Primary and secondary fusion pole-mounted switch fault studying and judging method based on edge calculation

Abstract

The invention discloses a primary and secondary fusion pole-mounted switch fault studying and judging method based on edge computing, and relates to the technical field of power system fault detection and diagnosis. Primary side electric parameters and secondary side state data are synchronously collected under the unified time scale of edge computing nodes, a primary and secondary fusion time sequence data set is constructed, and the primary and secondary fusion pole-mounted switch fault studying and judging method based on edge computing is achieved; the problem that fine time sequence correlation analysis cannot be carried out due to asynchronous data time marks in a traditional method is solved fundamentally, and a reliable data foundation is laid for follow-up accurate analysis. The change trend of the primary response time difference sequence and the secondary response time difference sequence in multiple cycles is extracted and analyzed, progressive defects such as slow drifting of action time caused by aging and jamming of a switching mechanism or a control loop can be effectively recognized, early warning of latent faults is achieved, and the limitation that judgment is prone to missing in single-cycle threshold judgment is broken through.

Description

Technical Field

[0001] This invention relates to the field of power system fault detection and diagnosis technology, and in particular to a method for judging faults of pole-mounted switches based on edge computing and primary and secondary fusion. Background Technology

[0002] With the rapid development of smart grid and distribution network automation technologies, status perception and intelligent fault assessment of power distribution equipment have become core aspects of improving power supply reliability. Pole-mounted switches, as key equipment for segmentation, interconnection, and fault isolation in distribution networks, directly impact fault handling speed and user experience in terms of operational reliability. Traditional fault monitoring technologies primarily rely on information reported by protection devices within substations or fault indicators in the master station system. This information is mostly action conclusion signals (such as "overcurrent trip") or simple binary states (such as "open / close"), lacking a deep characterization of the fine-grained timing correlation between the changes in primary electrical quantities and the response behavior of the secondary control circuit. In recent years, primary and secondary integration technology has become an important development direction for intelligent pole-mounted switches. By integrating current / voltage sensors and intelligent terminals into the switch body, it achieves on-site digital acquisition of both electrical and non-electrical quantities. Simultaneously, the rise of edge computing has made real-time data processing and intelligent analysis possible on the equipment side, driving the evolution of fault assessment models from "centralized analysis at the master station" to "on-site decision-making at the edge," aiming to shorten the fault assessment chain and improve real-time response.

[0003] However, existing applications based on primary and secondary fusion and edge computing still have significant limitations in achieving accurate fault diagnosis. First, at the data level, most solutions only focus on identifying primary-side fault characteristics (such as sudden current changes) or simply reporting secondary-side status signals, lacking millisecond-level synchronous correlation analysis of continuous transient processes on the primary side and secondary-side control responses under a unified time scale. Due to communication delays during data upload to the master station and clock asynchrony between different devices, traditional methods struggle to accurately capture and quantify the inherent time difference between the "start of electrical quantity change" and the "mechanical state response," which is precisely the key basis for judging the health status of the control loop and whether the relay's operating characteristics have deteriorated.

[0004] Secondly, in terms of judgment logic, existing methods are mostly based on threshold comparisons at a single point or in a single cycle, such as judging whether the current exceeds a set value or whether the switch operates at the expected time. They lack longitudinal trend analysis of the consistency of the primary-secondary response timing relationship over multiple consecutive operating cycles. When there are progressive defects such as jamming in the switching mechanism or oxidation of relay contacts, the behavior is characterized by a slow drift in the action time rather than complete failure, making it very easy to miss in single-cycle detection.

[0005] Finally, in terms of fault scenario coverage, traditional methods are insufficient in assessing the serious hidden danger of "abnormal secondary circuit causing switch failure to operate". They can often only report the result of "overcurrent failure", and cannot distinguish at the edge side whether it is due to different reasons such as power failure of control power supply, open circuit of relay coil, or mechanical jamming of mechanism, resulting in a lack of targeted operation and maintenance in the future. Summary of the Invention

[0006] The purpose of this section is to outline some aspects of the embodiments of the present invention and to briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this section, the abstract and title of the invention. Such simplifications or omissions shall not be used to limit the scope of the present invention.

[0007] In view of the aforementioned existing problems, the present invention is proposed.

[0008] To address the aforementioned technical problems, this invention provides the following technical solution: In the edge computing node corresponding to the pole-mounted switch, continuous sampling data of primary-side electrical parameters and secondary-side state data are synchronously collected, and local time stamps of a unified format are added to form a primary-secondary fusion time-series dataset; based on the primary-secondary fusion time-series dataset, the time difference sequence between the start time of the primary-side electrical parameter change and the secondary-side state response time is extracted in the edge computing node, and the changing trend of the time difference sequence within a continuous operating cycle is analyzed to obtain a time misalignment feature result reflecting the primary-secondary response relationship; based on the change amplitude and duration range of the primary-side electrical parameters within the current operating cycle, the theoretical operating state range of the pole-mounted switch within the corresponding cycle is determined, and a consistency comparison is performed with the actual state response of the secondary side, outputting a state consistency discrimination result; when the state consistency discrimination result indicates a multi-cycle consistency deviation, and no effective secondary-side action is detected within the corresponding cycle, the no-action range is reverse-verified in conjunction with the time misalignment feature result, completing the pole-mounted switch fault assessment and generating an edge-side assessment conclusion.

[0009] As a preferred embodiment of the present invention, the formation of the primary and secondary fusion time-series dataset includes: continuously sampling the primary side voltage and current at a fixed sampling period in the edge computing node corresponding to the pole-mounted switch, and binding each primary side electrical parameter sampling value with the corresponding local time identifier to form primary side electrical parameter time-series data; in the same edge computing node, event-triggered acquisition of secondary side switch status quantities and protection trigger signals, and when a status change is detected, attaching a local time identifier in the same format as the primary side electrical parameter time-series data to the corresponding secondary side status response data; using the local time identifier as an alignment reference, sorting and merging the primary side electrical parameter time-series data and the secondary side status response data by time to construct a primary and secondary fusion time-series dataset containing a primary side electrical parameter feature value sequence and a secondary side status change record.

[0010] As a preferred embodiment of the present invention, in the primary and secondary fusion time series dataset, for each primary side electrical parameter change record characterized by the current or voltage mutation exceeding the corresponding change detection threshold, the time position of the first subsequent secondary side state change record is marked, and the corresponding time difference is written into the primary and secondary fusion time series dataset.

[0011] In a preferred embodiment of the present invention, the analysis of the changing trend of the time difference sequence within a continuous running cycle includes: dividing the time difference sequence into continuous running cycles of a preset fixed duration according to the local time identifier; based on a preset rolling window length, sequentially sliding the time difference sequence to extract multiple local sub-sequences within a time period covered by at least two of the continuous running cycles; calculating the arithmetic mean and standard deviation of the time difference contained in each local sub-sequence to form a time difference average sequence and a time difference standard deviation sequence; performing least squares linear fitting on the time difference average sequence to obtain the slope of the fitted line; calculating the arithmetic mean of all standard deviations in the time difference standard deviation sequence, and recording it as the fluctuation benchmark value; if the absolute value of the slope is greater than a preset slope threshold, it is determined that the time difference sequence has a monotonic changing trend; otherwise, it is determined that there is no significant monotonic trend; if the fluctuation benchmark value is less than or equal to a preset fluctuation threshold, it is determined that the fluctuation degree of the time difference sequence is low; otherwise, it is determined that the fluctuation degree of the time difference sequence is high.

[0012] As a preferred embodiment of the present invention, the time misalignment characteristic results include at least four types: positive trend drift with low volatility, negative trend drift with low volatility, unstable trend with high volatility, and no obvious trend with low volatility.

[0013] In a preferred embodiment of the present invention, the determination of the theoretical operating state range of the pole-mounted switch within the corresponding cycle includes: extracting the instantaneous current sampling sequence and the instantaneous voltage sampling sequence within the current operating cycle from the primary side electrical parameter timing data; calculating the current power frequency RMS value sequence and the voltage power frequency RMS value sequence based on the current instantaneous sampling sequence and the voltage instantaneous sampling sequence, respectively; calculating the sudden change amount within a preset short-time analysis window based on the current power frequency RMS value sequence; comparing the sudden change amount with a dynamic overcurrent threshold value; and recording the corresponding value when the sudden change amount continuously exceeds the dynamic overcurrent threshold value for a preset minimum duration. The start time is the overcurrent initiation time. When the sudden change in voltage drops below the dynamic overcurrent threshold, the corresponding time point is recorded as the overcurrent end time. The dynamic overcurrent threshold is the nth percentile of the current frequency RMS value data of this line under normal load in the same historical period. n is a constant. The voltage frequency RMS value sequence is compared with the preset voltage undervoltage judgment threshold. The period when the voltage is continuously lower than the preset voltage undervoltage judgment threshold is marked as the voltage undervoltage period. The interval from the overcurrent initiation time to the overcurrent end time is logically ANDed with the voltage undervoltage period, and the overlapping part of the two on the time axis is taken as the theoretical operating state interval.

[0014] As a preferred embodiment of the present invention, the output of the state consistency discrimination result includes: extracting interval data from the current frequency RMS value sequence and voltage frequency RMS value sequence corresponding to the theoretical operating state interval, respectively, with the overcurrent initiation time as the starting point and the overcurrent end time as the ending point, and detecting all continuous segments in the current frequency RMS value sequence that exceed the dynamic overcurrent threshold value; treating each detected continuous segment as an independent primary-side electrical parameter change event, and recording the overcurrent initiation time and the corresponding current mutation amount to form a primary event sequence to be compared; extracting timestamps within a preset buffer period before and after the theoretical operating state interval from the secondary-side state response data. All state change records are collected, and protection trigger signal records are filtered out to form a secondary event sequence to be compared. Each primary-side electrical parameter change event in the primary event sequence to be compared is matched with the secondary event sequence to be compared in terms of time correlation: for each primary-side electrical parameter change event, if at least one protection trigger signal record can be matched within a dynamic tolerance time window centered on the overcurrent start time, the primary-side electrical parameter change event is determined to be triggered effectively; otherwise, it is determined to be a missing trigger. If all primary-side electrical parameter change events are triggered effectively, the output is completely consistent; if there are missing trigger events, the output is locally inconsistent or seriously deviates based on the proportion or distribution characteristics of the missing trigger events.

[0015] As a preferred embodiment of the present invention, the reverse verification includes: for each primary-side electrical parameter change event determined to be missing in the state consistency discrimination result, extracting all secondary-side state change records within the corresponding time period from the primary-secondary fusion time-series dataset, and determining whether they contain protection trigger signal records; simultaneously, based on the local logs of the edge computing node, checking whether there are communication interruption records or clock synchronization anomaly alarm records within the same time period; if a protection trigger signal record is found, it is determined that the primary-side electrical parameter change event has actually been triggered, and the state of the corresponding event is corrected from missing trigger to valid trigger; if no protection trigger signal record is found, but a communication interruption or clock synchronization anomaly record is found, it is determined that the missing trigger is caused by a secondary circuit or clock anomaly, and a corresponding auxiliary judgment conclusion is generated; based on the review results of all missing trigger events, updating the state consistency discrimination result, and combining the auxiliary judgment conclusion, generating the edge-side judgment conclusion used to indicate the fault type or anomaly cause.

[0016] The beneficial effects of this invention are as follows: By synchronously collecting primary-side electrical parameters and secondary-side state data under a unified time scale at edge computing nodes, this invention constructs a primary-secondary fusion time-series dataset, fundamentally solving the problem that traditional methods cannot perform fine-grained time-series correlation analysis due to asynchronous data time scales, thus laying a reliable data foundation for subsequent accurate analysis. By extracting and analyzing the changing trends of the primary and secondary response time difference sequences over multiple periods, this invention can effectively identify progressive defects such as slow drift in action time caused by aging or jamming in switching mechanisms or control loops, achieving early warning of latent faults and overcoming the limitation of easy omission in single-period threshold judgment.

[0017] Furthermore, by comparing the theoretical action range with the actual response and combining the time misalignment characteristics to perform reverse verification of the failure to operate, the root cause of the "switch failure" type of fault can be preliminarily determined at the edge side. This effectively distinguishes whether the electrical quantity has not reached the threshold, the control loop is abnormal, or the mechanical fault, greatly improving the accuracy of fault diagnosis and the pertinence of operation and maintenance decisions, while reducing the analysis burden and communication dependence of the main station. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a flowchart illustrating a primary and secondary fusion method for fault assessment of a pole-mounted switch based on edge computing, as shown in this invention. Detailed Implementation

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0020] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort should fall within the scope of protection of this invention.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] According to an embodiment of the present invention, in combination Figure 1 The flowchart shown illustrates a method for fault assessment of pole-mounted switches based on edge computing, involving primary and secondary fusion. S1: In the edge computing node corresponding to the pole-mounted switch, continuously sampled data of primary side electrical parameters and secondary side status data are collected synchronously, and local time identifiers in a unified format are attached to form a primary and secondary fused time series dataset.

[0023] S1.1: In the edge computing node corresponding to the pole-mounted switch, the primary side voltage and current are continuously sampled according to a fixed sampling period (e.g., 128 points per power frequency cycle), and each primary side electrical parameter sampling value is bound to the corresponding local time identifier to form primary side electrical parameter timing data.

[0024] Specifically, after each sampling is completed, the current time is immediately read from the unified local high-precision clock source (such as a real-time clock module based on a crystal oscillator), a time stamp with microsecond-level precision is generated, and the time stamp is bound to the instantaneous value of the sample to form a sequence of original sampling points of primary side electrical parameters with precise time stamps.

[0025] S1.2: In the same edge computing node, the secondary side switch status quantities (such as open / close position) and protection trigger signals (such as overcurrent protection action output) are acquired in an event-triggered manner, and when a status change is detected, a local time identifier in the same format as the primary side electrical parameter timing data is added to the corresponding secondary side status response data.

[0026] The acquisition is event-triggered, meaning it only starts when a change in input level is detected. When a state change event occurs on any channel, the system immediately interrupts the current task, reads the precise time of the event from the same unified local high-precision clock source, and generates a local time identifier in the same format as the primary-side sampling data. This identifier is then bound to the recorded channel number and the changed state value to form a secondary-side state response event record.

[0027] It can be seen that by mandating that primary-side sampling and secondary-side event acquisition share the same physical clock source and follow a unified time signature format, the homogeneity and comparability of the two heterogeneous data streams on the time scale are fundamentally guaranteed. Even if there is a clock deviation between the edge computing node and the upper-level system, the relative timing relationship between the primary and secondary data within the node is absolutely accurate.

[0028] S1.3: Using the local time identifier as the alignment reference, sort and merge the time series data of the primary side electrical parameters and the secondary side state response data in time to construct a primary-secondary fusion time series dataset containing the sequence of primary side electrical parameter feature values ​​(such as the effective value sequence) and the records of secondary side state changes.

[0029] For high-frequency primary sampling points, a feature value sequence can be generated through preprocessing (such as calculating the effective value per cycle) before insertion to reduce the amount of data and highlight the changing characteristics. This operation constructs a time-domain coherent unified timeline that includes continuous primary features and discrete secondary event points.

[0030] S1.4: In the primary and secondary fusion time-series dataset, the primary-side electrical parameter characteristic values ​​(such as the effective value of current) are scanned. For each primary-side electrical parameter change record characterized by a sudden change in current or voltage exceeding the corresponding change detection threshold (set according to the statistical fluctuation range of electrical parameter characteristic values ​​during historical normal operation, for example, taking the mean of historical data plus three times the standard deviation, which can be set according to actual conditions), the time position of the first subsequent secondary-side state change record is marked, and the corresponding time difference is written into the primary and secondary fusion time-series dataset. For example, after marking the starting point of the corresponding event as the primary-side electrical parameter change record, the first secondary-side state change record of type "protection trigger signal" is searched backward in the globally ordered sequence, starting from the time identifier of the change record. The difference between the time identifiers of these two records is the primary-secondary response time difference of this event.

[0031] S2: Based on the primary and secondary fusion time series dataset, extract the time difference sequence between the start time of the change of the primary side electrical parameters and the time of the secondary side state response in the edge computing node, and analyze the changing trend of the time difference sequence in the continuous operation cycle to obtain the time misalignment feature result reflecting the primary-secondary response relationship.

[0032] S2.1: Divide the time difference sequence into continuous operating cycles with a preset fixed duration according to the local time identifier.

[0033] Specifically, the process iterates through all labeled primary-side electrical parameter change records with time differences in the primary and secondary fusion time-series dataset, extracting the local time identifier and its corresponding time difference value for each record to form an original time difference event sequence. Then, based on the preset fixed-duration (e.g., 5 minutes) continuous running cycle boundary, all time difference events are divided into corresponding cycle buckets. Each running cycle outputs a subset containing all time difference data within that cycle.

[0034] S2.2: Based on a preset scrolling window length (e.g., covering the most recent 100 events), the time difference sequence is sequentially slid over a time period covered by at least two of the consecutive running cycles to capture multiple local subsequences.

[0035] By ensuring that the window span covers multiple operating cycles, the extracted local features are guaranteed to reflect cross-cycle, relatively long-term dynamic behavior, rather than random fluctuations within a single cycle.

[0036] S2.3: Calculate the arithmetic mean and standard deviation of the time difference for each local subsequence to form a time difference mean sequence and a time difference standard deviation sequence (the mean represents the typical level of the primary and secondary response time differences within the window period, while the standard deviation characterizes the degree of discrete fluctuation around the typical level).

[0037] As the window slides, two new derived sequences are obtained: the time difference mean sequence and the time difference standard deviation sequence. These two sequences quantify the changes in time difference behavior over time from the two dimensions of "central tendency" and "dispersion," respectively.

[0038] S2.4: Perform least squares linear fitting on the time difference average sequence to obtain the slope of the fitted line. This slope quantifies the average rate of change of the typical level of time difference over time (positive growth, negative growth, or basically unchanged). Calculate the arithmetic mean of all standard deviations in the time difference standard deviation sequence, denoted as the volatility baseline value, which characterizes the average intensity of time difference volatility over the entire analysis period.

[0039] Furthermore, if the absolute value of the slope is greater than a preset slope threshold, the time difference sequence is determined to have a monotonic trend; otherwise, it is determined that there is no significant monotonic trend.

[0040] If the fluctuation benchmark value is less than or equal to the preset fluctuation threshold, the fluctuation level of the time difference sequence is determined to be low; otherwise, the fluctuation level of the time difference sequence is determined to be high.

[0041] Among them, the time misalignment characteristic results include at least four types: positive trend drift with low volatility, negative trend drift with low volatility, unstable trend with high volatility, and no obvious trend with low volatility.

[0042] S3: Based on the change range and duration of the primary side electrical parameters in the current operating cycle, determine the theoretical operating state range of the pole-mounted switch in the corresponding cycle, and compare it with the actual state response of the secondary side to output the state consistency judgment result.

[0043] S3.1: The theoretical operating state range of the pole-mounted switch within the corresponding cycle includes: S3.1.1: Extract the instantaneous current sampling sequence and the instantaneous voltage sampling sequence within the current operating cycle from the primary side electrical parameter timing data.

[0044] S3.1.2: Based on the instantaneous current sampling sequence and the instantaneous voltage sampling sequence, the current power frequency RMS value sequence and the voltage power frequency RMS value sequence are calculated respectively.

[0045] Specifically, recursive average filtering or similar algorithms are used to calculate the abrupt change in the effective value of the current at power frequency within a short-time analysis window (such as 3 cycles).

[0046] S3.1.3: Based on the current power frequency effective value sequence, calculate the sudden change within a preset short-time analysis window (e.g., 3 power frequency cycles); compare the sudden change with the dynamic overcurrent threshold value; when the sudden change continuously exceeds the dynamic overcurrent threshold value to reach a preset minimum duration (simulating the inverse time limit or time limit characteristic of protection), record the corresponding start time as the overcurrent start time; when the sudden change falls back and is lower than the dynamic overcurrent threshold value, record the corresponding time point as the overcurrent end time.

[0047] The dynamic overcurrent threshold value is taken as the nth percentile of the effective current frequency value of this line under normal load in recent historical data (e.g., rolling statistics of the same period in the past 24 hours); where n is a preset constant, such as 95. This threshold value is not fixed, but is adaptively adjusted according to the daily load level of the line.

[0048] S3.1.4: To closely approximate actual protection principles (e.g., overcurrent protection needs to operate when there is no significant voltage abnormality), the system performs parallel analysis of the voltage power frequency RMS value sequence. The voltage power frequency RMS value sequence is compared with a preset voltage undervoltage judgment threshold (e.g., 0.2 times the rated voltage), and the period when the voltage is consistently below the preset voltage undervoltage judgment threshold is marked as the voltage undervoltage period.

[0049] S3.1.5: Perform a logical AND operation between the overcurrent start time and the overcurrent end time and the voltage loss time period, and take the overlapping part of the two on the time axis as the theoretical operating state interval.

[0050] It can be seen that the present invention, through the integrated logic judgment of voltage blocking, makes the "theoretical operating range" calculated on the edge side closer to the behavior logic of the actual protection device, laying a solid and accurate foundation for meaningful comparison with the actual operation on the secondary side, and solving the problem that the theoretical model is too simplified in the traditional method, resulting in low reliability of the comparison results.

[0051] S3.2: Output state consistency judgment result: S3.2.1: Extract interval data from the current frequency RMS value sequence and voltage frequency RMS value sequence corresponding to the theoretical operating state interval, respectively, with the overcurrent start time as the starting point and the overcurrent end time as the ending point, and detect all continuous segments in the current frequency RMS value sequence that exceed the dynamic overcurrent threshold value.

[0052] S3.2.2: Each detected continuous segment is abstracted into an independent primary-side electrical parameter change event, and the overcurrent initiation time and corresponding current surge are recorded to form a primary event sequence to be compared. Simultaneously, from the secondary-side state response data, all state change records with timestamps within a preset buffer period before and after the theoretical operating state interval are extracted, and protection trigger signal records are selected to form a secondary event sequence to be compared. The preset buffer period before and after the theoretical operating state interval can be set as a fixed proportion of the interval's duration (e.g., 50%).

[0053] S3.2.3: Perform time-related matching between each primary-side electrical parameter change event in the primary event sequence to be compared and the secondary event sequence to be compared: For each primary-side electrical parameter change event, if at least one protection trigger signal record can be matched within a dynamic tolerance time window centered on the overcurrent start time, the primary-side electrical parameter change event is determined to be triggered effectively; otherwise, it is determined to be a missing trigger.

[0054] The dynamic tolerance time window does not use a fixed time tolerance, but rather dynamically determines the tolerance window based on the generated time misalignment feature results. For example, if the feature result is "high volatility and unstable trend," a larger tolerance window is automatically used to cover the uncertainty; if it is "no obvious trend and low volatility," a smaller standard tolerance window is used. The specific window setting needs to be determined according to the actual situation, and this embodiment of the invention does not impose a unique limitation. This dynamic window is used to search and match for each single event.

[0055] S3.2.4: Based on the matching results of all primary-side electrical parameter change events, the state consistency judgment result is generated comprehensively: if all primary-side electrical parameter change events are triggered effectively, the output is "completely consistent"; if there are missing trigger events, the output is "partially inconsistent" or "severely deviated" based on the proportion or distribution characteristics of the missing trigger events. For example, if only some events are missing, the output is "partially inconsistent" based on the proportion of missing events or whether they are concentrated in a specific time period; if most or key events are missing, the output is "severely deviated". The setting of the proportion or specific time period can be set according to actual needs, and this embodiment is not unique.

[0056] It can be seen that by comparing event-based and sequence-based data, the interval comparison is transformed into a precise point-to-point verification, which can accurately pinpoint which specific fault feature failed to trigger a response, greatly improving the granularity of the analysis. By introducing a dynamic tolerance window linked to time misalignment features, the comparison process becomes adaptive, automatically relaxing the criteria when the temporal relationship itself is unstable, avoiding misjudgments caused by inherent system fluctuations, and enhancing the overall robustness of the method. Furthermore, by outputting tiered consistency judgment results, richer decision-making basis is provided for subsequent processing (for example, "partial inconsistency" may indicate the need for further observation, while "serious deviation" may directly trigger an alarm), achieving refined and operational diagnostic conclusions.

[0057] S4: When the state consistency judgment result indicates that there is a multi-cycle consistency deviation and no effective action of the secondary side is detected in the corresponding cycle, the inactive interval is reverse-verified in combination with the time misalignment feature result to complete the pole-mounted switch fault judgment and generate the edge side judgment conclusion.

[0058] S4.1: For each primary-side electrical parameter change event determined as having a missing trigger in the state consistency discrimination result, extract all secondary-side state change records within the corresponding time period from the primary and secondary fusion time-series dataset, and determine whether they include protection trigger signal records; simultaneously, based on the local logs of the edge computing node, check whether there are communication interruption records or clock synchronization anomaly alarm records within the same time period. These two searches are aligned based on a unified local time identifier.

[0059] Specifically, for each primary side electrical parameter change event marked as "trigger missing", a preset review buffer period is first extended forward and backward based on the overcurrent start time and overcurrent end time as the core, which is dynamically adjusted based on the time misalignment feature results, thereby forming a complete time interval to be searched, and then the above two search tasks are performed in parallel.

[0060] After the retrieval is completed, the electrical quantity evidence (protection trigger signal records) and system status evidence (communication / clock anomaly log records) are encapsulated and strongly associated with the primary-side electrical parameter change event ID that triggered the retrieval, forming a multi-dimensional evidence package for that specific event. The multi-dimensional evidence package contains metadata such as the original timestamps, event descriptions, and confidence levels (e.g., log record levels) of various types of evidence, providing complete and structured input for subsequent logical reasoning.

[0061] It should be noted that by constructing a multi-dimensional evidence package around a suspicious event, a single abnormal electrical quantity signal is examined within a broader system context. This mechanism enables the method to acquire key auxiliary information from the device's own operating status (such as "whether communication was interrupted at the time"), which is not available in traditional fault information systems. This provides direct data support for distinguishing between a genuine protection failure and a "false failure" caused by external factors (such as signal transmission failure or timestamp errors), greatly enhancing the comprehensiveness and depth of fault analysis conclusions.

[0062] S4.2: If a protection trigger signal record is found, it is determined that the primary side electrical parameter change event has actually been triggered and the status of the corresponding event is corrected from trigger missing to trigger valid; if no protection trigger signal record is found, but a communication interruption or clock synchronization abnormality record is found, it is determined that the trigger missing is caused by the secondary circuit or clock abnormality, and a corresponding auxiliary judgment conclusion is generated.

[0063] Specifically, for each event's multi-dimensional evidence package, the engine applies rules sequentially for judgment: First, check if there are any protection trigger signal records whose timestamps fall within the dynamic tolerance window. If so, trigger rule one, determine that "the action signal has been generated but was missed in the initial comparison due to time tolerance or window misalignment", and accordingly correct the comparison status of the event from "trigger missing" to "trigger valid", while recording the reason for the correction; If no protection trigger signal is found, check for communication interruption or clock synchronization anomaly logs. If found, trigger rule two, determining that "the action signal was not correctly recorded or aligned due to secondary circuit communication failure or time base inaccuracy," and generate a specific auxiliary judgment conclusion, such as "there was a communication interruption during the relevant time period, and the signal is suspected to be lost."

[0064] S4.3: Based on the review results of all missing trigger events, update the state consistency judgment result, and combine it with the auxiliary judgment conclusion to generate the edge-side judgment conclusion used to indicate the fault type or abnormal cause.

[0065] The final assessment includes: the updated state consistency judgment result (e.g., corrected from "severe deviation" to "partial inconsistency"), the generated set of auxiliary assessment conclusions, and the time misalignment feature results. Based on this information, the final edge-side assessment conclusion is generated according to the preset conclusion template.

[0066] This conclusion is not a single label, but a structured output that includes at least: the object of analysis (e.g., switch ID), the analysis period, the main conclusion (e.g., "there is a risk of failure to operate on the secondary side" or "the anomaly is caused by clock asynchrony"), a summary of key evidence (e.g., "no protection action was detected in 2 out of 3 overcurrent events, and there were no communication anomalies during the period"), and a confidence level. This conclusion is sent directly through the communication interface of the edge computing node or triggers a local alarm.

[0067] As can be seen, through rule-driven automated reasoning, this invention achieves intelligent attribution of complex anomaly scenarios, freeing maintenance personnel from manual correlation analysis of massive amounts of data. By generating structured and hierarchical edge-side judgment conclusions, the output not only points out "what the problem is" (e.g., failure to operate), but also attempts to explain "why it might happen" (e.g., clock malfunction), and provides a key chain of evidence, making the conclusions highly credible and practically applicable to guiding maintenance actions. This achieves a complete transformation from raw data to high-value decision-making information at the data source side, effectively supporting the rapid location and precise handling of distribution network faults.

[0068] The present invention also includes one or more processors and a memory.

[0069] The memory is used to store operable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations, including the flow of a primary and secondary fusion pole-mounted switch fault assessment method based on edge computing as described in the foregoing embodiments, particularly... Figure 1 The flowchart of the method is shown.

[0070] Other aspects disclosed in the embodiments of the present invention also propose a computer-readable medium for storing software including instructions executable by one or more computers, which, upon execution, cause the one or more computers to perform operations including the flow of a primary and secondary fusion pole-mounted switch fault assessment method based on edge computing as described in the foregoing embodiments, particularly... Figure 1 The flowchart of the method is shown.

[0071] It should be recognized that embodiments of the present invention may be implemented or carried out by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable storage medium.

[0072] The method can be implemented using standard programming techniques, including a non-transitory computer-readable storage medium configured with a computer program in the computer program, wherein the storage medium is configured such that the computer operates in a specific and predefined manner.

[0073] Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system; however, if required, the program can be implemented in assembly or machine language.

[0074] In any case, the language can be either compiled or interpreted.

[0075] Furthermore, for this purpose, the program can run on programmed application-specific integrated circuits.

[0076] The processes described herein (or variations and / or combinations thereof) can be executed under the control of one or more computer systems configured with executable instructions, and can be implemented by hardware or a combination thereof as code (e.g., executable instructions, one or more computer programs, or one or more applications) that commonly executes on one or more processors. The computer program includes a plurality of instructions executable by one or more processors.

[0077] Furthermore, the method can be implemented in any suitable computing platform, including but not limited to personal computers, minicomputers, mainframes, workstations, networked or distributed computing environments, standalone or integrated computer platforms, or in communication with charged particle tools or other imaging devices.

[0078] Various aspects of the present invention can be implemented in machine-readable code stored on a non-transitory storage medium or device, whether portable or integrated into a computing platform, such as a hard disk, optical read and / or write storage medium, RAM, ROM, etc., such that it can be read by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the processes described herein.

[0079] Furthermore, machine-readable code, or parts thereof, can be transmitted via wired or wireless networks.

[0080] When such media includes instructions or programs that combine with a microprocessor or other data processor to implement the steps described above, the invention described herein includes these and other different types of non-transitory computer-readable storage media.

[0081] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for fault assessment of pole-mounted switches based on edge computing and primary / secondary fusion, characterized in that: include: In the edge computing node corresponding to the pole-mounted switch, continuous sampling data of primary side electrical parameters and secondary side status data are collected synchronously, and local time identifiers of uniform format are attached to form a primary and secondary fused time series dataset. Based on the primary and secondary fusion time series dataset, the time difference sequence between the start time of the change of primary side electrical parameters and the time of the secondary side state response is extracted in the edge computing node, and the changing trend of the time difference sequence in the continuous operation cycle is analyzed to obtain the time misalignment feature result reflecting the primary-secondary response relationship. Based on the variation range and duration of the primary side electrical parameters in the current operating cycle, the theoretical operating state range of the pole-mounted switch in the corresponding cycle is determined, and a consistency comparison is made with the actual state response of the secondary side, and the state consistency judgment result is output. When the state consistency judgment result indicates that there is a multi-cycle consistency deviation and no effective action of the secondary side is detected in the corresponding cycle, the inactive interval is reverse-verified in combination with the time misalignment feature result to complete the pole-mounted switch fault judgment and generate the edge side judgment conclusion.

2. The method for fault assessment of a pole-mounted switch based on edge computing as described in claim 1, characterized in that: The formation of the primary and secondary fusion time series dataset includes: In the edge computing node corresponding to the pole-mounted switch, the primary side voltage and current are continuously sampled according to a fixed sampling period, and each primary side electrical parameter sampling value is bound to the corresponding local time identifier to form primary side electrical parameter time series data; In the same edge computing node, the secondary-side switch status and protection trigger signals are acquired in an event-triggered manner, and when a status change is detected, a local time identifier in the same format as the primary-side electrical parameter timing data is added to the corresponding secondary-side status response data. Using the local time identifier as the alignment reference, the time series data of the primary side electrical parameters and the state response data of the secondary side are sorted and merged by time to construct a primary-secondary fusion time series dataset containing the sequence of primary side electrical parameter feature values ​​and the record of secondary side state changes.

3. The method for fault assessment of a pole-mounted switch based on edge computing, as described in claim 2, is characterized in that: In the primary and secondary fusion time series dataset, for each primary-side electrical parameter change record characterized by a current or voltage surge exceeding the corresponding change detection threshold, the time position of the first subsequent secondary-side state change record is marked, and the corresponding time difference is written into the primary and secondary fusion time series dataset.

4. The method for fault assessment of a pole-mounted switch based on edge computing, as described in claim 3, is characterized in that: The analysis of the changing trend of the time difference sequence within a continuous operating cycle includes: The time difference sequence is divided into continuous operating cycles with a preset fixed duration based on the local time identifier. Based on a preset scrolling window length, the time difference sequence is sequentially slid over a time period covered by at least two consecutive running cycles to capture multiple local subsequences. Calculate the arithmetic mean and standard deviation of the time differences contained in each local subsequence to form a time difference mean sequence and a time difference standard deviation sequence; The average time difference sequence is linearly fitted using the least squares method to obtain the slope of the fitted line; the arithmetic mean of all standard deviations in the average time difference sequence is calculated and denoted as the fluctuation baseline value. If the absolute value of the slope is greater than a preset slope threshold, the time difference sequence is determined to have a monotonic trend; otherwise, it is determined that there is no significant monotonic trend. If the fluctuation benchmark value is less than or equal to the preset fluctuation threshold, the fluctuation level of the time difference sequence is determined to be low; otherwise, the fluctuation level of the time difference sequence is determined to be high.

5. The method for fault assessment of a pole-mounted switch based on edge computing as described in claim 4, characterized in that: The time misalignment characteristics include at least four types: positive trend drift with low volatility, negative trend drift with low volatility, unstable trend with high volatility, and no obvious trend with low volatility.

6. The method for fault assessment of a pole-mounted switch based on edge computing as described in claim 5, characterized in that: The theoretical operating state range of the on-board switch within the corresponding period includes: Extract the instantaneous current sampling sequence and the instantaneous voltage sampling sequence within the current operating cycle from the primary side electrical parameter timing data; Based on the instantaneous current sampling sequence and the instantaneous voltage sampling sequence, the current power frequency RMS value sequence and the voltage power frequency RMS value sequence are calculated respectively; Based on the current power frequency effective value sequence, calculate the sudden change within a preset short-time analysis window; compare the sudden change with the dynamic overcurrent threshold value; when the sudden change continuously exceeds the dynamic overcurrent threshold value for a preset minimum duration, record the corresponding start time as the overcurrent start time; when the sudden change falls back and is lower than the dynamic overcurrent threshold value, record the corresponding time point as the overcurrent end time. The dynamic overcurrent threshold value is taken as the nth percentile of the effective value of the current frequency of this line under normal load in the same historical period; where n is a constant. The voltage power frequency effective value sequence is compared with a preset voltage undervoltage judgment threshold, and the period when the voltage is continuously lower than the preset voltage undervoltage judgment threshold is marked as the voltage undervoltage period. The interval from the start of overcurrent to the end of overcurrent is logically ANDed with the voltage loss period, and the overlapping part of the two on the time axis is taken as the theoretical operating state interval.

7. The method for fault assessment of a pole-mounted switch based on edge computing, as described in claim 6, is characterized in that: The output of the state consistency determination result includes: From the current frequency RMS value sequence and voltage frequency RMS value sequence corresponding to the theoretical operating state interval, extract the interval data with the overcurrent start time as the starting time and the overcurrent end time as the ending time, and detect all continuous segments in the current frequency RMS value sequence that exceed the dynamic overcurrent threshold value. Each detected continuous segment is treated as an independent primary side electrical parameter change event, and the overcurrent initiation time and the corresponding current surge are recorded to form a primary event sequence to be compared. From the secondary side state response data, extract all state change records within a preset buffer period before and after the theoretical action state interval, and filter out protection trigger signal records to form a secondary event sequence to be compared. Each primary-side electrical parameter change event in the primary event sequence to be compared is matched with the secondary event sequence to be compared in terms of time correlation: for each primary-side electrical parameter change event, if at least one protection trigger signal record can be matched within a dynamic tolerance time window centered on the overcurrent start time, the primary-side electrical parameter change event is determined to be triggered effectively; otherwise, it is determined to be a missing trigger. If all primary side electrical parameter change events are triggered and valid, the output will be completely consistent; if there are missing trigger events, the output will be partially inconsistent or severely deviated depending on the proportion or distribution characteristics of the missing trigger events.

8. The method for fault assessment of a pole-mounted switch based on edge computing as described in claim 7, characterized in that: The reverse verification process includes: For each primary-side electrical parameter change event that is determined to be missing in the state consistency discrimination result, extract all secondary-side state change records within the corresponding time period from the primary and secondary fusion time series dataset, and determine whether they contain protection trigger signal records. At the same time, based on the local logs of the edge computing nodes, check whether there are any communication interruption records or clock synchronization abnormality alarm records within the same time period; If a protection trigger signal record is detected, it is determined that the primary side electrical parameter change event has actually been triggered and the corresponding event status is corrected from trigger missing to trigger valid. If no protection trigger signal record is found, but a communication interruption or clock synchronization abnormality record is found, it is determined that the missing trigger is caused by a secondary circuit or clock abnormality, and a corresponding auxiliary judgment conclusion is generated. Based on the review results of all missing events, the state consistency judgment result is updated, and combined with the auxiliary judgment conclusion, the edge-side judgment conclusion is generated to indicate the fault type or abnormal cause.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the edge computing-based primary and secondary fusion pole-mounted switch fault assessment method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the edge computing-based primary and secondary fusion pole-mounted switch fault assessment method according to any one of claims 1 to 8.

Images