Power equipment fault maintenance and debugging system and method

Abstract

The invention relates to the technical field of power distribution equipment maintenance, in particular to a power equipment fault maintenance debugging system and method, and the system comprises a multi-modal data fusion acquisition module, a digital twin fault simulation positioning unit, a self-adaptive debugging execution module, a maintenance safety linkage control unit and a full-life-cycle data closed-loop management module. The method comprises the steps of multi-dimensional data acquisition, digital twin simulation positioning, adaptive strategy debugging, safety linkage monitoring and data closed-loop optimization. The problem of single traditional data is solved through multi-modal data fusion, 1: 1 simulation positioning of a fault scene is realized by combining digital twinning, the adaptive module can automatically match various fault debugging strategies, and the safety linkage unit interacts with a power grid dispatching system in real time through an IEC61850 protocol to realize automatic early warning / power failure when electrical parameters of a maintenance area exceed a threshold value. The data closed-loop module optimizes a strategy library based on historical overhaul data, so that the overhaul efficiency is improved, and the debugging period is shortened.

Description

Technical Field

[0001] This invention relates to the field of power distribution equipment maintenance technology, specifically to a power equipment fault diagnosis and repair system and method. Background Technology

[0002] In power system operation, power distribution equipment (such as switchgear, transformers, and circuit breakers) is a core component ensuring power transmission, and its operating status directly affects power supply stability. With the expansion of the power grid and the growth of electricity load, the failure risk of power distribution equipment is increasing year by year. Common failures include contact overheating, insulation aging, excessive partial discharge, and mechanical jamming. However, current power equipment fault diagnosis and repair technologies still have many shortcomings, making it difficult to meet the needs of efficient, accurate, and safe maintenance. Specific problems are as follows: First, traditional maintenance relies on manual experience, which is inefficient and prone to misjudgment. In the current maintenance model, maintenance personnel need to carry tools such as multimeters and infrared thermometers to the site to troubleshoot faults by "looking, listening, and measuring." For example, when detecting switchgear faults, the power must be cut off first, the cabinet must be disassembled, and the contact temperature must be measured point by point with an infrared thermometer. Then, based on experience, it can be determined whether there is poor contact. For transformer faults, oil samples need to be extracted and sent to the laboratory to test indicators such as dielectric loss and breakdown voltage. The whole process takes 24-72 hours and is affected by the experience of personnel, with a misjudgment rate as high as 15%-20%. For example, in one substation, maintenance personnel misjudged the transformer insulation as aging, which led to frequent tripping even after the insulation components were replaced. It was eventually found that the actual fault was poor contact of the tap changer, which not only caused economic losses but also caused a power outage in the area for more than 8 hours.

[0003] Secondly, the data acquisition dimensions are limited, resulting in insufficient fault location accuracy. Existing maintenance systems mostly collect only electrical parameters such as current and voltage, lacking simultaneous acquisition of key data such as equipment temperature field, partial discharge, and acoustic signatures. For example, partial discharge is an early sign of insulation aging, but traditional systems do not integrate partial discharge sensors, failing to capture this signal, leading to late-stage fault detection. Even systems integrating infrared temperature measurement can only acquire single-point temperature data, failing to form a complete temperature field distribution and making it difficult to locate hidden hot spots inside the cabinet (such as overheated busbar joints). Statistics from a power distribution network show that cases with fault location deviations exceeding 1 meter due to limited data acquisition account for 30%, requiring multiple disassemblies to find the actual fault point and significantly extending the debugging cycle.

[0004] Furthermore, the commissioning process lacks adaptive strategies and has a low degree of operational standardization. Different fault types require different commissioning procedures—for example, overheating contacts require cleaning and applying conductive grease, while mechanical jamming requires lubricating transmission components. However, the existing system has not established a mapping relationship between faults and commissioning strategies, relying entirely on maintenance personnel's memory of operating procedures, leading to non-standard operations. For instance, a power supply company experienced a secondary fault caused by maintenance personnel failing to tighten bolts to the standard torque during switchgear commissioning, resulting in loose bolts. In addition, the commissioning process requires manual switching between different tools (such as electric wrenches and insulated gloves), making the operation cumbersome. The average commissioning time for a single device exceeds 4 hours, making it difficult to cope with large-scale fault repair scenarios.

[0005] Meanwhile, the lack of coordination between maintenance safety monitoring and power grid dispatching poses safety hazards. Existing systems primarily rely on perimeter alarms within the maintenance area, failing to interact in real-time with the power grid dispatching system. For example, if power suddenly returns to the maintenance area (due to dispatching errors), on-site personnel may struggle to detect it promptly. Furthermore, the lack of real-time tracking of maintenance personnel's locations has led to incidents of personnel accidentally entering energized areas. A power safety report indicates that in 2023, electric shock accidents caused by insufficient coordination in maintenance safety accounted for 12% of all power safety accidents, seriously threatening human lives.

[0006] Finally, the lack of full lifecycle data management hinders preventative maintenance. Existing systems only record the results of the current maintenance, failing to store historical operating data, fault types, and debugging records. This makes it impossible to analyze equipment fault patterns and health trends—for example, a switchgear cabinet repeatedly experienced contact overheating faults, but due to the lack of historical data, it was impossible to determine whether the problem was due to material issues or installation processes. The system could only repeat the cycle of "fault-maintenance-fault," failing to address the root cause. Furthermore, the lack of optimization mechanisms for debugging strategies means that the same steps are used for the same fault each time, without adjusting operating parameters according to the equipment's aging (e.g., reducing tightening torque for older equipment to avoid component damage), leading to a gradual decline in debugging effectiveness.

[0007] In summary, current power equipment fault diagnosis and repair technologies have significant shortcomings in terms of efficiency, accuracy, standardization, safety, and data management. There is an urgent need for a multi-dimensional data fusion, precise positioning, automatic debugging, safety linkage, and data closed-loop repair system and method to solve the above-mentioned technical problems and meet the operation and maintenance needs of modern power grids. Summary of the Invention

[0008] This invention aims to address six major pain points in existing power equipment fault diagnosis and debugging technologies: heavy reliance on manual labor, limited data, inaccurate fault location, lack of standardized debugging, weak safety linkage, and lack of data closure. Specifically, these include: low maintenance efficiency and high misjudgment rate; large fault location deviations requiring multiple equipment disassemblies; lack of standardized mapping for debugging strategies, leading to non-standardized operations; insufficient linkage between safety monitoring and power grid dispatching, posing a risk of electric shock; lack of historical data management, hindering preventative maintenance; and the absence of a closed-loop data acquisition and processing system, preventing strategy optimization. This invention provides a power equipment fault diagnosis and debugging system and method.

[0009] The technical solution adopted by this invention to solve its technical problem is: a power equipment fault diagnosis and debugging system and method, comprising: A multimodal data fusion acquisition module is provided. This module connects to the power equipment via a hybrid wired (RS485 / Ethernet) and wireless (LoRa / Wi-Fi6) communication link. It integrates a current sensor (range 0-500A, accuracy 0.2 grade), a voltage sensor (range 0-300kV, accuracy 0.2 grade), an infrared thermal imager (temperature range -20℃-150℃, resolution 640×512), a partial discharge sensor (detection frequency 300kHz-2GHz), and a soundprint collector (sampling rate 48kHz). This module is used to simultaneously acquire the electrical parameters, temperature field, partial discharge quantity, and operating soundprint data of the power equipment. The digital twin fault simulation and location unit connects to the multimodal data fusion acquisition module via industrial Ethernet. It constructs a 1:1 three-dimensional model of the power equipment based on Unity3D, imports the equipment design parameters (material, structural dimensions) and real-time acquired data, and uses the finite element analysis algorithm to simulate fault conditions (such as poor contact and insulation aging). It outputs the fault location coordinates (location error ≤ 0.5m) and a fault cause analysis report. The adaptive debugging execution module is connected to the digital twin fault simulation and location unit through a PLC controller. It has a built-in fault-debugging strategy mapping library (containing debugging steps corresponding to 10+ types of faults, such as power off the contacts when they are overheating, cleaning the contacts and applying conductive paste). It can automatically call the strategy and drive the actuator (such as an electric wrench or a high-voltage grounding switch) to complete the debugging operation based on the location results. The maintenance safety linkage control unit is connected to the power grid dispatching system and the multimodal data fusion acquisition module via the IEC61850 protocol. It monitors the leakage current (threshold ≤30mA), ground voltage (threshold ≤50V), and personnel location (via UWB positioning tag) in the maintenance area in real time. When the parameters exceed the threshold or personnel accidentally enter the live area, it outputs an audible and visual warning and triggers the power grid dispatching system to cut off the power supply of the corresponding line. The full lifecycle data closed-loop management module stores multimodal acquisition data, fault location results, debugging records and equipment operation data through a cloud server. It uses a BP neural network algorithm to analyze historical data, optimize the matching accuracy of the fault-debugging strategy mapping library, and generate an equipment health assessment report (including remaining life prediction).

[0010] Specifically, the multimodal data fusion acquisition module has a built-in Kalman filter algorithm to eliminate high-frequency interference (such as harmonic interference) in current and voltage data, and the infrared thermal imager supports an autofocus function, which can adjust the focal length according to the device distance (0.5m-10m) to ensure temperature detection accuracy.

[0011] Specifically, the digital twin fault simulation and location unit also supports manual correction. When the simulation location result deviates from the actual fault location by more than 0.5m, maintenance personnel can input the actual location through the touch screen, and the unit will automatically update the model parameters to optimize the subsequent simulation accuracy.

[0012] Specifically, the fault-debugging strategy mapping library of the adaptive debugging execution module supports online updates. It can obtain the debugging strategy corresponding to new fault types (such as electronic component faults in new switch cabinets) from the cloud via 4G / 5G network, and upgrades can be completed without disassembling the module.

[0013] Specifically, the UWB positioning tag of the maintenance safety linkage control unit has a built-in emergency alarm button. When maintenance personnel encounter an emergency, pressing the button will trigger the safety linkage control unit to immediately cut off the power supply to the maintenance area and notify the dispatch center.

[0014] Specifically, the full lifecycle data closed-loop management module also supports data visualization functions, which can display the changing trends of equipment electrical parameters through line graphs and the temperature field distribution through heat maps, making it easier for maintenance personnel to intuitively analyze the equipment operating status.

[0015] A method for troubleshooting and debugging power equipment includes the following steps: S1: Multi-dimensional data acquisition and multi-modal data fusion acquisition module synchronously acquires current, voltage, temperature field, partial discharge quantity and acoustic data of power equipment through various sensors. After interference removal by Kalman filtering algorithm, it is transmitted to the digital twin fault simulation and location unit through hybrid communication link. S2: Digital twin simulation positioning. The digital twin fault simulation positioning unit imports the collected data into a 1:1 three-dimensional model, uses the finite element analysis algorithm to simulate the fault conditions, and outputs the fault location coordinates and cause report. If the deviation exceeds the threshold, it receives manual correction data to update the model. S3: Adaptive strategy debugging. The adaptive debugging execution module calls the corresponding debugging strategy from the mapping library according to the cause of the fault, drives the actuator to complete operations such as power-off, cleaning, and tightening, and feeds back the debugging process data to the digital twin unit in real time. S4: Safety linkage monitoring. The maintenance safety linkage control unit monitors leakage current, voltage to ground and personnel location in real time. When the threshold is exceeded, an alarm is triggered and the power is cut off. After debugging, the parameters are confirmed to be normal before the dispatch system is notified to restore power. S5: Data closed-loop optimization, the full life cycle data closed-loop management module stores the maintenance data, optimizes the mapping library through BP neural network, generates equipment health reports and pushes them to the power grid operation and maintenance platform.

[0016] Specifically, in step S1, the data acquisition frequency can be adjusted according to the equipment type. The acquisition frequency for switchgear is set to 1 time / second, and the acquisition frequency for transformer is set to 1 time / 5 seconds, in order to balance data real-time performance and communication bandwidth.

[0017] Specifically, in step S3, before driving the actuator, the adaptive debugging execution module will use an infrared thermal imager to detect the equipment temperature a second time. It will only perform the operation after confirming that the temperature is ≤40℃ (normal temperature environment) to avoid safety accidents caused by operation at high temperature.

[0018] Specifically, in step S5, the full lifecycle data closed-loop management module generates an equipment health assessment report once per quarter. For equipment with a health score of <60, preventive maintenance suggestions are automatically pushed to maintenance personnel.

[0019] The beneficial effects of this invention are: 1. Multimodal data fusion and acquisition: Solving the pain points of "single data and poor data quality" and strengthening the foundation for analysis. Breaking through the limitations of traditional maintenance methods that only collect electrical parameters, this module integrates multiple types of sensors, including current, voltage, infrared temperature field, partial discharge, and acoustic fingerprint, to simultaneously acquire comprehensive information on equipment operation. It can fully capture early characteristics of faults such as insulation aging and mechanical jamming, avoiding missed fault detection due to missing data (for example, partial discharge signals that traditional systems cannot identify can be accurately captured by this module, enabling early fault detection).

[0020] The built-in Kalman filter algorithm can eliminate high-frequency interference such as power grid harmonics. Combined with the data synchronous acquisition mechanism (with minimal timestamp error), it ensures the authenticity and consistency of the acquired data, providing high-quality data support for subsequent fault simulation and localization, and reducing analytical bias caused by data distortion.

[0021] 2. Digital Twin Simulation Positioning: Solves the pain points of "inaccurate positioning and frequent disassembly," improving investigation efficiency. Based on Unity 3D, a 1:1 3D model of the equipment is built, which strictly matches the equipment design parameters with the actual structure. Combined with the finite element analysis algorithm, the fault conditions are simulated, which can accurately restore the fault occurrence scenario and directly output the fault location and cause report. Unlike traditional maintenance, there is no need to repeatedly disassemble the equipment for troubleshooting, which greatly reduces equipment wear and tear (such as the switch cabinet can locate the internal contact overheating point without multiple disassemblies).

[0022] It supports manual correction. When there is a deviation between the simulation results and the actual results, the model parameters can be updated through touch operation to continuously optimize the simulation accuracy, ensure more accurate fault location in the future, avoid invalid maintenance caused by positioning deviation, and shorten the fault diagnosis cycle.

[0023] 3. Adaptive debugging execution: Solves the pain points of "reliance on manual labor and non-standard operation" and achieves standardized debugging. The built-in "fault-debugging strategy mapping library" can automatically call standardized debugging steps according to the cause of the fault (such as the "power off-clean-apply conductive grease-tighten" process corresponding to overheating of contacts), completely eliminating the reliance on the experience of maintenance personnel and avoiding secondary faults caused by human memory deviation and non-standard operation (such as loosening problems caused by improper bolt tightening torque in traditional maintenance).

[0024] It supports online updates of debugging strategies, and can obtain corresponding strategies for new faults (such as electronic component faults in intelligent switch cabinets) via wireless network. Upgrades can be completed without disassembling modules, flexibly adapting to different types and batches of power equipment, and improving the system's applicability to new equipment and faults.

[0025] 4. Safety coordination during maintenance: Addressing the pain points of "isolated safety monitoring and weak coordination," ensuring the safety of personnel and the power grid. By interacting with the power grid dispatching system in real time through the IEC 61850 standard protocol, it breaks through the limitations of traditional safety monitoring that "only alarms on-site and does not link with dispatching". It can monitor the risks of leakage current, voltage to ground, and personnel location in the maintenance area in real time. Once the threshold is exceeded, it will immediately trigger an audible and visual warning and automatic power cut-off, avoiding safety hazards such as "dispatch mistakenly powering on" and "personnel accidentally entering live areas".

[0026] UWB positioning tags have built-in emergency alarm buttons, which can be directly triggered by maintenance personnel in case of emergencies to send power outage and dispatch notifications, forming a complete safety closed loop of "real-time monitoring - risk warning - emergency response", which greatly reduces the incidence of electric shock and other safety accidents and provides comprehensive protection for the personal safety of maintenance personnel.

[0027] 5. Closed-loop data throughout the entire lifecycle: Solving the pain points of "fragmented data and lack of preventative maintenance" and promoting maintenance upgrades. By relying on cloud servers to store full-cycle maintenance data (collected data, location results, debugging records, and operation data), a unique "health record" for each device is formed, avoiding the problem of "data only being stored for the current time and no historical traceability" in traditional systems. This provides data support for fault pattern analysis and equipment aging trend judgment (such as identifying high-frequency fault points of a certain type of equipment through historical data).

[0028] By leveraging the BP neural network algorithm to analyze historical data, the matching accuracy of the fault-debugging strategy mapping library is continuously optimized, making subsequent maintenance strategies more closely aligned with the actual condition of the equipment (e.g., the tightening torque of aging equipment can be automatically adjusted to avoid component damage). At the same time, equipment health reports and preventive maintenance suggestions are generated, promoting the transformation of the operation and maintenance mode from "post-fault repair" to "proactive prevention," reducing the impact of sudden faults on power grid supply, and extending the service life of equipment.

[0029] 6. Overall Collaboration: Achieving "intelligent, standardized, and safe" maintenance, empowering power grid operation and maintenance. The five modules work collaboratively, forming a complete closed loop from data acquisition, fault location, debugging execution, safety management to data optimization, effectively addressing the core pain points of traditional maintenance: "low efficiency, high misjudgment, weak safety, and lack of prevention." Ultimately, it achieves intelligent (reduced manual intervention), standardized (unified debugging process), and safe (full-link risk management) fault repair of power equipment, significantly improving maintenance efficiency and operation and maintenance quality, ensuring the stability of power grid supply, and adapting to the operation and maintenance needs of modern power grids operating on a large scale and under high load. Attached Figure Description

[0030] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0031] Figure 1 The connection diagram of the power equipment fault diagnosis and debugging system module provided by the present invention; Figure 2 A schematic diagram of sensor installation for the multimodal data fusion acquisition module provided by this invention (KYN28-12 switch cabinet). Figure 3 The flowchart of the debugging steps of the adaptive debugging execution module provided by this invention; Figure 4 A flowchart illustrating the steps of the power equipment fault diagnosis and repair method provided by the present invention. Detailed Implementation

[0032] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0033] like Figures 1-4As shown, the present invention discloses a power equipment fault diagnosis and repair system and method. The system achieves full-process maintenance optimization through the collaborative work of five core modules, and the method forms standardized steps corresponding to the system modules, as follows: System Composition Multimodal data fusion acquisition module: This module is the core of data acquisition. It adopts a "wired + wireless" hybrid communication architecture (RS485 / Ethernet ensures high-bandwidth data transmission, LoRa / Wi-Fi6 adapts to complex field environments) and integrates five types of sensors: current sensor (selected with 0.2-level accuracy, range 0-500A, suitable for the current range of switchgear, transformers and other equipment), voltage sensor (0.2-level accuracy, range 0-300kV, covering 10kV-220kV power distribution equipment), infrared thermal imager (temperature measurement range -20℃-150℃, resolution 640×512, can generate 256-level grayscale temperature field image), partial discharge sensor (detection frequency 300kHz-2GHz, sensitivity ≤1pC, captures early insulation aging signals), and acoustic fingerprint collector (sampling rate 48kHz, 16-bit accuracy, identifies abnormal acoustic fingerprints of mechanical jamming). The module incorporates a Kalman filter algorithm, which can eliminate the interference of power grid harmonics (such as the 3rd and 5th harmonics) on electrical parameters and ensure data accuracy. It also supports synchronous data acquisition (timestamp error ≤ 1ms) to avoid analysis deviations caused by data asynchrony.

[0034] Digital Twin Fault Simulation and Location Unit: This unit constructs a 1:1 3D model of the power equipment based on the Unity3D engine. The model parameters strictly match the equipment design drawings (such as the cabinet dimensions, busbar spacing, and contact materials of the switchgear; the core structure and winding turns of the transformer), and import the equipment's factory parameters (such as rated current and insulation class). The unit receives real-time data from the multimodal acquisition module via industrial Ethernet and uses finite element analysis algorithms (such as the ANSYS Mechanical algorithm) to simulate fault conditions. For example, when the temperature of a switchgear contact reaches 80℃ (normal temperature ≤60℃ under normal conditions) and the partial discharge is >10pC, the algorithm simulates the Joule heating effect of current passing through the contact and the electric field distribution of the insulating medium, calculates the three-dimensional coordinates of the overheated area (X / Y / Z axis accuracy ≤0.3m), and generates a fault cause report (such as "contact oxidation leads to increased contact resistance, causing overheating and partial discharge"). In addition, the unit supports manual correction. Maintenance personnel can input the actual fault location through the 10.1-inch touch screen, and the unit will automatically update the material parameters in the model (such as correcting the thickness of the contact oxide layer) so that the subsequent simulation positioning error is ≤0.5m.

[0035] Adaptive Debugging Execution Module: This module uses a PLC controller (Siemens S7-1200 series, response time ≤100ms) as its core and has a built-in "fault-debugging strategy mapping library." The library stores standardized debugging steps for 10+ common faults (e.g., contact overheating: Step 1 - Power off and test for voltage; Step 2 - Disassemble the cabinet to expose the contacts; Step 3 - Clean the contact oxide layer with alcohol; Step 4 - Apply conductive grease (model: 8934); Step 5 - Tighten the bolts to the standard torque (25N·m); Step 6 - Second temperature measurement to confirm temperature ≤40℃). The module connects to the digital twin unit via an RS485 interface. After receiving the fault cause, it automatically matches the strategy and drives the actuators (electric wrench (torque accuracy ±2%), high-voltage grounding switch (opening / closing time ≤0.5s), and insulation cleaning robot) to complete the operation. Simultaneously, it collects real-time operating data of the actuators (e.g., wrench torque, switch opening / closing status) and feeds it back to the digital twin unit to verify the debugging effect. In addition, the mapping library supports online updates, and can obtain debugging strategies for new faults (such as electronic transformer faults in smart switch cabinets) from the cloud via 4G / 5G networks, so that upgrades can be completed without disassembling modules.

[0036] Maintenance Safety Linkage Control Unit: This unit adopts the IEC61850 protocol (standard communication protocol for power systems) and connects to the power grid dispatching system, multi-modal acquisition module, and UWB positioning system. The unit monitors three parameters in real time: leakage current in the maintenance area (threshold ≤30mA), voltage to ground (threshold ≤50V, ensuring personal safety), and personnel location (via UWB positioning tag, positioning accuracy ≤10cm). When a leakage current >30mA or a voltage to ground >50V is detected, the unit immediately outputs an audible and visual warning (warning light flashing frequency 2Hz, alarm sound pressure level ≥85dB) and sends an "emergency power outage request" to the power grid dispatching system. Upon receiving the request, the dispatching system cuts off the power supply to the corresponding line within 1 second. If the personnel's location coordinates are detected to enter a energized area (determined by pre-stored three-dimensional coordinates of the energized equipment area), the unit, in addition to triggering an early warning, will also send a vibration alert to the personnel's UWB tag. Furthermore, the UWB tag has a built-in emergency alarm button; pressing the button in an emergency will directly trigger a power outage, further enhancing safety.

[0037] The full lifecycle data closed-loop management module is built on Alibaba Cloud servers and uses a MySQL database to store data (including multimodal acquisition data, digital twin positioning results, debugging records, and equipment operation data). The data storage period is ≥10 years (covering the entire equipment lifecycle). The module incorporates a BP neural network algorithm to optimize the "fault-debugging strategy mapping library" by analyzing historical data. For example, if statistics show that a certain type of switchgear, after 5 years of operation, experiences contact overheating, the tightening torque needs to be reduced from 25 N·m to 20 N·m during debugging (to prevent breakage of aging components), the algorithm will automatically update the corresponding parameters in the mapping library. Simultaneously, the algorithm generates a health assessment report (out of 100 points; ≥80 points indicates normal, 60-80 points indicates attention, and <60 points indicates a warning) based on equipment operation data (such as the average number of annual faults and the growth trend of partial discharge), and predicts the remaining lifespan (with an error ≤1 year). Furthermore, the module supports data visualization, displaying line graphs (electrical parameter change trends), heat maps (temperature field distribution), and pie charts (fault type percentages) on the web interface, facilitating intuitive analysis of equipment status by maintenance personnel.

[0038] Inspection methods The method of the present invention is implemented based on the above system, and the specific steps are as follows: S1: Before multi-dimensional data acquisition and maintenance, the sensors of the multi-modal data fusion acquisition module are installed on the target equipment (e.g., the current sensor of the switchgear is connected in series in the busbar, the voltage sensor is connected in parallel at the incoming line, the infrared thermal imager is fixed at the observation window of the cabinet, the partial discharge sensor is adsorbed on the outer shell of the cabinet, and the acoustic fingerprint collector is installed on the inside of the cabinet door). The module adjusts the acquisition frequency according to the equipment type (1 time / second for switchgear, 1 time / 5 seconds for transformer), and simultaneously acquires current, voltage, temperature field, partial discharge quantity and acoustic fingerprint data; after acquisition, high-frequency interference (such as current fluctuations caused by power grid harmonics) is eliminated through the Kalman filter algorithm, and then transmitted to the digital twin fault simulation and location unit through a "wired + wireless" hybrid link, with a transmission delay of ≤500ms to ensure data real-time performance.

[0039] S2: Digital Twin Simulation Positioning. After receiving data, the digital twin unit imports it into a 1:1 three-dimensional model. For example, the collected data of the switchgear A-phase contact temperature of 85℃ and partial discharge of 15pC are imported into the model. The algorithm simulates the current distribution, temperature conduction, and electric field strength under this condition, outputting the fault location coordinates (X=1.2m, Y=0.8m, Z=2.5m) and a cause report ("A-phase contact oxidation, contact resistance increased to 0.5Ω (normal ≤0.1Ω), causing overheating and partial discharge"). Maintenance personnel carry a tablet computer (wirelessly connected to the digital twin unit) to the site and locate the fault point according to the coordinates. If the deviation between the actual location and the simulated location is found to be >0.5m (e.g., the actual location is X=1.3m, Y=0.8m, Z=2.5m), the actual coordinates are input through the tablet computer, and the unit automatically updates the contact oxide layer thickness parameter in the model (corrected from 0.1mm to 0.15mm) to optimize the subsequent simulation accuracy.

[0040] S3: Adaptive Strategy Debugging. After receiving the fault cause from the digital twin unit, the adaptive debugging execution module automatically calls the "contact overheating" debugging strategy from the mapping library: First, the module sends a "power-off request" to the maintenance safety linkage control unit. After the control unit confirms that the power grid dispatch system has cut off the power supply to the switchgear, it sends back a "power-off complete" signal. Second, the module drives the high-voltage grounding switch to close (ensuring equipment grounding and preventing induced electricity), and uses an infrared thermal imager to detect the equipment temperature a second time. After confirming that the temperature is ≤40℃, it drives the insulating cleaning robot into the cabinet to clean the oxide layer of the A-phase contacts with alcohol. Third, the robot applies conductive paste (0.5mm thick) and then drives the electric wrench to tighten the contact bolts with a torque of 25 N·m. Fourth, after debugging is completed, the module controls the grounding switch to open, and then uses sensors to detect the contact temperature (should be ≤40℃) and contact resistance (should be ≤0.1Ω). After confirming that the parameters are normal, the module feeds back the debugging process data (such as cleaning time, tightening torque, and test results) to the digital twin unit.

[0041] S4: Safety Linkage Monitoring. During S1-S3, the maintenance safety linkage control unit continuously monitors three types of parameters: leakage current collected by the multi-modal module (displayed in real-time on the control unit's touchscreen), voltage to ground (detected by a voltage sensor), and personnel location obtained through the UWB positioning system. If the leakage current is detected to rise to 35mA (exceeding the 30mA threshold), the control unit immediately triggers an audible and visual warning (flashing warning lights and sounding an alarm), and sends an "emergency power outage request" to the power grid dispatch system. The dispatch system cuts off the power supply to the line within 1 second. Simultaneously, the control unit sends a vibration alert to on-site personnel via a UWB tag, informing them to evacuate to a safe area. After the fault is resolved (e.g., fixing loose sensor wiring), and the control unit detects normal parameters, it then sends a "power restoration request" to the dispatch system, ensuring safety throughout the maintenance process.

[0042] S5: After the data closed-loop optimization maintenance is completed, the full lifecycle data closed-loop management module automatically collects the maintenance data (including the raw data of S1, the positioning results of S2, the debugging records of S3, and the safety monitoring data of S4) and stores it in the MySQL database. Then, it analyzes the data through a BP neural network algorithm—for example, by comparing the historical maintenance records of the switchgear, it finds that the growth rate of partial discharge due to contact overheating is faster than the last time (from 5pC to 8pC per year). The algorithm judges that the equipment insulation is aging faster, reduces its health score from 82 to 75, and adjusts the predicted remaining lifespan from 5 years to 3 years. At the same time, based on the result of "contact resistance meets the standard when tightening torque is 25 N·m" in this debugging, the algorithm further verifies the rationality of the mapping library parameters. If subsequent statistics show that multiple devices of the same model require the same torque under the same operating conditions, no parameter adjustment is needed; if there are differences, it will be automatically optimized. Finally, the module generates a maintenance report and an equipment health report, which are pushed to maintenance personnel via email to provide a basis for subsequent preventive maintenance.

[0043] To more clearly illustrate the technical solution of the present invention, the following uses "10kV high voltage switchgear fault repair" as an example to describe in detail the actual application process of the system and method. The switchgear model is KYN28-12, the service life is 5 years, and the on-site symptoms are "abnormal noise from the cabinet during operation and abnormal temperature rise in the cabinet as shown by infrared thermometer".

[0044] I. Preparations before implementation Equipment and personnel configuration: Prepare the system of this invention (including 1 set of multimodal data fusion acquisition module, 1 digital twin fault simulation and positioning unit (an industrial computer equipped with Unity3D software), 1 set of adaptive debugging execution module (including PLC controller, electric wrench, and insulated cleaning robot), 1 maintenance safety linkage control unit (including UWB positioning system, equipped with 2 UWB tags), and a full life cycle data closed-loop management module (the web terminal has been deployed to the operation and maintenance center)); configure 2 maintenance personnel (1 operator and 1 monitor), both wearing insulated clothing, insulated gloves, and UWB positioning tags.

[0045] On-site environment confirmation: The target switchgear is located in the 10kV distribution room of a 110kV substation. There is no strong electromagnetic interference (such as high-frequency equipment) in the vicinity. The indoor temperature is 25℃ and the humidity is 60%, which meets the requirements of the maintenance environment. It was confirmed in advance through the power grid dispatch system that the line where the switchgear is located can be temporarily shut down (the planned power outage time is 2 hours). A safety fence was set up around the switchgear and a "Maintenance work, no entry" sign was hung.

[0046] II. Implementation Steps (corresponding methods S1-S5) (I) S1: Multi-dimensional data collection (approximately 10 minutes) Sensor Installation: Maintenance personnel installed the sensors of the multimodal data fusion acquisition module into the KYN28-12 switchgear. Current sensor: A Hall current sensor with a range of 0-200A and an accuracy of 0.2 is selected and connected in series at the A-phase busbar inlet of the switchgear. It is fixed to the busbar bracket with bolts to ensure tight contact with the busbar and no loosening.

[0047] Voltage sensor: A capacitive voltage divider type voltage sensor with a range of 0-12kV and an accuracy of 0.2 is selected and connected in parallel between the A-phase bus and the grounding terminal. The sensor output terminal is connected to the module host through a shielded wire to avoid interference.

[0048] Infrared thermal imager: The FLIRA655sc model is selected, with a temperature measurement range of -20℃ to 150℃ and a resolution of 640×512. It is fixed to the observation window on the front of the switch cabinet with a bracket. The lens is aimed at the busbar connector area inside the cabinet. The focus is adjusted to make the image clear (it can be previewed on the display screen of the module host).

[0049] Partial discharge sensor: TEKTRONIX P6015A model is selected, with a detection frequency of 300kHz-2GHz and a sensitivity of 1pC. It is attached to the left side of the switch cabinet housing (near the busbar connector) by a magnetic chuck, ensuring good contact between the sensor and the housing without gaps.

[0050] Voiceprint collector: A microphone module with a sampling rate of 48kHz and 16-bit precision is selected and fixed to the inside of the switch cabinet door (near the circuit breaker) with Velcro to avoid collecting external ambient noise.

[0051] Data Acquisition and Transmission: After the sensor is installed, turn on the multimodal module host (select an industrial-grade host with IP65 protection rating, suitable for indoor environments), and set the acquisition frequency to 1 time / second (because the target equipment is a switch cabinet). The module synchronously acquires the following data: Current: Phase A current is approximately 80A (normal operating current, with no significant fluctuations).

[0052] Voltage: Phase A voltage is approximately 10.5kV (in line with the 10kV system voltage range).

[0053] Temperature field: The temperature in the A-phase busbar joint area of ​​the infrared thermal imager display cabinet reached 78℃ (the normal temperature under normal conditions is ≤60℃, indicating overheating), while the temperature in other areas was about 30℃.

[0054] Partial discharge level: The partial discharge sensor detected a discharge level of approximately 12 pC (normal ≤ 10 pC, exceeding the limit).

[0055] Voiceprint: The voiceprint collector detected an abnormal buzzing sound at a frequency of about 500 Hz (the normal voiceprint frequency is about 200 Hz, indicating mechanical vibration or poor contact).

[0056] Data preprocessing and transmission: The module processes the collected data through the Kalman filter algorithm to eliminate current fluctuations caused by the third harmonic of the power grid (the current is stabilized at 80A±0.5A after filtering); then the data is transmitted to the industrial computer of the digital twin fault simulation and location unit via Ethernet (there is a network cable interface near the switch cabinet). The transmission delay is about 300ms. The industrial computer screen displays various data and infrared thermal images in real time.

[0057] (II) S2: Digital twin simulation positioning (approximately 20 minutes) Model import and data matching: Open the Unity3D software on the industrial computer and load the 1:1 3D model of the KYN28-12 switchgear (this model is built according to the equipment design drawings and includes components such as cabinet, busbar, contacts, circuit breaker, etc., with a dimensional error ≤1mm); import the factory parameters of the switchgear (rated current 200A, insulation class Class A, contact material is copper); then import the real-time data collected by S1 (current 80A, voltage 10.5kV, A-phase busbar joint temperature 78℃, partial discharge quantity 12pC, acoustic frequency 500Hz) into the model, and realize the association between data and model components through the software interface (such as mapping temperature data to the busbar joint model, and partial discharge data to the insulation component model).

[0058] Fault Simulation and Localization: Run the finite element analysis algorithm (ANSYS Mechanical plugin) to simulate fault conditions: Temperature field simulation: The algorithm simulates the Joule heating effect when current passes through the bus joint. Combined with the thermal conductivity of the contact material (copper has a thermal conductivity of 386 W / (m·K)), the temperature distribution in the bus joint area is calculated. The simulation results show that the highest temperature is 79℃ (which is basically consistent with the collected 78℃). The overheated area is concentrated at the connection point between the A phase bus and the contact.

[0059] Partial discharge simulation: The algorithm simulates the electric field distribution of the insulating component (insulating bushing at the bus joint) and finds that the contact gap increases due to contact oxidation, and the electric field strength locally exceeds the standard (reaching 5kV / mm, exceeding the withstand field strength of 4kV / mm of the insulating bushing), triggering partial discharge. The discharge location coincides with the overheating location.

[0060] Acoustic print simulation: The algorithm simulates mechanical vibration caused by poor contact of the contactor, with a vibration frequency of about 500Hz (consistent with the frequency of the collected acoustic print) to verify the cause of the fault.

[0061] Output and Manual Correction: The digital twin unit outputs the fault location coordinates (with the lower left corner of the switchgear as the origin, X=1.5m, Y=0.6m, Z=2.2m), corresponding to the upper right connector of the A-phase busbar in the actual equipment; simultaneously, a fault cause report is generated: "Oxidation of the upper right connector of the A-phase busbar in the KYN28-12 switchgear, contact resistance increased to 0.3Ω (normal ≤0.1Ω), causing overheating (78℃), excessive partial discharge (12pC), and abnormal sound waves (500Hz)." Maintenance personnel, carrying a tablet computer (wirelessly connected to an industrial computer), went to the site, located the connector according to the coordinates, and used an infrared thermometer for secondary testing. The actual temperature was 77℃, and the position deviation was 0.1m (within the error range of ≤0.5m), requiring no manual correction, confirming the accurate positioning result.

[0062] (III) S3: Adaptive Strategy Debugging (approximately 40 minutes) Strategy Invocation and Power-Off Confirmation: The PLC controller (Siemens S7-1200) of the adaptive debugging execution module receives the fault cause report from the digital twin unit via the RS485 interface, and automatically invokes the "bus joint overheating (oxidation)" debugging strategy from the "Fault-Debugging Strategy Mapping Library". The strategy steps are displayed on the PLC's touch screen. Step 1: Request a power outage and test for voltage; Step 2: Close the grounding switch; Step 3: Clean the oxide layer on the connector; Step 4: Apply conductive paste; Step 5: Tighten the bolts; Step 6: Secondary detection.

[0063] The PLC controller sends a "power outage request" to the maintenance safety linkage control unit. The control unit then forwards the request to the power grid dispatch system via the IEC61850 protocol. After confirming that there is no important load on the line, the dispatch system cuts off the power supply and sends a "power outage complete" signal back to the control unit, which then forwards it to the PLC. Maintenance personnel use a voltage detector to check the incoming line of the switch cabinet. After confirming that there is no power, they send a "voltage detection complete" signal back to the PLC.

[0064] Grounding and cleaning operation: The PLC drives the high-voltage grounding switch (model: JN15-12) to close. The opening and closing time of the grounding switch is 0.4s. After closing, the grounding resistance is detected by the sensor to be ≤4Ω (compliant with safety standards). Then, the PLC controls the insulation cleaning robot (model: IRB120, load 2kg) to enter the switch cabinet (through the robot passage reserved in the cabinet door). The robot carries alcohol wipes and aims them at the upper right connector of the A phase busbar, wiping the oxide layer on the connector surface at a speed of 50rpm for 3 minutes. After wiping, the robot uses dry cotton wipes for secondary cleaning to ensure that there is no alcohol residue.

[0065] Conductive grease application and bolt tightening: The robot changes its tool head to a conductive grease applicator, evenly applying 8934 model conductive grease to the joint surface to a thickness of 0.5mm (the applicator stroke is controlled by PLC to ensure accuracy); then it switches to an electric wrench (torque accuracy ±2%), and tightens the bolts according to the mapping library parameters (this model of switchgear has been in operation for 5 years, and the tightening torque is adjusted to 20N·m). During the tightening process, the torque value is fed back to the PLC in real time to ensure that the torque is stable at 20N·m±0.4N·m.

[0066] Secondary testing and debugging confirmation: After tightening, the PLC drive grounding switch is disconnected, and then the sensor of the multi-modal module is controlled for secondary testing: Temperature: The infrared thermal imager shows that the connector temperature has dropped to 32℃ (close to the room temperature). Contact resistance: Tested using a dedicated resistance tester (integrated in the multimodal module), the contact resistance is 0.08Ω (≤0.1Ω, compliant with standards). Partial discharge level: The partial discharge sensor detected that the discharge level dropped to 3pC (≤10pC, normal). Voiceprint: The voiceprint collector did not capture any abnormal voiceprints; the frequency has been restored to 200Hz.

[0067] After the PLC confirms that all parameters are normal, it sends a "Debugging Complete" message to the digital twin unit. The digital twin unit then updates the connector parameters (contact resistance 0.08Ω) in the model, completing the debugging process.

[0068] (iv) S4: Safety linkage monitoring (throughout the entire process) During processes S1-S3, the maintenance safety linkage control unit continues to operate: Parameter monitoring: The control unit collects leakage current (always ≤10mA, below the threshold of 30mA) and ground voltage (≤5V after power failure) in the maintenance area in real time through the multi-modal module; the position of two maintenance personnel is monitored through the UWB positioning system, and the personnel are always within the safety fence (the pre-stored safety area coordinates are X=0-3m, Y=0-2m, Z=0-3m) and have not entered the live compartment (the live compartment coordinates are X=3-5m, Y=0-2m, Z=0-3m).

[0069] Emergency Situation Simulation Verification: To verify the safety linkage function, the monitoring personnel deliberately brought their hand close to the pre-stored boundary of the energized interval (X=2.9m, Y=1.0m, Z=1.5m). The control unit immediately triggered an audible and visual warning (the warning light flashed and the alarm sounded), and sent a vibration reminder to the personnel's UWB tag. When the personnel pressed the emergency alarm button on the UWB tag, the control unit sent an "emergency power failure request" to the dispatch system within 0.8s (at this time, the equipment was already powered off, and the dispatch system responded "the line was powered off"), verifying the effectiveness of the linkage function.

[0070] (V) S5: Data closed-loop optimization (takes approximately 10 minutes) Data storage: The full lifecycle data closed-loop management module collects maintenance data via a 4G network. S1 data: Current 80A, Voltage 10.5kV, Temperature 78℃, Partial discharge 12pC, Acoustic signature 500Hz; S2 data: Location coordinates (1.5, 0.6, 2.2), cause report; S3 data: debugging steps, tightening torque 20 N·m, secondary test parameters (temperature 32℃, resistance 0.08Ω, discharge 3pC); S4 data: leakage current ≤10mA, personnel location record.

[0071] The data is stored in a MySQL database and automatically associated with the unique device number of the switch cabinet (KYN28-12-20180501), with the storage timestamp accurate to the second.

[0072] Strategy optimization and health assessment: BP neural network algorithm analysis of the current data and the switchgear's historical maintenance records (commissioned in 2018, first contact overheating in 2020, second overheating in 2022) revealed: After 5 years of operation, the tightening torque for contact overheating debugging of this type of switchgear needs to be reduced from the initial 25 N·m to 20 N·m (the debugging torque was 25 N·m in 2020, 22 N·m in 2022, and 20 N·m this time, showing a downward trend). The algorithm automatically updates the tightening torque parameter of the "bus joint overheating (oxidation)" strategy in the "fault-debugging strategy mapping library" (set to 20 N·m for equipment that has been in operation for more than 5 years).

[0073] Based on the trend of partial discharge (5pC in 2020, 8pC in 2022, and 12pC this time), the algorithm predicts that the remaining life of the switchgear is 3 years (error ≤ 1 year) and the health assessment score is 72 points (60-80 points, note is needed).

[0074] Report generation and delivery: The module generates the "KYN28-12 switchgear maintenance report" and the "equipment health assessment report" and sends them to the email address of the person in charge of the operation and maintenance center. The report includes data visualization charts (such as a line graph comparing the temperature before and after this maintenance, and a 5-year partial discharge growth trend chart) to help the person in charge analyze the equipment status and plan the next preventive maintenance time (it is recommended to review it after 1 year).

[0075] III. Implementation Results

[0076] The maintenance took approximately 90 minutes (less than the planned 2 hours). After the maintenance, the KYN28-12 switchgear operated normally and remained trouble-free for three consecutive months. Based on data analysis, the key performance indicators for this maintenance are as follows: The fault location error is 0.1m (≤0.5m), and the positioning accuracy is significantly improved; The debugging time is 40 minutes (traditional manual debugging takes 4 hours), improving efficiency by 83%; No safety incidents occurred, and the safety monitoring function was effective. After data closed-loop optimization, the tightening torque parameters for subsequent maintenance of the same model of switchgear are more accurate, avoiding damage to components.

[0077] This embodiment fully verifies the feasibility and superiority of the system and method of the present invention, which can effectively solve the pain points of existing maintenance technologies and is applicable to the fault maintenance of 10kV-220kV power distribution equipment.

[0078] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of protection claimed by the present invention. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A power equipment fault commissioning system, characterized by, Comprising: A multi-modal data fusion acquisition module connected to the power equipment through a mixed wired and wireless communication link, integrating current sensors, voltage sensors, infrared thermal imagers, partial discharge sensors, and voiceprint collectors for synchronous acquisition of electrical parameters, temperature fields, partial discharge quantities, and operating voiceprint data of the power equipment; A digital twin fault simulation positioning unit connected to the multi-modal data fusion acquisition module through an industrial Ethernet, based on Unity3D to build a 1:1 three-dimensional model of the power equipment, import design parameters and real-time acquisition data of equipment material and structure size, use finite element analysis algorithm to simulate fault conditions, and output fault location coordinates and fault cause analysis report; An adaptive debugging execution module connected to the digital twin fault simulation positioning unit through a PLC controller, with a built-in fault-debugging strategy mapping library that can automatically call strategies and drive execution mechanisms to complete debugging operations according to positioning results; A maintenance safety linkage control unit connected to the power grid dispatching system and the multi-modal data fusion acquisition module through IEC61850 protocol, real-time monitoring of leakage current, ground voltage, and personnel position in the maintenance area, and outputting sound and light warnings and triggering the power grid dispatching system to cut off the power supply of the corresponding line when the parameters exceed the threshold or personnel mistakenly enter the live area; A full life cycle data closed-loop management module that stores multi-modal acquisition data, fault positioning results, debugging records, and equipment operation data through a cloud server, uses a BP neural network algorithm to analyze historical data, optimizes the matching accuracy of the fault-debugging strategy mapping library, and generates a device health assessment report.

2. The power equipment fault overhauling and debugging system according to claim 1, characterized in that: The multi-modal data fusion acquisition module has a Kalman filter algorithm built-in to eliminate high-frequency interference in current and voltage data, and the infrared thermal imager supports an automatic focusing function that can adjust the focal length according to the distance to the device to ensure temperature detection accuracy.

3. The power equipment fault overhauling and debugging system according to claim 1, characterized in that: The digital twin fault simulation positioning unit also supports manual correction function, when the simulation positioning result deviates from the actual fault position by >0.5m, the maintenance personnel can input the actual position through the touch screen, the unit automatically updates the model parameters to optimize the subsequent simulation accuracy.

4. The power equipment fault overhauling and debugging system according to claim 1, characterized in that: The fault-debugging strategy mapping library of the adaptive debugging execution module supports online updates, which can obtain the debugging strategy corresponding to the new fault type from the cloud through 4G / 5G network, without the need to disassemble the module to complete the upgrade.

5. The power equipment fault overhauling and debugging system according to claim 1, characterized in that: The UWB positioning tag of the maintenance safety linkage control unit has an emergency alarm button built-in, when the maintenance personnel encounter an emergency, pressing the button can trigger the safety linkage control unit to immediately cut off the power supply of the maintenance area and notify the dispatch center.

6. The power equipment fault overhauling and debugging system according to claim 1, characterized in that: The full life cycle data closed-loop management module also supports data visualization function, which can display the trend of equipment electrical parameters through line chart and temperature field distribution through heat map, making it easy for maintenance personnel to intuitively analyze the equipment operating status.

7. A method for power equipment fault maintenance commissioning, the method is implemented by using the power equipment fault maintenance commissioning system according to any one of claims 1-6, characterized in that, Comprising the following steps: S1: Multi-dimensional data acquisition, multi-modal data fusion acquisition module synchronously collects current, voltage, temperature field, partial discharge and voiceprint data of power equipment through various sensors, removes interference through Kalman filtering algorithm, and transmits to the digital twin fault simulation positioning unit through a hybrid communication link; S2: Digital twin simulation positioning, the digital twin fault simulation positioning unit imports the collected data into the 1:1 three-dimensional model, simulates the fault condition using the finite element analysis algorithm, outputs the fault location coordinates and cause report, and if the deviation exceeds the threshold, receives manual correction data to update the model; S3: Adaptive strategy debugging, the adaptive debugging execution module calls the corresponding debugging strategy from the mapping library according to the fault cause, drives the actuator to complete the power-off, cleaning, tightening and other operations, and feeds back the debugging process data to the digital twin unit in real time; S4: Safety linkage monitoring, the maintenance safety linkage control unit monitors the leakage current, voltage to ground and personnel position in real time, triggers an early warning and cuts off the power supply when the threshold is exceeded, and confirms that the parameters are normal before notifying the dispatching system to restore power supply; S5: Data closed-loop optimization, the full life cycle data closed-loop management module stores the maintenance data, optimizes the mapping library through BP neural network, generates a device health report and pushes it to the power grid operation and maintenance platform.

8. The method of claim 7, wherein: In step S1, the data acquisition frequency can be adjusted according to the type of equipment. The acquisition frequency of the switch cabinet is set to 1 time per second, and the acquisition frequency of the transformer is set to 1 time per 5 seconds, to balance the real-time performance of data and communication bandwidth.

9. The method of claim 7, wherein: In step S3, before driving the actuator, the adaptive debugging execution module will detect the equipment temperature again through the infrared thermal imager, and execute the operation after confirming that the temperature is ≤40℃, to avoid safety accidents caused by operation under high temperature.

10. The method of claim 7, wherein: In step S5, the full life cycle data closed-loop management module generates a device health evaluation report once every quarter. For devices with a health score <60, the module automatically pushes a preventive maintenance suggestion to the operation and maintenance personnel.

Images