Remote operation and maintenance management and control system of high and low voltage power distribution cabinet based on cloud computing

Abstract

This invention provides a cloud-based remote operation and maintenance management and control system for high and low voltage switchgear, covering the field of remote operation and maintenance management, and improving the accuracy of switchgear fault early warning and operation and maintenance efficiency. First, an edge intelligent sensing module collects sensor data and performs multi-dimensional correlation analysis based on static safety thresholds to identify abnormal trends with spatiotemporal correlation and generate preliminary early warnings. Next, a cloud platform predictive decision module drives a digital twin model for dynamic simulation, combining a prediction algorithm based on historical performance degradation curves to generate predictive maintenance strategies and dynamic safety thresholds. Finally, an adaptive control execution module uses dynamic thresholds to optimize subsequent analysis and triggers differentiated control actions, from monitoring and adjustment to emergency isolation, based on the strategy type. This invention, by constructing a cloud-edge collaborative closed-loop intelligent operation and maintenance architecture, realizes a transformation from passive response to proactive prediction in operation and maintenance, significantly improving the reliability and safety of power distribution system operation.

Description

Technical Field

[0001] This invention relates to the field of remote operation and maintenance management, specifically to a cloud-based remote operation and maintenance management and control system for high and low voltage power distribution cabinets. Background Technology

[0002] With the continuous expansion of power system scale and the increasing requirements for power supply reliability, the operation and maintenance management of high and low voltage switchgear, as key equipment for power distribution, is becoming increasingly important. Traditional switchgear operation and maintenance methods mainly rely on manual periodic inspections and passive fault response, which have problems such as long inspection cycles, delayed fault detection, high maintenance costs, and strong reliance on manual experience. This reactive maintenance mode is difficult to achieve early warning of faults, and problems are often only discovered after serious equipment failures occur, leading to an expansion of power outage areas and increased economic losses, which cannot meet the requirements of modern power systems for high reliability and intelligent operation and maintenance.

[0003] Existing technologies, such as IoT sensors and remote monitoring power distribution cabinet management systems, can collect real-time operating data such as electrical parameters and temperature of power distribution cabinets and achieve preliminary identification of abnormal states through threshold alarms. Compared with traditional manual inspections, these systems have the advantages of strong real-time performance and wide monitoring range. However, these systems generally have the following shortcomings: First, they use static threshold judgments, which cannot adapt to changes in the operating characteristics of power distribution cabinets under different load rates and environmental conditions, resulting in a high false alarm rate or a high risk of missed alarms. Second, they lack the ability to analyze the correlation of multi-dimensional data, only alarming for single parameter exceeding limits, making it difficult to identify complex fault modes caused by the coupling of multiple parameters. Third, they lack predictive maintenance mechanisms, failing to formulate maintenance strategies in advance based on the trend of equipment performance degradation, remaining at the fault response stage rather than the fault prevention stage. Summary of the Invention

[0004] To address the technical problems mentioned in the background section, this invention proposes a cloud-based remote operation and maintenance management and control system for high and low voltage power distribution cabinets.

[0005] Therefore, the technical solution adopted by the present invention is as follows:

[0006] A cloud-based remote operation and maintenance management and control system for high and low voltage switchgear, comprising:

[0007] M1, the edge intelligent sensing module, collects and preprocesses the sensor data of the power distribution cabinet, and performs multi-dimensional correlation analysis on the edge side of the preprocessed sensor data based on the preset static safety threshold. When the abnormal change trend of the two types of sensor data is identified to have spatiotemporal correlation, a preliminary fault warning event is generated.

[0008] M2, the cloud platform prediction and decision module, based on the preliminary fault warning events and preprocessed sensor data, drives a digital twin model synchronized with the power distribution cabinet to perform dynamic simulation, and combines a prediction algorithm that integrates the historical performance degradation curve of the power distribution cabinet to generate predictive maintenance strategies and dynamic safety thresholds.

[0009] M3, the adaptive control execution module, performs multi-dimensional correlation analysis in the next stage based on the dynamic safety threshold, and triggers differentiated remote control actions according to the type of the predictive maintenance strategy.

[0010] Furthermore, the sensing data includes current data, voltage data, and temperature data.

[0011] The multidimensional correlation analysis is divided into three levels: single-dimensional threshold determination, multidimensional correlation calculation, and comprehensive anomaly assessment.

[0012] Furthermore, the single-dimensional threshold determination is expressed as:

[0013]

[0014] in, Indicates the first Sensor data at time The degree to which the value deviates from the static safety threshold; Indicates the first Sensor data at time The value; Indicates the first Static security threshold for sensor data;

[0015] when Data that exceeds the threshold and is considered an abnormal state.

[0016] Based on the aforementioned abnormal state data exceeding the threshold, the multi-dimensional correlation calculation is performed, as follows:

[0017]

[0018] in, Indicates the first Type of sensor data and the first Correlation coefficients between sensor data; and They represent the first Class and First Sensor data at data points The value; and They represent the first Class and First The average value of the sensor type within the time window; This indicates the total number of data points within the time window.

[0019] Furthermore, the comprehensive anomaly assessment is expressed as follows:

[0020]

[0021] in, Indicates the first Type of sensor data and the first Sensor data at time The correlation anomaly score; and These represent the weighting coefficients for threshold deviation and correlation coefficient, respectively.

[0022] By comparing the correlation anomaly score with the set correlation anomaly score judgment threshold, a preliminary judgment of the spatiotemporal correlation of the abnormal change trend is made; further independent judgments of temporal correlation and spatial correlation are made.

[0023] Furthermore, the digital twin model is a virtual mapping of the physical entity of the power distribution cabinet on the cloud platform.

[0024] The dynamic simulation includes electrical simulation and thermodynamic simulation.

[0025] The electrical simulation is represented as follows:

[0026]

[0027] in, express The nodal admittance matrix of dimension , Indicates the number of nodes in the distribution cabinet; Represents the node voltage vector; Represents the node injected current vector;

[0028] The thermodynamic simulation uses a three-dimensional heat conduction equation to obtain the temperature field of the power distribution cabinet. , is represented as:

[0029]

[0030] in, Indicates the density of the material; Indicates specific heat capacity; Indicates thermal conductivity; Represents a spatial position vector; This indicates the heating power density of the internal heat source;

[0031] The historical performance degradation curve of the distribution cabinet is represented as follows:

[0032]

[0033] in, Indicates time Performance parameters; Indicates the performance parameters at the initial moment; This represents the asymptotically stable value of the performance parameter; This represents the decay rate constant.

[0034] Furthermore, the prediction algorithm outputs the distribution cabinet in the future time. The probability of internal failure , is represented as:

[0035]

[0036] in, This represents the failure probability obtained based on digital twin simulation; This represents the probability of failure predicted based on historical performance degradation curves. This indicates the severity score of the initial fault warning event. , and These represent the weighting coefficients of the three items respectively;

[0037] Based on the predicted failure probability, predictive maintenance strategies are generated, employing a hierarchical classification system divided into three main types and four levels.

[0038] The three main types are monitoring-enhanced strategies, early warning and notification strategies, and proactive intervention strategies.

[0039] The four levels are routine monitoring, key monitoring, planned maintenance, and emergency response.

[0040] Furthermore, the dynamic security threshold Represented as:

[0041]

[0042] in, Indicates the first Correction factor for current load rate of sensor data; This indicates a correction factor based on ambient temperature.

[0043] The load rate correction coefficient is generated by a piecewise function.

[0044] Furthermore, the dynamic security threshold replaces the original static security threshold, and an update cycle for the dynamic security threshold is set.

[0045] Based on the policy identifier code of the predictive maintenance strategy, the four levels of policies are delivered to the corresponding execution branches.

[0046] Compared with the prior art, the advantages of the present invention are as follows:

[0047] 1. This invention combines multi-dimensional correlation analysis at the edge with digital twin simulation and performance degradation curve prediction in the cloud, enabling the identification of potential risks and the generation of predictive maintenance strategies before failures occur, thus changing the traditional passive mode that relies on manual inspection and post-event handling.

[0048] 2. This invention abandons the single static threshold criterion, and performs comprehensive anomaly assessment by analyzing the spatiotemporal correlation between multi-source sensor data, and dynamically adjusts the safety threshold in combination with real-time operating status, so that the system can accurately identify complex fault modes and adapt to different load and environmental conditions, avoiding the false alarm or missed alarm problem caused by fixed thresholds in the prior art.

[0049] 3. This invention constructs an adaptive control execution mechanism that automatically triggers differentiated remote control actions based on the predicted fault risk level. It forms a hierarchical response system from monitoring parameter adjustment and alarm notification to load transfer and equipment isolation, realizing closed-loop management of the entire process from data perception, intelligent analysis, predictive decision-making to adaptive control, which significantly improves the power supply reliability and operation and maintenance efficiency of the power distribution system. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a flowchart of the remote operation and maintenance management and control system for high and low voltage distribution cabinets of the present invention;

[0052] Figure 2 This is a flowchart of the edge intelligent sensing module of the present invention;

[0053] Figure 3 This is a flowchart of the cloud platform prediction and decision-making module of the present invention. Detailed Implementation

[0054] To achieve the above objectives, the present invention provides a cloud-based remote operation and maintenance management and control system for high and low voltage power distribution cabinets. This system includes:

[0055] M1, the edge intelligent sensing module, collects and preprocesses sensor data from the power distribution cabinet. Based on a preset static safety threshold, it performs multi-dimensional correlation analysis on the preprocessed sensor data at the edge. When abnormal change trends of two types of sensor data are identified to have spatiotemporal correlation, a preliminary fault warning event is generated.

[0056] The edge intelligent sensing module, as the data acquisition and preliminary analysis layer, is responsible for acquiring multi-source heterogeneous sensor data from the power distribution cabinet in real time. The multi-source heterogeneous sensor data includes current data, voltage data, and temperature data, and performs data preprocessing and correlation analysis at the edge.

[0057] Multiple types of sensors, including current transformers, voltage transformers, and temperature sensors, are deployed at key locations in the power distribution cabinet; the electrical parameters are collected at a frequency of 10 times per second, and the temperature data is collected at a frequency of 1 time per second; the sensor data is transmitted to the edge computing gateway via fieldbus to form a multi-dimensional time series dataset.

[0058] The edge computing gateway preprocesses the received sensor data. First, it removes abnormal pulse interference using a median filtering algorithm. For electrical parameter data, it uses a sliding time window for smoothing, with the time window length set to 5 seconds. The preprocessed sensor data points are then normalized to a uniform dimension.

[0059] Based on the rated operating parameters of the distribution cabinet and national power industry standards, static safety thresholds are set for each type of sensor data; the current threshold is set to 120% of the rated current, the voltage threshold is set to ±10% of the rated voltage, and the temperature threshold is set separately for different areas within the distribution cabinet, with the temperature threshold at the busbar joint set to 75 degrees Celsius and the temperature threshold at the circuit breaker contact set to 85 degrees Celsius; these static safety thresholds constitute the basis for preliminary anomaly judgment.

[0060] The multi-dimensional correlation analysis at the edge aims to identify the spatiotemporal correlations between different types of sensor data and, combined with static security thresholds, determine abnormal states. The analysis process consists of three levels: single-dimensional threshold determination, multi-dimensional correlation calculation, and comprehensive anomaly assessment.

[0061] First, the preprocessed sensor data are compared with the corresponding static safety thresholds. Sensor data at time numerical value The threshold deviation is defined as:

[0062]

[0063] in, Indicates the first Sensor data at time The degree to which the value deviates from the static safety threshold; Indicates the first Static security threshold for sensor data; when This indicates that the data has exceeded the safety threshold and is marked as an abnormal state exceeding the threshold.

[0064] For all sensor data in the abnormal state exceeding the threshold, the temporal correlation between these data is calculated using a sliding time window method with a window length set to 60 seconds. The Pearson correlation coefficient is then calculated for the data sequences within the window.

[0065]

[0066] in, Indicates the first Type of sensor data and the first Correlation coefficients between sensor data; Indicates the first Sensor data at data points The value; Indicates the first The average value of the sensor type within the time window; This represents the total number of data points within the time window; when the absolute value of the correlation coefficient is greater than 0.7, the two types of sensor data are considered to be significantly correlated.

[0067] The comprehensive anomaly assessment integrates single-dimensional threshold judgment results with multi-dimensional correlation calculation results to construct a correlation anomaly scoring mechanism. For the two types of sensor data, a correlation anomaly score is defined as follows:

[0068]

[0069] in, Indicates the first Type of sensor data and the first Sensor data at time The correlation anomaly score; and These represent the weighting coefficients for threshold deviation and correlation coefficient, respectively; typical values ​​are... , The score takes into account whether the data exceeds the static safety threshold and whether there is a strong correlation between the two types of data.

[0070] The threshold for determining correlation anomaly scores is set as follows: ,when At that time, the judgment of the first Class and First The abnormal change trends of sensor data have spatiotemporal correlation; this comprehensive evaluation method ensures that static safety thresholds play a fundamental role in correlation analysis, while avoiding missed or false alarms that may be caused by a single indicator.

[0071] After completing the correlation anomaly score calculation, the spatiotemporal characteristics of the abnormal events are further verified. For sensor data combinations whose scores exceed the judgment threshold, the time difference of reaching the abnormal state is checked, and a time correlation window is defined. When the time difference is less than the time correlation window, it is confirmed that the anomalies of the two types of data are temporally correlated.

[0072] Simultaneously, spatial correlation is checked, that is, to verify whether the two types of sensors monitor the same device or electrically closely related devices; the electrical distance of the device to which the sensor belongs is queried through the topology database of the power distribution cabinet. The electrical distance is defined as the shortest electrical path length between two devices. When the electrical distance is less than 3 nodes, it is considered to have spatial correlation.

[0073] Based on the combined results of temporal and spatial correlation, only anomalies that simultaneously meet the criteria of exceeding the correlation anomaly score threshold, having a time difference within the correlation window, and having an electrical distance that satisfies the spatial proximity condition are identified as anomalous change trends with temporal and spatial correlation.

[0074] Once an abnormal trend with spatiotemporal correlation is identified, the edge intelligent perception module generates a preliminary fault warning event. This event includes the following information: the timestamp of the anomaly, the types and values ​​of the sensor data involved, the corresponding static safety threshold, the threshold deviation, the correlation coefficient, the correlation anomaly score, and the identification and location information of the associated power distribution cabinet equipment. The preliminary fault warning event is uploaded to the cloud platform prediction and decision-making module through an encrypted channel, providing complete contextual information for subsequent in-depth analysis and decision-making.

[0075] The M2 cloud platform prediction and decision module, based on the preliminary fault warning events and preprocessed sensor data, drives a digital twin model synchronized with the distribution cabinet to perform dynamic simulation. It also combines this with a prediction algorithm that integrates the historical performance degradation curves of the distribution cabinet to generate predictive maintenance strategies and dynamic safety thresholds.

[0076] The cloud platform's predictive decision module receives preliminary fault warning events and pre-processed sensor data uploaded from the edge side, and uses digital twin technology and predictive algorithms to generate accurate maintenance strategies and dynamic safety thresholds.

[0077] A digital twin model is a virtual mapping of the physical entity of a power distribution cabinet onto a cloud platform. This model includes the cabinet's three-dimensional geometry, electrical topology, thermodynamic characteristics, and mechanical properties. The electrical portion of the model uses the node admittance matrix method to describe the power grid structure, while the thermodynamic portion uses the finite element method to delineate the temperature field. The digital twin model maintains real-time synchronization with the power distribution cabinet, continuously updating its state variables by receiving sensor data.

[0078] Upon receiving an initial fault warning event, the cloud platform drives the digital twin model to perform dynamic simulation. During the simulation, preprocessed sensor data is used as the input boundary conditions for the model, while anomalies identified in the warning event are also introduced. The simulation employs a time-stepping method with a time step size of 0.1 seconds, covering the operational status for the next 30 minutes.

[0079] At the electrical simulation level, based on Kirchhoff's current law and voltage law, a set of electrical equations is established for each node of the distribution cabinet. A distribution cabinet network with nodes, node voltage vectors With node injected current vector The relationship between them is represented as follows:

[0080]

[0081] in, express The node admittance matrix is ​​dimensional; by solving this system of linear equations, the voltage distribution and branch current distribution of each node are obtained;

[0082] At the thermodynamic simulation level, a three-dimensional heat conduction equation is used to describe the evolution of the temperature field inside the distribution cabinet; temperature field The variation of time and space is described by a partial differential equation, expressed as:

[0083]

[0084] in, Indicates the density of the material; Indicates specific heat capacity; Indicates thermal conductivity; Represents a spatial position vector; This indicates the heating power density of the internal heat source; the internal heat source mainly includes Joule heating of the conductor and eddy current loss of the iron core.

[0085] To improve prediction accuracy, the cloud platform maintains a performance degradation database containing historical operating data of distribution cabinets. This database records the performance parameter changes of key components of the distribution cabinet throughout its entire lifecycle, including contact resistance growth curves, insulation resistance decrease curves, and mechanical characteristic degradation curves. For a specific key component, the degradation trend of its performance parameters over operating time is described using an exponential decay model, expressed as:

[0086]

[0087] in, Indicates time Performance parameters; Indicates the performance parameters at the initial moment; This represents the asymptotically stable value of the performance parameter; This represents the degradation rate constant; by performing regression analysis on historical data, the degradation rate constants of different types of components are determined, and historical performance degradation curves of the distribution cabinet are generated.

[0088] The prediction algorithm integrates digital twin simulation results and historical performance degradation curves, employing an improved long short-term memory neural network to achieve fault prediction. The network's input layer receives sensor data from multiple time steps and the simulation output of the digital twin model, the hidden layer extracts temporal features through a gating mechanism, and the output layer generates the fault probability distribution for future time periods.

[0089] For time The prediction algorithm outputs the distribution cabinet in the future time. The probability of internal failure , is represented as:

[0090]

[0091] in, This represents the failure probability obtained based on digital twin simulation; This represents the probability of failure predicted based on historical performance degradation curves. This indicates the severity score of the initial fault warning event. , and Let represent the weight coefficients of the three terms respectively, which sum to 0. The weight coefficients are determined through cross-validation of the training dataset.

[0092] Based on the predicted failure probability, the cloud platform generates predictive maintenance strategies, employing a hierarchical classification system divided into three main types and four levels. The three main types are monitoring and reinforcement strategies, early warning and notification strategies, and proactive intervention strategies. Based on these three main types, the strategies are further refined into four levels according to the specific numerical values ​​of the failure probability.

[0093] The first-level strategy is routine monitoring, coded as MT-L1, which is generated when the failure probability is less than 10% and belongs to the monitoring enhancement strategy.

[0094] The second-level strategy is key monitoring, coded as MT-L2, which is generated when the probability of failure is between 10% and 30%, and belongs to the monitoring enhancement strategy;

[0095] The third-level strategy is planned maintenance, coded as AL-L3. It is generated when the probability of failure is between 30% and 60%, and belongs to the early warning and notification type strategy.

[0096] The fourth-level strategy is emergency response, coded as IN-L4. It is generated when the probability of failure exceeds 60% and belongs to the proactive intervention strategy.

[0097] The preset static safety threshold cannot adapt to the changing characteristics of the distribution cabinet under different operating states and environmental conditions; therefore, the cloud platform calculates a dynamic safety threshold based on the digital twin simulation results and the current operating state. Sensor-like data, dynamic safety threshold Represented as:

[0098]

[0099] in, Indicates the first Correction factor for the current load rate of sensor data; This represents the correction factor based on ambient temperature; the load factor correction factor is determined through a piecewise function, expressed as:

[0100]

[0101] in, Indicates load rate;

[0102] The ambient temperature correction factor is determined by linear interpolation. When the ambient temperature is higher than 30 degrees Celsius, the correction factor decreases by 0.05 for every 5 degrees Celsius increase. The dynamic safety threshold is adjusted in real time according to the operating conditions, which can more accurately reflect the safety margin of the distribution cabinet.

[0103] M3, the adaptive control execution module, performs multi-dimensional correlation analysis in the next stage based on the dynamic safety threshold, and triggers differentiated remote control actions according to the type of predictive maintenance strategy.

[0104] The adaptive control execution module, acting as the execution layer, performs a new round of correlation analysis at the edge based on the dynamic security thresholds and predictive maintenance strategies issued by the cloud platform, and triggers corresponding remote control actions, thus achieving closed-loop management from prediction to execution.

[0105] The edge computing gateway receives dynamic security thresholds from the cloud platform, replacing the original static thresholds for the next stage of multi-dimensional correlation analysis; the update cycle of the dynamic security thresholds is set to 5 minutes to ensure that the thresholds can reflect changes in the operating status of the power distribution cabinet in a timely manner.

[0106] The multi-dimensional correlation analysis process based on dynamic security thresholds is consistent with the process described in Module 1, including three levels: single-dimensional threshold determination, multi-dimensional correlation calculation, and comprehensive anomaly assessment. The difference is that the determination benchmark is switched from static thresholds to dynamic thresholds, making the analysis results more consistent with the actual operating status of the power distribution cabinet. When a new abnormal change trend with spatiotemporal correlation is identified, the edge computing gateway generates an updated fault warning event, which is labeled as the result of dynamic threshold analysis and uploaded to the cloud platform for further verification and policy adjustment.

[0107] After receiving a predictive maintenance strategy, this module extracts and parses the strategy identifier code. The identifier code uses a structure of type code, level code, and timestamp. Type codes include MT (Monitoring Enhancement), AL (Early Warning Notification), and IN (Proactive Intervention), while level codes range from L1 to L4. Based on the type code, the strategy is delivered to the corresponding execution branch for differentiated remote control actions. Specifically,

[0108] For the MT-L1 strategy, inspection suggestions are written to the local task queue and synchronized with the operation and maintenance management system every 24 hours to upload accumulated tasks. The task queue adopts a priority queue structure, with MT-L1 tasks having the lowest priority.

[0109] For the MT-L2 strategy, firstly, the sensor sampling frequency is adjusted to twice the original frequency; then, supplementary lighting and video monitoring are activated; finally, a monitoring task work order is generated.

[0110] For the AL-L3 strategy, a comprehensive process is executed, including alarm notification, work order generation, spare parts inspection, and load transfer plan preparation. For load transfer plan preparation, a digital twin model interface is used for simulation verification. The simulation includes overload checks, voltage limit exceeding checks, transient impact checks, and protection action checks. After successful simulation, a detailed document containing operating procedures, safety precautions, and emergency plans is generated and submitted to the multi-level review process of the operation and maintenance management system. Once approved, the plan is marked as approved and awaits confirmation from operations and maintenance personnel regarding the timing of execution.

[0111] Once the maintenance personnel confirm the operation, they will execute the load transfer control, operating each circuit breaker in sequence with a 2-second interval between each operation. For each circuit breaker, they will send a tripping or closing pulse command. After the operation is completed, they will verify the position feedback signal and electrical parameters. During the transfer, they will continuously monitor power changes. If the power change exceeds 5% of the total power of the transferred load, they will record the abnormality and issue an alarm.

[0112] For IN-L4 policies, an emergency response mechanism should be activated immediately, including emergency alarm triggering, backup system startup, isolation scheme optimization, and isolation operation execution.

[0113] The emergency alarm is triggered by activating the on-site audible and visual alarm devices, emitting a flashing red light and a 90-decibel buzzer sound. Simultaneously, all relevant personnel are notified through multiple channels including telephone voice, SMS, email, and mobile push notifications. An emergency warning message is also sent and displayed with the highest priority on the dispatcher's workstation.

[0114] The backup system startup preheats the backup power supply system.

[0115] The isolation scheme optimization involves extracting a preliminary isolation plan and further refining it based on real-time power grid conditions. First, critical loads are identified. A depth-first search algorithm is used to traverse downstream nodes from the faulty node, marking the scope of the isolation impact. If the critical load is affected, a power supply protection process is initiated.

[0116] The isolation operation is executed by generating an execution plan that includes step number, operation object, operation type, pre-check items and post-verification items; before execution, the circuit breaker position, interlocking signal, energy storage status and control circuit are checked; after the check is passed, the operation command is sent and a 5-second timeout monitoring is initiated.

[0117] This invention proposes a cloud-based remote operation and maintenance management and control system for high and low voltage distribution cabinets. By constructing a three-layer closed-loop architecture of edge intelligent perception, cloud platform predictive decision-making, and adaptive control execution, it achieves a fundamental transformation in distribution cabinet operation and maintenance from passive response to proactive prediction and precise control. At the edge, multi-dimensional correlation analysis based on spatiotemporal correlation is introduced for preliminary fault warning. On the cloud platform, a predictive algorithm that integrates digital twin dynamic simulation and historical performance degradation curves is used to generate forward-looking maintenance strategies and dynamic safety thresholds. Finally, through an adaptive execution module, the dynamic thresholds are fed back to the edge optimization analysis criteria, and differentiated control actions, from parameter adjustment to emergency isolation, are triggered according to the strategy level, thereby forming an intelligent operation and maintenance ecosystem with self-learning and self-optimization capabilities.

[0118] In summary, this invention effectively addresses the three major pain points of existing technologies—high false alarms and missed alarms at static thresholds, lack of multi-parameter correlation analysis, and inability to perform predictive maintenance—through cloud-edge collaborative intelligent analysis, predictive decision-making, and closed-loop control. This significantly improves the safety, reliability, and operational efficiency of power distribution systems.

[0119] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A cloud-based remote operation and maintenance management and control system for high and low voltage distribution cabinets, characterized in that: The system includes: M1, the edge intelligent sensing module, collects and preprocesses the sensor data of the power distribution cabinet, and performs multi-dimensional correlation analysis on the edge side of the preprocessed sensor data based on the preset static safety threshold. When the abnormal change trend of the two types of sensor data is identified to have spatiotemporal correlation, a preliminary fault warning event is generated. M2, the cloud platform prediction and decision module, based on the preliminary fault warning events and preprocessed sensor data, drives a digital twin model synchronized with the power distribution cabinet to perform dynamic simulation, and combines a prediction algorithm that integrates the historical performance degradation curve of the power distribution cabinet to generate predictive maintenance strategies and dynamic safety thresholds. M3, the adaptive control execution module, performs multi-dimensional correlation analysis in the next stage based on the dynamic safety threshold, and triggers differentiated remote control actions according to the type of the predictive maintenance strategy. The sensing data includes current data, voltage data, and temperature data. The multidimensional correlation analysis is divided into three levels: single-dimensional threshold determination, multidimensional correlation calculation, and comprehensive anomaly assessment. The single-dimensional threshold determination is expressed as follows: in, Indicates the first Sensor data at time The degree to which the value deviates from the static safety threshold; Indicates the first Sensor data at time The value; Indicates the first Static security threshold for sensor data; when Data that exceeds the threshold and is considered an abnormal state. Based on the aforementioned abnormal state data exceeding the threshold, the multi-dimensional correlation calculation is performed, as follows: in, Indicates the first Type of sensor data and the first Correlation coefficients between sensor data; and They represent the first Class and First Sensor data at data points The value; and They represent the first Class and First The average value of the sensor type within the time window; This represents the total number of data points within the time window; The comprehensive anomaly assessment is expressed as follows: in, Indicates the first Type of sensor data and the first Sensor data at time The correlation anomaly score; and These represent the weighting coefficients for threshold deviation and correlation coefficient, respectively. Indicates the first Sensor data at time The degree to which the value deviates from the static safety threshold; By comparing the correlation anomaly score with the set correlation anomaly score judgment threshold, a preliminary judgment of the spatiotemporal correlation of the abnormal change trend is made; further independent judgments of temporal correlation and spatial correlation are made.

2. The cloud-based remote operation and maintenance management and control system for high and low voltage distribution cabinets according to claim 1, characterized in that, The digital twin model is a virtual mapping of the physical entity of the power distribution cabinet on a cloud platform. The dynamic simulation includes electrical simulation and thermodynamic simulation. The electrical simulation is represented as follows: in, express The nodal admittance matrix of dimension , Indicates the number of nodes in the distribution cabinet; Represents the node voltage vector; Represents the node injected current vector; The thermodynamic simulation uses a three-dimensional heat conduction equation to obtain the temperature field of the power distribution cabinet. , is represented as: in, Indicates the density of the material; Indicates specific heat capacity; Indicates thermal conductivity; Represents a spatial position vector; This indicates the heating power density of the internal heat source; The historical performance degradation curve of the distribution cabinet is represented as follows: in, Indicates time Performance parameters; Indicates the performance parameters at the initial moment; This represents the asymptotically stable value of the performance parameter; This represents the decay rate constant.

3. The cloud-based remote operation and maintenance management and control system for high and low voltage distribution cabinets according to claim 2, characterized in that, The prediction algorithm outputs the future time of the power distribution cabinet. The probability of internal failure , is represented as: in, This represents the failure probability obtained based on digital twin simulation; This represents the probability of failure predicted based on historical performance degradation curves. This indicates the severity score of the initial fault warning event. , and These represent the weighting coefficients of the three items respectively; Based on the predicted failure probability, predictive maintenance strategies are generated, employing a hierarchical classification system divided into three main types and four levels. The three main types are monitoring-enhanced strategies, early warning and notification strategies, and proactive intervention strategies. The four levels are routine monitoring, key monitoring, planned maintenance, and emergency response.

4. The cloud computing-based remote operation and maintenance management and control system for high and low voltage distribution cabinets according to claim 3, characterized in that, The dynamic security threshold Represented as: in, Indicates the first Correction factor for current load rate of sensor data; This indicates a correction factor based on ambient temperature. The load rate correction coefficient is generated by a piecewise function.

5. The cloud-based remote operation and maintenance management and control system for high and low voltage distribution cabinets according to claim 4, characterized in that, The dynamic security threshold replaces the original static security threshold, and a set update cycle for the dynamic security threshold is established. Based on the policy identifier code of the predictive maintenance strategy, the four levels of policies are delivered to the corresponding execution branches.

Images