Voltage sudden change identification and grid-connected protection control method for grid fault ride-through

Abstract

Embodiments of the present disclosure disclose a voltage mutation identification and grid-connected protection control method for power grid fault ride-through. A specific embodiment of the method comprises: in response to detecting a voltage mutation feature point identification instruction, receiving power system time series data; extracting a time series detection value sequence from the power system time series data; performing a change rate domain coarse screening operation on the time series detection value sequence to obtain a candidate feature point set; performing time clustering and boundary filtering on the candidate feature point set to obtain a feature point set; performing amplitude domain recursive fine screening on the feature point set to obtain a fault feature point set; generating a control instruction according to the fault feature point set, and sending the control instruction to a power system protection device to trigger a protection action or reset a protection state. The embodiment relates to the computer technical field, and can realize accurate identification of voltage mutation points under weak power grid slow-changing faults and high-noise environments, and convert the identification results into protection control instructions.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the field of computer technology, and more specifically to a method for voltage surge identification and grid-connected protection control for power grid fault ride-through. Background Technology

[0002] With the increasing penetration rate of new energy power generation, a large number of photovoltaic and wind power generation devices are being connected to the end of weak power grids. Weak power grids are characterized by small short-circuit capacity, low system inertia, and high impedance ratios, causing the grid voltage to exhibit dynamic characteristics drastically different from traditional strong power grids when subjected to disturbances. Among these, slow voltage fluctuation faults are one of the most representative fault modes under weak power grid conditions: when the output of new energy power generation equipment fluctuates or a non-metallic grounding occurs on a remote line, the grid connection point voltage does not instantly drop to its lowest point, but rather decreases slowly over tens or even hundreds of milliseconds. The drop depth is shallow, the rate of change is low, and the recovery phase also shows a slow upward trend. In the grid connection testing of new energy power generation equipment, the Low Voltage Ride Through (LVRT) test is a core component in evaluating whether the equipment meets grid connection standards (such as GB / T 32892 and GB / T 19964). This test is typically performed on a Hardware-in-the-Loop (HIL) platform. It generates voltage sag waveforms in real time and applies them to the controller of the device under test (DUT) to verify its ability to operate without disconnecting from the grid and its reactive power support capabilities during faults. Accurately identifying the entry and exit points of voltage surge events is a prerequisite for automatic fault waveform capture, transient and steady-state partitioning, and standard compliance assessment. Figure 4 As shown, this illustrates the link between a new energy power station and the public power grid via a high-impedance interconnection line, the occurrence of a gradual fault at the grid connection point, and the identification and grid connection protection control of voltage surges.

[0003] However, existing methods for voltage surge identification and grid-connected protection control often suffer from the following technical problems: Gradual faults render traditional rate-of-change detection methods ineffective. Traditional methods typically use fixed thresholds based on first- or second-order voltage differentials to detect abrupt changes. However, for gradual faults in weak power grids, the voltage change rate is much lower than that of sudden-drop faults. A fixed threshold that is too high may lead to missed detections, while a threshold that is too low may introduce significant noise and false alarms. While the sliding window bilateral mean difference transform can amplify the cumulative change amplitude of the gradual signal to some extent, its effectiveness is highly dependent on the window width: if the window is too small, the cumulative effect of the gradual signal is insufficient, resulting in a lower peak amplitude after transformation; if the window is too large, the peak of a sudden-drop fault is flattened, reducing the accuracy of time positioning. Existing methods typically use a single fixed window, which cannot simultaneously adapt to various fault types with vastly different transition durations.

[0004] Clustering feature points is challenging in high-noise environments. In hardware-in-the-loop platforms, high-frequency operation of power electronic switching devices, electromagnetic interference, and measurement noise introduce numerous spurious feature points into voltage time-series data. These spurious feature points are often densely distributed around genuine feature points along the time axis, and their amplitude of variation is on the same order of magnitude as that of genuine feature points in shallow faults. Traditional temporal clustering methods merge data based solely on time intervals, lacking effective utilization of feature point amplitude information. This makes it difficult to distinguish between dense clusters formed by genuine fault events and those formed by noise, resulting in the inclusion of numerous spurious events or the omission of weak signal genuine events in the clustering results.

[0005] The disconnect between identification results and protection control is a significant issue. Existing methods for identifying voltage surge points typically stop at the time stamp of the output characteristic point for subsequent offline analysis, failing to directly translate the identification results into control commands for protection devices. In hardware-in-the-loop testing, the identified fault start and end times need to be transmitted in real-time to the platform controller or the protection device of the device under test to trigger fault recording, mode switching, or protection actions. The lack of this closed-loop control link significantly diminishes the practical value of the identification results and fails to meet the real-time requirements of automated testing.

[0006] The information disclosed in this background section is only intended to enhance the understanding of the background of the present disclosure concept, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0007] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0008] Some embodiments of this disclosure propose a voltage surge identification and grid-connected protection control method for power grid fault ride-through, in order to solve the technical problems mentioned in the background section above.

[0009] In a first aspect, some embodiments of this disclosure provide a method for voltage surge identification and grid-connected protection control for power grid fault ride-through. The method includes: receiving power system time-series data in response to a detected voltage surge feature point identification command; extracting a time-series detection value sequence from the power system time-series data, wherein the time-series detection value sequence includes a timestamp and a detection value, and the detection value is a voltage per-unit value; performing a rate-of-change domain coarse screening operation on the time-series detection value sequence to obtain a candidate feature point set; performing time clustering and boundary filtering on the candidate feature point set to obtain a feature point set; performing an amplitude domain recursive fine screening on the feature point set to obtain a fault feature point set; generating a control command based on the fault feature point set; and sending the control command to the controller of a power system protection device, wherein the controller triggers a protection action at the moment the voltage surge occurs and resets the protection state at the moment the voltage surge is cleared, according to the control command.

[0010] Secondly, some embodiments of this disclosure provide a voltage surge identification and grid-connected protection control device for power grid fault ride-through. The device includes: a receiving unit configured to receive power system time-series data in response to a detected voltage surge feature point identification command; an extraction unit configured to extract a time-series detection value sequence from the power system time-series data, wherein the time-series detection value sequence includes a timestamp and a detection value, and the detection value is a voltage per-unit value; a coarse screening unit configured to perform a rate-of-change domain coarse screening operation on the time-series detection value sequence to obtain a candidate feature point set; a clustering unit configured to perform time clustering and boundary filtering on the candidate feature point set to obtain a feature point set; a fine screening unit configured to perform amplitude domain recursive fine screening on the feature point set to obtain a fault feature point set; and a generation unit configured to generate a control command based on the fault feature point set and send the control command to the controller of the power system protection device, wherein the controller triggers a protection action at the moment the voltage surge enters and resets the protection state at the moment the voltage surge clears.

[0011] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.

[0012] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.

[0013] The above-described embodiments of this disclosure have the following beneficial effects: The voltage mutation identification and grid-connected protection control method for power grid fault ride-through, through a combination of multi-level progressive screening and adaptive thresholding, achieves accurate identification of voltage mutation points in weak power grids with gradual fault changes and high noise environments, and directly converts the identification results into protection control commands. Specifically, the reasons for low accuracy in voltage mutation point identification, poor noise resistance, and disconnection from protection control are: traditional methods use fixed thresholds to detect change rates, which cannot adapt to the low change rate characteristics of gradual fault changes; single time clustering methods are difficult to distinguish between noise clusters and real event clusters; and the identification results remain at the offline analysis level, unable to directly drive the protection device to operate. Based on this, the voltage mutation identification and grid-connected protection control method for power grid fault ride-through, according to some embodiments of this disclosure, firstly, receives power system time-series data in response to the detected voltage mutation feature point identification command. This establishes a real-time data channel between the execution entity and the real-time simulator, providing a continuous and reliable source of raw data for subsequent online processing. Secondly, it extracts the time-series detection value sequence from the aforementioned power system time-series data. Therefore, the original data packets can be parsed into standardized time-series data that is strictly ordered by timestamps, eliminating the out-of-order effects caused by network transmission and buffering, and laying a data foundation for subsequent accurate time-series analysis. Then, a rate-of-change domain coarse screening operation is performed on the above-mentioned time-series detection value sequence to obtain a candidate feature point set. Thus, the cumulative change characteristics of slowly varying signals can be amplified by the sliding window bilateral mean difference transformation, and combined with an adaptive threshold, the problem of missed or false detections in slowly varying fault scenarios under a fixed threshold is avoided. Next, time clustering and boundary filtering are performed on the above-mentioned candidate feature point set to obtain a feature point set. Thus, density-aware clustering and confidence assessment can effectively distinguish between dense clusters formed by real fault events and dense clusters formed by noise, eliminating a large number of false feature points and retaining high-confidence boundary feature points as input for subsequent fine screening, solving the problem of difficult feature point clustering in high-noise environments. Finally, the above-mentioned feature point set is recursively fine-screened in the amplitude domain to obtain a tomographic feature point set. Therefore, by identifying amplitude faults layer by layer and recursively stripping strong signal feature points, the entry and clearing times of fault events can be accurately extracted from boundary feature points, further improving the accuracy and reliability of feature point localization. Finally, based on the aforementioned set of fault feature points, control commands are generated and sent to the controller of the power system protection device. The controller, according to the control commands, triggers protection actions at the entry time of voltage surges and resets the protection state at the clearing time of voltage surges. Thus, the identification results can be directly converted into executable commands for the protection device, forming a complete closed loop from data acquisition and feature identification to control execution, meeting the real-time and automation requirements of hardware-in-the-loop simulation testing.This implementation method achieves accurate identification of voltage abrupt changes in weak power grids and high-noise environments, and directly converts the identification results into protection control commands. For example... Figure 9 As shown, the traditional fixed threshold method exhibits missed detections at high threshold settings, false alarms at low threshold settings, and accurate detection results of the method disclosed in this disclosure at adaptive threshold settings. Attached Figure Description

[0014] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0015] Figure 1 This is a flowchart of some embodiments of the voltage surge identification and grid-connected protection control method for power grid fault ride-through according to the present disclosure; Figure 2 This is a schematic diagram of some embodiments of the voltage surge identification and grid connection protection control device for power grid fault ride-through according to the present disclosure; Figure 3 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure; Figure 4 This is a schematic diagram of a slow-change fault scenario in a weak power grid according to some embodiments of this disclosure; Figure 5 This is a schematic diagram of the interface generation method for voltage surge identification and grid-connected protection control according to some embodiments of the present disclosure; Figure 6 This is an overview diagram of detection results according to some embodiments of the present disclosure, which shows the detection value sequence, the time-enhanced detection value sequence, and the voltage surge entry time and voltage surge clearing time; Figure 7 This is a schematic diagram of intermediate results of a coarse screening operation in the rate of change domain according to some embodiments of this disclosure; Figure 8 This is a schematic diagram of intermediate results of amplitude domain recursive fine screening according to some embodiments of the present disclosure; Figure 9 These are comparative diagrams showing the technical effects of some embodiments of this disclosure. Detailed Implementation

[0016] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0017] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0018] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0019] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0020] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0021] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0022] refer to Figure 1 The diagram illustrates a flow 100 of some embodiments of a voltage surge identification and grid-connected protection control method, apparatus, electronic device, and computer-readable medium for power grid fault ride-through according to the present disclosure. The voltage surge identification and grid-connected protection control method for power grid fault ride-through includes the following steps: Step 101: In response to the detected voltage change feature point identification command, receive power system timing data.

[0023] In some embodiments, the execution entity (e.g., a server) of the voltage surge identification and grid-connected protection control method for power grid fault ride-through can receive power system time-series data in response to a detected voltage surge feature point identification command. In practice, the execution entity can receive power system time-series data from a real-time simulator via a data output interface in response to a detected voltage surge feature point identification command. This data output interface can be an Ethernet-based TCP / UDP communication interface or a shared memory interface. The power system time-series data can be a time-ordered data sequence output by the real-time simulator at a fixed sampling period, where each data point includes a timestamp and a voltage per-unit value. The fixed sampling period can be 50 microseconds, 100 microseconds, or 1 millisecond. The real-time simulator can be a real-time digital simulation device for power system hardware-in-the-loop testing.

[0024] Step 102: Extract the time series detection value sequence from the power system time series data.

[0025] In some embodiments, the execution entity can extract a sequence of time-series detection values ​​from the power system time-series data. The time-series detection values ​​in the sequence include timestamps and detection values, where the detection values ​​are per-unit voltage values. Each time-series detection value in the sequence corresponds to a sampling point, and the index of the sampling point in the sequence is called the sampling point index. For example, there are N sampling points, and the sampling point index is denoted as i. The above time-series detection value sequence can be denoted as: The timestamp of the i-th sampling point can be denoted as... The time interval between any two adjacent time-series detection values ​​can be denoted as: , , .

[0026] In some optional implementations of certain embodiments, the aforementioned execution entity may extract a sequence of time-series detection values ​​from the aforementioned power system time-series data through the following steps: The first step is to parse the data packet format of the aforementioned power system time-series data to obtain a data frame sequence. In practice, the aforementioned execution entity can unpack the received raw data packets according to pre-configured protocol parsing rules to reconstruct the individual data frames and obtain the data frame sequence. The aforementioned pre-configured protocol parsing rules can be the configuration protocol of the aforementioned real-time simulator. For example, the aforementioned pre-configured protocol parsing rules can be in binary data packet format or CSV text format.

[0027] The second step is to read the timestamp and voltage per-unit value from each data frame of the above data frame sequence as timing detection values ​​to obtain a timing detection value set.

[0028] The third step is to sort the above time-series detection value set in ascending order of timestamps to obtain the time-series detection value sequence.

[0029] Step 103: Perform a rate-of-change domain coarse screening operation on the time-series detection value sequence to obtain a set of candidate feature points.

[0030] In some embodiments, the execution entity may perform a rate-of-change domain coarse screening operation on the time-series detection value sequence to obtain a candidate feature point set. The candidate feature points in the candidate feature point set include sampling point index, timestamp, change amplitude, and enhanced change amplitude.

[0031] In some optional implementations of certain embodiments, the execution entity may perform a rate-of-change domain coarse screening operation on the time-series detection value sequence to obtain a set of candidate feature points through the following steps: The first step is to perform a sliding window bilateral mean difference transformation on the aforementioned time-series detection value sequence to obtain a time-enhanced detection value sequence. In practice, the execution entity can perform the following operations centered on each time-series detection value in the aforementioned time-series detection value sequence: First, determine the left-window average of all time-series detection values ​​within the left-side window of the aforementioned center and the right-window average of all time-series detection values ​​within the right-side window of the aforementioned center; Second, determine the difference between the right-window average and the left-window average as the aforementioned center time-enhanced detection value, thus obtaining the time-enhanced detection value sequence. The window time width of the aforementioned sliding window bilateral mean difference transformation can be 20 milliseconds. The left-side window can be the time interval covered by the aforementioned window time width extending forward from the aforementioned time-series detection value. The right-side window can be the time interval covered by the aforementioned window time width extending backward from the aforementioned time-series detection value. As an example, the aforementioned time-enhanced detection value sequence can be denoted as... .

[0032] The second step involves determining the change amplitude sequence based on the aforementioned time-series detection value sequence, and determining the enhanced change amplitude sequence based on the aforementioned time-series enhanced detection value sequence. In practice, the executing entity can determine the absolute value of the first-order difference of the aforementioned time-series detection value sequence to obtain the change amplitude sequence. Then, it determines the absolute value of the first-order difference of the aforementioned time-series enhanced detection value sequence to obtain the enhanced change amplitude sequence. As an example, determining the absolute value of the first-order difference of the aforementioned time-series detection value sequence can specifically be: determining the absolute value of the difference between each time-series detection value in the aforementioned time-series detection value sequence and the previous adjacent time-series detection value. As an example, the i-th change amplitude in the aforementioned change amplitude sequence can be denoted as... The i-th enhancement magnitude in the above enhancement magnitude sequence can be denoted as... .

[0033] The third step is to determine the adaptive judgment threshold based on the aforementioned enhanced change amplitude sequence. In practice, the executing entity can determine the mean and standard deviation of the enhanced change amplitude sequence. Then, the product of the standard deviation of the enhanced change amplitude sequence and a preset multiple, plus the sum of the mean, is used to determine the adaptive judgment threshold. The preset multiple can be any value between 2 and 5. As an example, the preset multiple can be 3. The adaptive judgment threshold can be denoted as... , R represents the aforementioned sequence of enhanced change magnitudes. This represents the mean of the aforementioned enhanced change magnitude sequence. k is the aforementioned preset multiple. This represents the standard deviation of the above-mentioned enhanced change magnitude sequence.

[0034] The fourth step involves determining a set of candidate feature points based on the aforementioned adaptive judgment threshold and the enhanced change amplitude sequence. In practice, the executing entity can iterate through the enhanced change amplitude sequence and identify the sampling point index, timestamp, corresponding change amplitude value in the change amplitude sequence, and corresponding enhanced change amplitude value of sampling points whose enhanced change amplitude value is greater than the aforementioned adaptive judgment threshold as candidate feature points, thus obtaining a set of candidate feature points. As an example, this set of candidate feature points can be denoted as C. i is the index of the sampling point. For timestamps. Let be the magnitude of the i-th enhancement. Let be the magnitude of the i-th change. This indicates the adaptive decision threshold.

[0035] In addressing the technical problems mentioned in the background section, the application scenario—voltage gradual change faults in new energy power generation equipment (such as photovoltaic inverters) under weak grid conditions—often presents the following technical challenges: Existing sliding window bilateral mean difference transformations use a fixed window width. For faults with drastically different transition durations (e.g., a sudden voltage drop within 50ms versus a gradual drop within 500ms), the fixed window cannot simultaneously accommodate both. If the window is too small, the cumulative change in gradual faults is underestimated; if the window is too large, the spike in sudden drop faults is flattened, reducing time-based positioning accuracy. Considering the following requirements for this application scenario: maintaining stable detection sensitivity for fault types with different transition durations; ensuring a high detection rate for gradual faults without sacrificing time-based positioning accuracy for sudden drop faults; and adaptively adapting to unknown fault types without requiring manual pre-setting of the window width, we have decided to adopt the following solution: Optionally, in some alternative implementations of certain embodiments, the execution entity may perform a rate-of-change domain coarse screening operation on the time-series detection value sequence to obtain a set of candidate feature points through the following steps: The first step involves generating at least one candidate window time width based on the aforementioned time-series detection value sequence, resulting in a group of candidate window time widths. These candidate window time widths cover a range from the shortest expected transition segment to the longest expected transition segment. For example, the candidate window time widths may include, but are not limited to, 5 milliseconds, 20 milliseconds, and 100 milliseconds, corresponding to typical timescales for sudden drop faults, medium transition duration faults, and gradual change faults, respectively.

[0036] The second step involves performing a sliding window bilateral mean difference transformation on the time series detection value sequence based on the time series time width of each candidate window in the aforementioned candidate window time width group, resulting in a time series enhancement detection value sequence group. In practice, the execution entity can perform the following operations for each candidate window time width in the aforementioned candidate window time width group: First, using each time series detection value in the aforementioned time series detection value sequence as the center, determine the left window average of all time series detection values ​​within the left window of the center and the right window average of all time series detection values ​​within the right window of the center; second, determine the difference between the right window average and the left window average as the time series enhancement detection value at the center, thus obtaining the time series enhancement detection value sequence. After traversing all candidate window time widths, a time series enhancement detection value sequence group is obtained. Each time series enhancement detection value sequence in the aforementioned time series enhancement detection value sequence group can be denoted as... .

[0037] The third step is to determine the enhancement variation amplitude sequence group based on the aforementioned time-series enhancement detection value sequence group. In practice, the execution entity can perform the following operations for each time-series enhancement detection value sequence in the aforementioned time-series enhancement detection value sequence group: First, determine the absolute value of the first-order difference of the aforementioned time-series detection value sequence to obtain the variation amplitude sequence; Second, determine the absolute value of the first-order difference of the aforementioned time-series enhancement detection value sequence to obtain the enhancement variation amplitude sequence. After traversing all time-series enhancement detection value sequences, the enhancement variation amplitude sequence group is obtained. The enhancement variation amplitude in the aforementioned enhancement variation amplitude sequence group can be denoted as... The magnitudes of change in the above sequence of magnitudes can be denoted as: .

[0038] The fourth step is to determine the adaptive judgment threshold group based on the aforementioned enhanced change amplitude sequence group. In practice, the execution entity can determine the mean and standard deviation of each enhanced change amplitude sequence in the aforementioned enhanced change amplitude sequence group, and then sum the product of the standard deviation of the enhanced change amplitude sequence and the aforementioned preset multiple with the mean of the enhanced change amplitude sequence to determine the adaptive judgment threshold corresponding to the enhanced change amplitude sequence, thus obtaining the adaptive judgment threshold group.

[0039] The fifth step involves determining the candidate feature point set based on the aforementioned adaptive judgment threshold group and the aforementioned enhanced change amplitude sequence group. In practice, the execution entity can perform the following operations for each candidate window time width: traverse the enhanced change amplitude sequence corresponding to the aforementioned candidate window time width, and determine the sampling point index, timestamp, corresponding change amplitude value in the change amplitude sequence, and corresponding enhanced change amplitude value in the aforementioned enhanced change amplitude sequence of sampling points whose enhanced change amplitude value is greater than the adaptive judgment threshold corresponding to the aforementioned candidate window time width as candidate feature points, thus obtaining the candidate feature point group corresponding to the aforementioned candidate window time width. After traversing all candidate window time widths, the candidate feature point set is obtained. Each candidate feature point group in the candidate feature point set corresponds one-to-one with a candidate window time width. As an example, the candidate feature point groups in the aforementioned candidate feature point set can be denoted as... j is the index of the candidate window's time width.

[0040] The sixth step is to merge all candidate feature points in the aforementioned candidate feature point set to obtain a merged candidate feature point set. In practice, the executing entity can aggregate the candidate feature points from all candidate feature point groups in the aforementioned candidate feature point set into one set, thus obtaining a merged candidate feature point set. Since different candidate window time widths may detect candidate feature points at the same timestamp, the merged set may contain multiple merged candidate feature points with the same timestamp. The elements in the aforementioned merged candidate feature point set are the merged candidate feature points.

[0041] Step 7: In response to the existence of merged candidate feature points with the same timestamp in the aforementioned merged candidate feature point set, retain the merged candidate feature point with the largest enhancement change among the merged candidate feature points with the same timestamp, thus obtaining the candidate feature point set. In practice, the executing entity can use each timestamp in the aforementioned merged candidate feature point set as a unit, retaining the one with the largest enhancement change among all merged candidate feature points under the aforementioned timestamp as the candidate feature point for that timestamp, and deleting the remaining merged candidate feature points, finally obtaining the candidate feature point set. As an example, the aforementioned candidate feature point set can be denoted as C. .like Figure 7 As shown, the temporal enhanced detection value sequence, enhanced change amplitude sequence, adaptive decision threshold, and candidate feature points are illustrated after the sliding window bilateral mean difference transformation.

[0042] The above-mentioned technical content and related content, as an inventive point of this disclosure, combined with "steps 104-106," solve the technical problem that "existing sliding window bilateral mean difference transformation uses a fixed window width, which cannot simultaneously account for faults with extremely large differences in transition duration (such as a sudden voltage drop within 50ms and a gradual voltage drop within 500ms). When the window is too small, the cumulative change amplitude of the gradual fault is underestimated; when the window is too large, the peak of the sudden drop fault is flattened, and the time positioning accuracy decreases." The factors leading to the above-mentioned technical problem are often as follows: a sliding window with a single time scale cannot simultaneously match the inherent time characteristics of both fast and slow faults; any fixed window width is essentially a compromise for a certain type of fault and cannot achieve optimal performance across all fault types. If the above factors are solved, it is possible to achieve stable and accurate detection performance across all fault types regardless of how the fault transition duration changes. To achieve this effect, firstly, based on the above-mentioned time-series detection value sequence, at least one candidate window time width is generated, resulting in a candidate window time width group. Therefore, multiple typical time scales from short to long transition segments can be covered, providing a window parameter basis for subsequent multi-scale parallel detection and avoiding the limitations of a single window. Second, based on the time width of each candidate window in the above candidate window time width group, a sliding window bilateral mean difference transformation is performed on the above time-series detection value sequence to obtain a time-series enhanced detection value sequence group. Thus, the original signal can be transformed separately under each candidate window time width, allowing short windows to keenly capture the steep changes of sudden-drop faults, and long windows to fully accumulate the gradual characteristics of slowly changing faults, forming a complementary information view. Third, based on the above time-series enhanced detection value sequence group, an enhanced change amplitude sequence group is determined. Thus, the local change intensity of the transformed signal can be calculated independently at each time scale, allowing fault characteristics at different scales to be quantitatively expressed, facilitating subsequent unified threshold comparison and screening. Fourth, based on the above enhanced change amplitude sequence group, an adaptive judgment threshold group is determined. Thus, an adaptive threshold adapted to the signal statistical characteristics at each time scale can be calculated separately, avoiding the problem of thresholds being too high or too low at certain scales due to the use of a globally uniform threshold. Fifth, based on the aforementioned adaptive threshold set and the aforementioned enhanced change amplitude sequence set, a candidate feature point set is determined. This allows for independent screening of feature points exceeding the threshold at each time scale, ensuring that both rapidly decreasing feature points detected in short windows and gradually changing feature points detected in long windows are included in the candidate range without mutual exclusion. Sixth, all candidate feature points in the aforementioned candidate feature point set are merged to obtain a merged candidate feature point set. This allows for the aggregation of detection results from multiple time scales, forming a comprehensive candidate set containing various fault features, avoiding the omission of any type of fault feature point due to improper window selection.Seventh, in response to the existence of merged candidate feature points with the same timestamp in the aforementioned merged candidate feature point set, the merged candidate feature point with the largest enhancement change amplitude among the merged candidate feature points with the same timestamp is retained, resulting in a candidate feature point set. Thus, the feature point with the strongest response (i.e., the most significant) at each time point can be retained in the multi-scale detection results, combining the advantages of multi-scale detection while avoiding duplicate counting and redundant information. Finally, combining "steps 104-106", it is possible to achieve accurate identification of all types of fault voltage abrupt changes from millisecond-level sudden drops to hundred-millisecond-level gradual drops in the scenario of slow-changing faults in weak power grids, through a combination of multi-scale window parallel detection and adaptive thresholding, and to drive the protection device to execute corresponding control actions.

[0043] Step 104: Perform temporal clustering and boundary filtering on the candidate feature point set to obtain the feature point set.

[0044] In some embodiments, the execution entity may perform temporal clustering and boundary filtering on the candidate feature point set to obtain the feature point set.

[0045] In addressing the technical problems mentioned in the background section, and considering the application scenario—when a hardware-in-the-loop platform performs low-voltage ride-through testing on new energy power generation equipment—the voltage time-series data contains a large amount of high-frequency noise caused by the high-frequency operation of power electronic switching devices and electromagnetic interference. This often leads to the following technical problems: existing time-based clustering methods rely solely on time intervals for merging, lacking effective utilization of feature point amplitude information, making it difficult to distinguish between dense clusters formed by real fault events and those formed by noise. This results in a large number of false events or omissions of weak signal real events in the clustering results. To meet the following requirements for this application scenario: effective differentiation between real event clusters and noise clusters, sufficient sensitivity to weak signal real events, and adaptive clustering to different noise levels without the need for preset fixed parameters, we have decided to adopt the following solution: In some optional implementations of certain embodiments, the aforementioned execution entity may perform temporal clustering and boundary filtering on the aforementioned candidate feature point set through the following steps: The first step is to sort the candidate feature point set in ascending order by timestamp to obtain the candidate feature point sequence. In practice, the executing entity can use the timestamp as the key to sort the candidate feature point set in ascending order to obtain the candidate feature point sequence.

[0046] The second step is to determine the time interval between each candidate feature point in the above candidate feature point sequence and the previous candidate feature point, thus obtaining a time interval sequence. In practice, the execution entity can start from the second candidate feature point in the above candidate feature point sequence and use the difference between the timestamp of each candidate feature point and the timestamp of the previous candidate feature point as the time interval, thus obtaining a time interval sequence.

[0047] The third step is to determine the clustering time threshold based on the aforementioned time interval sequence. This clustering time threshold can be the product of the median of the aforementioned time interval sequence and the aforementioned preset multiple, or the average value of the aforementioned time interval sequence.

[0048] The fourth step involves assigning candidate feature points in the above sequence whose time intervals are less than the aforementioned clustering time threshold to the initial cluster set. In practice, the execution entity can iterate through the candidate feature point sequence, starting from the first candidate feature point, and sequentially determine whether the time interval between each candidate feature point and its predecessor is less than the aforementioned clustering time threshold. If it is less than the threshold, the current candidate feature point is assigned to the cluster containing the previous candidate feature point; if it is not less than the threshold, an initial cluster is created, and the current candidate feature point is set as the first element of the initial cluster.

[0049] Fifth, perform the following operations on each initial cluster in the above initial cluster set: Step 1: Determine the average value of the enhancement variation amplitude in the initial cluster to obtain the average amplitude within the cluster.

[0050] Step 2: Determine the duration of the timestamps in the initial cluster to obtain the cluster duration.

[0051] Step 3: Determine the confidence level of the initial cluster based on the average amplitude within the cluster and the duration of the cluster. The confidence level can be the product of the average amplitude within the cluster and the duration of the cluster.

[0052] The sixth step involves identifying the initial clusters with confidence levels greater than or equal to a preset confidence threshold as valid clusters, thus obtaining the valid cluster set. The preset confidence threshold can be an empirical value calibrated based on historical test data, typically ranging from 0.1 to 0.5. For example, the preset confidence threshold could be 0.2.

[0053] Step 7: Merge all candidate feature points belonging to the same effective cluster in the above effective cluster set from the candidate feature point sequence into a first candidate feature point, obtaining an initial feature point set. As an example, during merging, the candidate feature point with the largest change can be retained as the first candidate feature point. The initial feature point in the above initial feature point set can be denoted as FP. `index` is the row number of the initial feature point in the aforementioned power system time series data. `ts` is the absolute timestamp of the initial feature point in the aforementioned power system time series data. `m` is the change magnitude value (equal to...). , i.e., the sequence of time detection values (Value). trigger is the name of the physical quantity field that triggers this feature point. As an example, trigger can be a voltage per-unit value.

[0054] Step 8: Perform boundary filtering on the initial feature point set to obtain a feature point set. In practice, the execution entity can process each initial feature point in the initial feature point set as follows: if the initial feature point is the earliest or latest candidate feature point in the corresponding valid cluster, determine the initial feature point as a feature point; if the initial feature point is not the earliest or latest candidate feature point in the corresponding valid cluster, determine the initial feature point as a first point to be confirmed. Finally, the feature points constitute the feature point set; the first points to be confirmed constitute the first point to be confirmed set.

[0055] In some alternative implementations of certain embodiments, in the implementation of step 104 above, the clustering time threshold in the third step can also be a fixed value preset by the user based on the sampling period and noise level of the power system time-series data (for example, it could be 0.03 seconds). The confidence filtering in steps five and six above can be replaced by minimum cluster size filtering, that is, initial clusters with fewer than the preset minimum cluster size (for example, it could be 3) are determined as invalid clusters. In step seven above, each valid cluster is merged into an initial feature point, and the candidate feature point with the largest change amplitude in the valid cluster is retained as the initial feature point; and the earliest / latest boundary splitting in step eight above can be omitted, that is, the voltage change event corresponding to the valid cluster is represented by the initial feature point.

[0056] The above-mentioned technical content and related content, as an inventive point of the embodiments of this disclosure, combined with "steps 105-106", solve the technical problem that "existing time clustering methods rely solely on time intervals for merging, lacking effective utilization of feature point amplitude information, making it difficult to distinguish between dense clusters formed by real fault events and dense clusters formed by noise, resulting in a large number of false events or omission of weak signal real events in the clustering results." The factors leading to the above technical problems are often as follows: real fault events and noise events can both form dense clusters on the time axis, and the two cannot be distinguished solely by the time dimension; moreover, noise clusters and weak signal real clusters overlap in amplitude distribution, making simple thresholding methods difficult to separate them. If the above factors are solved, it is possible to significantly reduce the false alarm rate while maintaining a high recall rate, achieving accurate clustering and boundary identification of feature points in high-noise environments. To achieve this effect, firstly, the above-mentioned candidate feature point set is sorted in ascending order by timestamp to obtain a candidate feature point sequence. Thus, the unordered candidate feature points can be arranged chronologically, providing an ordered data foundation for subsequent calculation of the time interval between adjacent feature points. Second, determine the time interval between each candidate feature point in the above candidate feature point sequence and the previous candidate feature point to obtain a time interval sequence. This allows quantifying the distribution density of candidate feature points on the time axis, providing a basis for adaptively determining the clustering granularity. Third, determine the clustering time threshold based on the above time interval sequence. This allows for automatic calculation of the clustering threshold based on the actual time distribution of candidate feature points, avoiding the problem of overly loose or tight clustering at different noise levels due to the use of a fixed threshold. Fourth, classify candidate feature points in the above candidate feature point sequence whose time intervals are less than the above clustering time threshold into initial clusters to obtain an initial cluster set. This allows for the initial aggregation of temporally adjacent candidate feature points into candidate event clusters, completing the first stage of coarse-grained clustering. Fifth, perform the following operations on each initial cluster in the above initial cluster set: Step 1, determine the average value of the enhancement variation amplitude in the above initial cluster to obtain the average amplitude within the cluster. This allows quantifying the signal strength level of the event corresponding to the cluster, providing a basis for distinguishing strong signal real events from weak signal noise events in terms of amplitude dimension. Step 2, determine the duration of the timestamp in the above initial cluster to obtain the cluster duration. Therefore, the temporal span of the cluster can be quantified, providing a basis for assessing the temporal plausibility of the event. Step three: Based on the average amplitude within the cluster and the duration of the cluster, the confidence level of the initial cluster is determined. Thus, information from both amplitude and time dimensions can be combined to quantify the probability that each initial cluster belongs to a real event. Sixth, initial clusters with a confidence level greater than or equal to a preset confidence threshold are identified as valid clusters, resulting in a valid cluster set. This allows for the removal of low-confidence clusters caused by noise, retaining only high-confidence real event clusters, significantly reducing the false alarm rate in subsequent processing. Seventh, temporal deduplication clustering is performed on the candidate feature point sequences to obtain an initial feature point set.Therefore, candidate feature points with the same or very close timestamps can be merged into a single feature point, eliminating redundancy and obtaining a concise and unique feature point list. Eighth, boundary filtering is performed on the initial feature point set to obtain a new feature point set. Thus, the earliest and latest boundary feature points with the earliest timestamps can be extracted from the effective clusters as the entry and exit times of fault events, providing high-quality input for subsequent amplitude-domain recursive screening. Finally, combining steps 105-106, the system can effectively distinguish between real event clusters and noise clusters in the high-noise environment of slowly changing faults in weak power grids, providing high-quality input for subsequent amplitude-domain recursive screening, and ultimately driving the protection device to execute corresponding control actions.

[0057] Step 105: Perform amplitude domain recursive screening on the feature point set to obtain the fault feature point set.

[0058] In some embodiments, the execution entity may perform amplitude domain recursive screening on the feature point set to obtain a fault feature point set.

[0059] In some optional implementations of certain embodiments, the aforementioned execution entity may perform amplitude-domain recursive screening of the aforementioned feature point set to obtain a tomographic feature point set through the following steps: The first step is to sort the feature point set in descending order of the magnitude of change, thus obtaining the feature point sequence. As an example, the feature point set can be denoted as F, and the feature points in the set can be denoted as FP, where FP = {index, ts, mag, trigger}. The magnitude of change in the feature point sequence can be denoted as M. ,and k is the index of the feature point in the above feature point sequence. m represents the variation range of the feature points in the above feature point sequence. n represents the number of feature points in the above feature point sequence. .

[0060] The second step is to perform the following recursive segmentation operation on the above feature point sequence to obtain the set of tomographic feature points: The first recursive segmentation operation determines the absolute value of the amplitude difference in the above feature point sequence, resulting in a sequence of absolute amplitude difference values. In practice, the executing entity can determine the absolute value of the difference between the amplitude of each feature point in the above feature point sequence and the amplitude of the previous feature point as the absolute value of the amplitude difference, thus obtaining a sequence of absolute amplitude difference values. As an example, the absolute value of the amplitude difference in the above sequence of absolute amplitude difference values ​​can be denoted as... , p is the index of the absolute value of the magnitude difference in the above sequence of absolute magnitude differences.

[0061] The second recursive segmentation operation involves determining the tomography threshold based on the aforementioned absolute value sequence of amplitude differences. In practice, the executing entity can determine the mean and standard deviation of the aforementioned absolute value sequence of amplitude differences. Then, the sum of the product of the standard deviation of the aforementioned absolute value sequence and the aforementioned preset multiple, and the aforementioned mean, is used to determine the tomography threshold. As an example, the aforementioned tomography threshold can be denoted as... , . The above is the absolute value sequence of amplitude differences, and k is the above preset multiple.

[0062] The third recursive segmentation operation involves locating faults in the feature point sequence based on the aforementioned fault threshold, thereby obtaining fault feature points. In practice, the executing entity can identify feature points in the feature point sequence whose absolute amplitude difference is greater than the aforementioned fault threshold as fault feature points. If no feature point in the feature point sequence has an absolute amplitude difference greater than the aforementioned fault threshold, the recursive segmentation operation stops. As an example, the aforementioned fault feature points can be denoted as P. .

[0063] The fourth recursive segmentation operation involves segmenting the aforementioned fault feature points into a strong signal feature point sequence and a weak signal feature point sequence. In practice, the executing entity can identify the feature points preceding the aforementioned fault feature points as strong signal feature point sequences and the feature points following the aforementioned fault feature points as weak signal feature point sequences. For example, the strong signal feature point sequence can be denoted as... The amplitude values ​​corresponding to the above strong signal feature point sequence can be denoted as: The above sequence of weak signal feature points can be denoted as... The amplitude values ​​corresponding to the above weak signal feature point sequence can be denoted as: j is the index of the last fault feature point among the above fault feature points. For example... Figure 8 As shown, it illustrates the descending sequence of change amplitudes, the absolute value of amplitude differences, the fault threshold, and the set of fault feature points.

[0064] The fifth recursive segmentation operation performs adaptive tomography verification on the aforementioned strong signal feature point sequence and the aforementioned weak signal feature point sequence to obtain the tomography verification result. In practice, the executing entity can determine the tomography ratio as the ratio of the change amplitude of the last strong signal feature point in the aforementioned strong signal feature point sequence to the change amplitude of the first weak signal feature point in the aforementioned weak signal feature point sequence. Then, the ratio of the change amplitude of each feature point in the aforementioned feature point sequence to the change amplitude of the previous feature point is determined to obtain the change amplitude ratio sequence, and the median of the change amplitude ratio sequence is determined to obtain the median of the change amplitude ratio sequence. Finally, in response to the tomography ratio being greater than the product of the median of the change amplitude ratio sequence and the adaptive coefficient, a verification result indicating that the tomography is valid is obtained; in response to the tomography ratio being less than or equal to the product of the median of the change amplitude ratio sequence and the adaptive coefficient, a verification result indicating that the tomography is invalid is obtained. The adaptive coefficient can be any value between 1.5 and 3. For example, the adaptive coefficient can be 2.5.

[0065] In the sixth recursive segmentation operation, in response to the above fault verification results indicating that the fault is valid, the above strong signal feature point sequence is saved to the fault feature point set, and the above weak signal feature point sequence is used as the feature point sequence to perform the above recursive segmentation operation.

[0066] The third step is to determine the feature points in the weak signal feature point sequence as the second points to be confirmed in response to the above fault verification results indicating that the fault is invalid. This results in obtaining the second set of points to be confirmed, and the above strong signal feature point sequence is saved to the fault feature point set.

[0067] Optionally, if the number of feature points in the above feature point sequence is less than the preset minimum number of samples (e.g., 3), or if no amplitude tomography feature points are detected (i.e., P is empty), the above recursive segmentation operation is stopped.

[0068] Optionally, in response to the above fault verification result indicating that the fault is invalid, and the above fault feature point set has already saved fault feature points (i.e., the current second feature point sequence is the layer below the identified real fault), all feature points in the current feature point sequence are saved to the set of points to be confirmed or directly discarded, so as to avoid noise feature points in the uniform noise layer being mistakenly saved to the fault feature point set.

[0069] Step 106: Based on the set of fault feature points, generate control commands and send the control commands to the controller of the power system protection device.

[0070] In some embodiments, the executing entity can generate control commands based on the set of fault feature points and send these control commands to the controller of the power system protection device. The controller, based on the control commands, triggers protection action at the moment a voltage surge occurs and resets the protection state when the voltage surge is cleared. Figure 6 As shown, the detection value sequence, the time-enhanced detection value sequence, and the voltage surge entry and voltage surge clearing times are illustrated.

[0071] In some optional implementations of certain embodiments, the aforementioned execution entity may generate control commands based on the aforementioned set of fault feature points and send the control commands to the controller of the power system protection device through the following steps: The first step is to read the fault feature point with the earliest timestamp from the above set of fault feature points as the voltage mutation entry time, and read the fault feature point with the latest timestamp as the voltage mutation clearing time.

[0072] The second step involves generating control instructions based on the voltage surge entry and clearance times. In practice, the executing entity can assemble the voltage surge entry and clearance times into control instructions using a predefined format. This predefined format can be JSON or a binary protocol format. The control instructions may include an instruction type field, a start time field, and an end time field. The instruction type field can be "protection action." The start time field can be a value taken from the voltage surge entry time, in microseconds, indicating the starting point for the protection device to begin executing the protection action. The end time field can be a value taken from the voltage surge clearance time, in microseconds, indicating the ending point for the protection device to stop executing the protection action or switch back to normal operation mode.

[0073] The third step involves sending the control command to the controller of the power system protection device via the aforementioned data output interface. In practice, the executing entity can send the control command to the controller of the power system protection device through the aforementioned data output interface to execute at least one of the following protection actions: initiating fault recording and triggering protection action. Initiating fault recording can involve the controller recording voltage and current waveform data backward for a preset duration (e.g., 10 milliseconds) from the moment the voltage surge occurs until the voltage surge clears, then continuing recording for another preset duration (e.g., 20 milliseconds) to form a complete fault waveform file. Triggering protection action can involve the controller sending a trip signal to the circuit breaker or grid-connected switch of the tested equipment at the moment the voltage surge occurs, or sending a closing signal at the moment the voltage surge clears, to simulate the actual operation of the protection device. The controller, based on the aforementioned control command, triggers protection action at the moment the voltage surge occurs and resets the protection state at the moment the voltage surge clears. Figure 5 As shown, a scheme generation interface for a voltage surge identification and grid-connected protection control method according to some embodiments of the present disclosure is illustrated.

[0074] The above-described embodiments of this disclosure have the following beneficial effects: The voltage mutation identification and grid-connected protection control method for power grid fault ride-through, through a combination of multi-level progressive screening and adaptive thresholding, achieves accurate identification of voltage mutation points in weak power grids with gradual fault changes and high noise environments, and directly converts the identification results into protection control commands. Specifically, the reasons for low accuracy in voltage mutation point identification, poor noise resistance, and disconnection from protection control are: traditional methods use fixed thresholds to detect change rates, which cannot adapt to the low change rate characteristics of gradual fault changes; single time clustering methods are difficult to distinguish between noise clusters and real event clusters; and the identification results remain at the offline analysis level, unable to directly drive the protection device to operate. Based on this, the voltage mutation identification and grid-connected protection control method for power grid fault ride-through, according to some embodiments of this disclosure, firstly, receives power system time-series data in response to the detected voltage mutation feature point identification command. This establishes a real-time data channel between the execution entity and the real-time simulator, providing a continuous and reliable source of raw data for subsequent online processing. Secondly, it extracts the time-series detection value sequence from the aforementioned power system time-series data. Therefore, the original data packets can be parsed into standardized time-series data that is strictly ordered by timestamps, eliminating the out-of-order effects caused by network transmission and buffering, and laying a data foundation for subsequent accurate time-series analysis. Then, a rate-of-change domain coarse screening operation is performed on the above-mentioned time-series detection value sequence to obtain a candidate feature point set. Thus, the cumulative change characteristics of slowly varying signals can be amplified by the sliding window bilateral mean difference transformation, and combined with an adaptive threshold, the problem of missed or false detections in slowly varying fault scenarios under a fixed threshold is avoided. Next, time clustering and boundary filtering are performed on the above-mentioned candidate feature point set to obtain a feature point set. Thus, density-aware clustering and confidence assessment can effectively distinguish between dense clusters formed by real fault events and dense clusters formed by noise, eliminating a large number of false feature points and retaining high-confidence boundary feature points as input for subsequent fine screening, solving the problem of difficult feature point clustering in high-noise environments. Finally, the above-mentioned feature point set is recursively fine-screened in the amplitude domain to obtain a tomographic feature point set. Therefore, by identifying amplitude faults layer by layer and recursively stripping strong signal feature points, the entry and clearing times of fault events can be accurately extracted from boundary feature points, further improving the accuracy and reliability of feature point localization. Finally, based on the aforementioned set of fault feature points, control commands are generated and sent to the controller of the power system protection device. The controller, according to the control commands, triggers protection actions at the entry time of voltage surges and resets the protection state at the clearing time of voltage surges. Thus, the identification results can be directly converted into executable commands for the protection device, forming a complete closed loop from data acquisition and feature identification to control execution, meeting the real-time and automation requirements of hardware-in-the-loop simulation testing.This implementation method achieves accurate identification of voltage abrupt changes in weak power grids and high-noise environments, and directly converts the identification results into protection control commands. For example... Figure 9 As shown, the traditional fixed threshold method exhibits missed detections at high threshold settings, false alarms at low threshold settings, and accurate detection results of the method disclosed in this disclosure at adaptive threshold settings.

[0075] Continue to refer to Figure 2 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of a voltage surge identification and grid-connected protection control device for power grid fault ride-through. These device embodiments are similar to... Figure 1 Corresponding to the method embodiments shown, the voltage surge identification and grid connection protection control device for power grid fault ride-through can be specifically applied to various electronic devices.

[0076] like Figure 2 As shown, a voltage surge identification and grid-connected protection control device 200 for power grid fault ride-through includes: a receiving unit 201, an extraction unit 202, a coarse screening unit 203, a clustering unit 204, a fine screening unit 205, and a generation unit 206. The receiving unit 201 is configured to receive power system time-series data in response to a voltage surge feature point identification command; the extraction unit 202 is configured to extract a time-series detection value sequence from the power system time-series data, wherein the time-series detection value sequence includes a timestamp and a detection value, and the detection value is a voltage per-unit value; the coarse screening unit 203 is configured to perform a rate-of-change domain coarse screening operation on the time-series detection value sequence to obtain a candidate feature point set; the clustering unit 204 is configured to perform time clustering and boundary filtering on the candidate feature point set to obtain a feature point set; the fine screening unit 205 is configured to perform amplitude domain recursive fine screening on the feature point set to obtain a fault feature point set; and the generation unit 206 is configured to generate a control command based on the fault feature point set and send the control command to the controller of the power system protection device, wherein the controller triggers a protection action at the moment a voltage surge occurs and resets the protection state at the moment the voltage surge is cleared, based on the control command.

[0077] It is understandable that the units described in the voltage surge identification and grid connection protection control device 200 for power grid fault ride-through are similar to those in the reference device. Figure 1 The steps in the described method correspond accordingly. Therefore, the operations, features, and beneficial effects described above for the method also apply to the voltage surge identification and grid connection protection control device 200 for power grid fault ride-through and the units contained therein, and will not be repeated here.

[0078] The following is for reference. Figure 3It shows a method suitable for implementing this disclosure. Figure 1 A schematic diagram of the structure of an electronic device 300 in some corresponding embodiments. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0079] like Figure 3 As shown, the electronic device 300 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory 302 or a program loaded from a storage device 308 into a random access memory 303. The random access memory 303 also stores various programs and data required for the operation of the electronic device 300. The processing unit 301, the read-only memory 302, and the random access memory 303 are interconnected via a bus 304. An input / output interface 305 is also connected to the bus 304.

[0080] Typically, the following devices can be connected to the input / output interface 305: input devices 306 including, for example, a touchscreen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; output devices 307 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 308 including, for example, magnetic tape, hard disk, etc.; and communication devices 309. Communication device 309 allows electronic device 300 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 3 An electronic device 300 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 3 Each box shown can represent a device or multiple devices as needed.

[0081] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 309, or installed from a storage device 308, or installed from a read-only memory 302. When the computer program is executed by the processing device 301, it performs the functions defined in the methods of some embodiments of this disclosure.

[0082] It should be noted that, in some embodiments of this disclosure, the computer-readable medium described above may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical fiber, a portable compact disk read-only memory, an optical storage device, a magnetic storage device, or any suitable combination thereof. In some embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0083] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0084] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to: receive power system time-series data in response to a voltage surge feature point identification instruction; extract a time-series detection value sequence from the power system time-series data, wherein the time-series detection value sequence includes a timestamp and a detection value, and the detection value is a voltage per-unit value; perform a rate-of-change domain coarse screening operation on the time-series detection value sequence to obtain a candidate feature point set; perform time clustering and boundary filtering on the candidate feature point set to obtain a feature point set; perform amplitude domain recursive fine screening on the feature point set to obtain a fault feature point set; generate a control instruction based on the fault feature point set, and send the control instruction to the controller of the power system protection device, wherein the controller triggers a protection action at the moment the voltage surge occurs and resets the protection state at the moment the voltage surge is cleared, according to the control instruction.

[0085] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0086] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0087] The units described in some embodiments of this disclosure can be implemented in software or in hardware. The described units can also be located in a processor.

[0088] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0089] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A method for voltage surge identification and grid-connected protection control during power grid fault ride-through, characterized in that, include: In response to a voltage surge feature point identification command, power system timing data is received; Extract a sequence of time-series detection values ​​from the power system time-series data, wherein the time-series detection values ​​in the sequence include timestamps and detection values, and the detection values ​​are voltage per-unit values; A rate-of-change domain coarse screening operation is performed on the time-series detection value sequence to obtain a set of candidate feature points; Temporal clustering and boundary filtering are performed on the candidate feature point set to obtain the feature point set; The set of feature points is subjected to amplitude domain recursive screening to obtain the set of fault feature points; Based on the set of fault feature points, a control command is generated and sent to the controller of the power system protection device. The controller triggers a protection action when a voltage surge occurs and resets the protection state when the voltage surge is cleared, according to the control command.

2. The method according to claim 1, characterized in that, The step of generating control commands based on the set of fault feature points and sending the control commands to the controller of the power system protection device includes: The fault feature point with the earliest timestamp is read from the set of fault feature points as the voltage mutation entry time, and the fault feature point with the latest timestamp is read as the voltage mutation clearing time. Control commands are generated based on the voltage surge entry time and the voltage surge clearing time; The control commands are sent to the controller of the power system protection device via the data output interface.

3. The method according to claim 1, characterized in that, The candidate feature points in the candidate feature point set include the sampling point index, timestamp, change magnitude, and enhanced change magnitude.

4. The method according to claim 1, characterized in that, The step of performing a rate-of-change domain coarse screening operation on the time-series detection value sequence to obtain a candidate feature point set includes: The time-series detection value sequence is subjected to a sliding window bilateral mean difference transform to obtain a time-enhanced detection value sequence; Based on the time-series detection value sequence, a change amplitude sequence is determined, and based on the time-series enhancement detection value sequence, an enhancement change amplitude sequence is determined; Based on the enhanced change amplitude sequence, an adaptive judgment threshold is determined; A set of candidate feature points is determined based on the adaptive judgment threshold and the enhanced change amplitude sequence.

5. The method according to claim 1, characterized in that, The extraction of the time-series detection value sequence from the power system time-series data includes: The data packet format of the power system time-series data is parsed to obtain the data frame sequence; The timestamp and voltage per-unit value are read from each data frame of the data frame sequence as timing detection values ​​to obtain a timing detection value set; The time-series detection value set is arranged in ascending order by timestamp to obtain the time-series detection value sequence.

6. The method according to claim 1, characterized in that, The step of performing amplitude-domain recursive screening on the feature point set to obtain the fault feature point set includes: The feature point set is sorted in descending order of variation magnitude to obtain a feature point sequence; The following recursive segmentation operation is performed on the feature point sequence to obtain the set of tomographic feature points: Determine the absolute value of the amplitude difference of the feature point sequence to obtain the absolute value sequence of amplitude differences; The fault threshold is determined based on the absolute value sequence of amplitude differences. Based on the fault threshold, the feature point sequence is subjected to fault localization to obtain fault feature points; Based on the fault feature points, the feature point sequence is divided into a strong signal feature point sequence and a weak signal feature point sequence; Adaptive tomography verification is performed on the strong signal feature point sequence and the weak signal feature point sequence to obtain the tomography verification results. In response to the fault verification result indicating that the fault is valid, the strong signal feature point sequence is saved to the fault feature point set, and the weak signal feature point sequence is used as the feature point sequence to perform the recursive segmentation operation. In response to the fault verification result indicating that the fault is invalid, the feature points in the weak signal feature point sequence are identified as the second points to be confirmed, thus obtaining the second set of points to be confirmed, and the strong signal feature point sequence is saved to the fault feature point set.

7. A voltage surge identification and grid-connected protection control device for power grid fault ride-through, characterized in that, include: The receiving unit is configured to receive power system timing data in response to a voltage surge feature point identification command. The extraction unit is configured to extract a sequence of time-series detection values ​​from the power system time-series data, wherein the time-series detection values ​​in the sequence include timestamps and detection values, and the detection values ​​are voltage per-unit values. The coarse screening unit is configured to perform a rate-of-change domain coarse screening operation on the time-series detection value sequence to obtain a set of candidate feature points; Clustering units are configured to perform temporal clustering and boundary filtering on the candidate feature point set to obtain a feature point set. The fine screening unit is configured to perform amplitude domain recursive fine screening on the set of feature points to obtain a set of fault feature points. The generation unit is configured to generate control commands based on the set of fault feature points, and to send the control commands to the controller of the power system protection device, wherein the controller triggers protection action at the moment of voltage surge and resets the protection state at the moment of voltage surge clearing, based on the control commands.

8. An electronic device, characterized in that, include: One or more processors; A storage device on which one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-6.

9. A computer-readable medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-6.

Images