Power grid digital twin modeling method and platform based on cloud-edge collaboration

Abstract

The present invention provides a power grid digital twin modeling method and platform based on cloud-edge collaboration, which relates to the field of digital processing technology, including: integrating a distributed sensor network, communicating with multiple real-time monitoring units, and monitoring indicators including current, voltage, and temperature; deploying edge computing servers to upload real-time monitoring data and quickly locate faults; interacting with edge computing servers and cloud processors, dynamically adjusting scheduling strategies, and building a power grid digital twin model; performing local marking to establish multiple local update spaces; matching multiple independent update cycles and priority mechanisms, and adaptively triggering multiple local update requests; performing self-learning iteration and reinforcement during operation to assist in the operation and maintenance management of the target power grid. The present invention solves the technical problems that the power grid dispatching strategies of the prior art are often difficult to adapt to the dynamic changes in the operation of the power grid, lack of intelligent means, and lead to waste of resources and system instability.

Description

Technical Field

[0001] The present invention relates to the field of digital processing technology, and in particular to a power grid digital twin modeling method and platform based on cloud-edge collaboration. Background Art

[0002] A large amount of data generated by various devices in the power grid needs to be collected and processed in real time to ensure the stability and security of the power grid operation. However, the data collection and processing of traditional power grid systems often have delays and cannot reflect the real status of the power grid in a timely manner. There are data transmission delays, slow data processing speeds, and insufficient data accuracy, which makes it difficult for the power grid system to quickly locate and respond to faults. In addition, power grid dispatching requires comprehensive load forecasting, power generation plans, equipment status and other factors to achieve economical and efficient power supply. However, traditional power grid dispatching strategies lack intelligence and dynamic adjustment capabilities, and cannot cope with various dynamic changes in power grid operation. There are rigid dispatching strategies and inability to respond to power grid load changes in real time, resulting in resource waste and system instability. Summary of the invention

[0003] This application provides a digital twin modeling method and platform for power grids based on cloud-edge collaboration, aiming to solve the technical problems that power grid dispatching strategies in the prior art are often difficult to adapt to dynamic changes in power grid operation, lack intelligent means, and lead to waste of resources and system instability.

[0004] The first aspect disclosed in the present application provides a method for modeling a digital twin of a power grid based on cloud-edge collaboration, the method comprising: integrating a distributed sensor network, the distributed sensor network being communicatively connected to a plurality of real-time monitoring units, the monitoring indicators corresponding to the plurality of real-time monitoring units comprising current, voltage, and temperature; deploying an edge computing server at a key node of a target power grid, the edge computing server integrating augmented reality technology to upload the real-time monitoring data corresponding to the plurality of real-time monitoring units and to perform rapid fault location; utilizing a 5G dedicated network to interact with the edge computing server and a cloud processor, and using a machine learning model to dynamically adjust the dispatching strategy of the target power grid to construct a digital twin model of the power grid, the digital twin model of the power grid comprising a power grid topology, a device standby status, operating parameters; based on the power grid digital twin model, in combination with the edge computing server and multiple real-time monitoring units associated with the mapping, local marking is performed in the power grid digital twin model to establish multiple local update spaces, and each of the multiple local update spaces has an overlapping part; based on the cloud processor, multiple independent update cycles and priority mechanisms are matched against the multiple local update spaces, and multiple local update requests corresponding to the multiple local update spaces are adaptively triggered, and the multiple local update requests are mapped one by one to the multiple local update spaces; during the operation of the target power grid, the multiple local update spaces and the multiple local update requests are used to perform self-learning iterative reinforcement on the power grid digital twin model, and assist in the operation and maintenance management of the target power grid.

[0005] The second aspect disclosed in the present application provides a power grid digital twin modeling platform based on cloud-edge collaboration, and the platform is used for the above-mentioned power grid digital twin modeling method based on cloud-edge collaboration, and the platform includes: a sensor network integration module, the sensor network integration module is used to integrate a distributed sensor network, and the distributed sensor network is communicatively connected to multiple real-time monitoring units, and the monitoring indicators corresponding to the multiple real-time monitoring units include current, voltage, and temperature; a rapid fault location module, the rapid fault location module is used to deploy an edge computing server at the key nodes of the target power grid, and the edge computing server integrates augmented reality technology to upload the real-time monitoring data corresponding to the multiple real-time monitoring units and perform rapid fault location; a digital twin model construction module, the digital twin model construction module is used to use the 5G dedicated network to interact with the edge computing server and the cloud processor, and use a machine learning model to dynamically adjust the scheduling strategy of the target power grid to construct a power grid digital twin model, the power grid digital The twin model includes a power grid topology, equipment status, and operating parameters; a local update space establishment module, which is used to perform local marking in the power grid digital twin model based on the power grid digital twin model, combined with the edge computing server and multiple real-time monitoring units associated with the mapping, and establish multiple local update spaces, and each of the multiple local update spaces has an overlapping part; a local update request triggering module, which is used to match multiple independent update cycles and priority mechanisms with the multiple local update spaces based on a cloud processor, and adaptively trigger multiple local update requests corresponding to the multiple local update spaces, and the multiple local update requests are mapped one-to-one with the multiple local update spaces; an operation and maintenance management module, which is used to perform self-learning iterative reinforcement on the power grid digital twin model with the multiple local update spaces and multiple local update requests during the operation of the target power grid, and assist in the operation and maintenance management of the target power grid.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] By integrating distributed sensor networks and edge computing servers, data such as current, voltage, and temperature are collected in real time, and edge computing technology is used for data processing and rapid fault location, which improves the real-time and accuracy of the data. By using 5G dedicated networks and machine learning models, the power grid dispatching strategy is dynamically adjusted, and the power grid digital twin model is updated in real time, making the power grid dispatching more intelligent and able to adapt to various dynamic changes in power grid operation, thereby improving the flexibility and efficiency of power grid operation. By performing local marking in the power grid digital twin model and establishing multiple local update spaces, local updates and resource allocation are performed more accurately, thereby improving the overall performance and resource utilization of the power grid. By matching multiple independent update cycles and priority mechanisms, update requests are adaptively triggered, local updates and resource allocation are optimized, and the overall performance of the system is improved. During the operation of the power grid, the power grid digital twin model is self-learned and iteratively strengthened, and the operation and maintenance management of the target power grid is assisted, thereby improving the intelligence level and adaptability of operation and maintenance management and reducing the complexity and uncertainty of management.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic diagram of a process flow of a power grid digital twin modeling method based on cloud-edge collaboration provided in an embodiment of the present application;

[0010] Figure 2 A schematic diagram of the structure of a power grid digital twin modeling platform based on cloud-edge collaboration provided in an embodiment of the present application.

[0011] Explanation of the accompanying drawings: sensor network integration module 10, rapid fault location module 20, digital twin model construction module 30, local update space establishment module 40, local update request triggering module 50, operation and maintenance management module 60. DETAILED DESCRIPTION

[0012] The embodiments of the present application provide a digital twin modeling method and platform for power grids based on cloud-edge collaboration, thereby solving the technical problems that power grid dispatching strategies in the prior art are often difficult to adapt to dynamic changes in power grid operation, lack intelligent means, and lead to waste of resources and system instability.

[0013] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically introduced below in conjunction with the accompanying drawings of the specification. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0014] Embodiment 1, as Figure 1 As shown, an embodiment of the present application provides a power grid digital twin modeling method based on cloud-edge collaboration, the method comprising:

[0015] An integrated distributed sensor network is communicatively connected with a plurality of real-time monitoring units, and the monitoring indicators corresponding to the plurality of real-time monitoring units include current, voltage, and temperature.

[0016] A distributed sensor network is a network composed of multiple sensor nodes, which are distributed in different locations of the power grid and are used to monitor the operation data of the power grid. The real-time monitoring units are used to collect sensor data in real time and perform preliminary processing. They are connected to the sensors and transmit the data to the edge computing server.

[0017] Select sensors suitable for the power grid environment, including current sensors, voltage sensors, and temperature sensors, and deploy these sensors at key nodes of the power grid, such as substations, transmission lines, and distribution cabinets. Connect each sensor to the corresponding real-time monitoring unit, which is used to monitor the sensor data. The monitoring indicators include current, voltage, and temperature.

[0018] An edge computing server is deployed at a key node of the target power grid, and the edge computing server integrates augmented reality technology to upload the real-time monitoring data corresponding to the multiple real-time monitoring units and perform rapid fault location.

[0019] Edge computing servers are used to analyze data in real time, reduce data transmission delays, and improve response speed. They have strong computing and storage capabilities and can quickly process large amounts of data. Identify key nodes in the target power grid, such as substations, distribution cabinets, and important transmission lines, determine the locations of these key nodes, and install edge computing servers at the determined key nodes. Edge computing servers integrate augmented reality technology, which is a technology that uses computer technology to superimpose virtual information on the real world. It can display virtual monitoring data, fault information, etc., to help operation and maintenance personnel understand the power grid status more intuitively.

[0020] The real-time monitoring unit uploads the collected real-time monitoring data to the edge computing server. The edge computing server processes the received data in real time and uses the processed data to perform rapid fault detection and locate the fault point.

[0021] The edge computing server interacts with the cloud processor by utilizing the 5G dedicated network, and a machine learning model is used to dynamically adjust the dispatching strategy of the target power grid to build a power grid digital twin model, which includes the power grid topology, equipment status, and operating parameters.

[0022] A dedicated network refers to a 5G network that is configured specifically for certain devices in a specific area to ensure the exclusivity and stability of network resources. A cloud processor is a high-performance processor deployed on a cloud computing platform that is used to receive, store, and analyze data from edge computing servers. Using a 5G dedicated network, the real-time monitoring data collected by the edge computing server and the processed fault location results are uploaded to the cloud processor, which receives the uploaded data and performs synchronous processing to ensure real-time data consistency.

[0023] Machine learning models learn through training data and make predictions and decisions in practical applications. In power grid dispatching, machine learning models are used to dynamically adjust dispatching strategies and optimize power grid operation. Machine learning models are trained based on historical data and real-time data. The training content includes load prediction, fault detection, and optimized dispatching of power grids. The trained machine learning models are deployed on cloud processors to process and analyze data uploaded by edge computing servers in real time, identify the operating status and potential problems of the power grid, and dynamically adjust the dispatching strategies of the power grid, including load distribution, power dispatching, and fault handling, based on the analysis results, to ensure the stability and efficiency of power grid operation.

[0024] The power grid digital twin model is a digital virtual model of the power grid, which reflects the topological structure, equipment status and operating parameters of the actual power grid. The power grid topological structure shows the connection relationship between various devices and lines in the power grid, the equipment status includes information such as the operating status and health status of the equipment, and the operating parameters include real-time operating data such as current, voltage, temperature, and load. The cloud processor continuously receives and processes new data, and updates the power grid digital twin model in real time to ensure that the model is consistent with the operating status of the actual power grid.

[0025] Based on the power grid digital twin model, combined with the edge computing server and multiple real-time monitoring units associated with the mapping, local marking is performed in the power grid digital twin model to establish multiple local update spaces, and each of the multiple local update spaces has an overlapping part.

[0026] In the digital twin model of the power grid, key nodes are marked to locate and identify key areas in the model. As marked areas, these marked areas represent important parts of the power grid. The real-time monitoring data collected by the edge computing server is mapped to the corresponding marked areas in the digital twin model of the power grid to ensure that each marked area in the model can reflect the actual operating status. Based on the marked areas in the digital twin model of the power grid, multiple local update spaces are defined. These update spaces represent areas in the power grid that need to be monitored and optimized. There are overlapping parts between each local update space, so that data and resources can be shared between different update spaces, helping to coordinate and optimize the operation of different update spaces. For example, the transmission line between two adjacent substations can be used as an overlapping area.

[0027] Based on the cloud processor, multiple independent update cycles and priority mechanisms are matched against the multiple local update spaces, and multiple local update requests corresponding to the multiple local update spaces are adaptively triggered, and the multiple local update requests are mapped one-to-one with the multiple local update spaces.

[0028] The independent update cycle is the update cycle of each local update space, which is determined based on the actual operation of the power grid and historical data analysis. The independent update cycle ensures that each update space can be updated and adjusted as needed to maintain the stable operation of the power grid. The priority mechanism assigns priorities to different local update spaces based on the key performance indicators of the power grid operation. The priority mechanism ensures that important areas and emergency situations are handled in a timely manner, improving the operation efficiency and reliability of the power grid.

[0029] The cloud processor receives data from the edge computing server and real-time monitoring unit, analyzes the operating status of the power grid, and determines the independent update cycle of each local update space; according to the key performance indicators of the power grid operation, such as downtime, equipment load rate, energy efficiency loss, etc., it assigns priorities to different local update spaces to ensure that important areas and emergencies are handled in a timely manner.

[0030] The local update request is used to trigger the update operation of the local update space. The local update request is generated based on the real-time monitoring data and trigger conditions to ensure efficient management and optimization of the power grid. Based on the real-time monitoring data and historical operation examples, the trigger conditions of multiple local update spaces are set. For example, when the current, voltage or temperature in a local update space exceeds the normal range, the corresponding update request is triggered. When the trigger condition is met, the cloud processor automatically generates and triggers the corresponding local update request to ensure that the local update space can respond and adjust in real time. The cloud processor coordinates the execution order of multiple local update requests based on the priority mechanism and independent update cycle to optimize the overall operation efficiency of the power grid.

[0031] Among them, multiple local update requests are mapped one-to-one with multiple local update spaces, ensuring that the corresponding relationship between the requests and the spaces is clear and the update operation can be accurately executed.

[0032] During the operation of the target power grid, the multiple local update spaces and the multiple local update requests are used to perform self-learning iterative reinforcement on the power grid digital twin model, and assist in the operation and maintenance management of the target power grid.

[0033] During the operation of the target power grid, current, voltage, temperature and other data are collected from multiple real-time monitoring units and edge computing servers, and the collected data are integrated into the corresponding local update space. According to real-time data and historical data, the machine learning model is continuously updated and optimized. According to the execution results of the local update request, the parameters of the model are adjusted and optimized so that the model can better adapt to the dynamic changes of the power grid. By analyzing the monitoring data and historical fault data, possible fault points are predicted, and prevention and maintenance are carried out in advance. According to the prediction results, the dispatching strategy of the power grid is optimized to improve the operation efficiency and reliability of the power grid. According to the actual operation of the power grid, resources are reasonably allocated and managed to ensure the stable operation of the power grid.

[0034] Further, a distributed sensor network is integrated, wherein the distributed sensor network is communicatively connected to a plurality of real-time monitoring units, and the method comprises:

[0035] Based on Zigbee wireless sensors, multimodal data are collected, and the monitoring indicators corresponding to the multimodal data include acoustic type indicators, vibration type indicators, temperature type indicators, humidity type indicators, and electromagnetic field type indicators; the multimodal data are preprocessed by a signal processing algorithm, and fuzzy logic fusion technology is used to perform data fusion of probabilistic reasoning to determine multiple independent update decisions corresponding to multiple local update spaces; based on the multiple independent update decisions and multiple independent update cycles corresponding to the multiple local update spaces, the priority mechanism is triggered to update.

[0036] Zigbee wireless sensors are short-range, low-rate self-organizing multi-hop networks based on the IEEE802.15.4 standard, which are composed of a large number of sensor nodes through wireless communication technology. They include acoustic sensors, vibration sensors, temperature sensors, humidity sensors, and electromagnetic field sensors. These sensors are deployed at key nodes of the power grid to collect multimodal data. Multimodal data is different types of data collected from multiple sensors, including acoustic data, vibration data, temperature data, humidity data, and electromagnetic field data. These data can provide comprehensive information on the monitored objects from multiple dimensions and angles.

[0037] Multimodal data are preprocessed through signal processing algorithms. Signal processing algorithms are algorithms used to process and analyze signals, including steps such as data cleaning, normalization, smoothing, and feature extraction. Signal processing algorithms can help improve data quality and extract useful information.

[0038] Data fusion for probabilistic reasoning using fuzzy logic fusion technology. Fuzzy logic fusion technology is an information processing technology used to handle uncertainty and fuzziness. By reasoning through fuzzy rules, it can handle fuzzy or uncertain information that is difficult to handle with traditional logic. Specifically, using fuzzy rules, by defining "if-then" rules, different types of data can be fused together. For example, if the temperature is high and the vibration is large, then the equipment may fail. Apply fuzzy rules to input data and generate fuzzy output through the reasoning mechanism.

[0039] Based on the results of fuzzy logic fusion, the states of multiple local update spaces are analyzed, for example, which local update spaces need to be updated and the priority of the update are determined, and multiple independent update decisions are generated.

[0040] Based on the generated multiple independent update decisions and multiple independent update cycles, the priority mechanism is updated in a triggered manner, that is, based on the priority mechanism and the actual situation, the update request is dynamically triggered and executed to ensure that high-priority requests are processed in a timely manner and adjusted according to actual needs. Specifically, based on the calculated priority, the corresponding update request is triggered, and these requests are processed from high to low priority. For example, high-priority device update requests will be processed first; the update request is dynamically adjusted according to the actual situation. For example, if a sudden failure occurs in a device in a certain area, its priority will be increased for emergency updates.

[0041] Furthermore, based on the multiple independent update decisions and the multiple independent update cycles corresponding to the multiple local update spaces, the priority mechanism is triggered to update, and the method includes:

[0042] Connect the data storage module to obtain historical operation instances, which include equipment failure data and equipment maintenance data; based on the historical operation instances, apply supervised learning methods to perform fault pattern recognition, combine time series analysis to perform fault prediction, add trigger conditions corresponding to the multiple local update spaces, and the trigger conditions are associated with key performance indicators of the target power grid, which include downtime, equipment load rate, and energy efficiency loss.

[0043] The data storage module is a database used to store and manage historical operation data, including equipment failure data and equipment maintenance data. The equipment failure data records detailed information when the equipment fails, including failure time, failure type, affected area, etc.; the equipment maintenance data records detailed information on equipment maintenance, including maintenance time, maintenance content, maintenance personnel, etc.

[0044] Supervised learning is a machine learning method that uses labeled data for training so that the model can predict the labels of unseen data. In power grid maintenance, supervised learning is used to identify equipment failure modes. Fault instances in historical operating data are labeled as different fault types and used as training data. The labeled historical fault data is used to train the supervised learning model to identify different failure modes. The performance of the model is evaluated through cross-validation methods to ensure that it can accurately identify failure modes.

[0045] Analyze the time series of equipment operation data to identify and predict equipment failure trends. Common methods include autoregressive moving average models, long short-term memory networks, etc., which train prediction models on historical time series data to predict possible future equipment failures.

[0046] Based on the results of fault mode identification and fault prediction, specific trigger conditions are defined. For example, when a key performance indicator of a device reaches a certain threshold, a maintenance operation is triggered. The trigger condition is associated with the key performance indicators of the target power grid. The key performance indicators include downtime, equipment load rate, and energy efficiency loss. The downtime is the total time that the device is shut down due to failure or maintenance, the equipment load rate is the ratio of the actual load of the device to its maximum load, and the energy efficiency loss is the energy efficiency loss of the device during operation.

[0047] Through fault mode identification and fault prediction, predictive maintenance strategies are implemented to enable maintenance operations to be performed in advance to reduce equipment failures and improve the overall performance of the power grid.

[0048] Furthermore, the power grid digital twin model is self-learned and iteratively strengthened to assist the operation and maintenance management of the target power grid, and the method includes:

[0049] Based on the historical operation examples, M normal power grid operation scenarios and N abnormal power grid operation scenarios are simulated; through the M normal power grid operation scenarios, the power grid digital twin model is self-learned and iteratively strengthened under the constraints of a regular trigger mode; through the N abnormal power grid operation scenarios, the power grid digital twin model is self-learned and iteratively strengthened under the constraints of an emergency trigger mode.

[0050] Based on the normal operation data in the historical operation instances, M normal operation scenarios are defined. These scenarios can represent the typical state of the power grid under normal operation conditions, such as normal operation of equipment, stable load, no emergencies, etc. By using power grid simulation software or simulation tools, a simulation environment is created to simulate these normal operation scenarios to ensure that various conditions and parameters in actual operation can be reproduced.

[0051] Based on the abnormal conditions in historical operation instances, N abnormal operation scenarios are defined. These scenarios include power grid failures, equipment failures, load abnormalities, etc. A simulation environment is created to simulate these abnormal operation scenarios to ensure that faults and abnormal conditions can be accurately reproduced.

[0052] Based on the data in the normal operation scenarios, a regular trigger mode is defined. The regular trigger mode represents the regular behavior of the power grid during normal operation, such as load change patterns, equipment operation cycles, etc. The normal operation scenario data is used to perform self-learning training on the power grid digital twin model. The model learns based on the normal operation data to improve the capture and simulation of normal operation patterns. During the self-learning process, the model parameters and structure are adjusted according to the feedback in the simulation data to improve the model's fit and prediction ability for normal operation scenarios.

[0053] Emergency trigger modes are response modes for abnormalities or failures in the power grid. These modes are used to describe the behavior and response strategies of the power grid in emergency situations. For example, the power grid may experience emergency trigger modes in situations such as sudden load increases, equipment failures, or natural disasters. The power grid digital twin model is self-learned using normal operation data. The model is adjusted based on its performance in abnormal scenarios to improve its responsiveness and accuracy in similar situations. Based on the results of self-learning, the model parameters and structure are iteratively optimized to better adapt to the emergency trigger mode. Multiple iterations are performed to gradually improve the model's ability to handle abnormal situations.

[0054] Furthermore, through the M normal power grid operation scenarios, the power grid digital twin model is subjected to self-learning iterative reinforcement under the constraints of a regular triggering mode, and the method includes:

[0055] Data mining technology is used to identify energy consumption patterns and set energy consumption cycle fluctuation channels. Based on the energy consumption cycle fluctuation channels, sliding mode control laws are added in combination with timing nodes to predict periodic changes in normal operating data. Under the constraints of the regular triggering mode and in combination with the periodic changes in the normal operating data, sliding mode control is used to perform adaptive enhancement training on the periodic changes of the digital twin model of the power grid.

[0056] Collect historical energy consumption data of the power grid, including power load, equipment energy consumption, environmental conditions, etc. Use data mining algorithms, such as cluster analysis, association rule mining, time series analysis, etc., to identify energy consumption patterns, including common load curves, equipment energy efficiency changes, periodic fluctuations, etc. By analyzing the identified energy consumption patterns, determine the periodic fluctuations of energy consumption, which may be caused by factors such as periodic operation of equipment, load changes or changes in environmental conditions, and establish an energy consumption cycle fluctuation channel to monitor and predict periodic changes in energy consumption.

[0057] Time series nodes are key time points that mark periodic changes in data. These time points correspond to fluctuation cycles of energy consumption, such as day-night changes, seasonal changes, etc. The sliding mode control law is used to deal with the uncertainty and nonlinearity of the system. Sliding mode control adjusts the dynamic behavior of the system by introducing sliding surfaces and control laws to achieve the desired performance. The sliding mode control law is applied to the energy consumption cycle fluctuation channel. The model combines the characteristics of periodic fluctuations and time series nodes to predict the periodic changes of normal operation data. According to the prediction results, the operation strategy of the power grid is adjusted and optimized to improve energy efficiency and system stability.

[0058] Under the constraints of the regular trigger mode, the model is trained in combination with the periodic changes of normal operating data. The sliding mode control is used to adaptively adjust the periodic changes of the model to improve the model's prediction and processing capabilities for normal operating data. The sliding mode control parameters are continuously updated and optimized to ensure that the model can adapt to the periodic changes under normal operating conditions.

[0059] Furthermore, based on the energy consumption cycle fluctuation channel, a sliding mode control law is added in combination with a time series node to predict the periodic changes of normal operation data. The method includes:

[0060] By using mixed integer programming, a membership degree between 0 and 1 is assigned to each parameter through a membership function to construct an energy-saving logic rule base, which is used to describe the energy consumption characteristics of the target power grid under different operating conditions; based on the energy-saving logic rule base, energy consumption is inferred by fuzzy logic operation, and the energy-saving optimization strategy corresponding to the energy consumption mode is configured; using closed-loop control logic, with the goal of maximizing energy efficiency and power supply reliability, combined with the energy-saving optimization strategy corresponding to the energy consumption mode, the operating state of the target power grid is continuously monitored and adjusted.

[0061] Mixed integer programming is a mathematical optimization method used to solve optimization problems with integer decision variables and continuous decision variables. By constructing the objective function and constraints, the decision variable values ​​that optimize the objective function are found. The membership function is used to describe the degree to which a parameter belongs to a fuzzy set. The value range is usually between 0 and 1. A membership degree of 1 means that it completely belongs to the set, 0 means that it does not belong to the set at all, and a value between 0 and 1 means partial membership.

[0062] Select a suitable membership function type, such as triangular membership function, trapezoidal membership function, Gaussian membership function, etc., to represent the membership degree of each parameter, use mixed integer programming algorithm to model the data, optimize the energy-saving strategy, the objective function can be the minimization of energy consumption, the constraints can include the operation restrictions of the power grid and the capacity restrictions of the equipment, and generate energy-saving logic rules based on the optimization results and membership functions. These rules describe how to adjust the power grid operation under different operating conditions to achieve energy saving.

[0063] The actual measured data value is converted into the membership value of the fuzzy set, and the fuzzified data is inferred using fuzzy rules to generate fuzzy reasoning results. Fuzzy rules usually include "if-then" statements, such as if the load is high and the temperature is high, then the energy consumption may be large, and the fuzzy reasoning results are converted into specific energy consumption prediction values. According to the energy consumption prediction value, energy-saving optimization strategies are formulated, such as adjusting the grid operation parameters, optimizing the load distribution, adjusting the equipment operation mode, etc., to reduce energy consumption.

[0064] Closed-loop control logic is an automated control strategy that monitors the output of the system in real time, compares the actual results with the expected goals, and makes adjustments based on the deviations. This process forms a feedback loop that aims to keep the system in the desired state. The monitoring system collects the operation data of the power grid in real time, evaluates the operation status of the power grid, and detects whether it meets the requirements of energy efficiency and power supply reliability. By comparing with the preset goals, the deviation is identified, and according to the closed-loop control logic, adjustment measures are taken to correct the deviation. For example, if energy consumption is too high, energy-saving measures are added; if power supply reliability is reduced, the power grid configuration is optimized to improve stability. Continuously collect power grid operation data, evaluate the adjustment effect, and optimize energy efficiency and power supply reliability through continuous monitoring and adjustment.

[0065] Furthermore, through the N abnormal power grid operation scenarios, the power grid digital twin model is subjected to self-learning iterative reinforcement under the constraint of an emergency trigger mode, and the method includes:

[0066] Under the constraints of the emergency trigger mode, random abnormal factors are introduced, and the random abnormal factors include extreme meteorological factors; based on the random abnormal factors and the N abnormal power grid operation scenarios, a risk assessment framework is used to analyze potential negative impacts; the potential negative impacts are arranged in order from large to small, and the resource allocation of the overlapping part of each local update space in the multiple local update spaces is optimized, and the resource allocation includes routing power flow allocation and load allocation; after the resource allocation of the overlapping part of each local update space in the multiple local update spaces is optimized, according to the nonlinear characteristics of the random abnormal factors, nonlinear control is used to perform adaptive enhancement training on the random changes of the power grid digital twin model.

[0067] Random abnormal factors refer to factors that are difficult to predict but may have a significant impact on the system. These factors include extreme meteorological events such as storms, floods, extremely high or low temperatures, etc. Introducing these random abnormal factors into the digital twin model of the power grid is used to simulate and predict the performance of the power grid under extreme conditions, which can enhance the model's adaptability to these abnormal situations.

[0068] The risk assessment framework is a systematic method for identifying, analyzing and evaluating potential risks, including steps such as risk identification, risk analysis, and risk assessment. Specifically, it identifies possible risks caused by random abnormal factors, such as extreme weather events that may cause damage to power equipment and power outages. It analyzes the possibility and impact of these risks, and evaluates the impact of different abnormal factors on power grid operations. For example, storms may cause line short circuits, and extreme high temperatures may cause equipment overheating. It quantifies the impact of risks, including possible losses and consequences, calculates the potential economic losses, equipment damage, and service interruption time caused by these abnormal factors, and obtains potential negative impacts.

[0069] The identified potential negative impacts are ranked from most severe to least severe. This ranking is based on the results of risk assessment, including the degree of impact, probability of occurrence and potential losses. The overlapping areas in multiple local update spaces are determined and resources are allocated to these areas. Specifically, routing power flow distribution is performed to optimize the path of power flow and ensure sufficient power supply in high-risk areas to ensure that power can be effectively transmitted when responding to potential risks; load distribution is performed. According to the risk ranking, load distribution is adjusted to give priority to load demand in high-risk areas to balance the load of the power grid and reduce risks, avoid load overload and equipment damage.

[0070] Random abnormal factors have nonlinear characteristics, which means that their impact cannot be accurately described by linear models. Nonlinear characteristics include the irregularity and complexity of system responses. For example, extreme weather conditions may trigger nonlinear damage modes of equipment, or sudden loads may cause nonlinear power system responses.

[0071] Nonlinear control aims to deal with the nonlinear characteristics existing in the system. Through nonlinear control, the control strategy can be adjusted to adapt to the complex behavior of the system. Specifically, simulation training is performed using power grid operation scenarios containing random abnormal factors to test the response of the nonlinear control model. Based on the simulation results, the control strategy is adjusted to better adapt to the random abnormal factors, including adjusting control parameters and optimizing control algorithms. The nonlinear control method is used to train and optimize the power grid digital twin model. In a constantly changing environment, the power grid digital twin model can effectively cope with various uncertainties and changes by training and adjusting the control strategy.

[0072] In summary, the power grid digital twin modeling method based on cloud-edge collaboration provided by the embodiment of the present application has the following technical effects:

[0073] By integrating distributed sensor networks and edge computing servers, data such as current, voltage, and temperature are collected in real time, and edge computing technology is used for data processing and rapid fault location, which improves the real-time and accuracy of the data. By using 5G dedicated networks and machine learning models, the power grid dispatching strategy is dynamically adjusted, and the power grid digital twin model is updated in real time, making the power grid dispatching more intelligent and able to adapt to various dynamic changes in power grid operation, thereby improving the flexibility and efficiency of power grid operation. By performing local marking in the power grid digital twin model and establishing multiple local update spaces, local updates and resource allocation are performed more accurately, thereby improving the overall performance and resource utilization of the power grid. By matching multiple independent update cycles and priority mechanisms, update requests are adaptively triggered, local updates and resource allocation are optimized, and the overall performance of the system is improved. During the operation of the power grid, the power grid digital twin model is self-learned and iteratively strengthened, and the operation and maintenance management of the target power grid is assisted, thereby improving the intelligence level and adaptability of operation and maintenance management and reducing the complexity and uncertainty of management.

[0074] Embodiment 2 is based on the same inventive concept as the power grid digital twin modeling method based on cloud-edge collaboration in the previous embodiment. Figure 2 As shown, the embodiment of the present application provides a power grid digital twin modeling platform based on cloud-edge collaboration, and the platform includes:

[0075] A sensor network integration module 10, wherein the sensor network integration module 10 is used to integrate a distributed sensor network, wherein the distributed sensor network is communicatively connected to a plurality of real-time monitoring units, wherein the monitoring indicators corresponding to the plurality of real-time monitoring units include current, voltage, and temperature; a rapid fault location module 20, wherein the rapid fault location module 20 is used to deploy an edge computing server at a key node of a target power grid, wherein the edge computing server integrates augmented reality technology to upload the real-time monitoring data corresponding to the plurality of real-time monitoring units and perform rapid fault location; a digital twin model construction module 30, wherein the digital twin model construction module 30 is used to utilize a 5G dedicated network to interact between the edge computing server and the cloud processor, and to dynamically adjust the dispatching strategy of the target power grid using a machine learning model to construct a power grid digital twin model, wherein the power grid digital twin model includes a power grid topology, equipment status, and operating parameters; and a local update space establishment module. 40. The local update space establishment module 40 is used to perform local marking in the power grid digital twin model based on the power grid digital twin model, combined with the edge computing server and multiple real-time monitoring units associated with the mapping, and establish multiple local update spaces. Each of the multiple local update spaces has an overlapping part; the local update request trigger module 50 is used to match multiple independent update cycles and priority mechanisms with the multiple local update spaces based on a cloud processor, and adaptively trigger multiple local update requests corresponding to the multiple local update spaces. The multiple local update requests are mapped one-to-one with the multiple local update spaces; the operation and maintenance management module 60 is used to perform self-learning iterative reinforcement on the power grid digital twin model with the multiple local update spaces and multiple local update requests during the operation of the target power grid, and assist in the operation and maintenance management of the target power grid.

[0076] Furthermore, the platform also includes a priority mechanism update module to perform the following operation steps:

[0077] Based on Zigbee wireless sensors, multimodal data are collected, and the monitoring indicators corresponding to the multimodal data include acoustic type indicators, vibration type indicators, temperature type indicators, humidity type indicators, and electromagnetic field type indicators; the multimodal data are preprocessed by a signal processing algorithm, and fuzzy logic fusion technology is used to perform data fusion of probabilistic reasoning to determine multiple independent update decisions corresponding to multiple local update spaces; based on the multiple independent update decisions and multiple independent update cycles corresponding to the multiple local update spaces, the priority mechanism is triggered to update.

[0078] Furthermore, the platform also includes a trigger condition adding module to perform the following operation steps:

[0079] Connect the data storage module to obtain historical operation instances, which include equipment failure data and equipment maintenance data; based on the historical operation instances, apply supervised learning methods to perform fault pattern recognition, combine time series analysis to perform fault prediction, add trigger conditions corresponding to the multiple local update spaces, and the trigger conditions are associated with key performance indicators of the target power grid, which include downtime, equipment load rate, and energy efficiency loss.

[0080] Furthermore, the platform also includes a self-learning iterative reinforcement module to perform the following operation steps:

[0081] Based on the historical operation examples, M normal power grid operation scenarios and N abnormal power grid operation scenarios are simulated; through the M normal power grid operation scenarios, the power grid digital twin model is self-learned and iteratively strengthened under the constraints of a regular trigger mode; through the N abnormal power grid operation scenarios, the power grid digital twin model is self-learned and iteratively strengthened under the constraints of an emergency trigger mode.

[0082] Furthermore, the platform also includes an adaptive enhancement training module to perform the following steps:

[0083] Data mining technology is used to identify energy consumption patterns and set energy consumption cycle fluctuation channels. Based on the energy consumption cycle fluctuation channels, sliding mode control laws are added in combination with timing nodes to predict periodic changes in normal operating data. Under the constraints of the regular triggering mode and in combination with the periodic changes in the normal operating data, sliding mode control is used to perform adaptive enhancement training on the periodic changes of the digital twin model of the power grid.

[0084] Furthermore, the platform also includes an operation status monitoring module to perform the following operation steps:

[0085] By using mixed integer programming, a membership degree between 0 and 1 is assigned to each parameter through a membership function to construct an energy-saving logic rule base, which is used to describe the energy consumption characteristics of the target power grid under different operating conditions; based on the energy-saving logic rule base, energy consumption is inferred by fuzzy logic operation, and the energy-saving optimization strategy corresponding to the energy consumption mode is configured; using closed-loop control logic, with the goal of maximizing energy efficiency and power supply reliability, combined with the energy-saving optimization strategy corresponding to the energy consumption mode, the operating state of the target power grid is continuously monitored and adjusted.

[0086] Furthermore, the platform also includes an enhanced training module to perform the following steps:

[0087] Under the constraints of the emergency trigger mode, random abnormal factors are introduced, and the random abnormal factors include extreme meteorological factors; based on the random abnormal factors and the N abnormal power grid operation scenarios, a risk assessment framework is used to analyze potential negative impacts; the potential negative impacts are arranged in order from large to small, and the resource allocation of the overlapping part of each local update space in the multiple local update spaces is optimized, and the resource allocation includes routing power flow allocation and load allocation; after the resource allocation of the overlapping part of each local update space in the multiple local update spaces is optimized, according to the nonlinear characteristics of the random abnormal factors, nonlinear control is used to perform adaptive enhancement training on the random changes of the power grid digital twin model.

[0088] Through the above-mentioned detailed description of the power grid digital twin modeling method based on cloud-edge collaboration in this specification, technical personnel in this field can clearly understand the power grid digital twin modeling platform based on cloud-edge collaboration in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0089] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A power grid digital twin modeling method based on cloud-edge collaboration, characterized in that: The method comprises: An integrated distributed sensor network is communicatively connected to a plurality of real-time monitoring units, and the monitoring indicators corresponding to the plurality of real-time monitoring units include current, voltage, and temperature; Deploy an edge computing server at a key node of the target power grid, wherein the edge computing server integrates augmented reality technology to upload the real-time monitoring data corresponding to the multiple real-time monitoring units and perform rapid fault location; Using a 5G dedicated network, the edge computing server interacts with the cloud processor, and a machine learning model is used to dynamically adjust the dispatching strategy of the target power grid to build a power grid digital twin model, which includes a power grid topology, equipment status, and operating parameters; Based on the power grid digital twin model, in combination with the edge computing server and multiple real-time monitoring units associated with the mapping, local marking is performed in the power grid digital twin model to establish multiple local update spaces, and each of the multiple local update spaces has an overlapping part; Based on the cloud processor, matching multiple independent update cycles and priority mechanisms with the multiple local update spaces, adaptively triggering multiple local update requests corresponding to the multiple local update spaces, wherein the multiple local update requests are mapped one-to-one with the multiple local update spaces; During the operation of the target power grid, the multiple local update spaces and the multiple local update requests are used to perform self-learning iterative reinforcement on the power grid digital twin model, and assist the operation and maintenance management of the target power grid; Wherein, a distributed sensor network is integrated, and the distributed sensor network is communicatively connected with a plurality of real-time monitoring units, including: Based on Zigbee wireless sensors, multimodal data is collected, and the monitoring indicators corresponding to the multimodal data include acoustic type indicators, vibration type indicators, temperature type indicators, humidity type indicators, and electromagnetic field type indicators; Preprocessing the multimodal data by using a signal processing algorithm, and performing data fusion of probability reasoning using fuzzy logic fusion technology to determine multiple independent update decisions corresponding to multiple local update spaces; Based on the multiple independent update decisions and the multiple independent update cycles corresponding to the multiple local update spaces, triggering an update to the priority mechanism; Based on the multiple independent update decisions and the multiple independent update cycles corresponding to the multiple local update spaces, the priority mechanism is triggered to update, including: Connecting to the data storage module to obtain historical operation instances, wherein the historical operation instances include equipment failure data and equipment maintenance data; Based on the historical operation instances, supervised learning methods are applied to identify fault patterns, and time series analysis is combined to predict faults. Trigger conditions corresponding to the multiple local update spaces are added. The trigger conditions are associated with key performance indicators of the target power grid, and the key performance indicators include downtime, equipment load rate, and energy efficiency loss.

2. The power grid digital twin modeling method based on cloud-edge collaboration according to claim 1, characterized in that: The power grid digital twin model is self-learned and iteratively strengthened to assist the operation and maintenance management of the target power grid. The method includes: Based on the historical operation examples, simulating M normal power grid operation scenarios and N abnormal power grid operation scenarios; Through the M normal power grid operation scenarios, the power grid digital twin model is subjected to self-learning iterative reinforcement under the constraints of a regular triggering mode; Through the N abnormal power grid operation scenarios, the power grid digital twin model is subjected to self-learning iterative reinforcement under the constraints of the emergency trigger mode.

3. The power grid digital twin modeling method based on cloud-edge collaboration as claimed in claim 2, characterized in that: Through the M normal power grid operation scenarios, the power grid digital twin model is subjected to self-learning iterative reinforcement under the constraints of a regular triggering mode, and the method includes: Use data mining technology to identify energy consumption patterns and set up energy consumption cycle fluctuation channels; Based on the energy consumption cycle fluctuation channel, a sliding mode control law is added in combination with the timing nodes to predict the periodic changes of normal operation data; Under the constraints of the regular triggering mode, combined with the periodic changes of the normal operating data, sliding mode control is used to perform adaptive enhancement training on the periodic changes of the power grid digital twin model.

4. The power grid digital twin modeling method based on cloud-edge collaboration as claimed in claim 3 is characterized in that: Based on the energy consumption cycle fluctuation channel, a sliding mode control law is added in combination with a time series node to predict the periodic changes of normal operation data. The method includes: Using mixed integer programming, assigning a degree of membership between 0 and 1 to each parameter through a membership function, and constructing an energy-saving logic rule base, wherein the energy-saving logic rule base is used to describe the energy consumption characteristics of the target power grid under different operating conditions; Based on the energy-saving logic rule base, the energy consumption is inferred by fuzzy logic operation, and the energy-saving optimization strategy corresponding to the energy consumption mode is configured; By using closed-loop control logic, with the goal of maximizing energy efficiency and power supply reliability, combined with the energy-saving optimization strategy corresponding to the energy consumption mode, the operating status of the target power grid is continuously monitored and adjusted.

5. The power grid digital twin modeling method based on cloud-edge collaboration as claimed in claim 2, characterized in that: Through the N abnormal power grid operation scenarios, the power grid digital twin model is subjected to self-learning iterative reinforcement under the constraint of an emergency trigger mode, and the method includes: Under the constraint of the emergency trigger mode, random abnormal factors are introduced, and the random abnormal factors include extreme meteorological factors; Based on the random abnormal factors and the N abnormal grid operation scenarios, a risk assessment framework is used to analyze potential negative impacts; Arrange the potential negative impacts in descending order, and optimize resource allocation of the overlapping portion of each local update space in the multiple local update spaces, wherein the resource allocation includes routing power flow allocation and load allocation; After optimizing the resource allocation of the overlapping part of each local update space in the multiple local update spaces, adaptive enhancement training is performed on the random changes of the power grid digital twin model using nonlinear control according to the nonlinear characteristics of the random abnormal factors.

6. A power grid digital twin modeling platform based on cloud-edge collaboration, characterized by: For implementing the power grid digital twin modeling method based on cloud-edge collaboration according to any one of claims 1 to 5, the platform comprises: A sensor network integration module, wherein the sensor network integration module is used to integrate a distributed sensor network, wherein the distributed sensor network is communicatively connected with a plurality of real-time monitoring units, and the monitoring indicators corresponding to the plurality of real-time monitoring units include current, voltage, and temperature; A fast fault location module, the fast fault location module is used to deploy an edge computing server at a key node of the target power grid, the edge computing server integrates augmented reality technology to upload the real-time monitoring data corresponding to the multiple real-time monitoring units and perform fast fault location; A digital twin model construction module, wherein the digital twin model construction module is used to utilize a 5G dedicated network to interact with the edge computing server and the cloud processor, and to dynamically adjust the dispatching strategy of the target power grid using a machine learning model to construct a power grid digital twin model, wherein the power grid digital twin model includes a power grid topology, device status, and operating parameters; A local update space establishment module, wherein the local update space establishment module is used to perform local marking in the power grid digital twin model based on the power grid digital twin model, in combination with the edge computing server and multiple real-time monitoring units associated with the mapping, and establish multiple local update spaces, wherein each of the multiple local update spaces has an overlapping portion; A local update request triggering module, the local update request triggering module is used to adaptively trigger a plurality of local update requests corresponding to the plurality of local update spaces based on a cloud processor, matching a plurality of independent update cycles and a priority mechanism with the plurality of local update spaces, wherein the plurality of local update requests are mapped one-to-one with the plurality of local update spaces; An operation and maintenance management module, wherein the operation and maintenance management module is used to perform self-learning iterative reinforcement on the power grid digital twin model using the multiple local update spaces and the multiple local update requests during the operation of the target power grid, and to assist in the operation and maintenance management of the target power grid.

Images