Network reference time based zero-crossing edge type identification method, system and apparatus

Abstract

The present application relates to the technical field of power line communication, and particularly relates to a zero-crossing edge type identification method, system and device based on network reference time, which comprises the following steps: collecting zero-crossing collection data and equipment identification information of a target station of a power line communication system according to a preset time period, generating a sub-group identification based on the equipment identification information; obtaining a pre-constructed sub-group edge information table, performing lookup matching on the sub-group identification and the sub-group edge information table, and obtaining an edge reference based on the lookup matching result; judging whether the edge reference and the zero-crossing sampling edge type are consistent, if the judgment result is inconsistent, correcting the zero-crossing sampling edge type, and obtaining the edge type of the target station based on the corrected zero-crossing sampling edge type; otherwise, taking the zero-crossing sampling edge type as the edge type of the target station. The present application can correct the network reference time, and improve the robustness of the zero-crossing edge type identification.

Description

Technical Field

[0001] This invention relates to the field of power line communication technology, and more specifically to a method, system, and apparatus for identifying zero-crossing edge types based on network reference time. Background Technology

[0002] In power line communication systems of low-voltage distribution networks, the Central Coordinator (CCO) needs to identify the physical phases (including phases A, B, and C) connected to each station (STA) to support functions such as differentiated bandwidth allocation and transformer substation identification for the three-phase power lines. Zero-Crossing Signal and Network Time Base (NTB) are the primary bases for physical phase identification. Protocol standards such as Q / GDW 12087 define the concept and basic acquisition methods of NTB, but do not specify the detailed processing methods for NTB data.

[0003] In existing technologies, phase identification using NTB (Network Timer) suffers from periodic overflow and wraparound due to the NTB being calculated by a 32-bit counter. Current solutions directly use the raw NTB data for phase matching without detecting and compensating for wraparound points, leading to discontinuities in data near these points and incorrect phase matching results.

[0004] Existing solutions do not extract the power line cycle as a physical quantity from NTB data, and therefore cannot automatically identify the 50Hz / 60Hz power grid frequency. This results in phase matching parameters that are not compatible with exported products or specific power grid scenarios. Furthermore, zero-crossing detection circuits from different manufacturers have inherent time delay differences, and existing solutions do not compensate for these differences, leading to phase identification errors when multiple manufacturers' equipment are networked together. Summary of the Invention

[0005] The purpose of this invention is to provide a method, system, and apparatus for identifying zero-crossing edge types based on network reference time, in order to solve the technical problem of lack of rule correction when identifying zero-crossing edge types using NTB data.

[0006] In a first aspect, the present invention provides a method for identifying zero-crossing edge types based on network reference time, comprising the following steps: According to a preset time period, zero-crossing data and equipment identification information of the target site of the power line communication system are collected, and sub-group identifiers are generated based on the equipment identification information; the zero-crossing data includes site address identifier, zero-crossing sampling edge type and network reference timestamp. Obtain the pre-built subgroup edge information table, match the subgroup identifier with the subgroup edge information table, and obtain the edge reference based on the match result; Determine whether the edge reference and the zero-crossing sampling edge type are consistent. If the determination result is inconsistent, correct the zero-crossing sampling edge type and obtain the edge type of the target station based on the corrected zero-crossing sampling edge type; otherwise, use the zero-crossing sampling edge type as the edge type of the target station.

[0007] The significant advantage of this invention is that, addressing the inherent time delay differences in zero-crossing detection circuits from different manufacturers, and the lack of compensation and calibration in existing solutions for these differences, which leads to misjudgment of edge types when multiple manufacturers' devices are networked together, this solution collects device identification information from target sites and generates fine-grained sub-group identifiers. Devices with the same hardware characteristics and firmware logic are grouped into the same sub-group. Then, a pre-built sub-group edge information table is used for lookup and matching to obtain an edge reference that matches the actual characteristics of the sub-group. This eliminates the reliance on a single fixed value of the system or the default type of the central coordinator when determining the edge reference, fundamentally avoiding the problem of inconsistent sampling edge directions caused by differences in zero-crossing detection circuits and firmware logic of different devices.

[0008] Meanwhile, to address the issue of existing solutions directly using raw network reference time data for phase matching and the problem of uncompensated counter wrapping leading to erroneous matching results, this solution establishes a zero-crossing acquisition mechanism based on the network reference timestamp and includes a consistency judgment step between the edge reference and the zero-crossing sampling edge type. When the two are determined to be inconsistent, the zero-crossing sampling edge type is proactively corrected instead of directly using the original sampling result with deviations. This effectively eliminates the interference of sampling data ambiguity caused by network reference time counter wrapping and other anomalies on edge type determination, and improves the robustness of zero-crossing edge type recognition.

[0009] Furthermore, this solution only requires matching pre-built information tables and simple logical judgments to complete edge type identification and correction, without the need for manual configuration of each device or reliance on specific calibration procedures. This significantly improves the automation level and accuracy of zero-crossing edge type identification in power line communication systems in multi-vendor equipment mixed networking scenarios, and provides accurate and unified edge type basic data for the reliable determination of subsequent physical phases.

[0010] Secondly, the present invention provides a zero-crossing edge type identification system based on network reference time, for implementing the zero-crossing edge type identification method based on network reference time as described above, including: The power line communication system data acquisition module is used to collect zero-crossing acquisition data and equipment identification information at the target site; The device grouping module is used to generate sub-group identifiers based on device identification information; The group lookup module is used to perform a lookup and match between the subgroup identifier and the subgroup edge information table, and to obtain the edge reference based on the lookup and match results; The edge type determination module is used to determine whether the edge reference and the zero-crossing sampling edge type are consistent; The zero-crossing sampling edge type correction module is used to correct the zero-crossing sampling edge type when the edge reference and the zero-crossing sampling edge type judgment are inconsistent. The edge type output module is used to output the corrected zero-crossing sampling edge type, or the zero-crossing sampling edge type as the edge type of the target station.

[0011] Thirdly, the present invention provides a zero-crossing edge type identification device based on network reference time, including a memory, a processor, and a computer program stored in the memory. The zero-crossing edge type identification device based on network reference time is located in a central coordinator. The processor executes the computer program to implement the steps of the above-described zero-crossing edge type identification method based on network reference time. Attached Figure Description

[0012] Figure 1 This is a flowchart of the zero-crossing edge type identification method based on network reference time in an embodiment of the present invention. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, a clear and complete description will be provided below in conjunction with the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the protection scope of the present invention.

[0014] See appendix Figure 1 The zero-crossing edge type identification method based on network reference time, as shown, includes the following steps: S1. Collect zero-crossing acquisition data and equipment identification information of the target site of the power line communication system according to the preset time period. The zero-crossing acquisition data includes the site address identifier, zero-crossing sampling edge type and network reference timestamp; generate sub-group identifiers based on the equipment identification information. Zero-crossing sampling edge types include four types: invalid, falling edge, rising edge, and double edge. The invalid state indicates that the edge type of the station has not yet been collected and determined; the rising edge indicates that the zero-crossing signal is triggered when the voltage crosses zero from negative to positive; the falling edge indicates that the voltage is triggered when the voltage crosses zero from positive to negative.

[0015] When acquiring the network reference timestamp, edge signals in the power line communication system can be collected using a zero-crossing acquisition module. Specifically, the zero-crossing acquisition module includes an edge detection circuit, a digital filtering circuit, and a network reference timestamp acquisition register. The edge detection circuit independently captures the rising and falling edges of each phase and converts the two edge signals into a unified pulse signal through board-level hardware. The specific structure and connection method of the edge detection circuit, numerical filtering circuit, and network reference timestamp acquisition register are existing technologies and will not be described in detail here.

[0016] When acquiring zero-cross data from target sites, the central coordinator can actively collect or gather zero-cross data proactively reported by each site. Active collection of zero-cross data includes: The central coordinator sends a collection request to the target site through the zero-crossing acquisition protocol. After receiving the request, the target site continuously collects network reference timestamps of several zero-crossing zero-crossing points, as well as site address identifiers and zero-crossing sampling edge types to form zero-crossing acquisition data, and reports it to the central coordinator.

[0017] The central coordinator initiates data collection requests at the following times: proactively collecting data to establish initial edge records when a new device joins the network; periodically collecting data for uncalibrated devices during network idle periods; and triggering verification data collection when inconsistent edge types are found among devices within a certain group. When a new device joins the network, if there are already records for the same group, the group's edge type is used as the initial prediction value. The central coordinator proactively initiates zero-crossing data collection, allowing the edge type to be predicted as the initial value based on existing group records when a new device joins the network, eliminating the need to wait for the first service to trigger data collection and shortening the phase recognition cycle for new devices.

[0018] The number of network reference timestamps collected is adaptively adjusted based on the signal-to-noise ratio (SNR) between the central coordinator and the target site. Specifically, when the SNR is greater than a preset threshold, a first sample value is used as the number of network reference timestamps collected to reduce the overhead during collection; otherwise, a second sample value is used as the number of network reference timestamps collected to improve the fault tolerance capability of zero-crossing phase identification, wherein the first sample value is less than the second sample value. In this embodiment, the preset threshold is 10dB, and the sample value is... The three physical phase zero-crossing signals are arranged alternately on the time axis with a period of 3. If the number of samples is an even multiple of 3, there are three possible phase alignment methods, leading to ambiguity in phase assignment. Using... The odd number of acquisitions can eliminate alignment ambiguity and ensure the uniqueness of phase determination.

[0019] The first and second sample values ​​are obtained by... The value of the first sample value is adjusted to adjust the final sampled value. Preferably, in this embodiment, the first sampled value is... The second sample value Specifically, the first sampling value consists of 19 network reference timestamps, and the second sampling value consists of 43 network reference timestamps. In other embodiments, other sampling values ​​can be selected according to the actual situation of the power line communication system. The sampling quantity n is adaptively adjusted according to the signal-to-noise ratio. After sampling, the network reference time data is subjected to interval variance checks and outlier removal to ensure the quality of the collected data and improve the reliability of phase determination.

[0020] The collected network reference timestamps are transmitted using the differential encoding format specified in the protocol standard. The two differential values ​​are stored in 3 bytes, with each value being a 12-bit differential value (the increment relative to the previous sampling point). After receiving the data, the original timestamp is restored by accumulating the data using the standard network reference timestamp.

[0021] The device identification information includes the supplier identifier, chip identifier, and firmware version number of the target site device. The specific details of obtaining the device's supplier identifier, chip identifier, and firmware version number are existing technologies and will not be elaborated upon here. The steps for generating the device group identifier code and sub-group identifier based on the device identification information include: When generating sub-group identifiers based on device identification information, the supplier identifier, chip identifier, and firmware version number are concatenated using a bitwise OR operation to form a 48-bit integer value. During concatenation, the supplier identifier is placed in the first 16 bits, the chip identifier in the middle 16 bits, and the firmware version number in the last 16 bits. Sub-group identifiers are used to identify finer-grained sub-groups, that is, a set of devices manufactured by a specific supplier using a specific chip under a specific firmware version.

[0022] Device families are divided into subgroups based on firmware version, each maintaining its own edge type. Sampling edge behavior is consistent within the same firmware version (determined by firmware-controlled zero-crossing detection logic). Subgroups are pre-assigned according to firmware version upon network access, fundamentally avoiding inconsistencies in edge characteristics between devices with the same firmware version. Different firmware versions do not affect each other, and devices automatically assign to new subgroups after firmware upgrades.

[0023] S2. Obtain the pre-built sub-group edge information table, match the sub-group identifier with the sub-group edge information table, and obtain the edge reference based on the match result. The sub-group edge information table is maintained by the central coordinator and is specifically stored in the persistent storage area of ​​the central coordinator. The sub-group edge information table is used to record the zero-crossing sampling edge type of each sub-group. The fields in the sub-group edge information table include sub-group identifier, edge type, and confidence state. The sub-group identifier is used to represent a sub-group of a device. The edge type is used to represent the direction of the zero-crossing sampling edge sampled by the sub-group. When the edge type value is 0, it represents a rising edge. When the edge type value is 1, it represents a falling edge. The confidence state is used to indicate whether the currently recorded edge type has been confirmed. When the confidence state value is 0, it means that it is pending confirmation, that is, the edge type has not been finally verified. When the confidence state value is 1, it means that it has been confirmed, that is, the edge type has been verified for reliability.

[0024] When searching and matching the subgroup identifier with the subgroup edge information table, the subgroup identifier is used as the key. The records in the subgroup edge information table are traversed one by one and compared with the subgroup identifier. If the subgroup identifier of a record in the subgroup edge information table matches the subgroup identifier, the search is successful and the edge type of the search-successful record is returned; otherwise, the search is unsuccessful.

[0025] When constructing an edge reference based on the search matching results, if the search is successful, the edge type in the search-successful record is used as the edge reference; otherwise, the preset edge type is used as the edge reference. Preferably, in this embodiment, the preset edge type is selected from the edge type of the central coordinator itself. The edge type of the central coordinator itself is determined according to the hardware configuration during system initialization and stored in the edge type register of the central coordinator.

[0026] S3. Determine whether the edge reference and the zero-crossing sampling edge type are consistent. If the determination result is inconsistent, correct the zero-crossing sampling edge type and obtain the edge type of the target station based on the corrected zero-crossing sampling edge type; otherwise, use the zero-crossing sampling edge type as the edge type of the target station. The steps for correcting the zero-crossing sampling edge type include: S301. Detect the wraparound point of the network reference time series and compensate the network reference time series based on the wraparound point to obtain the compensated network reference time series. The network reference time is a 32-bit unsigned counter. When the count value of the collected network reference time reaches the maximum value, its count value will wrap around to 0, causing the network reference time series to have a non-monotonic decreasing point. Therefore, it is necessary to detect the position of the wraparound point of the network reference time series and compensate the maximum value of the data accumulation counter after the wraparound point so that the data series restores the monotonically increasing continuity.

[0027] Specifically, when detecting the loopback point, the network reference time series is traversed, and when the loopback point in the network reference time series is... The value is greater than the first. The value, and the first value When the value is not 0, this is the first... The position corresponding to each value is the wraparound point. The 32-bit unsigned counter of the network reference time counts in microseconds, and its overflow period is 71 minutes. Therefore, the acquisition period of a single network reference time is usually much smaller than the overflow period of the counter. At most, only one wraparound will occur during the acquisition process. Therefore, only one wraparound compensation is needed to restore the network reference time series to a monotonically increasing continuous series.

[0028] In step S301, after obtaining the monotonically increasing compensation network baseline time series, linear regression is performed on the compensation network baseline time series, and outliers in the compensation network baseline time series are removed based on the linear regression results and an adaptive bias threshold. Specifically, this includes the following steps: A regression line is fitted based on the baseline time series of the compensated network. Specifically, when fitting the regression line, the virtual sampling sequence number of the compensated network baseline time is used as the independent variable, and the compensated network baseline time is used as the dependent variable. The slope of the regression line is the electric field period, and the formula for calculating the slope of the regression line is: In the formula, Let be the slope of the regression line. To compensate for the network reference time series One sampling point, To compensate for the arithmetic mean of all sample numbers in the network's baseline time series, To compensate for the network reference time series A network reference time value, This is to compensate for the arithmetic mean of the network reference time values ​​in the network reference time series.

[0029] The intercept of the regression line is the initial value of the network's baseline time. The formula for calculating the regression intercept is: In the formula, The intercept of the regression line.

[0030] In practical implementation, integer amplification can be performed using an integer amplification factor to avoid floating-point operations. The regression line is obtained by fitting using the least squares method. The specific fitting content of the regression line is existing technology and will not be elaborated here.

[0031] The deviation of each compensation network from the regression line in the baseline time series of the compensation network is calculated using the following formula: In the formula, To compensate for the deviation of the network reference time from the regression line, The first time series of the compensation network reference time series The predicted value of each sampling point on the regression line. This is for taking the absolute value.

[0032] The proportion of the deviation values ​​of the base time in the base time series of the compensation network that are greater than the deviation threshold is calculated. When the proportion of the deviation values ​​of the base time in the compensation network that are greater than the deviation threshold is greater than the preset deviation proportion, the data quality of the base time series of the compensation network is determined to be unqualified. At this time, the base time series of the compensation network is discarded and the base time series of the network is re-acquired. Otherwise, the base time in the compensation network that is greater than the deviation threshold is removed. The deviation threshold is the proportion of the regression slope. Preferably, in this embodiment, the deviation threshold is 10% and the preset deviation proportion is 10%.

[0033] Linear regression was performed again on the compensation network reference time series after the removal process to obtain a more accurate compensation network reference time series and power line period.

[0034] In step S301, outliers in the compensation network reference time series are removed. Specifically, the mean and standard deviation of each compensation network reference time deviation value are calculated. When the difference between the compensation network reference time value and the mean of the deviation value is greater than or equal to twice the standard deviation, the compensation network reference time value is considered an outlier and is removed from the compensation network reference time series.

[0035] S302. Extract the power line period based on the reference time series of the compensation network, convert the power line period into the initial frequency, and obtain the grid frequency by comparing the initial frequency with the frequency threshold. When extracting the power line period, the slope of the regression line fitted in step S301 is used as the power line period. Specifically, the physical meaning of the slope of the regression line is the count value of the network reference time corresponding to a complete cycle of alternating current, which is the reciprocal of the grid frequency. The power line period can be converted into the initial frequency by performing the reciprocal operation.

[0036] When determining the grid frequency, if the initial frequency is less than or equal to the frequency threshold, the network reference time period is longer and its frequency is lower; in this case, the grid frequency is determined to be 50Hz. Otherwise, the grid frequency is determined to be 60Hz. Preferably, in this embodiment, the frequency threshold is 57Hz.

[0037] S303. Obtain the reference network reference time series, determine the phase state of the network reference time series based on the reference network reference time series and the compensation network reference time series, and perform periodic offset correction on the network reference time series based on the phase state to obtain the network reference time correction sequence; and obtain the corrected zero-crossing sampling edge type based on the network reference time correction sequence.

[0038] When obtaining the reference network reference time, the network reference time of the central coordinator is collected according to the preset time period and a reference network reference time series is constructed. The preprocessing steps of the reference network reference time series are the same as the steps of compensating for network reference time series and handling outliers in step S301, and will not be described again here.

[0039] When performing phase matching based on the reference network reference time series and the compensation network reference time series, the following steps are included: The difference between the reference network time series and the network time series is calculated to obtain the difference sequence. Specifically, the difference sequence is obtained by subtracting the reference network time series and the network time series point by point. The specific content of calculating the difference between the reference network time series and the network time series is existing technology and will not be described in detail here.

[0040] The network reference time count value corresponding to the power grid frequency cycle is obtained based on the power grid frequency. The difference sequence is then subjected to half-cycle modulo operation based on this network reference time count value to obtain the half-cycle difference. This process eliminates the repetition caused by integer half-cycles and extracts the remaining deviation within one half-cycle. The formula for calculating the half-cycle difference is as follows: In the formula, The difference between half a cycle. It is a difference sequence. For modulo operation, The network reference time count value corresponds to the power grid frequency cycle, wherein the network reference time count value is determined by the clock frequency of the counter. The counter counts at a fixed frequency, and the network reference time count value within one cycle is the ratio of the frequency to the power grid frequency in the formula. Preferably, in this embodiment, when the power grid frequency is 50Hz, the network reference time count value corresponding to the power grid frequency cycle is 500000; when the power grid frequency is 60Hz, the network reference time count value corresponding to the power grid frequency cycle is 416667.

[0041] The full-cycle difference is obtained by performing full-cycle modulo operation on the difference sequence based on the network reference time count value corresponding to the power grid frequency cycle. This eliminates the repetition caused by an integer number of full cycles and extracts the residual deviation within a full cycle. The formula for calculating the full-cycle difference is as follows: In the formula, This is the difference over the entire period.

[0042] When the half-cycle difference is less than the cycle threshold, the phase state of the base time series of the compensation network is determined to be in phase; where the cycle threshold is one-sixth of the half-cycle, specifically the cycle threshold is about 30° phase angle.

[0043] When the difference between the full-cycle difference and the half-cycle difference is less than the cycle threshold, the phase state of the compensation network reference time series is determined to be reversed.

[0044] When the network reference time series is corrected by periodic offset based on phase state to obtain the network reference time correction sequence, specifically when the phase state is reversed, the compensation network reference time series is corrected. The zero-crossing signal generates two zero-crossing points per cycle in the AC power grid—a rising edge zero-crossing and a falling edge zero-crossing—separated by half a cycle (180° phase) on the time axis. When the target station and the comparison reference use different sampling edges (one rising edge, the other falling edge), the network reference timestamp collected by the station is offset by half a cycle relative to the reference. Since the three physical phases (A / B / C phases) of the low-voltage distribution network are arranged at one-third cycle intervals on the time axis, a half-cycle offset can cause a misalignment of one phase in the physical phase determination result (e.g., phase A is misjudged as phase B). Therefore, when the edge types are inconsistent, the physical phase determination result of the station is corrected by half a cycle. Specifically, this is achieved by flipping the inverted phase flag of the station; after correction, the edge type of the station is updated to the edge type collected in this acquisition, and the physical phase update information of the station is broadcast. For three-phase equipment, due to the complex phase relationships between the phases, half-cycle correction may lead to misjudgment. Therefore, the correction is skipped and left for separate data acquisition and calibration.

[0045] Step S3 further includes grouped edge learning and batch correction. Specifically, edge type learning is performed on devices with the same sub-group identifier target site, and batch correction is performed on the target sites located in that sub-group identifier based on the edge type learning results. The zero-cross-sampling edge characteristic common to devices from the same supplier, using the same chip, and with the same firmware version is represented by the first site in that sub-group whose edge type data is collected. Since sub-groups are divided by firmware version, devices in the same sub-group have consistent sampling edge behavior, fundamentally avoiding the problem of inconsistent edge characteristics among devices with the same firmware version. The specific steps include: The zero-cross sampling edge types of two devices with the same subgroup identifier are compared. If the zero-cross sampling edge types of the two devices are consistent, it is determined whether the edge type recorded by this subgroup is consistent with the edge type of the central coordinator itself. If they are consistent, no batch correction is needed; if they are inconsistent, batch correction is triggered. If the zero-cross sampling edge types of the two devices are inconsistent, the current sampling zero-cross sampling edge type is determined to be sampling noise, and the zero-cross sampling edge types of the two devices with the same subgroup identifier are re-collected. If the zero-cross sampling edge types of the two devices with the same subgroup identifier are still inconsistent after re-collection, the subgroup is marked as suspicious, and each device in the subgroup is sampled individually. Since devices with the same firmware version have consistent sampling edge behavior (determined by the zero-crossing detection logic controlled by the firmware), the main purpose of cross-validation is to eliminate sampling noise, rather than to handle hardware discreteness. Before batch correction, it is required that the sampling results of at least two devices in the same group are consistent to trigger the correction, avoiding the miscorrection of the entire group due to sampling noise from a single device. Compared with the scheme without cross-validation, this only adds one sampling, while still retaining the efficiency advantage of saving N-2 samplings. Residual deviations are gradually identified and corrected through regular inspections, representing a reasonable balance between efficiency and tolerance.

[0046] When performing batch corrections, a list of station devices is obtained by traversing all online stations in the power line communication system. When the network size is larger than a preset threshold, three-phase filtering is performed first, followed by sub-group filtering, to quickly eliminate a large number of irrelevant devices. When the network size is smaller, sub-group filtering is performed first for precise location.

[0047] The site device list is filtered to exclude devices currently performing data acquisition operations to avoid self-interference; and since the central coordinator itself does not participate in phase correction, it is also necessary to exclude the central coordinator itself from the site device list.

[0048] For each device in the filtered list of site devices, perform the following steps in sequence: Obtain the device identification information of the device to construct the sub-group identifier of the device. The specific steps for constructing the sub-group identifier are the same as those in step S1, and will not be repeated here.

[0049] The system determines whether the subgroup identifier of the device meets the correction conditions. If it does, a half-cycle correction is performed on the device; otherwise, no half-cycle correction is performed. The correction conditions include: the subgroup identifier being the same as the one currently being corrected in the batch; the current device not being a three-phase device; and the current device's zero-crossing sampling edge type being invalid. This ensures that during half-cycle correction, only devices with the same subgroup as the newly confirmed subgroup are processed, while avoiding incorrect correction of three-phase devices. Furthermore, if the device's zero-crossing sampling edge type is not invalid, it indicates that the device has already been individually calibrated, preventing duplicate correction of that device.

[0050] Batch correction uses a four-layer filtering mechanism: sub-group identifier matching, three-phase device exclusion, calibrated device protection, and self-exclusion to precisely limit the correction scope. The filtering priority is dynamically adjusted according to the network size to optimize traversal efficiency. The specific details of the half-cycle correction are the same as those of the network reference time correction sequence obtained by the cycle offset correction in step S3, and will not be repeated here.

[0051] Because subgroups are divided by firmware version, different firmware versions within the same device family correspond to different subgroup records. Each subgroup independently maintains its edge type and confidence level, and performs cross-validation and batch correction independently, without affecting the others. The firmware version number can be obtained when the device joins the network, and the subgroup is determined at the time of network access, without waiting for data collection and subsequent judgment. When the device firmware is upgraded, its subgroup affiliation is automatically updated to the subgroup corresponding to the new firmware version, requiring re-collection and confirmation of the edge type.

[0052] After the batch correction is completed, the central coordinator relies on the existing periodic inspection mechanism (part of normal network operation and maintenance) to conduct spot checks on the corrected devices. If, in rare cases, inconsistencies are found between devices with the same firmware version (such as rare cases like internal firmware logic anomalies), the device is individually sampled and calibrated, and the subgroup is marked as suspicious. Subsequently, the sampling of this subgroup is changed to be done on a per-device basis.

[0053] This invention forms a collaborative closed loop through its various steps: firmware version pre-grouping, cross-validation for noise reduction, batch correction for efficiency improvement, and periodic inspection as a fallback. Subgroups are pre-assigned according to firmware version upon network access, fundamentally avoiding inconsistencies between devices with the same firmware version; cross-validation eliminates acquisition noise; batch correction improves efficiency; and periodic inspection covers the very few abnormal cases. Each mechanism has a clear division of labor and a progressive hierarchy, collectively ensuring the accuracy and efficiency of phase identification.

[0054] Since the same site may have both individual edge types (determined through individual data collection) and group edge types (obtained through group learning), this scheme parses the edge type of each site in the following priority order: When the edge type of the target site is not invalid, it means that the site has been individually calibrated and its edge type has been determined through an independent acquisition process. In this case, the edge type of the site itself is used directly. This result has the highest priority, taking precedence over any grouping or default configuration.

[0055] If the edge type field of a site is invalid, it means that the site has not been individually calibrated. The system then checks if a valid record exists in the subgroup edge information table for the subgroup to which the site belongs. If it does, the edge type corresponding to that subgroup is used. This result has secondary priority and is used for unified configuration of devices within the same subgroup.

[0056] If neither of the above two conditions is met—that is, the station has not been individually calibrated and its subgroup has no valid record in the edge information table—then the system's default edge type is used. This ensures that all devices can still function normally without any prior information. Preferably, in this embodiment, the default edge type is rising edge.

[0057] This three-tiered priority mechanism establishes a coordination relationship between individual calibration and group learning. Devices that are calibrated individually use the individual edge to ensure accuracy; devices that are not individually calibrated but belong to a group that has already been learned use the group edge to improve efficiency by leveraging the results of group learning; devices that are neither individually calibrated nor covered by group learning use the default edge to provide a reasonable backoff benchmark. The learned group uses the group edge as the comparison benchmark, while the unlearned group uses the central coordinator edge as the comparison benchmark. The comparison benchmark dynamically switches according to the learning status, ensuring the accuracy of the learned group while providing a reasonable default comparison benchmark for the unknown group. Furthermore, each record in the subgroup edge information table maintains a confidence flag to indicate its confirmation status. When the confidence level is 0, it is marked as pending confirmation and can only be used as the initial prediction value for new devices, without triggering batch correction; when the confidence level reaches 1 (confirmed through cross-validation), it is marked as confirmed and batch correction can be triggered. Subgroups of different firmware versions maintain their confidence levels independently and do not affect each other.

[0058] Compared to existing technologies, this solution eliminates the need for manual configuration or device-by-device calibration. During normal operation, the central coordinator automatically learns the zero-crossing edge types of different device groups, enabling automatic identification of physical phases across vendors. Furthermore, when the edge type of a certain group is found to be inconsistent with that of the central coordinator for the first time, after cross-validation, a single operation corrects the phase information of all devices in that group that have not been individually calibrated across the entire network. This reduces the phase correction of N devices in the same group from N acquisitions to 2 acquisitions plus 1 traversal, significantly reducing bandwidth overhead and time consumption. This solution enables accurate physical phase determination, supporting multiple upper-level applications: improved transformer area identification accuracy, preventing cross-transformer area interference; phase-based differentiated bandwidth allocation, optimizing three-phase load balancing; and three-phase imbalance detection, providing data support for power grid management.

[0059] This invention also aims to provide a zero-crossing edge type identification system based on network reference time, for implementing the zero-crossing edge type identification method based on network reference time as described above, including: The power line communication system data acquisition module is used to collect zero-crossing acquisition data and equipment identification information at the target site; The device grouping module is used to generate sub-group identifiers based on device identification information; The group lookup module is used to perform a lookup and match between the subgroup identifier and the subgroup edge information table, and to obtain the edge reference based on the lookup and match results; The edge type determination module is used to determine whether the edge reference and the zero-crossing sampling edge type are consistent; The zero-crossing sampling edge type correction module is used to correct the zero-crossing sampling edge type when the edge reference and the zero-crossing sampling edge type judgment are inconsistent. The edge type output module is used to output the corrected zero-crossing sampling edge type, or the zero-crossing sampling edge type as the edge type of the target station.

[0060] The present invention also aims to provide a zero-crossing edge type identification device based on network reference time, including a memory, a processor, and a computer program stored in the memory. The zero-crossing edge type identification device based on network reference time is located in a central coordinator. The processor executes the computer program to implement the steps of the above-described zero-crossing edge type identification method based on network reference time.

[0061] 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 description is only a specific implementation method of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the scope of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for identifying zero-crossing edge types based on network reference time, characterized in that, Includes the following steps: According to a preset time period, zero-crossing data and equipment identification information of the target site of the power line communication system are collected, and sub-group identifiers are generated based on the equipment identification information; the zero-crossing data includes site address identifier, zero-crossing sampling edge type and network reference timestamp. Obtain the pre-built subgroup edge information table, match the subgroup identifier with the subgroup edge information table, and obtain the edge reference based on the match result; Determine whether the edge reference and the zero-crossing sampling edge type are consistent. If the determination result is inconsistent, correct the zero-crossing sampling edge type and obtain the edge type of the target station based on the corrected zero-crossing sampling edge type. Otherwise, the zero-cross-sampling edge type will be used as the edge type of the target site.

2. The zero-crossing edge type identification method based on network reference time according to claim 1, characterized in that, The methods for collecting zero-crossing data from target sites in the power line communication system include: Active collection: A collection request is sent to the target site. After receiving the request, the target site continuously collects several zero-crossing zero-crossing points, network reference timestamps, site address identifiers, and zero-crossing sampling edge types to form zero-crossing collection data. Passive data collection: Collecting zero-cross-data actively reported by each station.

3. The zero-crossing edge type identification method based on network reference time according to claim 2, characterized in that, The timing for initiating the active data collection includes: When a new device joins the network, data is actively collected to establish an initial edge record; during network idle periods, data is collected periodically for uncalibrated devices; and verification data is triggered when inconsistent edge types are found among devices within a sub-group.

4. The zero-crossing edge type identification method based on network reference time according to claim 1, characterized in that, The number of network reference timestamps in the collected zero-cross data is adaptively adjusted based on the signal-to-noise ratio between the central coordinator and the target site. When the signal-to-noise ratio is greater than a preset threshold, the first sample value is used as the number of times the network reference timestamp is collected; otherwise, the second sample value is used as the number of times the network reference timestamp is collected, wherein the first sample value is less than the second sample value.

5. The zero-crossing edge type identification method based on network reference time according to claim 1, characterized in that, The step of correcting the zero-crossing sampling edge type includes: The loop points of the network reference time series are detected, and the network reference time series is compensated based on the loop points to obtain the compensated network reference time series. The power line period is extracted based on the reference time series of the compensation network, the power line period is converted into the initial frequency, and the grid frequency is obtained by comparing the initial frequency with the frequency threshold. Obtain the reference network reference time series, determine the phase state of the network reference time series based on the reference network reference time series and the compensation network reference time series, and perform periodic offset correction on the network reference time series based on the phase state to obtain the network reference time correction series. The corrected zero-crossing sampling edge type is obtained based on the network reference time correction sequence.

6. The zero-crossing edge type identification method based on network reference time according to claim 5, characterized in that, After obtaining the compensated network reference time series by compensating the network reference time series based on the wraparound point, the process also includes performing linear regression on the compensated network reference time series: A regression line was fitted based on the baseline time series of the compensation network; Calculate the deviation of each compensation network from the regression line in the baseline time series of the compensation network; The proportion of deviation values ​​of the compensation network reference time in the statistical compensation network reference time series that are greater than the deviation threshold is determined. When the proportion of deviation values ​​of the compensation network reference time that are greater than the deviation threshold is greater than the preset deviation proportion, the data quality of the compensation network reference time series is determined to be unqualified. At this time, the compensation network reference time series is discarded and the network reference time series is re-acquired. Otherwise, remove the compensation network reference time from the compensation network reference time series that exceeds the deviation threshold; Linear regression was then performed again on the baseline time series of the compensation network after the removal process.

7. The zero-crossing edge type identification method based on network reference time according to claim 5, characterized in that, The determination of the phase state of the network reference time series based on the reference network reference time series and the compensated network reference time series includes: The difference between the reference network time series and the network time series is calculated to obtain the difference series; The network reference time count value corresponding to the power frequency cycle of the power grid is obtained based on the power grid frequency. The half-cycle difference value is obtained by taking the modulus of the difference sequence based on the network reference time count value. The full-cycle difference value is obtained by taking the modulus of the difference sequence based on the network reference time count value. When the half-cycle difference is less than the cycle threshold, the phase state of the compensation network reference time series is determined to be in phase; when the difference between the full-cycle difference and the half-cycle difference is less than the difference threshold, the phase state of the compensation network reference time series is determined to be out of phase.

8. The zero-crossing edge type identification method based on network reference time according to claim 1, characterized in that, The step of correcting the zero-crossing sampling edge type also includes grouped edge learning and batch correction, specifically including the following steps: The zero-cross sampling edge types of two devices with the same subgroup identifier are compared. If the zero-cross sampling edge types of the two devices are consistent, it is determined whether the edge type recorded in the subgroup is consistent with the edge type of the central coordinator itself. If they are consistent, no batch correction is required; if they are inconsistent, batch correction is triggered. The system iterates through all online sites in the power line communication system to obtain a list of site devices, and then filters the list of site devices. The filtering rules include excluding devices that are currently performing data acquisition operations and the central coordinator. For each device in the filtered list of site devices, perform the following steps in sequence: Obtain the device identification information and construct sub-group identifiers; Determine whether the subgroup identifier meets the correction conditions. If it does, perform half-cycle correction on the device; otherwise, do not perform half-cycle correction on the device. The correction conditions include that the subgroup identifier is the same subgroup identifier in the current batch correction, the current device is not a three-phase device, and the zero-crossing sampling edge type of the current device is invalid.

9. A zero-crossing edge type identification system based on network reference time, used to implement the zero-crossing edge type identification method based on network reference time as described in any one of claims 1-8, characterized in that, include: The power line communication system data acquisition module is used to collect zero-crossing acquisition data and equipment identification information at the target site; The device grouping module is used to generate sub-group identifiers based on device identification information; The group lookup module is used to perform a lookup and match between the subgroup identifier and the subgroup edge information table, and to obtain the edge reference based on the lookup and match results; The edge type determination module is used to determine whether the edge reference and the zero-crossing sampling edge type are consistent; The zero-crossing sampling edge type correction module is used to correct the zero-crossing sampling edge type when the edge reference and the zero-crossing sampling edge type judgment are inconsistent. The edge type output module is used to output the corrected zero-crossing sampling edge type, or the zero-crossing sampling edge type as the edge type of the target station.

10. A zero-crossing edge type identification device based on network reference time, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The zero-crossing edge type identification device based on network reference time is located in the central coordinator, and the processor executes the computer program to implement the steps of the zero-crossing edge type identification method based on network reference time as described in any one of claims 1-8.

Images