Switch cabinet state prediction and maintenance system based on digital twinning

Abstract

The invention discloses a switch cabinet state prediction and maintenance system based on digital twinning, and relates to the technical field of power equipment state monitoring and maintenance, and the system comprises a data collection module which collects the operation and environment parameters of a switch cabinet in real time; the digital twin modeling module is used for building physical-virtual dynamic mapping by combining a three-dimensional structure and real-time data; the state prediction module is used for predicting performance degradation of key elements within 72 hours by using deep learning and generating early warning when the performance degradation exceeds a threshold value; the maintenance decision module is used for generating a differentiation scheme according to early warning and historical records, and calculating the cost-benefit ratio to select the optimal scheme; the execution control module drives the maintenance equipment to execute and feed back the progress; the data management module is used for storing full-life-cycle data in a distributed manner and supporting multi-dimensional retrieval; and the security monitoring module is used for monitoring communication and data security and disconnecting a connection and storing logs when the communication and data are abnormal. According to the invention, monitoring is comprehensive, prediction is accurate, an efficient maintenance scheme is generated according to fault priorities, data are safe and traceable, switch cabinet faults are reduced, and operation and maintenance cost is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power equipment state monitoring and maintenance, and particularly relates to a switch cabinet state prediction and maintenance system based on digital twinning. BACKGROUND

[0002] As the core equipment of the power distribution system, the running stability of the switch cabinet directly determines the reliability of the power supply. However, the traditional switch cabinet state monitoring and maintenance mode has significant technical defects. The traditional monitoring mainly relies on manual inspection, and the inspection cycle is usually set to 1-3 months, which is difficult to capture the dynamic changes of the equipment operation parameters in real time. For example, the bus joint temperature may rise above 60℃ in a short time during the load peak, exceeding the safety threshold, but it cannot be discovered in time within the manual inspection interval, which may cause local overheating and burnout. The vibration parameter monitoring relies more on the judgment of the operation and maintenance personnel, and cannot quantify the vibration frequency and amplitude. The hidden faults such as bolt loosening are often discovered after the equipment makes abnormal noise, which has caused a certain degree of mechanical damage. At the same time, the traditional monitoring only covers basic parameters such as temperature and current, ignoring the related influence of humidity, air pressure and component insulation resistance in the cabinet. In a high temperature and high humidity environment, the insulation resistance decay rate is accelerated, but the lack of comprehensive monitoring leads to frequent partial discharge faults. The occurrence rate of such faults in the rainy season is 3-4 times that in the traditional dry environment.

[0003] Although there have been attempts to apply early digital twinning technology in the field of switch cabinets, the model functions are single and the prediction accuracy is insufficient, which is difficult to support the actual maintenance needs. Most early digital twinning models only realize the static three-dimensional structure mapping of the switch cabinet, and cannot realize the real-time synchronization of dynamic parameters such as temperature field and electric field of the physical entity. The "disconnection" between the virtual model and the physical equipment leads to a large deviation between the prediction results and the actual running state. Even if some models have temperature field simulation functions, they do not consider the coupling effect of load fluctuation and environmental humidity on temperature distribution. For example, under the condition of 80% rated load and 90% humidity superposition, the deviation between the hot spot temperature predicted by the model and the actual value is often more than 10℃, which cannot provide reliable basis for fault prediction. In addition, the traditional fault prediction is mostly based on fixed empirical formula, without considering individual differences such as component material and operating environment.

[0004] The extensive maintenance decision and data management further exacerbate the inefficiency of switch cabinet operation and maintenance. The traditional maintenance scheme lacks a priority sorting mechanism. Regardless of the fault impact range and component importance, it is executed in the mode of "reporting first and processing first". For example, the main bus temperature anomaly and the slight insulation aging of the branch bus are treated equally, which easily causes delay in handling critical faults. The maintenance records are stored in paper or simple electronic documents, which has the risk of record loss and tampering. When the same fault occurs subsequently, the effectiveness of the historical maintenance measures cannot be traced, leading to repeated trial and error. At the same time, the calibration of sensors relies on manual periodic execution, and the calibration period is as long as 6 months. During this period, the sensor drift may cause a measurement error of more than 5℃, directly affecting the credibility of the monitoring data. The digital twin model is prone to "mismatch" after long-term operation, that is, the deviation between the virtual parameters and the physical entity exceeds the threshold. However, due to the lack of automatic detection and calibration mechanism, the model gradually loses its guiding significance and finally returns to the traditional maintenance mode. These problems together lead to high operation and maintenance cost, high fault rate and low maintenance efficiency of the traditional switch cabinet, which is difficult to meet the demand of modern power distribution system for intelligent and refined operation and maintenance. SUMMARY

[0005] The switch cabinet state prediction and maintenance system based on digital twinning proposed by the present application solves the problems mentioned in the prior art.

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme: a switch cabinet state prediction and maintenance system based on digital twinning, comprising: A data acquisition module acquires the operating state parameters and environmental parameters of the switch cabinet. This module is arranged with temperature, vibration and air pressure sensors in the busbar room, cable room and circuit breaker room of the switch cabinet, and with temperature and humidity sensors on the outer wall of the cabinet body. The collected data is transmitted to the system hub through industrial Ethernet. A digital twinning modeling module constructs a dynamic mapping between the physical entity and the virtual model based on the three-dimensional structure parameters of the switch cabinet and the real-time data of the data acquisition module. The virtual model synchronizes the temperature field distribution, vibration mode and air pressure change of the physical entity in real time, and realizes dynamic update of 30 frames per second through lightweight rendering technology. A state prediction module analyzes the historical data and real-time data output by the digital twinning model using a deep learning algorithm, predicts the performance degradation trend of the key components of the switch cabinet within the next 72 hours, and generates a warning signal when the predicted value exceeds the preset threshold. A maintenance decision module generates a differentiated maintenance scheme based on the warning signal of the state prediction module and the historical maintenance records. For emergency warning, the shutdown maintenance strategy is adopted, and for general warning, the live repair strategy is adopted. At the same time, the cost-benefit ratio of the maintenance scheme is calculated, and the scheme with the highest cost performance is selected preferentially. An execution control module receives instructions from the maintenance decision module and drives the maintenance equipment to perform operations, the module feeds back the operation execution progress in real time, suspends the execution and issues a prompt when the operation deviation exceeds the allowed range; A data management module stores the full life cycle data of the switch cabinet, realizes data query and backup using a distributed database architecture, and supports retrieval by component type, time interval, and fault type; A security monitoring module monitors the communication state and data transmission security of the system, automatically cuts off abnormal connections and saves logs when unauthorized access or data tampering is detected, and classifies the parameter modification authority of the digital twin model, allowing only authorized personnel to configure and adjust.

[0007] Further, it also includes a state health assessment unit, which quantifies the performance state by calculating the health index of key components. The calculation method of the health index is H=1-(a1×T deviation+a2×F fault+a3×D degradation) / 3, where H is the health index, T deviation is the deviation rate of the measured temperature from the rated value, F fault is the ratio of the number of historical faults to the total number of operations, and D degradation is the ratio of the current value of the insulation resistance to the initial value. The health index is recalculated every 5 minutes, and when H is less than 0.6, it is determined to be a sub-healthy state, triggering preventive maintenance; when H is less than 0.3, it is determined to be a fault risk state, triggering emergency maintenance.

[0008] Further, the state prediction module introduces a mechanical wear accumulation coefficient when predicting the mechanical life of the circuit breaker, which is calculated as W=k1×N operation×(P load / P rated)×(T environment / 25), where W is the mechanical wear accumulation coefficient, k1 is the material wear coefficient, N operation is the total number of circuit breaker opening and closing, P load is the actual operating load, P rated is the rated load, and T environment is the average temperature in the cabinet. The traditional life prediction model is corrected by W, and the predicted value is compared with the actual operation state of the circuit breaker every 3 months. If the deviation exceeds 3%, the k1 value is recalibrated.

[0009] Further, the digital twin modeling module also includes a multi-physical field coupling analysis unit, which simulates the temperature field, electric field and stress field distribution of the switch cabinet under different loads and environmental conditions. The triggering condition for coupling analysis is that the load exceeds 80% of the rated value, the environmental humidity exceeds 85% RH, or the vibration frequency suddenly changes by more than 5Hz. The simulation results are displayed in real time in the virtual model in the form of color cloud map, and the simulation data is exported as Excel report, including the parameter values and risk assessment results of each key point.

[0010] Further, the maintenance decision module adopts a dynamic priority ranking mechanism when generating the maintenance scheme, comprehensively considers the element importance, fault influence range and maintenance cost, the importance coefficient is classified according to the influence of the element on the power distribution system; the influence range coefficient is calculated according to the number of circuits that may be affected by the fault, and the cost coefficient is calculated by the formula C coefficient = 1- (actual maintenance cost / budget maintenance cost), the actual maintenance cost includes spare parts cost, labor cost and power loss, and the budget maintenance cost is the average cost of the same type of maintenance preset by the system, the comprehensive priority = (importance coefficient x 0.5) + (influence range coefficient x 0.3) + (cost coefficient x 0.2), the priority value range is 1-5, when multiple faults are simultaneously warned, the system generates a maintenance order according to the comprehensive priority.

[0011] Further, the data acquisition module adopts a sensor self-calibration mechanism, which automatically calibrates the temperature sensor and vibration sensor every 24 hours. During the calibration process of the temperature sensor, the digital twin model calculates the theoretical temperature value of each monitoring point based on the heat conduction equation, and compares the measured value with the theoretical value. The calibration of the vibration sensor is based on the correlation between the natural frequency of the element and the operating load, and the theoretical amplitude and frequency of vibration are calculated and compared with the measured data of the sensor. The calibration data and correction record are kept for 1 year, supporting the query of historical calibration trend by sensor number. When the calibration deviation of a certain sensor exceeds 3% for 3 times in a row, a replacement warning is issued 60 days in advance.

[0012] Further, the execution control module and the maintenance equipment adopt wireless mesh network communication, and the communication protocol selects IEEE802.11s. The automatic inspection robot is equipped with a high-definition camera and an infrared thermal imager. When bolt loosening or element overheating is detected, the position coordinates and fault type are transmitted to the execution control module in real time. The execution control module optimizes the allocation of maintenance equipment according to the fault location. The tightening tool adopts an electric torque wrench, which sets the target torque according to the bolt specification pre-stored in the digital twin model. The torque sensor transmits the actual torque value to the execution control module every 0.1 second. When the deviation exceeds 3%, the output power is automatically adjusted. After the operation is completed, the tightening tool feeds back the actual torque value and operation time to the system. The digital twin model marks the corresponding bolt position as "maintained" and updates the maintenance record.

[0013] Further, the state prediction module further comprises an environmental adaptability correction unit, when the humidity outside the cabinet exceeds 85% or the temperature is lower than -5 DEG C, the prediction result of the insulation resistance is corrected, and the corrected insulation resistance prediction value is: R correction = R prediction * (1-0.01* (H environment - 85)) * (1-0.02* (-5-T environment )), wherein R correction is the corrected insulation resistance prediction value, R prediction is the initial prediction value, H environment is the humidity outside the cabinet, T environment is the temperature outside the cabinet, the corrected prediction value is compared with the insulation resistance measured value every hour, if the deviation exceeds 10%, the correction coefficient is adjusted again, when H environment exceeds 95% or T environment is lower than -15 DEG C, the cabinet dehumidification device or heating device is started in advance, the humidity in the cabinet is controlled to be below 75%, the temperature is controlled to be above 5 DEG C, the insulation resistance decay rate is reduced from the source, and the environmental intervention measure is associated with the corrected prediction result.

[0014] Further, the data management module stores key maintenance records by adopting a blockchain technology, each record comprises a maintenance time, an operator ID, a replacement spare part model, element state parameters before and after maintenance and effect evaluation, distributed nodes of the blockchain are deployed on a local server and a cloud platform of the system, PoA consensus mechanism is adopted between the nodes, a smart contract predefines verification rules of the maintenance records, when the records are uploaded, it is automatically verified whether all mandatory fields are included, data query permissions are classified according to roles, the retention period of the maintenance records is consistent with the life cycle of the switch cabinet, when the switch cabinet is scrapped, the related records are automatically archived to a blockchain archiving node, the integrity of the maintenance records is automatically audited every quarter through the smart contract, and if the missing rate exceeds 2%, an audit early warning is sent to the administrator.

[0015] Further, the safety monitoring module further comprises an anomaly detection unit of a digital twin model, the unit identifies anomalies by comparing the deviation of key parameters of the virtual model and the physical entity, anomaly detection is performed every 10 minutes, when the deviation of a certain parameter exceeds a threshold value for two times in succession, it is determined that the model is mismatched, a model calibration program is automatically started, the calibration program firstly selects data in the last 24 hours and eliminates abnormal data, then parameters of the digital twin model are retrained based on the effective data, the training process adopts a gradient descent method, and the deviation between the model output and the physical entity parameter is reduced to within the threshold value until the calibration is completed, after the calibration is completed, the system continuously monitors the parameter deviation for three times, if all are less than or equal to the threshold value, it is determined that the calibration is successful, otherwise, an artificial intervention process is started, and a technical personnel is informed to check three-dimensional structure parameters of model construction.

[0016] Compared with the prior art, the present application has the following beneficial effects: In terms of condition monitoring, the system achieves comprehensive coverage and deep correlation of switchgear operating parameters through multi-sensor layout and digital twin multiphysics field coupling analysis. It not only collects basic parameters such as temperature, vibration and humidity in real time, but also simulates the coupled distribution of temperature field, electric field and stress field, intuitively presenting the superposition relationship between hot spot area and high risk area, avoiding the failure omission caused by traditional single parameter monitoring. For example, bus joint failure with superposition of high temperature and high stress is easily overlooked in the traditional mode, but this system can identify and warn in advance, greatly reducing the probability of hidden faults.

[0017] The improved accuracy of condition prediction is one of the system's core advantages. The system quantifies component status through a health assessment unit, and combines mechanical wear accumulation coefficients with environmental adaptability corrections to fully consider individual differences in component materials, operating loads, and ambient temperature and humidity, avoiding the coarse predictions of traditional empirical formulas. At the same time, the model is regularly compared and calibrated with actual operating conditions to ensure long-term stability and reliability of prediction results. For example, the prediction of circuit breaker mechanical life is no longer affected by deviations caused by materials and the environment, providing a precise basis for selecting maintenance timing, avoiding resource waste caused by over-maintenance, and preventing sudden failures caused by insufficient maintenance.

[0018] The intelligent decision-making and resource optimization capabilities of the system significantly improve operational efficiency. The system employs a dynamic priority ranking mechanism, allocating resources based on component importance, fault impact, and maintenance costs. This ensures that faults in critical components such as main busbars and circuit breakers are handled with priority, avoiding delays caused by traditional "disordered maintenance." Simultaneously, it generates differentiated solutions including operation steps and spare parts lists, reducing reliance on the experience of maintenance personnel. Even novices can efficiently perform maintenance operations, significantly shortening fault handling time.

[0019] In terms of data management and system reliability, the application of blockchain technology ensures that maintenance records are tamper-proof and traceable, solving the problems of lost or tampered records in traditional systems, and facilitating subsequent fault analysis and responsibility determination. The sensor self-calibration and digital twin model anomaly detection mechanism can ensure the reliability of data and models without frequent manual intervention, reducing human error and lowering the workload of maintenance personnel.

[0020] Overall, this invention achieves intelligent management of the entire process of switchgear "monitoring-prediction-decision-maintenance" through digital twin technology. This not only improves the stability and reliability of equipment operation and extends the service life of components, but also significantly reduces operation and maintenance costs and fault handling time. It is adapted to the development needs of unmanned and refined operation and maintenance in modern power distribution systems, and provides a scalable solution for intelligent management of power equipment. Attached Figure Description

[0021] Figure 1A schematic block diagram of a switch cabinet state prediction and maintenance system based on digital twinning proposed by the present application is shown in the figure; Figure 2 A comparison line graph of breaker health prediction accuracy under different operating times is shown in the figure; Figure 3 A comparison column graph of insulation resistance prediction error under different environmental humidities is shown in the figure; Figure 4 A comparison line graph of long-term operation parameter mismatch rate of the digital twinning model is shown in the figure; Figure 5 A comparison column graph of maintenance response time for different fault types is shown in the figure. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0023] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise" and the like indicate the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application.

[0024] In addition, the terms "first", "second" are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited. In addition, the terms "mounting", "connecting", "connection" should be broadly understood, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium, or it can be the communication between two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances, and the present application will be further described in detail with reference to the drawings.

[0025] REFERENCE Figures 1 to 5A switch cabinet state prediction and maintenance system based on digital twinning, comprising: A data acquisition module for real-time acquisition of operating state parameters and environmental parameters of the switch cabinet, the operating state parameters including busbar temperature, circuit breaker opening and closing time, cable joint temperature, cabinet internal air pressure and vibration frequency, and the environmental parameters including cabinet internal humidity, cabinet external temperature and humidity and atmospheric pressure; the module is provided with temperature sensors, vibration sensors and air pressure sensors in the busbar room, cable room and circuit breaker room of the switch cabinet, and a temperature and humidity sensor on the outer wall of the cabinet body, with a sampling frequency of 10 Hz, and the collected data is transmitted to the system hub through an industrial Ethernet; A digital twinning modeling module for constructing dynamic mapping of physical entities and virtual models based on three-dimensional structure parameters of the switch cabinet and real-time data of the data acquisition module, the three-dimensional structure parameters including cabinet body size, component layout and connection relationship, the virtual model synchronizing the temperature field distribution, vibration mode and air pressure change of the physical entity in real time, and dynamic updating of 30 frames per second being realized through lightweight rendering technology, with support for detail magnification and parameter query of local structure; A state prediction module for analyzing historical data and real-time data output by the digital twinning model using a deep learning algorithm to predict the performance degradation trend of key components of the switch cabinet within the next 72 hours, the key components including circuit breakers, busbar joints and insulation support components, the prediction content including temperature anomaly probability, mechanical operation response delay time and insulation resistance decay amplitude, and a warning signal being generated when the prediction value exceeds a preset threshold; A maintenance decision module for generating a differentiated maintenance plan including maintenance timing, operation steps and required spare parts list based on the warning signal of the state prediction module and historical maintenance records, adopting a shutdown maintenance strategy for emergency warnings and an on-line repair strategy for general warnings, and simultaneously calculating the cost-benefit ratio of the maintenance plan to preferentially select the plan with the highest cost performance; An execution control module for receiving instructions of the maintenance decision module and driving maintenance equipment to perform operations, the maintenance equipment including an automatic inspection robot, an on-line cleaning device and a fastening tool, the module being capable of real-time feedback of operation execution progress and pausing execution and issuing a prompt when operation deviation exceeds the allowable range; A data management module for storing full life cycle data of the switch cabinet, including design parameters, manufacturing information, operation records, maintenance records and scrap information, realizing fast query and backup of data using a distributed database architecture, and supporting multi-dimensional retrieval according to component type, time interval and fault type; A safety monitoring module for monitoring the communication state and data transmission security of the system, automatically cutting off abnormal connections and saving logs when unauthorized access or data tampering is detected, and performing hierarchical management of parameter modification authority of the digital twinning model, allowing only authorized personnel to configure and adjust.

[0026] In the application, a state health degree evaluation unit is also included, which quantifies the performance state by calculating the health degree index of the key element, and the calculation method of the health degree index is H=1-(a1×T deviation+a2×F failure+a3×D degradation) / 3, wherein H is the health degree index, the value range is 0-1, a1, a2 and a3 are weight coefficients and the sum is 1, T deviation is the deviation rate of the temperature measured value and the rated value, F failure is the ratio of the historical failure times and the total running times, and D degradation is the ratio of the current value and the initial value of the insulation resistance. The weight coefficients a1, a2 and a3 are dynamically adjusted according to the element type, the circuit breaker is significantly affected by temperature on mechanical life, a1 is 0.4, a2 is 0.3, and a3 is 0.3; the bus joint is easy to cause large-area power failure due to failure, a1 is 0.3, a2 is 0.4, and a3 is 0.3; the insulation support directly affects the insulation performance due to degradation, a1 is 0.3, a2 is 0.3, and a3 is 0.4. The health degree index is recalculated every 5 minutes, when H is less than 0.6, it is determined that the state is sub-health, and preventive maintenance is triggered, and the specific measures include live cleaning of the element surface dust, detection of the insulation resistance and recording of the data; when H is less than 0.3, it is determined that the state is a failure risk state, and emergency maintenance is triggered, and the process is that the system first sends a load transfer request to the power monitoring system, locks the switch cabinet operation permission after the load is cut off, and pushes a maintenance work order including spare part models (such as adaptive insulation pads and fastening bolts) to the operation and maintenance terminal, to ensure that the maintenance personnel carry the correct tools to the scene.

[0027] In the application, when predicting the mechanical life of the circuit breaker, the mechanical wear cumulative coefficient is introduced, and the calculation method of the mechanical wear cumulative coefficient is W=k1×N operation×(P load / P rated)×(T environment / 25), wherein W is the mechanical wear cumulative coefficient, k1 is the material wear coefficient, N operation is the total number of circuit breaker opening and closing, P load is the actual running load, P rated is the rated load, and T environment is the average temperature in the cabinet. The material wear coefficient k1 is subdivided according to the contact material of the circuit breaker, the copper alloy contact has medium wear resistance, k1 is 0.8, the silver alloy contact has balanced conductivity and wear resistance, k1 is 1.0, and the tungsten alloy contact has strong high-temperature wear resistance, k1 is 1.2; N operation is counted according to normal opening and closing and failure opening and closing, the failure opening and closing (such as short-circuit tripping) causes intensified wear, and the single counting is calculated according to 2 times of normal opening and closing; P load is the average load for 30 consecutive minutes, to avoid calculation deviation caused by instantaneous peak value; T environment is the 24-hour average temperature, when the temperature exceeds 40℃, the value of T environment is additionally increased by 2℃ for every 5℃ increase, to simulate the accelerated wear of mechanical parts caused by high temperature. The coefficient corrects the traditional life prediction model, so that the prediction error of the mechanical life of the circuit breaker is controlled within 5%, and the predicted value is compared with the actual operation state (such as the change of opening and closing response time) of the circuit breaker every 3 months, if the deviation exceeds 3%, the k1 value is recalibrated to ensure long-term stability of the prediction accuracy.

[0028] In the present application, the digital twin modeling module further includes a multi-physical field coupling analysis unit that can simulate the temperature field, electric field and stress field distribution of the switch cabinet under different loads and environmental conditions. When simulating the temperature field, combine the component power consumption data (such as busbar loss P = I²R, I is the actual current, R is the busbar resistance) and the cabinet body ventilation and heat dissipation path, calculate the temperature distribution under different ventilation hole wind speeds (0.5-3m / s) through the finite element method, and focus on marking the hot spot area where the temperature exceeds 60℃; the electric field simulation focuses on the busbar joint, insulation support and other parts prone to local discharge, calculates the electric field strength distribution, when the electric field strength exceeds 3kV / mm, automatically marks the local discharge risk level (1-5 levels), and predicts the discharge starting time; stress field simulation for cabinet structure and component fixed parts, considering the coupling effect of vibration load (frequency 10-50Hz, amplitude 0.1-0.5mm) and temperature deformation, calculate the stress value of the bolt fastening part, when the stress exceeds 80% of the material yield strength, predict the bolt loosening time. The triggering condition of coupling analysis is that the load exceeds 80% of the rated value, the environmental humidity exceeds 85%RH or the vibration frequency mutation exceeds 5Hz, the simulation results are displayed in real time in the form of color cloud chart in the virtual model, support the superposition analysis of temperature field and electric field, stress field, after superposition, can directly view the coincidence degree of hot spot area and high stress area, provide multi-dimensional basis for maintenance decision, at the same time, the simulation data can be exported as Excel report, including the parameter value and risk assessment result of each key point.

[0029] In the application, the maintenance decision module adopts a dynamic priority ranking mechanism when generating a maintenance scheme, and comprehensively considers the element importance, fault influence range and maintenance cost. The importance coefficient is classified according to the influence of the element on the power distribution system, the main bus has an importance coefficient of 5 (the highest level) because it carries the overall current, the branch bus has an importance coefficient of 3, the circuit breaker has an importance coefficient of 4, and the insulation support has an importance coefficient of 2; the influence range coefficient is calculated according to the number of loops that may be affected by the fault, and when 1 loop is affected, the coefficient is 1.0, when 2-5 loops are affected, the coefficient is 1.2, and when more than 5 loops are affected, the coefficient is 1.5; the cost coefficient is calculated by the formula C coefficient = 1- (actual maintenance cost / budget maintenance cost), the actual maintenance cost includes spare parts cost, labor cost and power loss, and the budget maintenance cost is the average cost of the same type of maintenance preset by the system. The comprehensive priority = (importance coefficient x 0.5) + (influence range coefficient x 0.3) + (cost coefficient x 0.2), the priority value range is 1-5, the higher the value, the higher the maintenance priority. When multiple faults are simultaneously warned, the system generates a maintenance order according to the comprehensive priority, for example, the main bus temperature anomaly (priority 4.8) is prior to the branch bus insulation aging (priority 3.2); if the maintenance resources (such as operation and maintenance personnel, equipment) are insufficient, the maintenance tasks with priority above 3 are preferentially allocated, and at the same time, the low-priority tasks are re-evaluated every hour for fault deterioration, and if the importance coefficient is improved (such as the number of loops affected by the branch bus fault increases), the priority is adjusted to ensure that critical faults are not missed.

[0030] In the application, the data acquisition module adopts a sensor self-calibration mechanism, which automatically calibrates the temperature sensor and the vibration sensor every 24 hours. During the calibration process of the temperature sensor, the digital twin model calculates the theoretical temperature value of each monitoring point based on the heat conduction equation (Q = cmΔT, Q is the heat dissipation power of the element, c is the specific heat capacity of the material, m is the mass of the element, and ΔT is the temperature change), and compares the measured value with the theoretical value; the calibration of the vibration sensor is based on the correlation between the natural frequency of the element (such as the natural frequency of the bus being 15-25 Hz and the natural frequency of the circuit breaker being 20-30 Hz) and the operating load, and the theoretical amplitude and frequency of vibration are calculated and compared with the measured data of the sensor. When the deviation between the measured value and the theoretical value of the sensor is within 2-5℃ (temperature) or 3%-5% (vibration), the system automatically generates a correction coefficient (correction value = theoretical value - measured value) to real-time correct the subsequent collected data; when the deviation exceeds 5℃ (temperature) or 10% (vibration), the sensor is determined to be abnormal, and the system automatically switches to the standby sensor of the corresponding monitoring point, and pushes a sensor replacement work order to the operation and maintenance platform, the work order includes the sensor model (such as temperature sensor PT100 and vibration sensor IEPE type), installation location and replacement step diagram. The calibration data and correction records are kept for 1 year, supporting historical calibration trend query by sensor number, when a certain sensor has a calibration deviation of more than 3% for 3 times in a row, a replacement warning is issued 60 days in advance to avoid data interruption caused by sudden failure.

[0031] In the present application, the execution control module and the maintenance equipment use wireless mesh network communication, the communication protocol is selected as IEEE802.11s, the transmission rate is ≥100Mbps, and the communication delay is ≤10ms, which ensures the real-time performance of multi-device collaborative operation. The automatic inspection robot is equipped with a high-definition camera and an infrared thermal imager. When the bolt loosening (the bolt angle deviation based on image recognition exceeds 15°) or the component overheating (the temperature exceeds 70℃) is detected, the position coordinates (based on the internal preset two-dimensional code positioning of the switch cabinet, the accuracy is ±5mm) and the fault type are transmitted to the execution control module in real time. The execution control module optimizes the distribution of maintenance equipment according to the fault position, for example, the bolt loosening in different areas of the same cabinet body. One tightening tool is dispatched to process in turn according to the path planning (shortest path algorithm), avoiding repeated movement of multiple devices. The tightening tool uses an electric torque wrench. The target torque is set according to the bolt specifications (such as M12 bolt rated torque 35N・m, M10 bolt 20N・m) pre-stored in the digital twin model. The torque sensor transmits the actual torque value to the execution control module every 0.1 second. When the deviation exceeds 3%, the output power is automatically adjusted to ensure that the final torque error does not exceed ±5%. After the operation is completed, the tightening tool feeds back the actual torque value and the operation time to the system. The digital twin model marks the corresponding bolt position as “maintained” and updates the maintenance record. The whole process does not require manual intervention, and the maintenance efficiency is improved by more than 4 times compared with the traditional manual method.

[0032] In the application, the state prediction module further comprises an environmental adaptability correction unit, when the humidity outside the cabinet exceeds 85% or the temperature is lower than -5 DEG C, the prediction result of the insulation resistance is corrected, the corrected insulation resistance prediction value is R correction =R prediction x(1-0.01x(H environment-85))x(1-0.02x(-5-T environment)), wherein R correction is the corrected insulation resistance prediction value, R prediction is the initial prediction value, H environment is the humidity outside the cabinet, and T environment is the temperature outside the cabinet. The humidity correction adopts a segmented coefficient, the coefficient is 0.01 when H environment is 85%-90%, the coefficient is promoted to 0.015 when H environment is 90%-95%, and the coefficient is 0.02 when H environment is above 95%, the accelerated weakening of moisture on the insulation performance in a high-humidity environment is simulated; the temperature correction is also segmented, the coefficient is 0.02 when T environment is -5 DEG C to -10 DEG C, the coefficient is 0.03 when T environment is -10 DEG C to -15 DEG C, and the coefficient is 0.04 when T environment is below -15 DEG C, the characteristics of insulation material embrittlement and resistance reduction caused by low temperature are considered. The corrected prediction value is compared with the measured value of the insulation resistance (obtained by the insulation detection sensor of the data acquisition module) every hour, if the deviation exceeds 10%, the correction coefficient is adjusted again; when H environment exceeds 95% or T environment is lower than -15 DEG C, the cabinet dehumidification device or heating device is started in advance, the humidity in the cabinet is controlled below 75%, and the temperature is controlled above 5 DEG C, the insulation resistance decay rate is reduced from the source, and the environmental intervention measures are associated with the corrected prediction result, forming a closed-loop control of "prediction-intervention-re-prediction".

[0033] In the application, the data management module stores key maintenance records by using a blockchain technology, each record contains maintenance time, operator ID, replacement spare part model, element state parameters (such as temperature and insulation resistance) before and after maintenance and effect evaluation (qualified / needs to be improved). The distributed nodes of the blockchain are deployed on a local server (3 nodes) and a cloud platform (2 nodes) of the system, a PoA (proof of authority) consensus mechanism is used between the nodes, the block generation time is less than or equal to 10 seconds, and the data synchronization efficiency is ensured; the verification rules of the maintenance records are preset in the smart contract, when the records are uploaded, it is automatically verified whether all the required fields are included, if the "state parameters after maintenance" are missing, the uploading request is returned and the complete information is required to be supplemented; after the records are uploaded, a unique hash value is generated and is associated with the hash value of the previous block, so that the data cannot be tampered. The data query authority is classified according to the roles, the operation and maintenance personnel can only query the switch cabinet maintenance records responsible by themselves, the department administrator can query all the records in the jurisdiction area, and the system administrator has the full query authority; the retention period of the maintenance records is consistent with the life cycle of the switch cabinet (usually 15-20 years), when the switch cabinet is scrapped, the related records are automatically archived to the blockchain archiving node, and long-term tracing is supported. At the same time, the integrity of the maintenance records is audited automatically by the smart contract every quarter, if the missing rate exceeds 2%, an audit warning is sent to the administrator, and the continuity and reliability of the maintenance data are ensured.

[0034] In the present application, the safety monitoring module further comprises an anomaly detection unit of the digital twin model, which identifies anomalies by comparing the key parameter deviations of the virtual model and the physical entity. The deviation threshold values of different elements and parameters are differentiated: the circuit breaker temperature deviation threshold value is 4℃, the busbar temperature deviation threshold value is 5℃, the circuit breaker vibration frequency deviation threshold value is 1.5Hz, the cabinet vibration frequency deviation threshold value is 2Hz, the busbar room air pressure deviation threshold value is 4kPa, and the cable room air pressure deviation threshold value is 5kPa. Anomaly detection is performed every 10 minutes. When the deviation of a certain parameter exceeds the threshold value for two consecutive times, it is determined that the model is mismatched and the model calibration program is automatically started. The calibration program first selects the data of the last 24 hours, eliminates abnormal data such as sensor failure and transient interference (such as parameter mutation caused by lightning), and retains valid data (accounting for ≥80%); then the parameters of the digital twin model (such as the thermal conductivity coefficient of the temperature field and the damping coefficient of the vibration mode) are retrained based on the valid data. The training process uses the gradient descent method, and the iteration number is ≥100 times until the deviation between the model output and the physical entity parameter is reduced to within the threshold value; after calibration, the system continuously monitors the parameter deviation for three times, and if they are all ≤threshold value, it is determined that the calibration is successful, otherwise the manual intervention process is started, and the technical personnel are notified to check the three-dimensional structure parameters of the model construction (such as whether the element size and position are consistent with the actual situation), to ensure that the virtual model and the physical entity are long-term high-precision mapping.

[0035] The specific implementation of the system is further illustrated by two embodiments as follows: Embodiment 1: Industrial park 10kV high-voltage switch cabinet (large load fluctuation, environmental temperature -5℃~45℃, dust concentration 0.5~1mg / m³) 1. System module refinement configuration and operation process (1) Data acquisition module For the 10kV high-voltage switch cabinet (model KYN28-12, a total of 8 units, including busbar room, circuit breaker room, and cable room) in the industrial park, the data acquisition module adopts a "multi-region redundant layout": one PT100 temperature sensor (precision ±0.5℃, range -20℃~150℃) is arranged at the top and middle of the busbar room to monitor the busbar temperature; an IEPE vibration sensor (range ±50g, frequency response 0~5kHz) is installed on the side wall of the circuit breaker room to collect the opening and closing vibration frequency; an SHT31 temperature and humidity sensor (precision ±2%RH, range 0~100%RH) and an MPX5100 air pressure sensor (precision ±1kPa, range 0~100kPa) are arranged at the bottom of the cable room; a dust sensor (model PMS5003, range 0~1000μg / m³) is installed on the outer wall of the cabinet. The sampling frequency of all sensors is 10Hz, and the data is transmitted to the Siemens S7-1500 PLC through Profinet industrial Ethernet with a data transmission delay of ≤50ms. Typical acquisition data: busbar temperature 42℃ (rated maximum 65℃), circuit breaker opening and closing time 0.08s (rated ≤0.1s), cabinet humidity 65%RH, cabinet temperature 32℃, dust concentration 0.8mg / m³, vibration frequency 18Hz.

[0036] (2) Digital twin modeling module (including multi-physical field coupling) A three-dimensional model of the switch cabinet is established based on SolidWorks, with a cabinet size of 1800mm×800mm×600mm and component layout restored according to actual installation position (busbar spacing 120mm, circuit breaker distance from busbar 300mm). The multi-physical field coupling analysis unit starts under the following conditions: load ≥80% rated value (1000A), humidity ≥80%RH, or vibration frequency mutation ≥5Hz. During temperature field simulation, the busbar loss P=I²R=800²×0.005Ω=3200W (I=800A, R=0.005Ω), the air speed through the vent is 1.2m / s, and finite element calculation shows that the busbar joint temperature is 58℃ (hot spot area), the cabinet temperature in other areas is 35~45℃, and the simulation results are marked with a red cloud map; the electric field simulation focuses on the busbar joint (spacing 20mm), and the calculated electric field strength is 2.8kV / mm (not exceeding the threshold value of 3kV / mm), with a partial discharge risk level of 2; the stress field simulation focuses on the bolt fastening position (M12 bolt, torque 35N・m), with a vibration load frequency of 20Hz and an amplitude of 0.3mm, and the calculated stress value is 220MPa (material yield strength 280MPa, not exceeding the threshold value of 80%). The coupling analysis results are updated every 15 minutes, the virtual model is rendered by Unity with a dynamic synchronization of 30 frames / s, and the real-time values of temperature, electric field, and stress can be viewed by double-clicking on the busbar joint.

[0037] (3) State prediction module (including formula application) Health assessment: the health of the circuit breaker (silver alloy contact) is calculated, a1=0.4, a2=0.3, a3=0.3; T deviation=(42℃-35℃) / 35℃=0.2 (rated temperature 35℃), F fault=2 times / 500 times=0.004 (2 times of fault opening and closing in 500 times of operation), D degradation=500MΩ / 1000MΩ=0.5 (initial insulation resistance 1000MΩ); Substitute the formula H=1-(0.4×0.2+0.3×0.004+0.3×0.5) / 3=1-(0.08+0.0012+0.15) / 3=1-0.2312 / 3≈0.923 (health status).

[0038] Circuit breaker mechanical life prediction: k1=1.0 (silver alloy contact), N operation=500 times (including 2 times of fault opening and closing, counted as 4 times, total 502 times), P load=800A, P rated=1000A, T environment=35℃ (24 hours average, not exceeding 40℃); Substitute the formula W=1.0×502×(800 / 1000)×(35 / 25)=502×0.8×1.4=562.24. The traditional model predicts the life of 10000 times, and after W correction, it predicts 10500 times. Compared with the opening and closing response time (0.08s→0.085s) after 6 months of actual operation, the deviation is 2.3% (≤5%).

[0039] Environmental adaptability correction: humidity outside the cabinet in rainy season is 90% (not exceeding 85%), no correction is needed; T environment=-3℃ in winter (not lower than-5℃), insulation resistance prediction value R prediction=800MΩ, no correction is needed; If the humidity is 92% and the temperature is-6℃, then R correction=800×(1-0.01×(92-85))×(1-0.02×(-5+6))=800×0.93×0.98≈729MΩ.

[0040] (4) Maintenance decision and execution control Maintenance decision: the temperature of the main bus is 58℃ (close to the threshold of 60℃), the importance coefficient is 5, the influence range coefficient is 1.5 (affecting 3 circuits), the maintenance cost is 800 yuan (budget 1000 yuan, C coefficient=1-800 / 1000=0.2); Comprehensive priority=5×0.5+1.5×0.3+0.2×0.2=2.5+0.45+0.04=2.99 (priority level 3), generate a live cleaning plan, the spare parts list includes insulation brush, dust collection device, and the maintenance time is scheduled at the next day's low load (2:00-4:00).

[0041] Execution Control: The automatic inspection robot (model AGV-02) receives instructions via a wireless mesh network, locates the bus joint (QR code accuracy ±3mm), and verifies the temperature at 57℃ using an onboard infrared thermal imager; the live cleaning device (3000r / min speed) is activated, and the cleaning time is 10 minutes; during execution, the torque sensor provides real-time feedback, and the cleaning brush pressure deviation is ≤5%. After completion, the virtual model updates the bus joint's "maintained" status, and the data management module uploads the record to the blockchain (5 nodes synchronized, hash value 0x7a2f...).

[0042] (5) Data management and security monitoring The data management module uses a MySQL distributed database to store the design parameters (material: Q235 steel), manufacturing information (manufactured in May 2023), operation records (1 record every 10 seconds), and maintenance records (stored on a blockchain) for eight switchgear units. The security monitoring module detects an unauthorized IP address attempting to access the system, automatically disconnects the connection, and saves the log. The digital twin model is verified every 10 minutes; the virtual busbar temperature of 58℃ deviates from the measured 57℃ by 1℃ (≤4℃ threshold), indicating no abnormalities. The sensors self-calibrate every 24 hours; the theoretical temperature sensor value is 42℃, and the measured temperature is 41.8℃, a deviation of 0.2℃ (≤2℃), requiring no correction.

[0043] 2. Performance Comparison Data Table 1: Comparison of Operation and Maintenance Performance of 10kV Switchgear in Industrial Parks Evaluation index Traditional maintenance system Invention system Differential core reason Breaker life prediction error ±22% ±2.3% Mechanical wear coefficient W correction Hot spot fault discovery time 4.5 hours 15 minutes Multi-physical field coupling monitoring Maintenance resource waste rate 35% 8% Dynamic priority ranking Sensor data deviation rate ±5.8% ±0.5% Automatic self-calibration mechanism Annual fault occurrence rate 9.2% 1.1% Full-process predictive maintenance Table 1 Table 1 is based on one year of operational data from 8 switchgear units. Traditional systems do not consider the influence of contact material and load, resulting in a circuit breaker lifespan prediction error of ±22%, and hotspot faults requiring 4.5 hours of manual inspection to detect. This invention, through W-correction and multi-physics monitoring, reduces the error to ±2.3%, and faults are located within 15 minutes. Traditional maintenance, due to disordered resource allocation, results in 35% waste of spare parts and manpower; this invention's dynamic prioritization reduces the waste rate to only 8%. Sensor self-calibration reduces data deviation from ±5.8% to ±0.5%, and the annual failure rate from 9.2% to 1.1%, significantly adapting to the operational needs of industrial parks with large load fluctuations and complex environments, reducing downtime losses (single downtime losses exceeding 100,000 yuan).

[0044] Example 2: 0.4kV low-voltage switchgear for urban substation (rainy and humid environment, stable load, ambient humidity 60%~95%RH) 1. Detailed configuration and operation process of system modules (1) Data acquisition module (including sensor self-calibration) Urban distribution station 0.4 kV low-voltage switch cabinet (model GGD, 12 units, for residential areas), the data acquisition module focuses on environmental parameter monitoring: 2 SHT35 temperature and humidity sensors (master-slave redundancy) are arranged in each area (busbar room, cable room) of the cabinet, a BME280 composite sensor (temperature and humidity + air pressure) is installed outside the cabinet, a DS18B20 temperature sensor is installed in the circuit breaker room (precision ±0.2℃), and a vibration sensor (model ADXL345) is arranged at the bottom of the cabinet. The sampling frequency is 10 Hz, and the data is transmitted to the gateway through LoRa (distance 500 m, transmission rate 50 kbps). Typical data: bus temperature 38℃, circuit breaker opening and closing time 0.06s, cabinet humidity 88%RH (rainy season), cabinet temperature 25℃, air pressure 100.8kPa, vibration frequency 12Hz, insulation resistance 600MΩ. The sensor self-calibration is performed every 24 hours: the temperature sensor theoretical value is 38℃ (digital twin calculation), the measured value is 37.9℃, the deviation is 0.1℃ (≤2℃); the vibration sensor theoretical frequency is 12Hz, the measured value is 12.1Hz, the deviation is 0.8% (≤5%), no correction is needed; the humidity deviation of the main sensor is 6%RH, the slave sensor is automatically switched, and the Kalman filter fusion deviation is 1.2%RH.

[0045] (2) Digital twin modeling and multi-physical field coupling The size of the three-dimensional model is 2200mmx1000mmx800mm, and the element layout is designed according to the power distribution demand of the residential area (busbar load current 2000A, circuit breaker 8-way outgoing line). The multi-physical field coupling trigger condition is: humidity ≥85%RH, air pressure ≤99kPa or load ≥70% rated value (1400A). Temperature field simulation: busbar loss P=1400²x0.002Ω=3920W, ventilation hole wind speed 0.8m / s, calculated cable joint temperature 45℃ (no hot spot); electric field simulation insulation support (spacing 50mm), electric field strength 1.2kV / mm (safety threshold 2kV / mm); stress field simulation cabinet door hinge, vibration frequency 15Hz, amplitude 0.2mm, stress value 180MPa (yield strength 235MPa). Additional simulation of the effect of humidity on insulation in a humid environment, when the humidity in the cabinet is 88%RH, the moisture absorption of the insulation support surface is calculated to be 0.3g / m², and the electric field strength increases to 1.5kV / mm. The virtual model uses a blue cloud to mark the moisture absorption area.

[0046] (3) State prediction module Insulation resistance correction: humidity outside the cabinet is 92% and temperature is 25℃ (not lower than -5℃) in the rainy season, R predicted = 500 MΩ; substitute into the formula R corrected = 500 × (1-0.01 × (92-85)) × (1-0.02 × (-5-25)) = 500 × 0.93 × 1.6 = 500 × 1.488 = 744 MΩ correction: temperature is not lower than -5℃, the second term coefficient is 1, so R corrected = 500 × (1-0.01 × 7) = 500 × 0.93 = 465 MΩ, which deviates from the actual value of 450 MΩ by 3.3% (≤10%).

[0047] Health assessment: insulation support a1 = 0.3, a2 = 0.3, a3 = 0.4; T deviation = (38℃-30℃) / 30℃≈0.267, F failure = 0 / 800 = 0, D degradation = 450 MΩ / 1000 MΩ = 0.45; H = 1-(0.3 × 0.267 + 0.3 × 0 + 0.4 × 0.45) / 3 = 1-(0.0801 + 0 + 0.18) / 3≈1-0.2601 / 3≈0.913 (healthy state).

[0048] Mechanical wear prediction: low-voltage circuit breaker (copper alloy contact k1 = 0.8), N operation = 800 times (no failure opening and closing), P load = 1400 A, P rated = 2000 A, T environment = 25℃; W = 0.8 × 800 × (1400 / 2000) × (25 / 25) = 0.8 × 800 × 0.7 × 1 = 448, the corrected life prediction is 12000 times, which deviates from the actual operation by 2.1%.

[0049] (4) Maintenance decision and execution control Maintenance decision: insulation support humidity 88%RH (risk level 2), importance coefficient 2, influence range coefficient 1.0 (1 way out), maintenance cost 500 yuan (budget 600 yuan, C coefficient = 1-500 / 600≈0.167); comprehensive priority = 2 × 0.5 + 1.0 × 0.3 + 0.167 × 0.2 = 1.0 + 0.3 + 0.0334 = 1.333 (priority level 1), generate dehumidification maintenance scheme, start cabinet dehumidification device (model TE-05, power 120W), maintenance time is set at low load (0:00-2:00).

[0050] Execution control: automatic inspection robot (model SLAM-01) locates insulation support, dehumidification device wind speed is adjusted to 1.5 m / s, humidity in the cabinet is reduced to 65%RH after 2 hours of operation; wireless mesh network communication delay is 8ms during execution, torque tool error is ±3% when tightening bolts, data is uploaded to blockchain (3 local nodes + 2 cloud nodes) after completion, recording maintenance time, humidity value before and after dehumidification.

[0051] (5) Safety monitoring and model calibration The safety monitoring module detected a virtual model cable joint temperature of 45℃, while the actual measured temperature was 43℃, a deviation of 2℃ (≤5℃ threshold), indicating no abnormality. After a rainstorm, the virtual air pressure value of the model was 100kPa, while the actual measured value was 99.5kPa, a deviation of 0.5kPa (≤5kPa). Automatic calibration was initiated: 24-hour valid data (92% effectiveness) was filtered, and the air pressure field parameters (atmospheric pressure correction coefficient) were retrained. After 120 iterations, the deviation decreased to 0.2kPa. Blockchain maintenance record audits are performed quarterly, with a record missing rate of 0.5% (≤2%) for the 12 switchgear units, showing no signs of tampering.

[0052] 2. Performance Comparison Data Table 2: Comparison of Operation and Maintenance Performance of 0.4kV Switchgear in Urban Substations Evaluation index Traditional maintenance system Invention system Differential core reason Insulation resistance prediction error ±18% ±3.3% Environmental adaptability correction Moisture environment fault occurrence rate 12.5% 0.8% Multi-physical field humidity coupling monitoring Sensor calibration manual cost 800 yuan / month 50 yuan / month Automatic self-calibration mechanism Model long-term mismatch rate 28% 1.2% Automatic calibration and anomaly detection Maintenance work order completion efficiency 4 hours / single 1.5 hours / single Automatic execution and path optimization Table 2 Table 2 is based on one year of operational data from 12 switchgear units. Traditional systems do not consider the impact of humidity on insulation, resulting in a prediction error of ±18% and a failure rate of 12.5% ​​during humid seasons. This invention, through environmental correction and humidity-coupled monitoring, reduces the error to ±3.3% and the failure rate to only 0.8%. Traditional sensor calibration requires monthly on-site visits by personnel, costing 800 yuan / month; this invention's automatic calibration costs only 50 yuan / month. The long-term model mismatch rate is reduced from 28% to 1.2%, and maintenance work order completion efficiency is improved by 62.5%, adapting to the needs of urban substations in rainy and humid environments requiring efficient operation and maintenance, ensuring reliable power supply for residents.

[0053] refer to Figure 2 This figure visually demonstrates the long-term stability of the health assessment method of this invention. Traditional systems calculate health based solely on a single temperature parameter, neglecting factors such as the number of faults and insulation degradation. As operating time increases, parameter drift causes accuracy to drop from 95% in one month to 65% in 12 months, making it difficult to reliably guide maintenance in the long term. This invention uses the health formula H=1-(a1×T deviation + a2×F fault + a3×D degradation) / 3, dynamically adjusting weighting coefficients to suit component types (e.g., circuit breaker a1=0.4, a2=0.3, a3=0.3), and combining this with periodic calibration, ensuring that accuracy remains stable within the 94%-98% range. Even after 12 months of operation, accuracy still reaches 94%, avoiding the "long-term prediction inaccuracy" problem of traditional systems and providing a continuous and reliable quantitative basis for the full life-cycle maintenance of circuit breakers.

[0054] refer to Figure 3This figure highlights the advantages of environmental adaptability correction of the present application. The traditional system does not consider the influence of humidity on insulation performance. As humidity increases, the insulation resistance decays rapidly, and the prediction error increases from ±3% at 60% RH to ±22% at 95% RH, which easily leads to missed judgment of insulation failure in high humidity environment. The present application uses the environmental correction formula R correction = R prediction × (1-0.01×(H environment -85))×(1-0.02×(-5-T environment) to adjust the humidity correction coefficient (0.01 for 85%-90% RH, 0.02 for above 95% RH), so that even in a 95% RH high humidity environment, the error is only ±5%.

[0055] Reference Figure 4 This figure verifies the effectiveness of the model anomaly detection and calibration of the present application. The traditional digital twin model lacks an automatic calibration mechanism. After long-term operation, the deviation between virtual parameters and physical entities continues to expand, with a 6-month mismatch rate of 20% (exceeding the 10% warning threshold) and a 10-month mismatch rate of 35%. The model gradually loses its guiding significance. The present application uses the anomaly detection unit (trigger calibration when the temperature deviation exceeds 4°C and the vibration frequency deviation exceeds 2Hz) to check the parameters every 10 minutes and automatically retrain the model (iteration ≥100 times) with 24 hours of valid data when the deviation exceeds the threshold, so that the mismatch rate is always controlled within the range of 2%-6%, and the 10-month mismatch rate is only 6%, which is much lower than the warning threshold, avoiding the "running failure" problem of traditional models.

[0056] Reference Figure 5 This chart clearly shows the efficiency advantages of dynamic priority sorting of the present application. The traditional system has no fault priority mechanism. Regardless of the impact range, it is processed in the order of reporting. The response time of the main bus overheating is as long as 60 minutes, which easily leads to large-scale power outages. The present application uses a comprehensive priority formula (importance × 0.5 + impact range × 0.3 + cost × 0.2). The main bus has the highest priority with an importance coefficient of 5 and an impact range coefficient of 1.5, and the response time is only 15 minutes. The branch bus (importance 3) has the second priority, and the response time is 30 minutes, which ensures quick processing of critical faults and reasonable allocation of resources. The response time of the cabinet vibration (importance 1) is similar to that of the traditional system (35 minutes), avoiding resource waste, improving maintenance efficiency, and reducing power outages caused by faults.

[0057] The above is only a preferred embodiment of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can make equivalent substitutions or changes within the technical scope disclosed by the present application according to the technical solutions and inventive concepts of the present application, which should be covered within the protection scope of the present application.

Claims

1. A switchgear status prediction and maintenance system based on digital twins, characterized in that, include: The data acquisition module collects the operating status parameters and environmental parameters of the switchgear. Temperature, vibration and air pressure sensors are installed in the busbar compartment, cable compartment and circuit breaker compartment of the switchgear, respectively, and temperature and humidity sensors are installed on the outer wall of the cabinet. The collected data is transmitted to the system central hub via industrial Ethernet. The digital twin modeling module, based on the three-dimensional structural parameters of the switch cabinet and the real-time data from the data acquisition module, constructs a dynamic mapping between the physical entity and the virtual model. The virtual model synchronizes the temperature field distribution, vibration modes, and air pressure changes of the physical entity in real time, and achieves dynamic updates of 30 frames per second through lightweight rendering technology. The status prediction module uses deep learning algorithms to analyze historical and real-time data output by the digital twin model to predict the performance degradation trend of key components of the switchgear in the next 72 hours. When the predicted value exceeds the preset threshold, an early warning signal is generated. The maintenance decision module generates differentiated maintenance plans based on the early warning signals from the status prediction module and historical maintenance records. For emergency early warnings, a shutdown maintenance strategy is adopted, and for general early warnings, a live-line maintenance strategy is adopted. At the same time, the cost-benefit ratio of the maintenance plan is calculated, and the plan with the highest cost-effectiveness is selected first. The execution control module receives instructions from the maintenance decision module and drives the maintenance equipment to perform operations. This module provides real-time feedback on the operation execution progress. When the operation deviation exceeds the allowable range, the execution is paused and a prompt is issued. The data management module stores the entire lifecycle data of the switchgear, and adopts a distributed database architecture to realize data query and backup. It supports retrieval by component type, time interval and fault type. The security monitoring module monitors the system's communication status and data transmission security. When unauthorized access or data tampering is detected, it automatically disconnects the abnormal connection and saves the log. At the same time, it implements hierarchical management of the parameter modification permissions of the digital twin model, allowing only authorized personnel to make configuration adjustments.

2. The switchgear status prediction and maintenance system based on digital twins according to claim 1, characterized in that, It also includes a health assessment unit, which quantifies the performance status by calculating the health index of key components. The health index is calculated as H = 1 - (a1 × T deviation + a2 × F failure + a3 × D degradation) / 3, where H is the health index, T deviation is the deviation rate between the measured temperature value and the rated value, F failure is the ratio of the number of historical failures to the total number of operations, and D degradation is the ratio of the current value of insulation resistance to the initial value. The health index is recalculated every 5 minutes. When H is less than 0.6, it is judged to be in a sub-healthy state, triggering preventive maintenance. When H is less than 0.3, it is determined to be a fault risk state, triggering emergency maintenance.

3. The switchgear status prediction and maintenance system based on digital twins according to claim 1, characterized in that, The state prediction module introduces a mechanical wear accumulation coefficient when predicting the mechanical life of the circuit breaker. The calculation method is W=k1×Noperation×(PLoad / PRated)×(TEnvironment / 25), where W is the mechanical wear accumulation coefficient, k1 is the material wear coefficient, Noperation is the total number of circuit breaker opening and closing operations, PLoad is the actual operating load, PRated is the rated load, and TEnvironment is the average temperature inside the cabinet. The traditional life prediction model is corrected by W. At the same time, the predicted value is compared with the actual operating state of the circuit breaker every 3 months. If the deviation exceeds 3%, the k1 value is recalibrated.

4. The switchgear status prediction and maintenance system based on digital twins according to claim 1, characterized in that, The digital twin modeling module also includes a multiphysics coupling analysis unit, which simulates the temperature, electric and stress field distributions of the switchgear under different loads and environmental conditions. The triggering conditions for coupling analysis are that the load exceeds 80% of the rated value, the ambient humidity exceeds 85%RH, or the vibration frequency changes by more than 5Hz. The simulation results are displayed in real time in the virtual model in the form of a color cloud map. At the same time, the simulation data is exported as an Excel report, which includes the parameter values ​​of each key point and the risk assessment results.

5. The switchgear status prediction and maintenance system based on digital twins according to claim 1, characterized in that, When generating maintenance plans, the maintenance decision module adopts a dynamic priority ranking mechanism, which comprehensively considers the importance of components, the scope of fault impact, and maintenance costs. The importance coefficient is graded according to the impact of components on the power distribution system. The impact range coefficient is calculated based on the number of circuits that may be affected by the fault. The cost coefficient is calculated using the formula C coefficient = 1 - (actual maintenance cost / budgeted maintenance cost). The actual maintenance cost includes spare parts costs, labor costs, and power outage losses. The budgeted maintenance cost is the average cost of the same type of maintenance preset by the system. The comprehensive priority = (importance coefficient × 0.5) + (impact range coefficient × 0.3) + (cost coefficient × 0.2). The priority value ranges from 1 to 5. When multiple faults are warned at the same time, the system generates a maintenance order according to the comprehensive priority.

6. The switchgear status prediction and maintenance system based on digital twins according to claim 1, characterized in that, The data acquisition module employs a sensor self-calibration mechanism, automatically calibrating the temperature and vibration sensors every 24 hours. During temperature sensor calibration, the digital twin model calculates the theoretical temperature value of each monitoring point based on the heat conduction equation, comparing the measured sensor values ​​with the theoretical values. Vibration sensor calibration calculates the theoretical vibration amplitude and frequency based on the correlation between the component's natural frequency and the operating load, comparing it with the sensor's measured data. Calibration data and correction records are retained for one year, and historical calibration trends can be queried by sensor number. When a sensor's calibration deviation exceeds 3% for three consecutive times, a replacement warning is issued 60 days in advance.

7. The switchgear status prediction and maintenance system based on digital twins according to claim 1, characterized in that, The execution control module and maintenance equipment communicate via a wireless mesh network using the IEEE 802.11s protocol. The automated inspection robot is equipped with a high-definition camera and an infrared thermal imager. When it detects loose bolts or overheated components, it transmits the location coordinates and fault type to the execution control module in real time. The execution control module optimizes the allocation of maintenance equipment based on the fault location. The fastening tool is an electric torque wrench. The target torque is set according to the bolt specifications pre-stored in the digital twin model. The torque sensor transmits the actual torque value to the execution control module every 0.1 seconds. When the deviation exceeds 3%, the output power is automatically adjusted. After the operation is completed, the fastening tool feeds back the actual torque value and operation time to the system. The digital twin model marks the corresponding bolt position with a "maintained" status and updates the maintenance record.

8. The switchgear status prediction and maintenance system based on digital twins according to claim 1, characterized in that, The state prediction module also includes an environmental adaptability correction unit. When the external humidity exceeds 85% or the temperature is below -5℃, the predicted insulation resistance is corrected. The corrected insulation resistance prediction value is: R_correction = R_prediction × (1 - 0.01 × (H_environment - 85)) × (1 - 0.02 × (-5 - T_environment)), where R_correction is the corrected insulation resistance prediction value, R_prediction is the initial prediction value, H_environment is the external humidity, and T_environment is the external temperature. The corrected prediction value is compared with the measured insulation resistance value every hour. If the deviation exceeds 10%, the correction coefficient is readjusted. When H_environment exceeds 95% or T_environment is below -15℃, the system starts the dehumidification device or heating device in advance to control the internal humidity below 75% and the temperature above 5℃, thereby reducing the insulation resistance decay rate from the source. At the same time, the environmental intervention measures are correlated with the corrected prediction results.

9. The switchgear status prediction and maintenance system based on digital twins according to claim 1, characterized in that, The data management module uses blockchain technology to store key maintenance records. Each record includes maintenance time, operator ID, replacement spare parts model, component status parameters before and after maintenance, and effect evaluation. The distributed nodes of the blockchain are deployed on the system's local server and cloud platform, and the nodes use a PoA consensus mechanism. The smart contract presets the verification rules for maintenance records. When a record is uploaded, it automatically verifies whether it contains all required fields. Data query permissions are hierarchical according to roles. The retention period of maintenance records is consistent with the life cycle of the switch cabinet. When the switch cabinet is scrapped, the relevant records are automatically archived to the blockchain archive node. The completeness of maintenance records is automatically audited quarterly through the smart contract. If the missing rate exceeds 2%, an audit warning is issued to the administrator.

10. The switchgear status prediction and maintenance system based on digital twins according to claim 1, characterized in that, The security monitoring module also includes an anomaly detection unit for the digital twin model. This unit identifies anomalies by comparing the deviations of key parameters between the virtual model and the physical entity. Anomaly detection is performed every 10 minutes. When the deviation of a certain parameter exceeds the threshold twice consecutively, it is determined to be a model mismatch anomaly, and the model calibration program is automatically started. The calibration program first filters the data from the most recent 24 hours to remove abnormal data; then, it retrains the parameters of the digital twin model based on the valid data. The training process uses the gradient descent method until the deviation between the model output and the physical entity parameters is reduced to within the threshold. After calibration, the system continuously monitors the parameter deviation three times. If all deviations are ≤ the threshold, the calibration is considered successful; otherwise, a manual intervention process is initiated to notify technicians to check the three-dimensional structural parameters of the model.

Images