Low-voltage power distribution cabinet arc fault whole cycle management method

Abstract

The application discloses a kind of low-voltage distribution cabinet arc fault whole cycle management and control methods, belong to low-voltage distribution and industrial internet of things technical field, solve the defects of existing technology early warning lag, false alarm rate is high, emergency disposal network failure, no whole cycle closed-loop management and control.This method uses the double redundancy architecture of edge local priority and cloud platform bottom, including six steps of preposition system configuration and grid responsibility binding, multi-source data acquisition and edge preprocessing, super-early three-stage early warning and preposition intervention, fatal arc fault cloud edge collaborative emergency disposal, fault closed-loop operation and management and control, whole cycle data archiving and model iteration optimization, supporting working condition feature distinguishing rule and multi-source composite verification logic.The application can realize arc fault early identification for 1 to 3 weeks, false alarm rate is controlled within 0.1%, completes local emergency disposal within 100ms and network failure is not invalid, forms arc fault whole life cycle closed-loop management and control, and greatly reduces electrical fire and equipment damage risk.

Description

Technical Field

[0001] This invention relates to the field of low-voltage power distribution technology, specifically to the fields of industrial Internet of Things, intelligent early warning of electrical faults, emergency safety management and intelligent operation and maintenance technology, and is particularly applicable to the closed-loop management of the entire life cycle of arc faults in low-voltage complete switchgear / distribution boxes. Background Technology

[0002] Low-voltage switchgear is a core device in power systems responsible for power distribution, circuit control, and load protection. It is widely used in civil buildings, industrial production, commercial complexes, and new energy power plants. Its operational safety directly affects power supply stability and the safety of life and property. In actual operation, issues such as loose wiring, poor contact, insulation aging, and overload short circuits can easily lead to arcing faults inside the switchgear. The core temperature of an arc can reach 6000-8000℃, igniting surrounding insulation materials and cable busbars within milliseconds, thus causing electrical fires, equipment explosions, and electric shock accidents. It is the leading cause of disasters in low-voltage power distribution systems.

[0003] Currently, protection technologies for arc faults in distribution cabinets are mainly divided into two categories: one is post-event protection technology centered on arc fault protection devices (AFDD) and molded case circuit breakers, which can only perform trip protection after an arc fault occurs and cannot provide early warning at the incipient stage of the fault. It is a passive protection, and often the equipment has already been damaged and power outages have occurred by the time it is activated; the other is pre-alarm technology centered on arc light, current, and temperature detection, which can issue a warning shortly before the arc occurs, but still has significant technical defects.

[0004] Among existing publicly available technologies, CN121041627A discloses a fire prevention system for low-voltage distribution cabinets with arc detection function. This system collects parameters from multiple sources, preprocesses them to construct a fire prevention model, classifies risk levels, and executes tiered responses and status feedback. However, this technical solution has several drawbacks: First, all decision-making and execution commands rely entirely on cloud-based models, lacking local edge protection capabilities. Emergency response is completely ineffective when equipment is out of the network, creating a critical safety blind spot. Second, risk classification can only be performed after an arc fault occurs, failing to achieve ultra-early identification of faults in their nascent stages, resulting in severely delayed warnings. Third, it lacks a mechanism to distinguish between normal nonlinear loads and faulty arcs, making it highly susceptible to harmonic interference from conventional loads such as frequency converters, charging piles, and LED lights, leading to a high false alarm rate in field applications. Fourth, it only achieves a single fire prevention closed loop from detection to execution and unidirectional feedback, lacking a comprehensive design for subsequent operation and maintenance, emergency repair management, and product optimization feedback, resulting in severe disconnect between processes.

[0005] Furthermore, a pre-alarm system for arc flash protection in medium and low voltage switchgear, disclosed in publication number CN222482363U, achieves pre-alarm and protection actions through the acquisition of arc flash, arc sound, and current signals. This technical solution can only achieve pre-alarm for a very short time before arcing through arc sound signals, and cannot achieve ultra-early fault identification 1 to 3 weeks in advance. Moreover, it lacks any design for full-process control, closed-loop operation and maintenance, and long-term optimization, and can only achieve single-point hardware protection, failing to solve the root cause problem of arc faults.

[0006] Therefore, developing a full-cycle management and control method that can achieve ultra-early and accurate early warning of arc faults, millisecond-level highly reliable emergency response, closed-loop operation and maintenance management throughout the entire process, and long-term root cause optimization has become an urgent technical problem to be solved in this field. Summary of the Invention

[0007] To overcome the aforementioned deficiencies of existing technologies, the present invention aims to provide a full-cycle management method for arc faults in low-voltage distribution cabinets. This method overcomes the limitations of existing single-point passive protection and constructs a full-lifecycle closed-loop system encompassing early warning and intervention, millisecond-level cloud-edge collaborative emergency response, post-event closed-loop operation and maintenance management, and full-cycle data archiving and model iterative optimization. Simultaneously, it addresses industry pain points such as delayed early warning, high false alarm rate, network outage failure, disconnected management, and inability to optimize at the root cause. While ensuring the safe operation of low-voltage distribution cabinets, it also considers power supply continuity, on-site operational safety, and long-term product optimization capabilities.

[0008] To achieve the above-mentioned objectives, the present invention adopts the following technical solution:

[0009] A method for full-cycle management and control of arc faults in low-voltage distribution cabinets includes the data acquisition steps, parameter preprocessing steps, risk classification steps, emergency execution steps, and status feedback steps common to existing technologies; it also includes the following full-cycle management and control steps:

[0010] S1: Pre-configuration system and grid-based responsibility binding: A unique electronic file is established for each low-voltage distribution cabinet, and the operation and maintenance responsibility system is bound according to the three-level architecture from regional grid to equipment to responsible person. The system is pre-configured with three-level early warning thresholds for arc faults, emergency response permissions, spare parts and other emergency resources; wherein the emergency response permissions are divided according to the importance level of the load, including automatic handling permissions for ordinary loads and dual-person review permissions for important loads.

[0011] S2: Multi-source data synchronous acquisition and edge preprocessing: The electrical parameters, temperature parameters, insulation arc characteristic parameters and environmental status parameters of the distribution cabinet are synchronously acquired through the sensing acquisition unit. The edge computing unit performs validity verification, filtering and noise reduction, and anti-jitter verification for at least three consecutive acquisition cycles on the acquired data, and extracts current zero-rest characteristics and harmonic change characteristics.

[0012] S3: Early warning and intervention for arc faults: Based on multi-source data composite verification logic and combined with working condition feature differentiation rules, three-level warnings are executed according to the time sequence of fault development. Each of the three-level warnings can only be triggered if at least two different dimensions of parameters are abnormal. After the three-level warnings are triggered, the corresponding intervention actions are automatically executed.

[0013] The operating condition characteristic differentiation rule is as follows: distinguish between normal nonlinear load and fault arc characteristics. The harmonic characteristics of normal load are stable and positively correlated with the load size, while the harmonic characteristics of fault arc are intermittent, irregular, and abrupt, accompanied by zero current characteristics.

[0014] The three-tiered early warning system is divided into three stages based on time sequence: Level 1 early warning, which is 1 to 3 weeks in advance; Level 2 warning, which is several hours to several days in advance, for the mid-term deterioration stage; and Level 3 emergency warning, which is several seconds to several minutes in advance, for the critical outbreak stage.

[0015] S4: Millisecond-level cloud-edge collaborative emergency response to fatal arc faults: Adopting a dual-redundancy architecture with edge local priority and cloud platform backup, after the composite triggering conditions of fatal arc faults are met, the edge computing unit performs safety handling actions locally in a closed loop within 100ms, and simultaneously reports the fault message to the cloud platform; The emergency handling logic of the edge computing unit does not rely on the cloud network, and can independently complete all safety handling actions in the event of a network outage, and locally encrypts and caches fault data, and automatically retransmits it to the cloud platform after the network is restored.

[0016] S5: Fault Closed-Loop Operation and Maintenance Emergency Repair Full-Process Control: Based on a pre-bound grid-based responsibility system, it realizes full-process digital management of fault multi-channel notification, intelligent work order dispatch, on-site safety mandatory control, and work order review closed loop.

[0017] S6: Full-cycle data archiving and model iteration optimization: The fault data, handling records, and operation and maintenance data in the full-process data are encrypted and archived. The causes of faults are statistically analyzed and fed back to product design and production process optimization. At the same time, the early warning feature identification logic and trigger threshold are continuously optimized based on the archived full-process data.

[0018] Furthermore, in step S1, the grid-based responsibility system pre-configures four levels of responsible personnel, including a first-level dedicated maintenance person, a second-level backup maintenance person, a third-level grid supervisor, and a fourth-level emergency command center duty officer, and simultaneously binds the contact information, response permissions, and on-duty schedule of each responsible person.

[0019] Further, in step S2, the electrical parameters include three-phase voltage, three-phase current, total harmonic distortion of current, odd harmonic components, residual current, and zero-sequence current; the temperature parameters include busbar terminal temperature, circuit breaker contact temperature, and ambient temperature inside the cabinet; the insulation arc characteristic parameters include arc sensor signal, AFDD arc fault signal, and ultrasonic partial discharge signal; and the environmental status parameters include relative humidity inside the cabinet, dew point temperature, cabinet door status, and water immersion signal.

[0020] Furthermore, in step S3, the specific triggering and execution logic of the three-level tiered early warning is as follows:

[0021] Level 1 warning: 1 to 3 weeks in advance, triggered by at least two of the following abnormalities: continuous increase in three-phase current imbalance, abnormal increase in current harmonic distortion rate, excessive three-phase temperature difference at the same location, continuous increase in residual current, and humidity in the cabinet approaching the condensation threshold. The actions include generating and pushing out a hidden danger warning order, encrypting the data collection frequency, initiating pre-linkage adjustment, and generating a planned maintenance work order within 72 hours.

[0022] Level 2 alarm: Several hours to several days in advance, triggered by at least two of the following abnormalities: zero-cycle phenomenon in current waveform, continuous increase in harmonic distortion rate, exponential increase in temperature at fault location, decrease in insulation resistance, and continuous detection of partial discharge signal. The actions include: locally activating low-frequency audible and visual alarm, pushing emergency alarm information through multiple channels, locking remote control permissions of equipment, and generating an emergency hazard handling work order within 24 hours.

[0023] Level 3 Emergency Alarm: Triggered several seconds to several minutes in advance based on at least two of the following abnormalities: continuous and severe distortion of current throughout the entire cycle, sudden and significant change in current, sudden rise in temperature at the fault location, detection of dense pulses of partial discharge, and continuous output of fault arc signal by AFDD. The actions include locally cutting off the power supply to the fault circuit, locking the closing circuit, activating the highest level audible and visual alarm, triggering the automatic fire extinguishing device, and generating an immediate emergency repair work order.

[0024] Furthermore, step S3 also includes a warning escalation and de-escalation mechanism: when the parameter abnormality continues to worsen, the warning level is automatically upgraded; when the parameter returns to the normal range and there are no abnormalities for 36 consecutive hours, the warning is automatically downgraded or lifted, and the warning change data throughout the entire process is recorded simultaneously.

[0025] Furthermore, in step S4, the composite triggering condition for the fatal arc fault must satisfy any two of the following: the arc sensor continuously detects the arc signal, the AFDD outputs the fatal arc fault signal, the circuit current instantaneously surges to 10 times or more of the rated current, and the cabinet temperature rises by more than 20°C within 10 seconds and exceeds the 120°C safety threshold.

[0026] Furthermore, in step S4, the security handling actions executed locally in closed loop by the edge computing unit include, in order of priority:

[0027] a. Accurately disconnect the upstream circuit breaker of the faulty circuit, and immediately lock the closing circuit after the trip is completed to prevent accidental closing and power supply.

[0028] b. Activate the highest level audible and visual alarm of the cabinet, lock the electromagnetic lock on the cabinet door, and prohibit unauthorized personnel from opening the door while it is powered on;

[0029] c. Trigger the miniature automatic fire extinguishing device inside the cabinet, which will shut down unnecessary loads in adjacent circuits.

[0030] d. Encrypt, lock, and cache the voltage / current waveform data for the 10 power frequency cycles before the fault and the 5 power frequency cycles after the fault, as well as the full sensor data and action execution records at the time of the fault;

[0031] e. Send encrypted fault messages to the cloud platform through the highest priority channel, and use a multiple retransmission mechanism to ensure cloud reception.

[0032] Furthermore, in step S4, the cloud platform's coordinated handling actions include: displaying fault information at the top of the monitoring screen pop-up window, highlighting fault locations on the GIS map, locking the highest level of remote control permissions for the device, automatically pushing emergency notification information to the customer's responsible person, and automatically generating emergency repair work orders and dispatching them to the pre-bound maintenance personnel.

[0033] Furthermore, in step S5, the full-process operation and maintenance management includes:

[0034] Multi-channel tiered notification and timeout escalation mechanism: After a fault is triggered, a strong reminder will be pushed to the first-level responsible person via APP / mini-program, voice call, SMS, and WeChat notification. If no order is accepted within 3 minutes, it will be pushed to the second-level responsible person. If no order is accepted within 8 minutes, it will be pushed to the grid supervisor and emergency duty officer. If no order is accepted within 20 minutes, the company-level emergency plan will be activated to force order dispatch.

[0035] On-site QR code check-in and mandatory safety condition confirmation: After arriving on site, maintenance personnel must scan the equipment's unique QR code to check in, and check off each item to confirm that the power is off, the power is tested, the grounding wire is connected, and the safety protection is in place before they can enter the maintenance process;

[0036] Full traceability management of the maintenance process: During the maintenance process, photos of the fault location, information on damaged components, and maintenance records must be uploaded. Temporary closing tests require dual-person authorization verification.

[0037] Work order review and customer confirmation: After the maintenance is completed, the insulation test and no-load / load operation test data are uploaded. After the technical supervisor reviews and approves the data, the work order is closed and the maintenance report is sent to the customer simultaneously.

[0038] Furthermore, in step S6, the fault cause classification includes four categories: design defects, manufacturing process problems, component quality problems, and improper customer use. Based on the classification results, a fault analysis report is generated and pushed to the R&D and production departments to optimize product structure, manufacturing process, and component selection in a targeted manner. Based on the archived fault data, a warning model is trained through machine learning algorithms to continuously optimize the feature recognition logic and trigger threshold.

[0039] The beneficial effects of this invention are as follows:

[0040] This invention establishes a complete process link from early warning before an event, risk control and disposal during an event, closed-loop operation and maintenance after an event, to root cause optimization and iteration. It completely solves the core defects of existing technologies, such as disconnected links and many blind spots in management and control, and upgrades arc fault management from single passive protection to full-process proactive management and root cause optimization.

[0041] This invention addresses the complete development pattern of arc faults, enabling ultra-early identification of fault buds 1 to 3 weeks in advance, far superior to the short-term warning before arcing that existing technologies can only achieve. At the same time, through operating condition feature differentiation rules, it accurately distinguishes the characteristic differences between normal nonlinear loads and faulty arcs. Combined with multi-source composite verification logic that is triggered only when at least two different dimensional parameters are abnormal, it completely solves the long-standing pain points of high false alarm rate and difficulty in field implementation in the industry.

[0042] This invention pushes emergency response logic 100% to the local edge computing unit, completing the entire security process within 100ms. It is completely independent of the cloud network and can still execute all security actions even in network outage scenarios, thus completely solving the fatal flaw of existing technologies that rely on the cloud and fail when the network is down. At the same time, through load-level permission configuration, it automatically handles ordinary loads and performs dual-person verification for important loads, minimizing economic losses caused by a one-size-fits-all power outage while ensuring equipment safety.

[0043] This invention is the first to integrate arc fault protection with intelligent operation and maintenance throughout the entire process. Through grid-based responsibility binding and time-lapse progressive escalation dispatch mechanism, it ensures that 100% of faults are responded to and handled by someone. At the same time, through the mandatory on-site safety condition confirmation mechanism, the requirements of the "Electric Power Safety Work Regulations" are embedded into the system process, which eliminates personal safety risks in on-site operations from a systemic perspective and solves the problems of low emergency coordination efficiency and lack of on-site operation control in existing technologies.

[0044] This invention, through full-cycle data archiving and fault cause classification analysis, directly feeds fault data back into product development and production processes, enabling targeted optimization of product structure, production processes, and component selection. This fundamentally reduces the recurrence rate of similar faults, forming a long-term closed loop of control, analysis, optimization, and iteration, providing data support and optimization direction for improving the reliability of low-voltage switchgear products. Attached Figure Description

[0045] Figure 1 This is a block diagram of the overall architecture of the low-voltage distribution cabinet arc fault full-cycle management and control system upon which the present invention is based;

[0046] Figure 2 This is an overall flowchart of the low-voltage distribution cabinet arc fault full-cycle control method of the present invention;

[0047] Figure 3 This is a block diagram of the arc fault early warning logic control for the present invention.

[0048] Figure 4 This is a logic control block diagram for the cloud-edge collaborative emergency response to fatal arc faults according to the present invention.

[0049] In the diagram: 1-Sensing and acquisition unit; 101-Electrical parameter acquisition module; 102-Temperature acquisition module; 103-Insulation arc characteristic acquisition module; 104-Environmental status acquisition module;

[0050] 2-Edge computing unit; 201-Data preprocessing module; 202-Operating condition feature differentiation module; 203-Graded early warning logic module; 204-Local emergency response module; 205-Local data caching module;

[0051] 3-Transmission network unit;

[0052] 4-Cloud Platform Unit; 401-Equipment Management Module; 402-Cloud Early Warning Engine Module; 403-Emergency Command Module; 404-Maintenance Work Order Module; 405-Data Storage and Analysis Module; 406-Access Control Module;

[0053] 5-Actuator unit; 501-Intelligent circuit breaker; 502-Audible and visual alarm; 503-Automatic fire extinguishing device; 504-Electromagnetic lock for cabinet door;

[0054] 6-Application Operation and Maintenance Unit; 601-Operation and Maintenance APP / Mini Program Module; 602-WEB Monitoring Dashboard Module; 603-Client Notification Module. Detailed Implementation

[0055] The technical solution of the present invention will be clearly and completely described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some preferred embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the protection scope of the present invention.

[0056] This embodiment applies to the XL-21 type low-voltage power distribution cabinet manufactured by the applicant. Its rated parameters are rated voltage Un=380V and rated current In=250A. It is installed in a local outdoor commercial plaza and is a typical commercial load, pre-configured with automatic handling permissions. For example... Figure 1 The control system shown in this embodiment includes a sensing and acquisition unit 1, an edge computing unit 2, a transmission network unit 3, a cloud platform unit 4, an execution mechanism unit 5, and an application operation and maintenance unit 6. All units communicate with each other and are logically linked.

[0057] Example 1: A method for full-cycle control of arc faults in low-voltage distribution cabinets, the specific execution steps are as follows:

[0058] Step S1: Configure the front-end system and bind grid-based responsibilities

[0059] Establish a unique electronic file for this distribution cabinet, with the equipment number HS-PDG-202400126. The file should include the equipment model and specifications, installation address / GIS location, primary system diagram, component list, rated parameters, factory information, historical operation and maintenance records, customer contact person and authorization information.

[0060] Based on the three-tier architecture of "local grid to HS-PDG-202400126 power distribution cabinet to responsible person", the operation and maintenance responsibility system is bound together, and four levels of responsible personnel are pre-configured: Level 1 dedicated operation and maintenance responsible person Zhang XX, Level 2 backup operation and maintenance responsible person Li XX, Level 3 grid supervisor Wang XX, and Level 4 emergency command center duty officer Zhao XX. The mobile phone number, enterprise WeChat account, operation and maintenance APP account, response permissions, and 7×24-hour on-duty schedule of each responsible person are bound simultaneously.

[0061] The system is pre-configured with a three-level early warning threshold for arc faults and emergency response permissions. Since the equipment is a general commercial load, it is granted automatic response permissions, meaning that when the triggering conditions for a fatal arc fault are met, the edge computing unit can automatically perform a power-off operation. At the same time, the system is pre-matched with the spare parts model corresponding to the equipment, the location of the nearest local spare parts warehouse, and a list of emergency repair vehicles and tools.

[0062] Furthermore, it is necessary to sign a "Remote Monitoring and Emergency Response Service Agreement" with the client in advance to clarify the division of responsibilities for emergency response, the scope of authorization for remote operation, and the access authority for on-site repairs, thereby avoiding compliance and legal risks.

[0063] Step S2: Synchronous acquisition of multi-source data and edge preprocessing

[0064] Through sensing and acquisition unit 1, full-dimensional data of the power distribution cabinet is collected synchronously according to a preset cycle. The specific acquisition configuration is as follows:

[0065] Electrical parameter acquisition module 101: adopts a 0.5-class multi-functional power meter with a sampling rate of 2kHz, and acquires three-phase voltage, three-phase current, total harmonic distortion of current, 3rd / 5th / 7th odd harmonic components, residual current, and zero-sequence current at a period of 100ms.

[0066] Temperature acquisition module 102: It adopts 3 wireless contact temperature sensors, which are installed on the terminals of the A / B / C phase busbars respectively, and acquire the terminal temperature at a 10-second cycle. At the same time, it adopts 1 temperature and humidity sensor to acquire the ambient temperature inside the cabinet.

[0067] Insulation arc feature acquisition module 103: adopts two point-type arc light sensors, which are installed in the busbar compartment and the circuit breaker compartment respectively; AFDD arc fault protector; ultrasonic partial discharge sensor, which acquires arc light signal, arc fault signal and partial discharge signal at a 1ms cycle;

[0068] Environmental status acquisition module 104: Acquires relative humidity inside the cabinet, dew point temperature, cabinet door status, and water immersion signal;

[0069] Edge computing unit 2 preprocesses the collected raw data sequentially, and the specific process is as follows:

[0070] Validity verification: Invalid data caused by sensor disconnection, over-range, or electromagnetic interference are removed. Invalid data only triggers sensor fault alarms and does not enter subsequent warning logic.

[0071] Filtering and noise reduction: A moving average filtering algorithm is used to remove spikes and punctuation caused by instantaneous power grid fluctuations and electromagnetic interference;

[0072] Anti-shake verification: All data must meet the condition of continuous abnormality for 3 consecutive acquisition cycles, with a default interval of 500ms / cycle, before entering the subsequent warning judgment logic to filter out instantaneous interference signals;

[0073] Feature extraction: Extract zero-wave characteristics and harmonic abrupt change characteristics of the current waveform to provide a basis for subsequent early warning judgment.

[0074] Step S3: Early warning and proactive intervention for arc faults

[0075] Edge computing unit 2, based on preprocessed valid data and combined with operating condition feature differentiation rules, executes a three-level hierarchical early warning system. All early warnings can only be triggered if at least two parameters in different dimensions are abnormal. The specific execution process is as follows:

[0076] Operating condition characteristic differentiation and verification: Edge computing unit 2 distinguishes between normal nonlinear loads and fault arc characteristics through a preset algorithm.

[0077] If the harmonic characteristics are stable and continuous, and positively correlated with the load size, it is determined to be a normal nonlinear load and no warning is triggered.

[0078] If the harmonic characteristics show intermittent and irregular abrupt changes, are not fixedly related to the load size, and are accompanied by zero current characteristics, it is judged as a precursor to a fault arc and enters the subsequent graded early warning judgment.

[0079] Level 3 Early Warning Implementation:

[0080] Level 1 Warning Triggering and Execution: During system operation, the system detects that the current imbalance of phase A in the distribution cabinet continuously rises to 16%, the current THDi continuously rises from the normal ≤5% to 13%, the temperature of phase A terminals is 9℃ higher than phases B / C, the temperature rises by 4℃ within 1 hour, and there is no load change. This meets two or more of the abnormal triggering conditions for a Level 1 warning, and the fault was identified 14 days in advance. The system automatically performs the following actions: generates an "Equipment Hidden Danger Warning Form" and pushes it to maintenance personnel and customer managers; increases the data collection frequency from 1 minute / time to 10 seconds / time; starts the cabinet's heating and dehumidification device; and generates a planned maintenance work order within 72 hours. During on-site inspection, maintenance personnel find that the bolts of phase A terminals are loose and the torque is not up to standard. After tightening them to the standard torque, the parameters return to normal, the system cancels the warning, and the work order is closed.

[0081] Level 2 Alarm Triggering and Execution: If the Level 1 warning is not handled in time and the parameters continue to deteriorate, the A-phase current waveform is detected to have a zero-cycle phenomenon at the power frequency, with more than 30 intermittent bursts per day, THDi continuously rises to 22%, the A-phase terminal temperature rises to 75℃, the temperature rises by 6℃ within 10 minutes, and the insulation resistance drops to 0.8MΩ, meeting two or more of the abnormal triggering conditions for a Level 2 alarm, and the fault deterioration is identified 36 hours in advance; the system will automatically perform the following actions: activate a low-frequency audible and visual alarm locally, push emergency alarm information through SMS, WeChat, and the maintenance APP, lock the remote control permissions of the equipment, and generate an emergency hazard handling work order within 24 hours;

[0082] Level 3 Emergency Alarm Triggering and Execution: If the Level 2 alarm is still not handled, the parameters further deteriorate, and a continuous and severe distortion of the A-phase current throughout the entire cycle is detected, with THDi soaring to 35%, the A-phase terminal temperature rising sharply by 15°C within 10 seconds and exceeding 105°C, the partial discharge sensor detects continuous dense pulses, and the AFDD outputs a continuous fault arc signal, meeting two or more abnormal triggering conditions for the Level 3 emergency alarm, and identifying the critical outbreak of the fault 90 seconds in advance; the system automatically executes the following actions: the edge computing unit disconnects the upstream circuit breaker of the A-phase fault circuit according to the pre-authorization, locks the closing circuit, activates the highest level audible and visual alarm, triggers the miniature aerosol fire extinguishing device in the cabinet, and the cloud platform immediately generates an emergency repair work order and initiates the operation and maintenance dispatch process;

[0083] Early warning escalation and de-escalation mechanism: When the abnormal parameters continue to deteriorate, the system automatically escalates from Level 1 early warning to Level 2 alarm and then to Level 3 emergency alarm; when the parameters return to the normal range and there are no abnormalities for 36 consecutive hours, the system automatically de-escalates or cancels the early warning, and records the early warning change data throughout the entire process.

[0084] Step S4: Millisecond-level cloud-edge collaborative emergency response to fatal arc faults

[0085] If the warning is not addressed in time, and the distribution cabinet experiences a phase-to-phase short circuit due to unauthorized capacity increases by the customer, resulting in a large number of continuous electric arcs, the system will execute the following emergency response procedures:

[0086] Composite trigger verification: The arc sensor continuously detects the arc signal, and at the same time the circuit current instantaneously surges to 2600A, which is 10.4 times the rated current, meeting the two composite trigger conditions for a fatal arc fault, and emergency response is immediately initiated;

[0087] Millisecond-level local closed-loop processing at the edge, including completion within 75ms, with no data loss even when the network is down:

[0088] a. Accurately disconnect the main circuit breaker of the power distribution cabinet, and immediately lock the closing circuit after the circuit is opened to physically prevent accidental closing and power supply.

[0089] b. Activate the 110dB high-decibel audible and visual alarm and flashing warning light in the cabinet, lock the electromagnetic lock on the cabinet door, and prohibit unauthorized personnel from opening the door while it is powered on;

[0090] c. Trigger the miniature aerosol fire extinguishing device inside the cabinet to shut down unnecessary loads in adjacent circuits and prevent the fault from spreading;

[0091] d. Locally encrypt, lock, and cache voltage / current waveform data for the 10 power frequency cycles before the fault and the 5 power frequency cycles after the fault, as well as all sensor data and action execution records at the time of the fault. The data cannot be tampered with.

[0092] e. The encrypted fault message is sent to the cloud platform unit 4 with the highest priority through the 4G transmission network unit 3, and a 3-retransmission mechanism is used to ensure cloud reception.

[0093] The cloud platform coordinates and processes the data, completing the process within 800ms of receiving the message.

[0094] a. The company's emergency command center's WEB monitoring screen displays the fault information in a pop-up window at the top, highlights the fault location on the GIS map, and continuously provides audio and visual reminders to on-duty personnel.

[0095] b. Mark the device as "critical failure emergency status" and lock the highest level of remote control access to the device;

[0096] c. Automatically retrieve equipment electronic records and generate a "Critical Failure Emergency Briefing";

[0097] d. Automatically send emergency notification text messages and WeChat messages to the customer's responsible person, clearly stating the fault situation, actions already taken, and information on the repair personnel;

[0098] e. Automatically generate standardized emergency repair work orders, pre-fill all fault information and spare parts list, and automatically dispatch them to the first-level maintenance person Zhang XX according to the pre-bound grid-based responsibility system.

[0099] Step S5: Closed-loop Fault Maintenance and Emergency Repair Process Control

[0100] Based on a pre-bound grid-based responsibility system, the system performs full-process operation and maintenance management, with the specific process as follows:

[0101] Multi-channel tiered notification and timeout escalation: After a fault is triggered, the system simultaneously sends a strong reminder to Zhang XX via the maintenance app, an automatic voice call, an emergency SMS, and a WeChat notification; if Zhang XX does not accept the order within 3 minutes, the system simultaneously pushes it to the backup person in charge, Li XX; if the order is still not accepted within 8 minutes, the system pushes it to the grid supervisor, Wang XX, and the emergency duty officer, Zhao XX. Zhao XX immediately contacts Zhang XX manually by phone to confirm that he is unable to respond due to on-site operations and coordinates with Li XX to take over the work order; if the order is still not accepted within 20 minutes, the system will activate the company-level emergency plan and forcibly assign the order to the emergency standby team;

[0102] After receiving an order, the system automatically pushes the optimal navigation route, equipment information, safe operating procedures, and spare parts information. At the same time, it pushes the information of the person who received the order and the estimated arrival time at the site in 40 minutes to the customer and the emergency command center.

[0103] Mandatory on-site safety control: After Li XX arrives at the site, he must scan the equipment's unique QR code to sign in and upload the site location and photos; the system will forcibly pop up a safety operation confirmation checklist, and he must check and confirm each item such as "the upper power supply is completely disconnected, the voltage test has been completed, the grounding wire has been installed, the warning sign has been hung, and qualified insulating protective equipment has been worn" before he can enter the maintenance process, so as to completely eliminate the risk of live work;

[0104] The entire maintenance process is managed with full traceability: During the maintenance process, Li XX uploads photos of the fault location, photos of the damaged circuit breaker and busbar, and maintenance records. Temporary closing tests require dual-person authorization verification through the system, which is remotely authorized by grid supervisor Wang XX.

[0105] Work order review and customer confirmation: After the maintenance is completed, Li XX uploads the insulation resistance test, no-load power-on test, and load operation test data to confirm that the equipment is normal; after the maintenance record is reviewed and approved by the technical supervisor, the work order is officially closed and archived in the equipment's electronic file; the system automatically sends a maintenance completion notification SMS and a complete maintenance report to the customer, and collects customer service feedback at the same time.

[0106] Step S6: Full-cycle data archiving and model iterative optimization

[0107] The system will encrypt and permanently archive all data related to this fault, including early warning data, fault waveforms, emergency response records, maintenance work orders, and test data, with a storage period of no less than 3 years.

[0108] The system automatically analyzed the cause of the fault and determined that it was caused by an overload short circuit due to the customer's unauthorized capacity increase. At the same time, the analysis found that there was room for improvement in the terminal anti-loosening design. A fault analysis report was generated and pushed to the R&D and production departments.

[0109] Based on the report, the R&D department optimized the anti-loosening structure of the wiring terminals, and the production department optimized the bolt torque control process and the withstand voltage test process, thereby reducing the recurrence rate of the same type of failure from the root cause.

[0110] Based on the archived data of this fault, a warning model was trained using machine learning algorithms. The warning feature recognition logic and trigger threshold under overload scenarios were optimized to continuously improve the warning accuracy and control the false alarm rate to within 0.1%.

[0111] Example 2: This example is applied to a low-voltage distribution cabinet in a tertiary hospital. As a critical load, it is pre-configured with dual-user verification permissions, prohibiting the system from automatically performing power-off operations. When the combined triggering conditions of a fatal arc fault are met, the edge computing unit only activates the highest-level audible and visual alarm, cabinet door locking, and fire extinguishing device activation, without performing an automatic power-off operation. The cloud platform immediately pushes the highest-level alarm to both the maintenance manager and the hospital's logistics manager. Only after dual-user verification and confirmation by both the maintenance manager and the hospital's electrical manager can the system remotely execute the tripping operation, maximizing the power supply continuity for the hospital's critical loads and preventing medical safety accidents caused by abnormal power outages.

[0112] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for full-cycle management and control of arc faults in low-voltage distribution cabinets, comprising data acquisition steps, parameter preprocessing steps, risk classification steps, emergency execution steps, and status feedback steps; characterized in that, It also includes the following full-cycle management steps: S1: Front-end system configuration and grid-based responsibility binding: Establish a unique electronic file for each low-voltage distribution cabinet, bind the operation and maintenance responsibility system according to the three-level architecture from regional grid to equipment to responsible person, and pre-configure emergency resources such as three-level early warning threshold for arc faults, emergency response authority, and spare parts; The emergency response permissions are divided according to the importance level of the load, including automatic response permissions for ordinary loads and dual-person review permissions for important loads; S2: Multi-source data synchronous acquisition and edge preprocessing: The electrical parameters, temperature parameters, insulation arc characteristic parameters and environmental status parameters of the distribution cabinet are synchronously acquired through the sensing acquisition unit. The edge computing unit performs validity verification, filtering and noise reduction, and anti-jitter verification for at least three consecutive acquisition cycles on the acquired data, and extracts current zero-rest characteristics and harmonic change characteristics. S3: Early warning and intervention for arc faults in the early stage: Based on multi-source data composite verification logic and combined with the rules for distinguishing working conditions, three-level warnings are executed according to the time sequence of fault development. Each of the three-level warnings can only be triggered if at least two parameters in different dimensions are abnormal. After the three-level warnings are triggered, the intervention action is automatically executed. The operating condition characteristic differentiation rule is as follows: distinguish between normal nonlinear load and fault arc characteristics. The harmonic characteristics of normal load are stable and positively correlated with the load size, while the harmonic characteristics of fault arc are intermittent, irregular, and abrupt, accompanied by zero current characteristics. The three-level early warning system is divided into three stages according to time sequence: Level 1 early warning for the early incubation stage 1 to 3 weeks in advance, Level 2 warning for the mid-term deterioration stage several hours to several days in advance, and Level 3 emergency warning for the critical outbreak stage several seconds to several minutes in advance. S4: Millisecond-level cloud-edge collaborative emergency response to fatal arc faults: Adopting a dual-redundancy architecture with edge local priority and cloud platform backup, after the composite triggering conditions of fatal arc faults are met, the edge computing unit performs safety handling actions locally in a closed loop within 100ms, and simultaneously reports the fault message to the cloud platform; The emergency handling logic of the edge computing unit does not rely on the cloud network, and can independently complete all safety handling actions in the event of a network outage, and locally encrypts and caches fault data, and automatically retransmits it to the cloud platform after the network is restored; S5: Fault Closed-Loop Operation and Maintenance Emergency Repair Full-Process Control: Based on a pre-bound grid-based responsibility system, it realizes digital management of full-process data, including multi-channel fault notification, intelligent work order dispatch, mandatory on-site safety control, and work order review closed loop. S6: Full-cycle data archiving and model iteration optimization: The fault data, handling records, and operation and maintenance data in the full-process data are encrypted and archived. Then, the causes of the faults are statistically analyzed and fed back to the product design and production process optimization. At the same time, the early warning feature identification logic and trigger thresholds are continuously optimized based on the archived full-process data.

2. The method for full-cycle control of arc faults in low-voltage distribution cabinets according to claim 1, characterized in that, In step S1, the grid-based responsibility system is pre-configured with four levels of responsible personnel, including a first-level dedicated operation and maintenance person, a second-level backup operation and maintenance person, a third-level grid supervisor, and a fourth-level emergency command center duty officer. The contact information, response permissions, and on-duty schedule of each person are simultaneously bound.

3. The method for full-cycle control of arc faults in low-voltage distribution cabinets according to claim 1, characterized in that, In step S2 The electrical parameters include three-phase voltage, three-phase current, total harmonic distortion of current, odd harmonic components, residual current, and zero-sequence current. The temperature parameters include busbar terminal temperature, circuit breaker contact temperature, and cabinet ambient temperature. The insulating arc characteristic parameters include arc sensor signals, AFDD arc fault signals, and ultrasonic partial discharge signals; The environmental status parameters include the relative humidity inside the cabinet, dew point temperature, cabinet door status, and water immersion signal.

4. The method for full-cycle control of arc faults in low-voltage distribution cabinets according to claim 1, characterized in that, In step S3, the three-level early warning includes: Level 1 warning: 1 to 3 weeks in advance, triggered by at least two of the following abnormalities: continuous increase in three-phase current imbalance, abnormal increase in current harmonic distortion rate, excessive three-phase temperature difference at the same location, continuous increase in residual current, and humidity in the cabinet approaching the condensation threshold. The actions include generating and pushing out a hidden danger warning order, encrypting the data collection frequency, initiating pre-linkage adjustment, and generating a planned maintenance work order within 72 hours. Level 2 alarm: Several hours to several days in advance, triggered by at least two of the following abnormalities: zero-cycle phenomenon in current waveform, continuous increase in harmonic distortion rate, exponential increase in temperature at fault location, decrease in insulation resistance, and continuous detection of partial discharge signal. The actions include: locally activating low-frequency audible and visual alarm, pushing emergency alarm information through multiple channels, locking remote control permissions of equipment, and generating an emergency hazard handling work order within 24 hours. Level 3 Emergency Alarm: Triggered several seconds to several minutes in advance based on at least two of the following abnormalities: continuous and severe distortion of current throughout the entire cycle, sudden and significant change in current, sudden rise in temperature at the fault location, detection of dense pulses of partial discharge, and continuous output of fault arc signal by AFDD. The actions include locally cutting off the power supply to the fault circuit, locking the closing circuit, activating the highest level audible and visual alarm, triggering the automatic fire extinguishing device, and generating an immediate emergency repair work order.

5. The method for full-cycle control of arc faults in low-voltage distribution cabinets according to claim 1 or 4, characterized in that, Step S3 also includes an early warning escalation and de-escalation mechanism: when the parameter abnormality continues to worsen, the early warning level is automatically upgraded; when the parameter returns to the normal range and there are no abnormalities for 36 hours, the early warning is automatically downgraded or lifted, and the early warning change data of the entire process is recorded simultaneously.

6. The method for full-cycle control of arc faults in low-voltage distribution cabinets according to claim 1, characterized in that, In step S4, the combined triggering condition for the fatal arc fault must satisfy any two of the following: The arc sensor continuously detects arc signals, the AFDD outputs a fatal arc fault signal, the circuit current instantly surges to 10 times the rated current or more, and the temperature inside the cabinet rises by more than 20°C within 10 seconds and exceeds the 120°C safety threshold.

7. The method for full-cycle control of arc faults in low-voltage distribution cabinets according to claim 1, characterized in that, In step S4, the security handling actions executed locally in closed loop by the edge computing unit include, in order of priority: a. Accurately disconnect the upstream circuit breaker of the faulty circuit, and immediately lock the closing circuit after the trip is completed to prevent accidental closing and power supply. b. Activate the highest level audible and visual alarm of the cabinet, lock the electromagnetic lock on the cabinet door, and prohibit unauthorized personnel from opening the door while it is powered on; c. Trigger the miniature automatic fire extinguishing device inside the cabinet, which will shut down unnecessary loads in adjacent circuits. d. Encrypt, lock, and cache the voltage / current waveform data for the 10 power frequency cycles before the fault and the 5 power frequency cycles after the fault, as well as the full sensor data and action execution records at the time of the fault; e. Send encrypted fault messages to the cloud platform through the highest priority channel, and use a multiple retransmission mechanism to ensure cloud reception.

8. The method for full-cycle control of arc faults in low-voltage distribution cabinets according to claim 1, characterized in that, In step S4, the cloud platform's coordinated processing actions include: The system features a pop-up window on the monitoring screen displaying fault information, a GIS map highlighting fault locations, locking the highest level of remote control access to the equipment, automatically pushing emergency notifications to the customer's responsible person, and automatically generating emergency repair work orders and assigning them to the pre-bound maintenance personnel.

9. The method for full-cycle control of arc faults in low-voltage distribution cabinets according to claim 1, characterized in that, In step S5, the full-process operation and maintenance management includes: Multi-channel tiered notification and timeout escalation mechanism: After a fault is triggered, a strong reminder is pushed to the first-level responsible person via APP / mini-program software, voice call, SMS, and WeChat notification. If no order is accepted within 3 minutes, it is pushed to the second-level responsible person. If no order is accepted within 8 minutes, it is pushed to the grid supervisor and emergency duty officer. If no order is accepted within 20 minutes, the company-level emergency plan is activated to force order dispatch. On-site QR code scanning / photo check-in and mandatory safety condition confirmation: After arriving on site, maintenance personnel take a photo to check in, then scan the equipment's unique QR code to check in. Only after checking and confirming that the power is off, the power is tested, the grounding wire is connected, and the safety protection is in place can they enter the maintenance process. Full traceability management of the maintenance process: During the maintenance process, photos of the fault location, information on damaged components, and maintenance records must be uploaded. Temporary closing tests require dual-person authorization verification. Work order review and customer confirmation: After the maintenance is completed, the insulation test and no-load / load operation test data are uploaded. After the technical supervisor reviews and approves the data, the work order is closed and the maintenance report is sent to the customer simultaneously.

10. The method for full-cycle control of arc faults in low-voltage distribution cabinets according to claim 1, characterized in that, In step S6, the fault cause classification includes four categories: design defects, production process problems, component quality problems, and improper customer use. Based on the classification results, a fault analysis report is generated and pushed to the R&D and production departments to optimize product structure, production process and component selection in a targeted manner. Based on archived fault data, an early warning model is trained using machine learning algorithms, and the feature recognition logic and trigger thresholds are continuously optimized.

Images