve, then uses multi-scale decomposition to precisely locate the event boundaries, and finally recalculates the event parameters based on the most accurate boundary data and performs a reliability assessment. Therefore, this method not only detects transient events but also provides high-precision event reports containing accurate start and end times, duration, and recovery characteristics, providing reliable data support for subsequent power grid fault diagnosis and system maintenance. Attached Figure Description

[0052] Figure 1 This is a schematic flowchart of the voltage sag detection method based on fast amplitude tracking of a single-phase inverter provided in the first embodiment of the present invention;

[0053] Figure 2 This is a schematic diagram of the voltage sag detection device based on rapid amplitude tracking of a single-phase inverter provided in the second embodiment of the present invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] Reference Figure 1 The first embodiment of the present invention provides a voltage sag detection method based on fast amplitude tracking of a single-phase inverter, comprising the following steps:

[0056] S11: Obtain the output voltage signal of the single-phase inverter, perform frequency domain analysis and preliminary filtering, and obtain the voltage amplitude sequence after preliminary filtering;

[0057] S12, Based on the voltage amplitude sequence after preliminary filtering, perform dynamic tracking processing and determine candidate voltage sag events to obtain an event candidate list;

[0058] S13, For each candidate event in the event candidate list, perform signal decomposition and event boundary detail fusion to obtain the actual voltage sag event confirmation result;

[0059] S14. Based on the confirmation result of the real voltage sag event, compare it with the pre-established voltage sag event database. If the similarity is higher than the preset similarity threshold, it is determined to be a recurring pattern event, and the event duration sequence is calculated and obtained.

[0060] S15, perform recovery boundary matching based on the event duration sequence. If the matching is successful, confirm the recovery time and perform sequence fusion to obtain the recovery time sequence.

[0061] S16, perform correlation logic calculations based on the event duration sequence and the recovery time sequence to generate a final voltage sag event report.

[0062] In step S11, the output voltage signal of the single-phase inverter is acquired, and frequency domain analysis and preliminary filtering are performed to obtain the voltage amplitude sequence after preliminary filtering, including:

[0063] The output voltage signal is acquired from the single-phase inverter;

[0064] The output voltage signal is digitized to obtain a digitized voltage signal sequence;

[0065] Based on the digitized voltage signal sequence, frequency domain analysis is performed to obtain a frequency domain characteristic spectrum.

[0066] High-frequency noise components are extracted from the frequency domain feature spectrum. If the high-frequency noise components exceed a preset noise threshold, the high-frequency noise components are filtered out to obtain a pre-filtered voltage amplitude sequence.

[0067] First, the output voltage signal of the single-phase inverter is acquired and subjected to frequency domain analysis and preliminary filtering to obtain the voltage amplitude sequence after preliminary filtering. The specific implementation method is as follows: a high-speed analog-to-digital converter connected to the output terminal of the single-phase inverter is used for periodic sampling to acquire the instantaneous analog signal of the output voltage. The sampling frequency here must follow the Nyquist sampling theorem, which states that in order to reconstruct the original signal from the sampling point without distortion, the sampling frequency must be at least twice the highest frequency component in the signal to avoid signal distortion caused by frequency aliasing. Therefore, in this embodiment, based on the general understanding of grid harmonics and experimental statistics, the sampling frequency is set to 10kHz, thereby converting the continuous analog voltage signal into a discrete, digitized voltage signal sequence containing specific quantized amplitude and timestamp information.

[0068] It should be noted that the specific values ​​of the parameters described in this invention (such as sampling frequency, noise figure, descent percentage, similarity threshold, etc.) are recommended or typical values ​​determined based on common knowledge in the technical field, industry standards (such as IEC 61000-4-30), and extensive prior experimental verification. Those skilled in the art will fully understand that the above parameters can be adaptively adjusted according to the actual application scenario, the specific model of the single-phase inverter, the power grid environment, and the required balance between detection accuracy and real-time performance.

[0069] For example, the sampling frequency should be at least twice the highest frequency component of the signal to be analyzed. Considering the 50th harmonic (2.5kHz) which is usually of interest, choosing 10kHz provides sufficient margin. A noise figure of 0.02 means setting the high-frequency band noise energy threshold to 2% of the fundamental frequency energy amplitude, which achieves a practical balance between effectively filtering out anomalous noise and preserving normal harmonic components. The mathematical essence of all formulas is consistent with the general definitions in the relevant field, and for clarity, those skilled in the art can directly understand them as corresponding mathematical formulas.

[0070] Next, based on the digitized voltage signal sequence, a Fast Fourier Transform is applied to map the time-domain signal to the frequency domain. The specific operation process is as follows: the digitized voltage signal sequence is taken as a finite-length input vector, and the original time-domain signal is decomposed into a linear combination of a series of sine and cosine wave components of different frequencies. The output of this transformation is a set of complex numbers, where the magnitude and phase angle of each complex number represent the amplitude and phase of the corresponding frequency component, respectively. By extracting the amplitude corresponding to each frequency point, the time-domain signal is mapped to the frequency domain, resulting in a frequency domain characteristic spectrum that can intuitively reflect the energy distribution of the signal at each frequency point. This spectrum is a data structure with frequency as the horizontal axis and amplitude as the vertical axis, clearly showing the amplitude information of the fundamental frequency (e.g., 50Hz) and its harmonics (e.g., 150Hz, 250Hz, etc.).

[0071] Subsequently, high-frequency noise components are extracted from the frequency domain characteristic spectrum. Specifically, a total high-frequency noise amplitude is obtained by calculating the sum of squares of the amplitudes of each frequency point in a specified high-frequency band (e.g., all frequency points above 500Hz). This total high-frequency noise amplitude is then compared with a preset noise threshold. The preset noise threshold is determined based on statistical analysis of operating data from a large number of single-phase inverters under different loads and electromagnetic environments, combined with industry standards for harmonic distortion in the power system. It is set as a specific percentage of the fundamental frequency amplitude. The calculation method is as follows: the noise threshold equals the square of the amplitude at the fundamental frequency multiplied by a preset noise figure. The noise figure is set based on: firstly, an empirical value obtained after statistical analysis of operating data from a large number of single-phase inverters under different loads and electromagnetic environments, ensuring the universality of the threshold; secondly, it is combined with industry standards for harmonic distortion in the power system, which typically limit the harmonic content generated by grid-connected equipment to a specific range. Therefore, in this embodiment, the coefficient is set to 0.0004, which strikes a balance between effectively filtering out abnormal noise and avoiding over-filtering of harmonics within the standard allowable range.

[0072] If the high-frequency noise energy exceeds the noise threshold, a digital low-pass filter is activated to filter the original digitized voltage signal sequence. The specific filtering process is as follows: First, a set of filter coefficients is determined based on a preset cutoff frequency (e.g., 200Hz) and filter order. Then, the digitized voltage signal sequence is convolved with the filter coefficients. Specifically, for each sampling point, its current value and several historical values ​​are weighted and summed with the corresponding filter coefficients to calculate a new output sampling point. By performing this convolution operation point-by-point on the entire sequence, a new time series is generated, in which frequency components above the cutoff frequency are effectively attenuated. If the threshold is not exceeded, no filtering operation is performed, and the original sequence is directly used as the output. Ultimately, regardless of whether filtering is performed, a pre-filtered voltage amplitude sequence with effectively suppressed high-frequency noise is obtained, providing a high-quality data foundation for subsequent dynamic tracking.

[0073] In step S12, based on the pre-filtered voltage amplitude sequence, dynamic tracking processing is performed and candidate voltage sag events are determined to obtain an event candidate list, including:

[0074] Based on the pre-filtered voltage amplitude sequence, state estimation filtering is performed to obtain a smooth amplitude tracking curve;

[0075] Based on the smooth amplitude tracking curve, the voltage change is analyzed and calculated within a continuous time window to obtain the amplitude decrease magnitude of each time window.

[0076] If the magnitude of the voltage drop in each time window is greater than a preset percentage decrease, then the event corresponding to each time window is determined as a candidate voltage sag event, and a candidate event list is obtained.

[0077] First, based on the pre-filtered voltage amplitude sequence, a state estimation filtering process is performed. In this embodiment, the Kalman filter algorithm is used, which is a recursive optimal estimation algorithm. The process first initializes the parameters and the state: the system's state vector is defined as a two-dimensional vector containing the current voltage amplitude and its rate of change; the state vector is initialized, usually using the amplitude of the first sampling point as the initial amplitude, and the rate of change is set to 0; at the same time, based on a large amount of statistical analysis of the inverter's historical output signal characteristics, the process noise covariance matrix (characterizing the uncertainty of the voltage change model) and the measurement noise covariance (characterizing the noise level of the measurement process) are pre-set.

[0078] After initialization, the system enters the prediction phase, predicting the state at the next time step using the state transition equation. This equation describes the physical model of the evolution of voltage amplitude and rate of change over time. Specifically, the state prediction formula is: the predicted state at time k equals the state transition matrix multiplied by the posterior state estimate at time k-1. Simultaneously, the covariance at that time step is predicted. Specifically, the predicted covariance at time k equals the state transition matrix multiplied by the posterior covariance estimate at time k-1, then multiplied by the transpose of the state transition matrix, and finally added to the process noise covariance matrix.

[0079] After the prediction is completed, the update phase begins. First, the Kalman gain is calculated. This gain is used to balance the confidence of the predicted and measured values. The specific calculation process is as follows: the Kalman gain equals the prediction covariance at time k multiplied by the transpose of the observation matrix, then multiplied by the inverse of the matrix (the observation matrix H multiplied by the prediction covariance at time k, then multiplied by the transpose of H, and finally the measurement noise covariance R). Here, the observation matrix is ​​used to map the state vector to the measurement space, and its value is [1, 0]. Subsequently, the state prediction is corrected based on this gain to obtain the optimal estimate for the current time. The specific correction operation is as follows: the posterior state estimate at time k equals the predicted state at time k, plus the Kalman gain multiplied by (the actual observed value at time k minus the observation matrix H multiplied by the predicted state at time k).

[0080] The covariance is updated to reflect the uncertainty of this estimation. The specific correction rule is as follows: the posterior covariance estimate at time k is equal to the identity matrix minus the Kalman gain multiplied by the observation matrix, and finally multiplied by the prediction covariance at time k. By recursively performing the above prediction and update operations on each sampling point in the initially filtered voltage amplitude sequence, the voltage amplitude components in the final output posterior state estimate are connected to obtain a smooth amplitude tracking curve that has been optimally estimated and whose noise has been further suppressed.

[0081] Next, based on the smoothed amplitude tracking curve, the voltage change is analyzed and calculated within a continuous time window using the sliding window method. The specific analysis process is as follows: First, a fixed-size time window is set, its size determined by the typical duration of voltage sag events in the power system and the requirements for real-time response, for example, 20 milliseconds. Second, the sliding step size of the window is set, typically one sampling point period to achieve the finest point-by-point analysis. At the start of the analysis, the time window is placed at the beginning of the smoothed amplitude tracking curve, and the data within the window is analyzed and calculated.

[0082] After each calculation, the time window is moved forward one step along the time axis, and the analysis and calculation are repeated for the data within the new window. This process is repeated until the entire amplitude tracking curve has been traversed and analyzed. Within each time window, the amplitude decrease is calculated using the following formula: Amplitude decrease equals (reference voltage minus minimum amplitude within the window) divided by the reference voltage, then multiplied by 100%. The reference voltage is the average effective voltage (RMS) over a relatively long period prior to the event, as defined by industry standards, and is used to represent the normal voltage level.

[0083] Finally, the calculated amplitude drop for each time window is compared with a preset percentage decrease. This preset percentage decrease is determined based on the International Electrotechnical Commission (IEC) and related power quality standards' definition of a voltage sag event (i.e., the effective voltage value drops below 90% of the rated value), and in this embodiment, it is set to 10%. If the amplitude drop for any time window is greater than the preset percentage decrease, the start time, duration, and minimum amplitude of that time window are recorded as a candidate voltage sag event and added to the event candidate list. If the result is not greater than, the voltage within that time window is considered normal. After analyzing the entire amplitude tracking curve, the final event candidate list will be used for subsequent interference identification processing.

[0084] In step S13, for each candidate event in the event candidate list, signal decomposition and event boundary detail fusion are performed to obtain the actual voltage sag event confirmation result, including:

[0085] Based on each candidate event in the event candidate list, perform multi-scale decomposition to obtain the multi-scale components of each candidate event;

[0086] Based on the multi-scale components of each candidate event, transient mode features are extracted to obtain the transient mode features of each candidate event;

[0087] The transient mode features of each candidate event are matched with preset load mutation features. If they do not match, the amplitude drop information and event boundary details are fused to obtain the actual voltage sag event confirmation result.

[0088] First, for each candidate event in the event candidate list, the original digitized voltage signal sequence corresponding to that event's time window is decomposed into a multi-scale decomposition. In this embodiment, this decomposition is implemented using discrete wavelet transform. The process is as follows: the Daubechies4 wavelet basis, which has shown excellent performance in transient signal analysis based on experimental comparisons, is selected, and an orthogonal mirror filter bank is constructed according to the Marat algorithm to decompose the signal step by step. Specifically, in the first-level decomposition, the signal passes through a high-pass filter and a low-pass filter simultaneously to obtain the high-frequency detail component (D1) and low-frequency approximation component (A1) of the first level, respectively. Subsequently, the low-frequency approximation component (A1) of the first level is used as input and fed back into the same filter bank for the second-level decomposition to obtain D2 and A2. This process iteratively executes the preset number of decomposition levels, decomposing the original signal into a series of low-frequency approximation components and high-frequency detail components at different frequency scales. The high-frequency detail components can accurately capture the instantaneous change characteristics of the signal, ultimately yielding the multi-scale components of each candidate event.

[0089] Next, based on the multi-scale components of each candidate event, transient mode features are extracted from its high-frequency detail components. These features are parameter vectors used to quantify the characteristics of transient processes. The specific extraction process is as follows: For the high-frequency detail component coefficient sequence of each decomposition level, firstly, the maximum value of the absolute value of the coefficient is found by traversing the sequence, and this value is used as a feature characterizing the transient energy intensity. Secondly, the energy entropy is calculated by first calculating the energy of each coefficient in the sequence, i.e., the square of the coefficient, and then dividing the energy of each coefficient by the total energy to obtain its energy proportion. Finally, the following formula is used: the energy entropy is equal to the sum of the negative energy proportions multiplied by the logarithms of the energy proportions to the base 2. This value is used to characterize the complexity of the transient signal. By calculating these parameters, a transient mode feature vector that can describe the transient behavior of each candidate event is obtained.

[0090] Subsequently, based on the transient mode characteristics of each candidate event, it is matched against a pre-defined load mutation feature template library. This template library is constructed by experimentally collecting and decomposing a large number of known voltage waveforms generated when large inductive loads (such as motors) are started or disconnected, and statistically analyzing the distribution range and typical values ​​of their transient mode characteristics. The matching process is achieved by calculating the Euclidean distance between the feature vector of the candidate event and each feature template in the template library, and the specific calculation process follows the definition of Euclidean distance.

[0091] If the Euclidean distance between the feature vector of a candidate event and any template in the template library is less than a preset matching threshold, then the event is determined to match the load mutation feature, identified as an interference event, and removed from the list. The matching threshold is determined based on statistical analysis of the Euclidean distance distribution of a large number of known matching samples (real interference) and non-matching samples (real transients), selecting a value that optimally balances the false negative rate and the false positive rate.

[0092] If the judgment result is that the Euclidean distance between the feature vector of a candidate event and any template in the template library is not less than the matching threshold, then the event is determined to be mismatched with the load mutation feature. At this time, the process of determining and fusing event boundary details is initiated. The specific process is as follows: taking the timestamp of the candidate event as the center, on the smooth amplitude tracking curve generated in step S12, an edge positioning method based on the analysis of the first and second derivatives of the signal is used to accurately locate the starting point of the voltage drop and the ending point of recovery to the stable state.

[0093] The specific positioning operation process is as follows: The smooth amplitude tracking curve is treated as a one-dimensional signal. First, its gradient sequence is obtained by calculating the first derivative of the signal. The starting point where the voltage begins to decrease is identified as the location where a significant negative peak appears in the gradient sequence. Second, the ending point where the voltage recovers to a stable state is identified by calculating the second derivative, where the second derivative exhibits a zero crossover and the first derivative tends to stabilize. These two time points, along with the amplitude decrease depth and duration during this period, are fused to form an event record containing complete boundary and amplitude information. Finally, the set of all event records that have passed the mismatch judgment and successfully fused with boundary details constitutes the confirmation result of the real voltage sag event. This result list excludes false sags caused by load mutations, ensuring the accuracy of subsequent analysis. The confirmation result of the real voltage sag event is a data structure containing confirmed real sag events and their precise boundary details (such as starting point, ending point, amplitude depth, etc.). In the subsequent step S14, obtaining the corresponding event boundary details involves extracting relevant information about specific events from this result.

[0094] In step S14, based on the confirmation result of the actual voltage sag event, a comparison is made with a pre-established voltage sag event database. If the similarity is higher than a preset similarity threshold, it is determined to be a recurring pattern event, and the event duration sequence is calculated and obtained, including:

[0095] Based on the confirmation results of the actual voltage sag events, feature vectors are extracted and compared with a preset voltage sag event database to obtain a comparison similarity score.

[0096] If the comparison similarity score is higher than the preset similarity threshold, it is determined to be a repetitive pattern event, and the corresponding event boundary details are obtained.

[0097] Based on the event boundary details, timestamp correlation analysis and event duration calculation are performed to obtain the event duration sequence.

[0098] First, based on the confirmed results of the actual voltage sag events, a feature vector is extracted for each confirmed event. This feature vector is a multi-dimensional array used to quantitatively describe the core characteristics of the event, and its dimensions include parameters such as the amplitude drop depth, waveform distortion rate during the sag, and transient mode characteristics determined in step S13. Simultaneously, a preset voltage sag event database is invoked. This database is constructed by storing a large number of feature templates of typical voltage sag events caused by specific reasons (such as specific types of load switching, grid-side faults, etc.) through long-term monitoring and historical data accumulation. Then, a cosine similarity algorithm is used to compare the feature vector of the current event with each feature template stored in the database one by one. The specific calculation process is as follows: the cosine similarity is equal to the dot product of the current event feature vector and the template feature vector divided by the product of the magnitudes of the two vectors. This calculation generates a similarity score between 0 and 1 for each template.

[0099] Next, the obtained similarity score is compared with a preset similarity threshold. This threshold is determined by statistical methods such as receiver operating characteristic (ROC) curve analysis, testing a large number of known matching and non-matching events, and selecting a balance point that maximizes the correct recognition rate while minimizing the false recognition rate. In this embodiment, it is set to 0.9. If the judgment result is that the similarity score is higher than the preset similarity threshold, the real event is judged as a repetitive pattern event corresponding to the template, and the event boundary details determined for the event in step S13 are directly obtained; if the judgment result is not higher, the event is considered a new pattern event, and no further processing is performed or it is marked for manual analysis.

[0100] Finally, based on the event boundary details of the recurring pattern events, timestamp correlation analysis and event duration calculation are performed. The specific calculation process is as follows: extract the event start timestamp and event end timestamp recorded in the event boundary details, and calculate using the following formula: event duration equals event end timestamp minus event start timestamp. For example, if the start timestamp is 10:00:03.100 and the end timestamp is 10:00:03.300, the calculated duration is 200 milliseconds. The calculated durations of all events identified as recurring pattern events are then aggregated to obtain the event duration sequence.

[0101] In step S15, based on the event duration sequence, recovery boundary matching is performed. If the matching is successful, the recovery time is confirmed, and sequence fusion is performed to obtain the recovery time sequence, including:

[0102] Based on the event duration sequence, the event recovery phase is located, and recovery boundary features are extracted to obtain the recovery boundary sequence;

[0103] Based on the recovery boundary features, a comparison is made with a pre-established recovery boundary database. If the similarity is higher than a preset recovery threshold, the match is successful, and the event boundary detail sequence is obtained.

[0104] Based on the event boundary detail sequence, timestamp fusion and correlation calculations are performed to obtain the recovered time series.

[0105] First, based on the timestamp information of each event recorded in the event duration sequence, the event recovery phase immediately following the end of the sag is located in the smooth amplitude tracking curve generated in step S12, and the waveform of this phase is analyzed to extract its recovery boundary features. The specific analysis and extraction process is as follows: for the voltage amplitude sequence of the recovery phase, the voltage recovery slope is calculated by linear regression fitting or by connecting the two points of recovery start and end. The specific calculation formula is: the voltage recovery slope is equal to (amplitude at the end of recovery minus amplitude at the start of recovery) divided by (time at the end of recovery minus time at the start of recovery).

[0106] Simultaneously, the transient fluctuation amplitude during the recovery process is calculated. The process involves first determining an ideal recovery trend line, and then calculating the root mean square value of the difference between the actual amplitude of each sampling point during the recovery phase and the amplitude of the corresponding point on the trend line. The recovery boundary feature is a vector used to quantitatively describe the dynamic characteristics of the recovery process, which consists of parameters such as the voltage recovery slope and transient fluctuation amplitude calculated above. The recovery boundary sequence of each event is obtained by calculating these parameters.

[0107] Next, based on the described recovery boundary features, it is compared with a pre-established recovery boundary database. This database is constructed similarly to the voltage sag event database and load mutation feature template library. It is built by extracting features (such as recovery slope and statistics of fluctuation amplitude) from known, clearly identifiable voltage recovery waveform segments in historical data. The feature templates also include feature vectors such as recovery slope and transient fluctuation amplitude. Those skilled in the art can continuously expand and optimize this database based on actual monitoring data. The comparison process uses a cosine similarity algorithm to calculate the similarity between the recovery boundary feature vector of the current event and the feature vectors of each template in the database. The specific calculation process follows the definition of cosine similarity.

[0108] If the calculated similarity score is higher than a preset recovery threshold (this threshold is set based on statistical methods such as ROC curve analysis to control the false match rate while ensuring a high recognition rate), then the match is considered successful, and the event boundary detail sequence determined for the event in step S13 is obtained.

[0109] If the judgment result is not higher than the threshold, the recovery mode is considered an unknown mode. In a preferred embodiment, manual verification and calculation are introduced at this time, allowing professionals to verify and confirm on the corresponding interface, thereby obtaining the recovery time series and acquiring the event boundary detail sequence determined for the event in step S13. Simultaneously, a self-learning mechanism is introduced to update the recovery boundary database. Once a new recovery mode with a clear cause is manually confirmed, its feature vector and recovery time data can be added to the recovery boundary database as a new template. In another embodiment, to simplify the calculation process, the unknown mode event can be ignored and no further processing is performed.

[0110] Finally, based on the obtained event boundary detail sequence, timestamp fusion and correlation calculations are performed. The specific process is as follows: The recovery start timestamp (i.e., the end of the sag) and recovery end timestamp (i.e., the moment the voltage amplitude re-enters and stabilizes within the normal range) are extracted from the event boundary detail sequence. The recovery time is calculated using the following formula: recovery time equals recovery end timestamp minus recovery start timestamp. The calculated recovery times for each event are then aggregated to obtain the final recovery time sequence.

[0111] In step S16, based on the event duration sequence and the recovery time sequence, correlation logic calculations are performed to generate a final voltage sag event report, including:

[0112] Based on the event duration sequence and the recovery time sequence, the optimized detection result sequence is obtained by matching and verifying with the smooth amplitude tracking curve and performing attribute fusion.

[0113] Based on the optimized detection result sequence, multi-scale component decomposition is performed to obtain a refined version of the event boundary detail sequence;

[0114] Based on the refined event boundary detail sequence, the refined event duration and refined recovery time are calculated, and multi-scale analysis is performed to quantify the stationarity of the recovery process, thereby obtaining the system reliability assessment details.

[0115] The final voltage sag event report is generated based on the refined event duration, the refined recovery time, and the system reliability assessment details.

[0116] First, based on the event duration sequence and the recovery time sequence, a matching verification and attribute fusion are performed with the smooth amplitude tracking curve generated in step S12. The specific process is as follows: using the start and end timestamps in the time sequence, the corresponding waveform segment is located on the smooth curve, and the amplitude decrease characteristic of the segment is verified to be consistent with the initial judgment. After the verification is passed, the specific fusion operation is as follows: a new data structure is created, and the duration parameters in the original time sequence (such as the initially calculated duration and recovery duration) are used as the basic fields. Then, the detailed physical attributes extracted from the waveform segment (such as the average amplitude during the descent and the minimum amplitude) are added as new fields to the data structure, thereby forming a more complete and optimized detection result sequence.

[0117] Next, based on the optimized detection result sequence, to further improve boundary accuracy, multi-scale component decomposition is performed on the event boundary. The specific application and correction process is as follows: First, the discrete wavelet transform described in step S13 is performed on the signal segment to obtain high-frequency detail component coefficients. Second, the instantaneous energy of this coefficient sequence is calculated, i.e., the square of each coefficient value. Then, the precise time point at which the energy value first experiences a sharp jump and exceeds a preset multiple (e.g., 3 times the standard deviation) of the background noise energy is detected in the instantaneous energy sequence is identified as the energy mutation point. Finally, the timestamp corresponding to this energy mutation point is used as a new, more precise boundary timestamp to update and correct the original boundary, thereby obtaining a refined version of the event boundary detail sequence accurate to the level of a single sampling point.

[0118] Then, based on the refined version of the event boundary detail sequence, the event duration and recovery time are recalculated (based on the already calculated "event duration sequence" and "recovery time sequence", optimized calculation is performed based on a more accurate boundary). That is, the updated and more accurate start and end timestamps are used for subtraction, and the stationarity of the recovery process is quantified. The specific multi-scale analysis process is as follows: the recovery stage waveform defined by the refined boundary is input as an independent signal segment, and the discrete wavelet transform described in step S13 is performed on it to obtain the high-frequency detail component coefficients of the recovery process at different scales.

[0119] Subsequently, the stationarity of the recovery process is quantified by calculating the energy variance of these high-frequency detail component coefficients. The smaller the energy variance, the more stable the recovery process and the higher the reliability. This quantification result is the system reliability assessment detail.

[0120] Finally, based on the recalculated event duration, recovery time, and system reliability assessment details, all these high-precision parameters are integrated according to a preset data structure. The specific integration process is as follows: a preset data structure template containing multiple fields is invoked. The structure of this template is set according to the general standards in the field of power quality monitoring and the data requirements of power grid operation and maintenance personnel for fault analysis. This template defines the format of the final report.

[0121] Then, the high-precision parameters obtained in the preceding steps are assigned one by one to the corresponding preset fields in the data structure template. The specific assignment operation is a direct key-value mapping process. The rule is as follows: each calculated or evaluated parameter value (e.g., a recovery duration of '150 milliseconds') is used as a 'value' and accurately filled into the semantically corresponding field in the template (e.g., a field named 'recovery_duration_ms'). This mapping rule predefines a unique correspondence between each parameter and the template field. After all fields are assigned, a structured and complete event record is formed, which is the final voltage sag event report, serving as the final output of the entire detection process.

[0122] In summary, this invention uses signal decomposition technology to deeply analyze the transient characteristics of events, effectively distinguishing between real voltage sags and transient interferences such as load surges, solving the problem of low detection reliability and accuracy in existing technologies, and achieving accurate identification of minute voltage sags.

[0123] Reference Figure 2 The second embodiment of the present invention provides a voltage sag detection device based on fast amplitude tracking of a single-phase inverter, comprising:

[0124] The signal acquisition and filtering module is used to acquire the output voltage signal of the single-phase inverter, perform frequency domain analysis and preliminary filtering, and obtain the voltage amplitude sequence after preliminary filtering.

[0125] The candidate event determination module is used to perform dynamic tracking processing and determine candidate voltage sag events based on the pre-filtered voltage amplitude sequence, and obtain an event candidate list.

[0126] The real event confirmation module is used to perform signal decomposition and event boundary detail fusion on each candidate event in the event candidate list to obtain the real voltage sag event confirmation result;

[0127] The duration calculation module is used to compare the actual voltage sag event confirmation result with a pre-established voltage sag event database. If the similarity is higher than a preset similarity threshold, it is determined to be a recurring pattern event, and the event duration sequence is calculated and obtained.

[0128] The recovery time calculation module is used to perform recovery boundary matching based on the event duration sequence. If the matching is successful, the recovery time is confirmed, and the sequence is fused to obtain the recovery time sequence.

[0129] The report generation module is used to perform correlation logic calculations based on the event duration sequence and the recovery time sequence to generate a final voltage sag event report.

[0130] It should be noted that the voltage sag detection device based on fast amplitude tracking of a single-phase inverter provided in this embodiment of the invention is used to execute all the process steps of the voltage sag detection method based on fast amplitude tracking of a single-phase inverter in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0131] This invention also provides an electronic device. The electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a voltage sag detection program based on fast amplitude tracking of a single-phase inverter. When the processor executes the computer program, it implements the steps in the various embodiments of the voltage sag detection method based on fast amplitude tracking of a single-phase inverter described above, for example... Figure 1 The step S11 shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above-described device embodiments, such as the signal acquisition and filtering module.

[0132] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0133] The electronic device may be a desktop computer, laptop, handheld computer, or smart tablet, etc. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. It may include more or fewer components than described above, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0134] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting all parts of the electronic device via various interfaces and lines.

[0135] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0136] Wherein, if the modules / units integrated in the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0137] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0138] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A voltage sag detection method based on fast amplitude tracking of a single-phase inverter, characterized in that, include: The output voltage signal of the single-phase inverter is acquired, and frequency domain analysis and preliminary filtering are performed to obtain the voltage amplitude sequence after preliminary filtering. Based on the pre-filtered voltage amplitude sequence, dynamic tracking processing is performed and candidate voltage sag events are determined to obtain a candidate event list; For each candidate event in the event candidate list, signal decomposition and event boundary detail fusion are performed to obtain the actual voltage sag event confirmation result; Based on the confirmation results of the actual voltage sag events, a comparison is made with the pre-established voltage sag event database. If the similarity is higher than the preset similarity threshold, it is determined to be a recurring pattern event, and the event duration sequence is calculated and obtained. Based on the event duration sequence, recovery boundary matching is performed. If the matching is successful, the recovery time is confirmed, and the sequence is fused to obtain the recovery time sequence. Based on the event duration sequence and the recovery time sequence, perform correlation logic calculations to generate a final voltage sag event report; The step of dynamically tracking and determining candidate voltage sag events based on the pre-filtered voltage amplitude sequence to obtain a candidate event list includes: Based on the pre-filtered voltage amplitude sequence, state estimation filtering is performed to obtain a smooth amplitude tracking curve; Based on the smooth amplitude tracking curve, the voltage change is analyzed and calculated within a continuous time window to obtain the amplitude decrease magnitude of each time window. If the magnitude of the drop in the amplitude of each time window is greater than the preset percentage drop, then the event corresponding to each time window is determined as a candidate voltage sag event, and an event candidate list is obtained. Specifically, the step of performing signal decomposition and event boundary detail fusion on each candidate event in the event candidate list to obtain the actual voltage sag event confirmation result includes: Based on each candidate event in the event candidate list, perform multi-scale decomposition to obtain the multi-scale components of each candidate event; Based on the multi-scale components of each candidate event, transient mode features are extracted to obtain the transient mode features of each candidate event; The transient mode features of each candidate event are matched with the preset load change features. If they do not match, the amplitude drop information and event boundary details are fused to obtain the actual voltage sag event confirmation result. The step of performing recovery boundary matching based on the event duration sequence, confirming the recovery time if a match is successful, and performing sequence fusion to obtain the recovery time sequence includes: Based on the event duration sequence, the event recovery phase is located, and recovery boundary features are extracted to obtain the recovery boundary sequence; Based on the recovery boundary features, a comparison is made with a pre-established recovery boundary database. If the similarity is higher than a preset recovery threshold, the match is successful, and the event boundary detail sequence is obtained. Based on the event boundary detail sequence, timestamp fusion and correlation calculation are performed to obtain the recovered time series; The step of performing correlation logic calculations based on the event duration sequence and the recovery time sequence to generate a final voltage sag event report includes: Based on the event duration sequence and the recovery time sequence, the optimized detection result sequence is obtained by matching and verifying with the smooth amplitude tracking curve and performing attribute fusion. Based on the optimized detection result sequence, multi-scale component decomposition is performed to obtain a refined version of the event boundary detail sequence; Based on the refined event boundary detail sequence, the refined event duration and refined recovery time are calculated, and multi-scale analysis is performed to quantify the stationarity of the recovery process, thereby obtaining the system reliability assessment details. The final voltage sag event report is generated based on the refined event duration, the refined recovery time, and the system reliability assessment details.

2. The voltage sag detection method based on fast amplitude tracking of a single-phase inverter according to claim 1, characterized in that, The process of acquiring the output voltage signal of the single-phase inverter, performing frequency domain analysis and preliminary filtering to obtain the preliminarily filtered voltage amplitude sequence includes: The output voltage signal is acquired from the single-phase inverter; The output voltage signal is digitized to obtain a digitized voltage signal sequence; Based on the digitized voltage signal sequence, frequency domain analysis is performed to obtain a frequency domain characteristic spectrum. High-frequency noise components are extracted from the frequency domain feature spectrum. If the high-frequency noise components exceed a preset noise threshold, the high-frequency noise components are filtered out to obtain a pre-filtered voltage amplitude sequence.

3. The voltage sag detection method based on fast amplitude tracking of a single-phase inverter according to claim 1, characterized in that, The process involves comparing the confirmed results of the actual voltage sag events with a pre-established voltage sag event database. If the similarity score is higher than a preset similarity threshold, the event is determined to be a recurring pattern event. The event duration sequence is then calculated and obtained, including: Based on the confirmation results of the actual voltage sag events, feature vectors are extracted and compared with a preset voltage sag event database to obtain a similarity score. If the comparison similarity score is higher than the preset similarity threshold, it is determined to be a repetitive pattern event, and the corresponding event boundary details are obtained. Based on the event boundary details, timestamp correlation analysis and event duration calculation are performed to obtain the event duration sequence.

4. A voltage sag detection device based on fast amplitude tracking of a single-phase inverter, characterized in that, The voltage sag detection method based on fast amplitude tracking of a single-phase inverter as described in any one of claims 1 to 3 includes: The signal acquisition and filtering module is used to acquire the output voltage signal of the single-phase inverter, perform frequency domain analysis and preliminary filtering, and obtain the voltage amplitude sequence after preliminary filtering. The candidate event determination module is used to perform dynamic tracking processing and determine candidate voltage sag events based on the pre-filtered voltage amplitude sequence, and obtain an event candidate list. The real event confirmation module is used to perform signal decomposition and event boundary detail fusion on each candidate event in the event candidate list to obtain the real voltage sag event confirmation result; The duration calculation module is used to compare the actual voltage sag event confirmation result with a pre-established voltage sag event database. If the similarity is higher than a preset similarity threshold, it is determined to be a recurring pattern event, and the event duration sequence is calculated and obtained. The recovery time calculation module is used to perform recovery boundary matching based on the event duration sequence. If the matching is successful, the recovery time is confirmed, and the sequence is fused to obtain the recovery time sequence. The report generation module is used to perform correlation logic calculations based on the event duration sequence and the recovery time sequence to generate a final voltage sag event report.

Images