Power supply fault diagnosis method and system for low-voltage power distribution network based on power failure and recovery logic

Abstract

The invention discloses a power failure and recovery logic-based power supply fault diagnosis method for a low-voltage power distribution network, which belongs to the technical field of power supply fault diagnosis and comprises the following steps of: converting a user unstructured power failure complaint into a standardized event tag by utilizing a natural language processing technology, synchronously aligning an ammeter power failure pulse and a branch switch action record, and performing power failure diagnosis; constructing a multi-dimensional space-time event sequence; a fault propagation logic network is established through power supply unit topology division, a fault diffusion path is deduced according to a power failure time sequence difference of adjacent units and a failure state of a protection device, and an initial fault point is reversely positioned. And meanwhile, the fault association probability between the units is dynamically corrected, and a weighted candidate fault chain is generated. And through multiple verification mechanisms, the fault propagation tree is reversely corrected. And finally, calculating the confidence coefficient, and screening out the fault chain with the optimal time-space consistency. The problems of multi-source information splitting, space-time correlation weakening and insufficient verification reliability in low-voltage power distribution network fault positioning are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply fault diagnosis, and in particular to a power supply fault diagnosis method and system for a low-voltage distribution network based on power outage and restoration logic. Background Art

[0002] The power outage and restoration logic for low-voltage distribution networks is a distribution system based on intelligent control technology. Its core logic uses real-time data collection of line status, load data, and fault signals, combined with automated terminals to rapidly locate faults, isolate faulted sections, and automatically restore power to non-faulty areas, thereby reducing the scope and duration of power outages. This system typically integrates remote communications, edge computing, and distribution automation technologies. It can intelligently switch power paths or coordinate distributed energy resources for supplemental power supply in the event of a short circuit, overload, or planned maintenance, significantly improving power supply reliability and user experience.

[0003] Traditional diagnostic methods based on power outage and restoration logic mainly rely on the timing matching of the power-off pulse signal of the smart meter and the branch switch action record, and determine the fault propagation path through rule reasoning. However, with the expansion of distribution network scale, the access of distributed power sources and the complexity of user-side equipment, existing technologies face significant bottlenecks: First, the fusion of multi-source heterogeneous data is insufficient, and the unstructured descriptions in user fault report texts (such as key information such as "voltage drop" and "smoke and abnormal noise") lack standardized processing. Interference factors such as meteorological environment are not included in the analysis framework, resulting in one-sided fault feature extraction; Second, the spatiotemporal correlation modeling is extensive. Traditional methods use the meter as the smallest unit for event alignment, but do not combine the power supply unit topology (such as the independent power supply switch division of the end user group). It is difficult to accurately capture the propagation direction and hierarchical relationship of the fault in the physical network, and it is easy to cause misjudgment due to cross-interference of the switch action timing; Third, there is a lack of diagnostic closed-loop verification. Existing technologies rely on one-way reasoning to generate fault hypotheses, and lack a dynamic correction mechanism for feedback data such as the protection signal restoration status and simulated power-off and decoupling experiments. The diagnostic results may deviate from the actual fault scenario due to the accumulation of timing deviations or false operation / failure of protection devices. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the purpose of this application is to provide a power supply fault diagnosis method for a low-voltage distribution network based on power outage and power restoration logic, realize the multi-source feature fusion of user fault reports, equipment signals and environmental data through semantic recognition and millisecond-level timeline alignment, construct a spatiotemporal propagation tree model based on the power supply unit topology to accurately characterize the fault path, dynamically optimize the diagnosis weight in combination with equipment health and environmental interference factors, and form a closed-loop verification mechanism by simulating power outage and power restoration instruction response delay detection, thereby solving the technical problems of multi-source information fragmentation, weakening of spatiotemporal correlation and insufficient verification reliability in low-voltage distribution network fault location.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a power supply fault diagnosis method for a low-voltage distribution network based on power outage and restoration logic, which comprises the following steps:

[0007] The system collects user-initiated power outage reporting signals, smart meter power outage pulse signals, branch switch operation records, ambient temperature and humidity, and weather warning information from the low-voltage distribution network. Keywords in user fault report texts are extracted through semantic recognition and marked as standardized event tags. The meter power outage pulse signals and branch switch operation records are aligned at the millisecond timestamp to generate a power outage and restoration event timeline covering the entire substation.

[0008] In a low-voltage distribution network, two or more power supply units are divided based on whether the end-user group has an independent power supply switch; based on the power outage and restoration event timeline, the spatiotemporal propagation direction of the power outage and restoration events of the power supply units is identified; based on the time sequence of power outage events and the lack of protection actions between adjacent power supply units, a fault propagation tree model is constructed with the power supply units as nodes and logical association rules as edges; the event timeline is traversed in reverse, and the power supply unit that first triggered the power outage signal is traced back as the suspected fault starting point;

[0009] Based on the fault propagation tree model, the equipment health factor, environmental interference factor and user fault reporting spatial density data are integrated to dynamically adjust the diagnostic weights between power supply units and generate candidate fault chains arranged in descending order of weight;

[0010] Check the unreset protection signal of the starting device of the fault chain, simulate the disconnection of the upstream switch to verify the consistency of the downstream power outage state, trigger the power restoration instruction to observe the response delay, and reversely inject the verification result into the fault propagation tree model; correct the time sequence of the power outage event based on the difference between the downstream power outage range and the simulated disconnection result; update the judgment condition of the protection action missing state based on the protection signal restoration state; adjust the deviation between the power restoration instruction triggering time and the actual power restoration time of the power supply unit to reduce it to a preset deviation value based on the response delay;

[0011] Based on the revised fault propagation tree model and verification results, the top three fault chains with the smallest power outage range differences among the candidate fault chains are extracted. Combined with the equipment health factor, environmental interference factor, and user fault reporting spatial density data, the comprehensive confidence of each fault chain is calculated. The fault chain with the highest comprehensive confidence and the power restoration response delay and timeline deviation less than the preset threshold is output as the final diagnosis result.

[0012] As a preferred solution of the power supply fault diagnosis method of the low-voltage distribution network based on power outage and restoration logic of the present invention, wherein: the generating of the candidate fault chain arranged in descending order of weight includes:

[0013] Based on the fault propagation tree model, the fault frequency, maintenance record, and aging degree of each power supply unit's equipment historical operation data are used to generate an equipment health score. For power supply units where equipment with a health score below a preset threshold is located, the initial diagnostic weight is increased by a preset ratio to obtain an adjusted diagnostic weight for the equipment health.

[0014] Real-time collection of environmental temperature and humidity data and matching with extreme weather types and geographical coverage in meteorological warnings to determine whether the area where the power supply unit is located is a high humidity area, a heavy rainfall warning area, or a thunderstorm warning area; when the area where the power supply unit is located is in a high humidity area, a heavy rainfall warning area, or a thunderstorm warning area, the power supply unit is assigned a corresponding environmental interference coefficient, and the environmental interference coefficient is multiplied by the current diagnostic weight of the power supply unit to obtain a diagnostic weight after superimposing the environmental interference;

[0015] Count the geographical distribution density of user fault reporting signals. For power supply units where fault reporting points are concentrated and overlap with the timeline of the power outage event, assign a spatial density gain coefficient based on the extent to which the density exceeds the preset threshold. Multiply the current gain coefficient by the current diagnostic weight of the power supply unit to obtain the adjusted diagnostic weight weighted by spatial density.

[0016] According to the adjusted diagnostic weight of comprehensive equipment health, environmental interference superposition weight and spatial density weight, the final diagnostic weights between power supply units in the fault propagation tree model are corrected step by step through weighted fusion, and candidate fault chains are generated, which are sorted from high to low according to the corrected weight values.

[0017] As a preferred solution of the power supply fault diagnosis method of the low-voltage distribution network based on power outage and restoration logic described in the present invention, wherein: the checking of the unrestored protection signal of the starting device of the fault chain, the simulation of disconnecting the upstream switch to verify the consistency of the downstream power outage state, and the triggering of the power restoration instruction to observe the response delay include:

[0018] Obtain the real-time telesignal signal of the candidate fault chain starting device, and confirm whether the fault chain starting device has an unreset protection signal based on the last action record;

[0019] Use the remote control function of the distribution automation system to temporarily disconnect the upstream switch in the candidate fault chain and monitor the voltage signals of all downstream smart meters in real time to verify whether the downstream power outage status is consistent with the topological path of the candidate fault chain;

[0020] A remote power restoration instruction is sent to the power restoration node of the candidate fault chain through the master station system, and the time difference between the instruction issuance time and the meter voltage recovery time is recorded as the response delay data. The response delay data is compared with the preset communication delay threshold, and the data of the response delay exceeding the limit is recorded.

[0021] As a preferred solution of the power supply fault diagnosis method of the low-voltage distribution network based on the power outage and restoration logic of the present invention, wherein: the correction of the time sequence of the power outage event according to the difference between the downstream power outage range and the simulated disconnection result includes calculating the time sequence offset of the power outage event between the power supply units according to the deviation between the downstream power outage range and the simulated disconnection result;

[0022] The timestamps of the power outage events of adjacent power supply units in the power outage and restoration event timeline are reordered based on the timing offset to generate an updated time sequence of the power outage and restoration events.

[0023] As a preferred embodiment of the power supply fault diagnosis method for a low-voltage distribution network based on power outage and restoration logic of the present invention, the updating of the determination condition of the protection action missing state according to the protection signal restoration state includes: when the protection signal restoration state is not restored, determining that the protection action missing state of the corresponding power supply unit is missing, and updating the determination condition of the protection action missing state to use the non-restoration of the protection signal as a necessary condition for triggering the protection action missing;

[0024] When the protection signal restoration state is restored, the time window length in the judgment condition of the protection action missing state is adjusted according to whether the power restoration response delay and the time line deviation are less than the preset threshold value until the time line deviation meets the preset requirement.

[0025] As a preferred solution of the power supply fault diagnosis method of the low-voltage distribution network based on the power outage and power restoration logic described in the present invention, wherein: the deviation of adjusting the triggering time of the power restoration instruction and the actual power restoration time of the power supply unit to be reduced to a preset deviation value includes, based on the response delay between the triggering time of the power restoration instruction and the actual power restoration time of the power supply unit, taking the preset deviation value as the adjustment target, by iteratively adjusting the timestamp deviation between the triggering time of the power restoration instruction and the actual power restoration time of the power supply unit, so that the deviation gradually converges to within the preset range of the preset deviation value.

[0026] As a preferred solution of the power supply fault diagnosis method of the low-voltage distribution network based on power outage and restoration logic described in the present invention, wherein: the calculation of the comprehensive confidence of each fault chain includes:

[0027]

[0028] Among them, C is the comprehensive confidence, H is the equipment health factor, E is the environmental interference factor, D is the spatial density data of user fault reporting, ΔS is the power outage range difference, Δt is the power restoration response delay deviation, λ is the delay deviation adjustment coefficient, and e is the power outage response delay deviation. -λΔt is the delay decay term.

[0029] Another object of the present invention is to provide a power supply fault diagnosis system for a low-voltage distribution network based on power outage and restoration logic.

[0030] To solve the above technical problems, the present invention provides the following technical solutions: a power supply fault diagnosis system for a low-voltage distribution network based on power outage and restoration logic, comprising: a data acquisition module, a model building module, a candidate fault chain generation module, a detection module, and a diagnosis result module;

[0031] The data acquisition module is used to collect user-initiated power outage reporting signals, smart meter power outage pulse signals, branch switch action records, ambient temperature and humidity, and weather warning information in the low-voltage distribution network. Keywords in user fault reporting texts are extracted through semantic recognition and marked as standardized event tags. The meter power outage pulse signals and branch switch action records are aligned according to millisecond-level timestamps to generate a power outage and restoration event timeline covering the entire substation.

[0032] The model construction module is used to divide a low-voltage distribution network into multiple power supply units based on whether the end-user group has an independent power supply switch; identify the spatiotemporal propagation direction of the power outage and restoration events of each power supply unit based on the power outage and restoration event timeline; construct a fault propagation tree model with the power supply units as nodes and logical association rules as edges based on the time sequence of power outage events and the lack of protection actions between adjacent power supply units; traverse the event timeline in reverse, and trace back to the power supply unit that first triggered the power outage signal as the fault starting point;

[0033] The candidate fault chain generation module, based on the fault propagation tree model, integrates the equipment health factor, environmental interference factor and user fault reporting spatial density data, dynamically adjusts the diagnostic weights between power supply units, and generates candidate fault chains arranged in descending order of weight;

[0034] The detection module is used to check the unreset protection signal of the candidate fault chain starting point device, simulate the disconnection of the upstream switch to verify the consistency of the downstream power outage state, trigger the power restoration instruction to observe the response delay, and reversely inject the verification result into the fault propagation tree model; correct the time sequence according to the difference between the downstream power outage range and the simulated disconnection result; update the judgment condition of the protection action missing state according to the protection signal restoration state; and adjust the deviation between the power restoration instruction triggering time and the actual power restoration time of the power supply unit to reduce it to a preset deviation value according to the response delay;

[0035] The diagnosis result module, based on the revised fault propagation tree model and verification results, extracts the top three fault chains with the smallest power outage range differences among the candidate fault chains, calculates the comprehensive confidence of each fault chain based on the equipment health factor, environmental interference factor and user fault reporting spatial density data, and outputs the fault chain with the highest comprehensive confidence and the power restoration response delay and timeline deviation less than a preset threshold as the final diagnosis result.

[0036] The present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the power supply fault diagnosis method of a low-voltage distribution network based on power outage and restoration logic when executing the computer program.

[0037] The present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the power supply fault diagnosis method of the low-voltage distribution network based on power outage and restoration logic are implemented.

[0038] The beneficial effects of the present invention are as follows: The present invention utilizes natural language processing technology to convert users' unstructured power outage complaints into standardized event labels, synchronously aligns meter power-off pulses with branch switch action records, and constructs a multi-dimensional spatiotemporal event sequence; establishes a fault propagation logic network through power supply unit topology division, derives the fault diffusion path based on the power outage timing difference of adjacent units and the failure status of the protection device, and reversely locates the initial fault point. At the same time, the device health, environmental interference factors, and user fault reporting spatial clustering characteristics are introduced to dynamically correct the fault association probability between units and generate a weighted candidate fault chain. Through a triple verification mechanism of protection signal status verification, switch simulation operation, and power restoration instruction response delay analysis, the temporal sequence relationship of the fault propagation tree and the protection action judgment conditions are reversely corrected, and closed-loop feedback is used to optimize the triggering timing of the power restoration instruction. Finally, by calculating the confidence level, the power outage range matching, device status, environmental interference, and user distribution density are comprehensively considered to screen out the fault chain with the best spatiotemporal consistency. This solves the technical problems of multi-source information fragmentation, weakened spatiotemporal correlation, and insufficient verification reliability in low-voltage distribution network fault location. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0040] Figure 1 The present invention provides an overall flow chart of a method for diagnosing power supply faults in a low-voltage distribution network based on power outage and restoration logic according to an embodiment of the present invention.

[0041] Figure 2 A schematic structural diagram of a power supply fault diagnosis system for a low-voltage distribution network based on power outage and restoration logic provided by one embodiment of the present invention.

[0042] In the figure, 201 is the data acquisition module model; 202 is the construction module; 203 is the candidate fault chain generation module; 204 is the detection module; and 205 is the diagnosis result module. DETAILED DESCRIPTION

[0043] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0044] Example 1, with reference to Figure 1 , which is the first embodiment of the present invention, provides a power supply fault diagnosis method for a low-voltage distribution network based on power outage and restoration logic, comprising:

[0045] S101 collects user-initiated power outage reporting signals, smart meter power outage pulse signals, branch switch action records, ambient temperature and humidity, and weather warning information in the low-voltage distribution network. Through semantic recognition, it extracts keywords from user fault reporting texts and marks them as standardized event tags. It aligns the meter power outage pulse signals with the branch switch action records according to millisecond timestamps to generate a power outage and restoration event timeline covering the entire substation.

[0046] Specifically, the power outage event signals actively reported by the user side, the power outage pulse signals generated by the smart meter, the action records of the branch switch protection device, as well as the environmental temperature and humidity sensor data and external meteorological warning information are first synchronously collected.

[0047] For example, when a user in a community reports a text description of "sudden power outage in the kitchen circuit" through a mobile terminal, the system uses natural language processing technology to extract keywords such as "sudden power outage" and "circuit abnormality" and converts them into a standardized event label of "instantaneous power outage - user side", thereby eliminating the semantic ambiguity caused by the colloquial description.

[0048] At the same time, the smart meters in the substation record voltage drop events with millisecond accuracy. For example, if the meter records that the voltage value returns to zero at 2023-05-10T14:23:05.832, the system will associate the action record of the branch switch at the same time. If it is found that the switch performs an overcurrent tripping action at 14:23:05.828, the causal relationship between the two is confirmed through timestamp alignment.

[0049] During this process, ambient temperature and humidity data (such as the 95% relative humidity reported by a hygrometer) is matched with the coordinates of the yellow thunderstorm warning area in the meteorological warning. If the station's geographic coordinates fall within the warning coverage area, an auxiliary tag "extreme weather impact" is generated. Ultimately, through the spatiotemporal integration of multi-dimensional data, a corresponding power outage and restoration event timeline is formed. The horizontal axis represents the time scale accurate to milliseconds, and the vertical axis displays events such as user fault reports, meter power outages, switch actions, and environmental status in a layered manner. For example, a user's power outage text message was recorded at 14:23:05.800, branch switch S07 was activated at 14:23:05.828, and the meter power outage pulse was recorded at 14:23:05.832. This constructs the event propagation path from switch action to user power outage.

[0050] S102, in the low-voltage distribution network, two or more power supply units are divided according to whether the end-user group has an independent power supply switch; based on the power outage and restoration event timeline, the spatiotemporal propagation direction of the power outage and restoration event of the power supply unit is identified; according to the time sequence of the power outage events between adjacent power supply units and the missing status of the protection action, a fault propagation tree model is constructed with the power supply unit as the node and the logical association rules as the edge; the event timeline is traversed in reverse, and the power supply unit that first triggered the power outage signal is traced back as the suspected fault starting point.

[0051] Specifically, the system first divides power supply units based on whether end-user groups are equipped with independent power supply switches. For example, in a residential community, where each building supplies power to internal users via independent branch switches, each building and its associated distribution facilities constitute a power supply unit. After division, the system analyzes the spatiotemporal propagation direction of power outage and restoration events in each power supply unit based on the power outage and restoration event timeline. If the power outage event timestamp for power supply unit A is earlier than the power outage event for adjacent power supply unit B, and there is a millisecond-level gap between the two on the timeline, the power outage is determined to have propagated from A to B.

[0052] For example, when the branch switch operates at 14:23:05.828, causing a power outage in Building 3, which it supplies, and then at 14:23:05.832, a voltage drop triggers a power-off pulse in the meter of Building 5 in the same substation, but the branch switch belonging to Building 5 does not operate, the system recognizes that the power outage event has spread from the power supply unit of Building 3 to the power supply unit of Building 5.

[0053] Based on this spatiotemporal relationship, a fault propagation tree model is constructed with power supply units as nodes and logical association rules as edges, according to the time sequence of power outages and the absence of protective action between adjacent power supply units. The logical association rules between adjacent power supply units are determined by the time sequence of power outages and the absence of protective action: if, after power supply unit X experiences a power outage, power supply unit Y experiences a power outage within a preset time window but its protective device does not operate, a fault propagation edge is established from X to Y.

[0054] For example, if power supply unit P (including switch K1) in an industrial park loses power at 10:15:20.500, and its downstream power supply unit Q (including switch K2) loses power at 10:15:20.520, and there is no operation record for K2, then it is determined that the fault may propagate along the P→Q path, and a corresponding logical edge is added to the model.

[0055] When traversing the event timeline in reverse, the system traces back the first power supply unit that triggered the power outage signal as the suspected fault starting point. For example, when the power outage pulse timestamp (09:30:05.100) of power supply unit C in the power outage event timeline of a certain substation is earlier than that of all other units, C is preferentially marked as the fault origin candidate.

[0056] S103 , based on the fault propagation tree model, integrating the equipment health factor, environmental interference factor and user fault reporting spatial density data, dynamically adjusting the diagnosis weights between power supply units, and generating candidate fault chains arranged in descending order of weight.

[0057] Specifically, the failure frequency, maintenance record and aging degree in the historical operation data of the equipment of each power supply unit are used to generate an equipment health score. For example, the score is generated by weighted summation. For the power supply unit where the equipment has a health score lower than the preset threshold, its initial diagnostic weight is increased by a preset ratio to obtain the diagnostic weight after the equipment health is adjusted.

[0058] Furthermore, real-time environmental temperature and humidity data are collected and matched with the extreme weather types and their geographical coverage in the meteorological warning to determine whether the area where each power supply unit is located belongs to a high humidity area, a heavy rainfall warning area, or a thunderstorm warning area;

[0059] When the power supply unit is located in a high humidity area, a heavy rainfall warning area, or a thunderstorm warning area, the power supply unit is assigned a corresponding environmental interference coefficient, and the environmental interference coefficient is multiplied by the current diagnostic weight of the power supply unit to obtain the diagnostic weight after superimposing the environmental interference;

[0060] Specifically, the environmental interference factor is dynamically corrected in weight through real-time meteorological data matching: the system compares the real-time data collected by the environmental temperature and humidity sensors (such as the hygrometer in the area where the power supply unit Q in a residential area continuously reports 98% relative humidity) with the extreme weather types in the meteorological warning (such as the red warning for thunderstorms covering the area of ​​118.5° to 119.2° east longitude and 32.1° to 32.5° north latitude) for geographic coordinates. If the location of the power supply unit Q (118.8° east longitude, 32.3° north latitude) falls within the warning coverage polygon, the system assigns it an environmental interference coefficient of 1.2 according to the preset mapping table.

[0061] For example, the power supply unit R in a coastal substation is located in the center of a typhoon path. The system detects that the wind speed in its area exceeds level 17 and a reinforced rainfall warning is issued, triggering an environmental interference coefficient of 1.5. This increases the diagnostic weight of the unit from 1.0 to 1.5, strengthening the representation of the inducing effect of extreme weather on equipment failures in the model.

[0062] Furthermore, the geographical distribution density of user fault reporting signals is counted, and for power supply units where the reporting points are concentrated and overlap with the timeline of the power outage event, a spatial density gain coefficient is assigned according to the amplitude by which the density exceeds the preset threshold. The gain coefficient is multiplied by the current diagnostic weight of the power supply unit to obtain the adjusted diagnostic weight weighted by the spatial density.

[0063] Specifically, the spatial density data of user fault reports are assigned a gain coefficient through geographic cluster analysis: the system counts the geographic coordinate distribution of user fault reporting signals and uses the kernel density estimation algorithm to identify spatial clustering hotspots.

[0064] For example, power supply unit S in an office building complex received 32 "voltage sag" fault text messages from users on the same floor between 10:05 and 10:15. Their geographic coordinates formed a high-density cluster with a diameter of 50 meters on the electronic map. The system calculated that the fault density in this area reached 8.5 times per square kilometer per minute, exceeding 70% of the preset threshold of 5 times per square kilometer per minute. Therefore, the system assigned a spatial density gain factor of 1.35.

[0065] For example, after a power outage occurred in power supply unit T in an urban village due to aging lines, the system detected 23 user fault reports within 15 minutes within a radius of 100 meters centered on the fault point. The spatial density gain coefficient increased to 1.4, causing the power supply unit to move up two places in the candidate fault chain.

[0066] It should also be noted that according to the adjusted diagnostic weight of the comprehensive equipment health, the environmental interference superposition weight and the spatial density weight, the final diagnostic weight between the power supply units in the fault propagation tree model is corrected step by step through weighted fusion to generate candidate fault chains sorted from high to low according to the corrected weight value.

[0067] Specifically, the system uses weighted fusion to achieve step-by-step correction of diagnostic weights. Taking power supply unit a in a certain industrial park as an example, its equipment health adjustment weight is 1.2. After weighted fusion with the thunderstorm weather environment interference coefficient of 1.3, the weight becomes 1.56. It is further corrected to 1.95 due to the user fault reporting density gain coefficient of 1.25. The adjacent power supply unit b only maintains a basic weight of 0.9 due to good equipment health (weight 0.9), no environmental interference (coefficient 1.0), and alarm dispersion (coefficient 1.0). By comparing the weighted results of all power supply units, the system generates a candidate fault chain arranged in descending order of 1.95>1.56>0.9, accurately locking in power supply unit a as the most likely source of the fault.

[0068] S104, checking the unreset protection signal of the starting device of the fault chain, simulating the disconnection of the upstream switch to verify the consistency of the downstream power outage state, triggering the power-on instruction to observe the response delay, and reversely injecting the verification result into the fault propagation tree model; correcting the time sequence of the power outage event according to the difference between the downstream power outage range and the simulated disconnection result; updating the judgment condition of the protection action missing state according to the protection signal restoration state; adjusting the deviation between the power-on instruction triggering time and the actual power-on time of the power supply unit to reduce it to a preset deviation value according to the response delay.

[0069] Specifically, first obtain the real-time telesignaling signal of the starting point device of the candidate fault chain, and check the protection signal restoration status after its last action record.

[0070] For example, if the system detects that circuit breaker K3 in an industrial park continues to send a telesignal indicating "overcurrent not reset" after recording an actuation, it determines that the device has an unreset protection signal, indicating that its protection device has failed to automatically return to normal operation. The system then marks the device as abnormally locked. This inspection process effectively identifies residual signals from false trips caused by mechanical jamming of the protection device or secondary circuit faults, preventing temporary faults from being misidentified as permanent.

[0071] Then, the remote control function of the distribution automation system is used to temporarily disconnect the upstream switch in the candidate fault chain, and the voltage status of the downstream smart meter is monitored simultaneously.

[0072] For example, when verifying the fault propagation path from power supply unit X to Y in a commercial area, the system remotely disconnects upstream switch Q12 in unit X. If the smart meters in downstream unit Y all register zero voltage within 300 milliseconds, the topological path is confirmed to fully match the actual outage range. Conversely, if 50% of the meters in unit Y remain energized, the topological association of the candidate fault chain is incorrect. This verification method actively intervenes in the grid's operating status to verify the physical connectivity of the fault propagation path within milliseconds, significantly improving the physical authenticity of fault location.

[0073] At the same time, the system sends a remote power restoration instruction to the power restoration node of the candidate fault chain, and accurately records the time difference between the time the instruction is issued and the time when the meter voltage is restored.

[0074] For example, the timestamp for receiving a power-restore command from power supply unit Z in a residential complex is 15:20:05:500, while the voltage recovery times of the associated meters are distributed between 15:20:06:200 and 15:20:07:800. The system calculates an average response delay of 1.3 seconds, exceeding the preset communication delay threshold of 0.5 seconds by 160%. This data indicates severe congestion in the distribution network communication channel or sluggish actuator operation. The system marks this abnormal delay data as a critical correction parameter. By comparing the response delay data of multiple candidate chains, it can effectively identify fault misassociations caused by communication delays.

[0075] Based on these verification results, the system performs multi-dimensional corrections. First, based on the deviation in the downstream outage range caused by the simulated disconnection of upstream switches, the system calculates the timing offset of each power supply unit's outage event. For example, if the simulated disconnection operation results in two fewer downstream units experiencing outages than predicted by the candidate chain, the system recalculates the offsets of the adjacent unit's outage timestamps, reversing the order of the timestamps 14:05:30.500 and 14:05:30.520 in the original timeline. This physically verified timing correction eliminates spurious causal relationships caused by asynchronous signal acquisition and ensures that the direction of event propagation strictly corresponds to the grid topology.

[0076] Secondly, the judgment criteria are updated based on the protection signal restoration status. For protection devices that have not been restored for a long time, the system marks them as requiring protection action to be missed. For example, if a substation outgoing switch remains in the "open but not restored" state for 15 minutes after the fault is cleared, the system automatically updates the judgment criteria for protection action missed for the power supply unit where the switch is located to "must detect a missing restoration signal," thereby enhancing the sensitivity of identifying device faults. For protection devices that have been restored, the time window is dynamically adjusted based on the matching degree of the power restoration response delay and the timeline deviation. For example, if the power restoration delay deviation during a verification is 180ms, the system shortens the time window for determining protection action missed from the default 200ms to 150ms, making the timing correlation conditions more stringent.

[0077] Finally, an iterative adjustment mechanism was used to optimize the synchronization accuracy of the power restoration instruction triggering moment: With a preset deviation value of 50ms as the target, the system recorded response delays of 320ms, 210ms, and 90ms in three consecutive power restoration operations, respectively. Each time, the triggering moment was automatically advanced by 80ms, 60ms, and 40ms, ultimately converging to a 45ms deviation for the fourth operation. This closed-loop feedback regulation effectively compensates for the inherent communication delays and equipment inertia of the distribution automation system, ensuring that the timing correspondence between remote control instructions and actual state changes meets millisecond-level requirements. After full-process verification and correction, the timeline deviation of the fault propagation tree model can be reduced to less than 20% of the original data, significantly improving the credibility of the fault diagnosis results.

[0078] S105: Based on the revised fault propagation tree model and verification results, the top three fault chains with the smallest power outage range differences among the candidate fault chains are extracted. Combined with the equipment health factor, environmental interference factor, and user fault reporting spatial density data, the comprehensive confidence of each fault chain is calculated. The fault chain with the highest comprehensive confidence and the power restoration response delay and timeline deviation less than the preset threshold is output as the final diagnosis result.

[0079] Specifically, the system first extracts the top three candidate fault chains with the smallest difference in power outage range from the modified fault propagation tree model. The difference value is calculated by comparing the absolute difference between the number of power outage units predicted by the candidate chain and the actual number of power outage units.

[0080] For example, when the actual power outage involves power supply units A, B, and C, if candidate chain 1 predicts a power outage from A→B→C, candidate chain 2 predicts a power outage from A→D→C, and candidate chain 3 predicts a power outage from B→A, the difference values ​​0, 2, and 1 are calculated respectively, and candidate chain 1 with a difference value of 0 is preferentially selected as the benchmark analysis object.

[0081] Then, based on the device health factor, environmental interference factor, and user fault reporting density data, the comprehensive confidence of each fault chain is calculated. The calculation formula is as follows:

[0082]

[0083] Among them, C is the comprehensive confidence, H is the equipment health factor, E is the environmental interference factor, D is the spatial density data of user fault reporting, ΔS is the power outage range difference, Δt is the power restoration response delay deviation, λ is the delay deviation adjustment coefficient, and e is the power outage response delay deviation. -λΔt is the delay decay term.

[0084] Finally, the fault chain with the highest comprehensive confidence and the power-on response delay and timeline deviation less than the preset threshold is output as the final diagnosis result.

[0085] Example 2 is the second embodiment of the present invention, which is different from the previous embodiment in that:

[0086] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0087] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0088] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0089] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art and a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0090] Example 3 is the third embodiment of the present invention. This embodiment provides a power supply fault diagnosis system for a low-voltage distribution network based on power outage and restoration logic, including: a data acquisition module 201, a model construction module 202, a candidate fault chain generation module 203, a detection module 204, and a diagnosis result module 205;

[0091] Data acquisition module 201 is used to collect user-initiated power outage reporting signals, smart meter power outage pulse signals, branch switch operation records, ambient temperature and humidity, and weather warning information in the low-voltage distribution network. It extracts keywords from user fault report texts through semantic recognition and marks them as standardized event tags. It aligns the meter power outage pulse signals with the branch switch operation records using millisecond-level timestamps to generate a power outage and restoration event timeline covering the entire distribution area.

[0092] Model construction module 202 is used to divide a low-voltage distribution network into multiple power supply units based on whether the end-user group has an independent power supply switch; identify the spatiotemporal propagation direction of the power outage and restoration events of each power supply unit based on the power outage and restoration event timeline; construct a fault propagation tree model with the power supply units as nodes and logical association rules as edges based on the time sequence of power outage events and the lack of protection actions between adjacent power supply units; traverse the event timeline in reverse, and trace back to the power supply unit that first triggered the power outage signal as the fault starting point;

[0093] The candidate fault chain generation module 203, based on the fault propagation tree model, integrates the equipment health factor, environmental interference factor and user fault reporting spatial density data, dynamically adjusts the diagnosis weights between power supply units, and generates candidate fault chains arranged in descending order of weight;

[0094] Detection module 204 is configured to check the unreset protection signal of the candidate fault chain starting point device, simulate disconnection of the upstream switch to verify the consistency of the downstream power outage state, trigger the power restoration instruction to observe the response delay, and reversely inject the verification results into the fault propagation tree model; correct the time sequence based on the difference between the downstream power outage range and the simulated disconnection result; update the judgment condition for the protection action missing state based on the protection signal reset state; and adjust the deviation between the power restoration instruction triggering time and the actual power restoration time of the power supply unit to reduce it to a preset deviation value based on the response delay;

[0095] The diagnosis result module 205 extracts the top three fault chains with the smallest difference in power outage range from the candidate fault chains based on the revised fault propagation tree model and verification results. It calculates the comprehensive confidence of each fault chain by combining the equipment health factor, environmental interference factor and user fault reporting spatial density data, and outputs the fault chain with the highest comprehensive confidence and the power restoration response delay and timeline deviation less than the preset threshold as the final diagnosis result.

[0096] 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 the present invention. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A power supply fault diagnosis method for a low-voltage distribution network based on power outage and restoration logic, characterized by: include, The system collects user-initiated power outage reporting signals, smart meter power outage pulse signals, branch switch operation records, ambient temperature and humidity, and weather warning information from the low-voltage distribution network. Keywords in user fault report texts are extracted through semantic recognition and marked as standardized event tags. The meter power outage pulse signals and branch switch operation records are aligned at the millisecond timestamp to generate a power outage and restoration event timeline covering the entire substation. In the low-voltage distribution network, two or more power supply units are divided according to whether the end-user group has an independent power supply switch; Based on the power outage and restoration event timeline, the spatiotemporal propagation direction of the power outage and restoration events of the power supply units is identified; based on the time sequence of power outage events between adjacent power supply units and the missing status of protection actions, a fault propagation tree model is constructed with the power supply units as nodes and logical association rules as edges; the event timeline is traversed in reverse, and the power supply unit that first triggered the power outage signal is traced back as the suspected fault starting point; Based on the fault propagation tree model, the equipment health factor, environmental interference factor and user fault reporting spatial density data are integrated to dynamically adjust the diagnostic weights between power supply units and generate candidate fault chains arranged in descending order of weight; Checking the unreset protection signal of the starting device of the fault chain, simulating the disconnection of the upstream switch to verify the consistency of the downstream power outage state, triggering the power restoration command to observe the response delay, and reversely injecting the verification results into the fault propagation tree model; correcting the time sequence of the power outage events based on the difference between the downstream power outage scope and the simulated disconnection result; and updating the judgment condition of the protection action missing state based on the protection signal restoration state; According to the response delay, the deviation between the triggering time of the power restoration instruction and the actual power restoration time of the power supply unit is adjusted to be reduced to a preset deviation value; Based on the revised fault propagation tree model and verification results, the top three fault chains with the smallest power outage range differences among the candidate fault chains are extracted. Combined with the equipment health factor, environmental interference factor, and user fault reporting spatial density data, the comprehensive confidence of each fault chain is calculated. The fault chain with the highest comprehensive confidence and the power restoration response delay and timeline deviation less than the preset threshold is output as the final diagnosis result.

2. The power supply fault diagnosis method for a low-voltage distribution network based on power outage and restoration logic according to claim 1 is characterized in that: The generating of candidate fault chains arranged in descending order of weight comprises: Based on the fault propagation tree model, the fault frequency, maintenance record, and aging degree of each power supply unit's equipment historical operation data are used to generate an equipment health score. For power supply units where equipment with a health score below a preset threshold is located, the initial diagnostic weight is increased by a preset ratio to obtain an adjusted diagnostic weight for the equipment health. Real-time collection of ambient temperature and humidity data and matching it with extreme weather types and geographic coverage in meteorological warnings to determine whether the area where the power supply unit is located belongs to a high humidity area, a heavy rainfall warning area, or a thunderstorm warning area; When the power supply unit is located in a high humidity area, a heavy rainfall warning area, or a thunderstorm warning area, the power supply unit allocates a corresponding environmental interference coefficient, multiplies the environmental interference coefficient by the current diagnostic weight of the power supply unit, and obtains the diagnostic weight after superimposing the environmental interference; Count the geographical distribution density of user fault reporting signals. For power supply units where fault reporting points are concentrated and overlap with the timeline of the power outage event, assign a spatial density gain coefficient based on the extent to which the density exceeds the preset threshold. Multiply the current gain coefficient by the current diagnostic weight of the power supply unit to obtain the adjusted diagnostic weight weighted by spatial density. According to the adjusted diagnostic weight of comprehensive equipment health, environmental interference superposition weight and spatial density weight, the final diagnostic weights between power supply units in the fault propagation tree model are corrected step by step through weighted fusion, and candidate fault chains are generated, which are sorted from high to low according to the corrected weight values.

3. The power supply fault diagnosis method for a low-voltage distribution network based on power outage and restoration logic according to claim 2, characterized in that: The checking of the unreset protection signal of the starting device of the fault chain, simulating the disconnection of the upstream switch to verify the consistency of the downstream power outage state, triggering the power restoration instruction and observing the response delay include: Obtain the real-time telesignal signal of the candidate fault chain starting device, and confirm whether the fault chain starting device has an unreset protection signal based on the last action record; Use the remote control function of the distribution automation system to temporarily disconnect the upstream switch in the candidate fault chain and monitor the voltage signals of all downstream smart meters in real time to verify whether the downstream power outage status is consistent with the topological path of the candidate fault chain; A remote power restoration instruction is sent to the power restoration node of the candidate fault chain through the master station system, and the time difference between the instruction issuance time and the meter voltage recovery time is recorded as the response delay data. The response delay data is compared with the preset communication delay threshold, and the data of the response delay exceeding the limit is recorded.

4. The power supply fault diagnosis method for a low-voltage distribution network based on power outage and restoration logic according to claim 3 is characterized in that: The correcting the time sequence of the power outage events according to the difference between the downstream power outage range and the simulated disconnection result includes calculating the time sequence offset of the power outage events between the power supply units according to the deviation between the downstream power outage range and the simulated disconnection result; The timestamps of the power outage events of adjacent power supply units in the power outage and restoration event timeline are reordered based on the timing offset to generate an updated time sequence of the power outage and restoration events.

5. The method for diagnosing power supply faults in a low-voltage distribution network based on power outage and restoration logic according to claim 4 is characterized in that: The updating of the determination condition of the protection action missing state according to the protection signal reset state includes, when the protection signal reset state is not reset, determining that the protection action missing state of the corresponding power supply unit is missing, and updating the determination condition of the protection action missing state to use the non-reset protection signal as a necessary condition for triggering the protection action missing; When the protection signal restoration state is restored, the time window length in the judgment condition of the protection action missing state is adjusted according to whether the power restoration response delay and the time line deviation are less than the preset threshold value until the time line deviation meets the preset requirement.

6. The method for diagnosing power supply faults in a low-voltage distribution network based on power outage and restoration logic according to claim 4, wherein: The adjustment of the deviation between the triggering moment of the power-restore instruction and the actual power-restore moment of the power supply unit to reduce it to a preset deviation value includes, based on the response delay between the triggering moment of the power-restore instruction and the actual power-restore moment of the power supply unit, taking the preset deviation value as the adjustment target, and iteratively adjusting the timestamp deviation between the triggering moment of the power-restore instruction and the actual power-restore moment of the power supply unit, so that the deviation gradually converges to within the preset range of the preset deviation value.

7. The method for diagnosing power supply faults in a low-voltage distribution network based on power outage and restoration logic according to claim 4, wherein: The calculation of the comprehensive confidence of each fault chain includes, Among them, C is the comprehensive confidence, H is the equipment health factor, E is the environmental interference factor, D is the spatial density data of user fault reporting, ΔS is the power outage range difference, Δt is the power restoration response delay deviation, λ is the delay deviation adjustment coefficient, and e is the power outage response delay deviation. -λΔt is the delay decay term.

8. A power supply fault diagnosis system for a low-voltage distribution network based on power outage and restoration logic, applying the power supply fault diagnosis method for a low-voltage distribution network based on power outage and restoration logic as claimed in any one of claims 1 to 7, characterized in that: include: Data acquisition module, model building module, candidate fault chain generation module, detection module and diagnosis result module; The data acquisition module is used to collect user-initiated power outage reporting signals, smart meter power outage pulse signals, branch switch action records, ambient temperature and humidity, and weather warning information in the low-voltage distribution network. Keywords in user fault reporting texts are extracted through semantic recognition and marked as standardized event tags. The meter power outage pulse signals and branch switch action records are aligned according to millisecond-level timestamps to generate a power outage and restoration event timeline covering the entire substation. The model building module is used to divide a plurality of power supply units in the low-voltage distribution network according to whether the end user group has an independent power supply switch; Based on the power outage and restoration event timeline, the spatiotemporal propagation direction of the power outage and restoration events of each power supply unit is identified; based on the time sequence of power outage events between adjacent power supply units and the missing status of protection actions, a fault propagation tree model is constructed with the power supply units as nodes and logical association rules as edges; the event timeline is traversed in reverse, and the power supply unit that first triggered the power outage signal is traced back to the fault starting point; The candidate fault chain generation module, based on the fault propagation tree model, integrates the equipment health factor, environmental interference factor and user fault reporting spatial density data, dynamically adjusts the diagnostic weights between power supply units, and generates candidate fault chains arranged in descending order of weight; The detection module is used to check the unreset protection signal of the candidate fault chain starting point device, simulate the disconnection of the upstream switch to verify the consistency of the downstream power outage state, trigger the power restoration instruction to observe the response delay, and reversely inject the verification result into the fault propagation tree model; correct the time sequence according to the difference between the downstream power outage range and the simulated disconnection result; and update the judgment condition of the protection action missing state according to the protection signal reset state; According to the response delay, the deviation between the triggering time of the power restoration instruction and the actual power restoration time of the power supply unit is adjusted to be reduced to a preset deviation value; The diagnosis result module, based on the revised fault propagation tree model and verification results, extracts the top three fault chains with the smallest power outage range differences among the candidate fault chains, calculates the comprehensive confidence of each fault chain based on the equipment health factor, environmental interference factor and user fault reporting spatial density data, and outputs the fault chain with the highest comprehensive confidence and the power restoration response delay and timeline deviation less than a preset threshold as the final diagnosis result.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the power supply fault diagnosis method of the low-voltage distribution network based on power outage and restoration logic according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the power supply fault diagnosis method of a low-voltage distribution network based on power outage and restoration logic according to any one of claims 1 to 7 are implemented.

Images