Intelligent Inspection Method and System for Power Grid Control Based on Digital Twin

Abstract

This invention provides a digital twin-based intelligent inspection method and system for power grid control. The method includes: Step 1: Constructing a multi-granularity digital twin model for all nodes; Step 2: Establishing a mapping relationship between the physical entities of the power grid control system and the digital twin model; Step 3: Connecting the inspection system with different nodes by integrating a full range of SSH protocols to obtain node data information; Step 4: Calculating the time difference between the clock source node and other nodes based on the obtained node data information, determining whether the time difference exceeds a set threshold, and adjusting the inspection cycle if the time difference exceeds the threshold; Step 5: Constructing a periodic wavelet-like function for predicting CPU utilization and a linear and exponential compound growth model for predicting storage space occupancy based on the obtained node data information, and optimizing resource allocation based on the prediction results. This invention has advantages such as high operating efficiency and high reliability.

Description

Technical Field

[0001] This invention relates to the technical field of equipment digitization and intelligence, specifically to a method and system for intelligent inspection of power grid control based on digital twins. Background Technology

[0002] With the continuous expansion of power system scale and increasing technological complexity, the operational efficiency and stability of power grid control systems face unprecedented challenges. Traditional power grid control methods mainly rely on manual remote inspections, which are not only time-consuming and labor-intensive but also have significant limitations. For example, the power company's intelligent dispatch control system (D5000 including backup dispatch) has 60 nodes, and each inspection takes 2 to 4 hours. This not only leads to untimely data collection but may also affect the safe operation of the power grid due to delayed detection of potential faults. Furthermore, due to the lack of an effective unified monitoring window, time deviations are difficult to detect quickly when nodes or NTP services go offline, thus affecting the accurate delivery of power grid data and the rapid judgment of tripping events.

[0003] In terms of data management and processing, the power grid control system generates a large amount of logs, temporary data, and anomaly analysis files during operation. If this data is not effectively managed and cleaned, it will consume a significant amount of system resources, causing a decrease in system response speed and even resulting in system freezes, severely interfering with the work efficiency and accident handling capabilities of on-duty personnel. Therefore, improving the inspection efficiency and data processing capabilities of the power grid control system has become an urgent problem to be solved.

[0004] In recent years, digital twin technology, as an emerging information technology tool, has attracted widespread attention due to its ability to deeply integrate the physical and digital worlds. In particular, applying digital twin technology to power grid control systems enables real-time monitoring and accurate simulation of power grid operation, providing strong technical support for fault detection, fault diagnosis, dispatch optimization, and status assessment. For example, in Chinese patent application CN 118536835 A, entitled "Intelligent Prediction and Management Platform Based on the Combination of Digital Twin and Large Model," the platform utilizes digital twin technology to accurately model and efficiently simulate complex systems, improving the system's intelligence level.

[0005] However, existing digital twin technologies still have some shortcomings when applied to power grid control systems. First, most research focuses on macro-level modeling and simulation, lacking refined modeling for specific application scenarios, especially research on multi-granularity digital twin models that allow for application (service) inspection. Second, regarding clock synchronization, although some studies have proposed solutions based on NTP services, they haven't provided specific methods for calculating clock deviations, which to some extent limits the effectiveness of digital twin technology in power grid control. Finally, for the large amounts of data generated by power grid control systems, existing technologies still have significant room for improvement in data processing efficiency and resource optimization. Summary of the Invention

[0006] The technical problem to be solved by this invention is to provide a digital twin-based intelligent inspection method and system for power grid control that is highly efficient and reliable. It can monitor key indicators such as CPU load rate and memory usage of equipment in the data center in real time. It can also optimize the overall resource allocation through in-depth data analysis and detect and warn of potential fault points in advance, thereby effectively improving the operating efficiency and security of the data center.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0008] A smart inspection method for power grid control based on digital twins includes the following steps:

[0009] Step 1: Construct a full-node, multi-granularity digital twin model;

[0010] The physical entities of the power grid control system are divided into multiple granularities, and digital twin model components corresponding to each granularity are created and loaded to construct a digital twin model.

[0011] Step 2: Establish the mapping relationship between the physical entities of the power grid control system and the digital twin model;

[0012] Information on all nodes of the physical entity of the power grid control system is stored in a configuration file in a hierarchical manner. The information in the configuration file is associated with the component information of the digital twin model through the node name, so as to realize the mapping at different granularities.

[0013] Step 3: Obtaining node data;

[0014] By integrating all versions of the SSH protocol, the inspection system can connect with different nodes and obtain node data information.

[0015] Step 4: Clock Deviation Detection and Early Warning;

[0016] The time difference between the clock source node and other nodes is calculated based on the acquired node data information. It is then determined whether the time difference exceeds the set threshold. If the time difference exceeds the threshold, the inspection cycle is adjusted.

[0017] Step 5: Resource prediction and optimization;

[0018] Based on the acquired node data, a periodic wavelet-like function is constructed to predict CPU utilization and a linear and exponential compound growth model is constructed to predict storage space occupancy. Resource allocation is then optimized based on the prediction results.

[0019] As a further improvement to the method of the present invention: in step 2, the method for implementing mapping at different granularities includes:

[0020] Step 201: Organize the information of all physical nodes of the power grid control system according to the hierarchy of the node's region, work location, computer room, and cabinet, and store the collected information in the configuration file;

[0021] Step 202: Name the component files using the node names, so that the component file name of each node is consistent with its name in the configuration file;

[0022] Step 203: Read and parse the node information in the configuration file. Based on the node name, read the corresponding component file, parse the parameters in the component file, and associate the information in the configuration file with the information in the component file through the node name.

[0023] Step 204: Load the component files in the corresponding folder through the access interface to realize the mapping between regional physical entities and digital twin models;

[0024] Step 205: Associate the information in the configuration file with the information in the component file through the node name to realize the mapping between node-level physical entities and digital twin models.

[0025] As a further improvement to the method of the present invention: step 2 also includes node state mapping, wherein the node state mapping is to set different component colors for different node-level components and region-level components to display the overall node status; the node-level component color setting is based on the inspection results of the application, clock, CPU, memory, and hard disk to set the color of the node-level components; the region-level component color setting is based on the inspection results and the node affiliation stored in the configuration file, and then a Boolean operation is performed to set the color of the region-level components.

[0026] As a further improvement to the method of the present invention: in step 3, the method for connecting the inspection system with different nodes includes:

[0027] Step 301: Use a socket to establish a Transmission Control Protocol (TCP) connection between the inspection system and the node using the node's IP address and port number. The socket is a network communication method.

[0028] Step 302: Send a request to the node to inquire about the supported SSH protocol version number through the established Transmission Control Protocol (TCP) connection;

[0029] Step 303: After finding the SSH protocol with the same version number, load the protocol and establish an SSH connection.

[0030] As a further improvement to the method of the present invention: in step 4, the method for calculating the clock deviation includes:

[0031] Step 401: The inspection system sends a request to the clock source node to inquire about the current time, and records the current time as the first timestamp;

[0032] Step 402: Use the current time of the information sent back by the clock source as the second timestamp, and use the current time of the information received by the inspection system from the clock source as the third timestamp;

[0033] Step 403: The inspection system sends a request to other nodes to inquire about the current time, and records the current time as the fourth timestamp;

[0034] Step 404: Use the current time of the information sent back by the other nodes as the fifth timestamp, and use the current time of the inspection system receiving the information sent back by the other nodes as the sixth timestamp;

[0035] Step 405: Calculate the time deviation between the clock source and the node based on the first, second, third, fourth, fifth, and sixth timestamps.

[0036] As a further improvement to the method of the present invention: the functional expression of the time deviation is:

[0037] ,

[0038] Among them, t 11 t represents the current time when the inspection system sends a time request to clock source node 1. 21 t represents the current time sent back by clock source node 1. 31 t represents the current time when the inspection system receives the time response sent back by clock source node 1. 1i t represents the current time when the inspection system sends a time request to other node i. 2i t represents the current time sent back by other node i. 3iThis represents the current time when the inspection system receives a time response from another node i.

[0039] As a further improvement to the method of the present invention: in step 5, the step of constructing a periodic wavelet-like function for predicting CPU utilization includes:

[0040] Step 501: Define a periodic wavelet function to describe the variation pattern of CPU utilization:

[0041] Step 502: Based on the operating characteristics of the power grid control system, estimate the range of values ​​for the parameters of the periodic wavelet function;

[0042] Step 503: Determine the parameter calculation step size by combining the required accuracy of the prediction results and the computing power of the inspection system;

[0043] Step 504: Fit the actual historical CPU usage data by combining multiple periodic wavelet functions, and obtain the optimal parameters of the periodic wavelet functions by minimizing the sum of squared errors;

[0044] Step 505: Predict CPU utilization using the sum of multiple periodic wavelet functions containing optimal parameters.

[0045] As a further improvement to the method of the present invention: in step 5, the method for constructing a linear and exponential compound growth model for predicting storage space occupancy includes:

[0046] Step 511: Construct the linear and exponential compound growth model function;

[0047] Step 512: Obtain historical data on node storage space occupancy rate, and process the historical data to obtain a dataset;

[0048] Step 513: Fit the dataset using the linear and exponential compound growth model function to obtain the parameter values ​​of the linear and exponential compound growth model function;

[0049] Step 514: Use the constructed linear and exponential compound growth model function to predict the node storage space occupancy rate at a future time. When the prediction result exceeds the threshold, shorten the inspection cycle and recommend corresponding handling solutions.

[0050] This invention also provides a power grid control intelligent inspection system based on digital twins, including...

[0051] The digital twin model construction and management module is used to divide the physical entities of the power grid control system into multiple granularities, create digital twin model components corresponding to each granularity, and load the components to construct the digital twin model.

[0052] The data communication and protocol adaptation module is used to acquire node data using a full range of secure communication protocols.

[0053] The clock skew monitoring and early warning module is used to calculate clock skew based on dual-node multi-timestamp.

[0054] The node resource prediction and analysis module is used to classify and predict node resources based on periodic wavelet-like models and compound growth models.

[0055] As a further improvement to the method of the present invention: the clock deviation monitoring and early warning module includes:

[0056] The timestamp recording unit is used to record the timestamps of time query requests sent to the clock source node and other nodes;

[0057] The clock skew calculation unit is used to calculate the time difference between the clock source and other nodes based on the timestamp;

[0058] The threshold comparison and early warning unit is used to compare with preset thresholds and mark and suggest actions for nodes that exceed the thresholds.

[0059] Compared with the prior art, the advantages of the present invention are as follows:

[0060] 1. This invention constructs a full-node, multi-granularity digital twin model to achieve comprehensive coverage of all levels and details of the power grid control system, ensuring the accuracy and completeness of the model. By establishing a mapping relationship between physical entities and digital twin models, real-time synchronization between physical entities and digital twin models is achieved. Combined with extensive node data acquisition capabilities, clock deviation detection and early warning mechanisms, and advanced resource prediction and optimization technologies, the invention comprehensively improves the inspection efficiency and reliability of the power grid control system, ensuring the stability and efficient operation of the system.

[0061] 2. By integrating a full range of SSH protocol versions, this invention enables nodes to automatically load the corresponding protocol based on the supported protocol version and establish a secure communication connection. This effectively solves the problem of data acquisition difficulties caused by protocol incompatibility in the prior art, improves the versatility and adaptability of the inspection system, and ensures the stability and reliability of data acquisition.

[0062] 3. This invention employs a dual-node, multi-timestamp calculation method to monitor clock deviation. Compared to traditional single-timestamp or simple time synchronization algorithms, it can more accurately calculate the time difference between the clock source and other nodes, promptly identify nodes whose clock deviation exceeds the threshold, and provide corresponding handling suggestions. This helps improve the clock synchronization accuracy of the power grid control system and ensures the stable operation of the system. Attached Figure Description

[0063] Figure 1This is a flowchart of the intelligent inspection method for power grid control based on digital twins, as described in an embodiment of the present invention.

[0064] Figure 2 This is a schematic diagram of the physical model of the power grid control system according to an embodiment of the present invention.

[0065] Figure 3 This is a schematic diagram of a digital twin model of a power grid control system according to an embodiment of the present invention.

[0066] Figure 4 This is a schematic diagram of the basic unit of the digital twin model of the power grid control system in an embodiment of the present invention.

[0067] Figure 5 This is a schematic diagram illustrating the principle of clock offset calculation based on dual-node multi-timestamp in an embodiment of the present invention. Detailed Implementation

[0068] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0069] like Figure 1 As shown, this embodiment of the intelligent inspection method for power grid control based on digital twins includes the following steps:

[0070] Step 1: Construct a full-node, multi-granularity digital twin model;

[0071] The physical entities of the power grid control system are divided into multiple granular levels. Digital twin model components are created for each granular level, and the components are loaded to construct the digital twin model.

[0072] In a specific application embodiment, constructing an application (service) inspectable multi-granularity digital twin model for all nodes of the power grid control system includes the following steps:

[0073] Step 101: Divide the physical entities of the power grid control system into multi-level granularity;

[0074] The power grid control system is divided into granular levels of 1-5 according to region, work location, computer room, cabinet, and node, such as Figure 2As shown, the 5-level granularity encompasses all nodes of the power grid control system, including information such as applications (services), clocks, CPUs, memory, and hard drives within each node. It is understandable that the granularity can be further refined or merged based on actual needs. For example, as the complexity of the power grid system increases, the existing 5-level granularity can be further refined by dividing the hardware modules within nodes and broader geographical areas, forming 6-level, 7-level, and other finer granularities for more precise monitoring and management; or some lower-level granularities can be appropriately merged to enhance macro-level control and highlight the display and analysis of key areas or critical links.

[0075] Step 102: Create digital twin model components corresponding to multi-level granular physical entities;

[0076] Based on geographical administrative divisions, create components corresponding to each administrative region, such as... Figure 3 As shown in (a); based on the nodes and their locations in the racks and server rooms, fabricate the corresponding components for each node, rack, and server room, such as... Figure 3 As shown in (b), the component is the basic unit of the digital twin model of the power grid control system. Each component includes information such as assembly parameters, geometric parameters, size parameters, and color parameters, such as... Figure 4 As shown, each component is saved as a separate file. The assembly parameters describe the positional relationships between components and can be modified according to actual conditions; the geometric parameters, size parameters, and color parameters describe the geometric shape, size, and color attributes of the components, respectively, and the color parameters can be set according to actual needs.

[0077] Step 103: Load components to build a digital twin model;

[0078] Components for building a regional-level granular digital twin model (hereinafter referred to as regional-level components) are placed in folder D1, and components for building a node-level granular digital twin model (hereinafter referred to as node-level components) are placed in folders G1~Gw, organized by computer room. Access interfaces FD1 and FG1~FGw are established for folders D1 and G1~Gw, respectively. The component files corresponding to each administrative region are loaded through access interface FD1, and the data in each component file is read. Based on the geometric, dimensional, and color parameters, a 3D model is drawn. The positional relationships between the 3D models are determined by assembly parameters, thus constructing a regional-level granular digital twin model of the power grid control system. Similarly, node-level granular digital twin models of the power grid control system are constructed through access interfaces FG1~FGw.

[0079] Step 2: Establish the mapping relationship between the physical entities of the power grid control system and the digital twin model;

[0080] Information on all nodes of the physical entity of the power grid control system is stored in a hierarchical manner in a configuration file. The information in the configuration file is associated with the component information of the digital twin model through the node name, so as to achieve mapping at different granularities.

[0081] In this embodiment, the method for implementing mapping at different granularities includes:

[0082] Step 201: Organize the information of all physical nodes of the power grid control system according to the hierarchy of the node's region, work location, computer room, and cabinet, and store the collected information in the configuration file; the information of all physical nodes of the power grid control system includes node name, IP address, port number, username, access password, etc.

[0083] Step 202: Name the component files using the node names so that the component file name of each node is consistent with its name in the configuration file.

[0084] Step 203: Read and parse the node information in the configuration file. Based on the node name, read the corresponding component file, parse the parameters in the component file, and associate the information in the configuration file with the information in the component file through the node name.

[0085] Step 204: Load the component files in the corresponding folder through the access interface to realize the mapping between regional physical entities and digital twin models;

[0086] Step 205: Associate the information in the configuration file with the information in the component file through the node name to realize the mapping between node-level physical entities and digital twin models.

[0087] Specifically, since the assembly parameters, geometric parameters, size parameters, and color parameters of components are already correlated with the spatial location, shape, size, and color of physical entities during manufacturing, the mapping between geographically granular physical entities and digital twin models can be achieved by loading component files from the corresponding folder through an access interface. Within the same data center, node names are unique; therefore, component files are named using nodes. By reading data from the configuration file, component file names, and information within the components, and then associating this information through the node names, the mapping between node-level physical entities and digital twin models can be achieved.

[0088] This embodiment establishes a precise mapping relationship between the physical entities of the power grid control system and the digital twin model. By using information such as configuration files and node names, various attributes of the physical entities are associated with the component information of the digital twin model, ensuring that the digital twin model can accurately reflect the state changes of the physical entities, thereby providing a more reliable basis for diagnosis and early warning.

[0089] It is understood that this embodiment can incorporate more dimensions of mapping information, such as device functional attributes, business logic relationships, and geospatial location. Through the fusion and mapping of multi-dimensional information, the inherent connection between physical entities and digital twin models can be more comprehensively reflected, providing a foundation for more complex analyses and application scenarios, such as fault propagation analysis based on business logic and resource optimization based on geospatial location.

[0090] In this embodiment, node state mapping is also included. Node state mapping sets different component colors for different node-level components and region-level components to display the overall node status. The node-level component color setting is based on the inspection results of the application, clock, CPU, memory, and hard disk. The region-level component color setting is based on the inspection results and the node affiliation stored in the configuration file, and then performs a Boolean operation to set the region-level component color.

[0091] In this embodiment, by using node status mapping and setting the color of components according to the inspection results, the status of all nodes can be displayed intuitively at both the regional and node levels. This allows maintenance personnel to understand the operation of the power grid control system at a glance, quickly locate abnormal nodes, and improve the efficiency of fault diagnosis.

[0092] It is understood that, in addition to the mapping methods described above, this embodiment can also utilize technologies such as image recognition and pattern recognition in artificial intelligence and machine learning to achieve automatic identification and mapping of physical entities and digital twin models. By automatically identifying features such as the appearance, markings, and location of equipment, the system can automatically associate corresponding digital twin model components, reducing the workload of manual configuration and intervention, and improving the accuracy and timeliness of mapping. Especially in large-scale complex power grid systems, this approach can significantly improve mapping efficiency.

[0093] Step 3: Obtaining node data;

[0094] By integrating all versions of the SSH protocol, the inspection system can connect with different nodes to obtain node data information.

[0095] In this embodiment, all versions of the SSH protocol are set in the same file S1, realizing the integration of all versions of secure communication protocols. The inspection system reads a specific portion of S1 as needed to load a particular version of the SSH protocol.

[0096] In this embodiment, the method for connecting the inspection system with different nodes includes:

[0097] Step 301: Use sockets to establish a Transmission Control Protocol (TCP) connection between the inspection system and the node using the node's IP address and port number. Sockets are a type of network communication.

[0098] Step 302: Send a request to the node to inquire about the supported SSH protocol version number through the established Transmission Control Protocol (TCP) connection.

[0099] Step 303: After finding the SSH protocol with the same version number, load the protocol and establish an SSH connection.

[0100] In this embodiment, the method for obtaining node data through a secure communication protocol includes:

[0101] Step 311: The inspection system sends a command to the node to query the operating system version through the established SSH connection.

[0102] Step 312: Based on the received node operating system version, the inspection system generates and sends an instruction to query detailed node information.

[0103] Step 313: Extract the required data from the node information based on the received node information and its operating system version.

[0104] This embodiment integrates the full range of SSH protocols, which can automatically load the corresponding protocol according to the protocol version supported by the node and establish a secure communication connection. This effectively solves the problem of data acquisition difficulties caused by protocol incompatibility in the prior art, improves the versatility and adaptability of the inspection system, and ensures the stability and reliability of data acquisition.

[0105] Step 4: Clock Deviation Detection and Early Warning;

[0106] The time difference between the clock source node and other nodes is calculated based on the acquired node data information. It is then determined whether the time difference exceeds the set threshold. If the time difference exceeds the threshold, the inspection cycle is adjusted.

[0107] In this embodiment, the method for calculating clock offset includes:

[0108] Step 401: The inspection system sends a request to the clock source node to inquire about the current time, and records the current time as the first timestamp;

[0109] Step 402: Use the current time of the information sent back by the clock source as the second timestamp, and use the current time of the information received by the inspection system from the clock source as the third timestamp;

[0110] Step 403: The inspection system sends a request to other nodes to inquire about the current time, and records the current time as the fourth timestamp;

[0111] Step 404: Use the current time of the information sent back by other nodes as the fifth timestamp, and use the current time of the inspection system receiving the information sent back by other nodes as the sixth timestamp;

[0112] Step 405: Calculate the time deviation between the clock source and the node based on the first, second, third, fourth, fifth, and sixth timestamps.

[0113] like Figure 5 As shown, in a specific application embodiment, the method for calculating clock skew based on dual-node multi-timestamp includes:

[0114] 1) The inspection system sends a request to clock source node 1 to query the current time. When sending the request, the inspection system records the current time as a timestamp t. 11 .

[0115] 2) Upon receiving the current time from clock source node 1, the inspection system records the current time as a timestamp t. 31 The current time sent back by clock source node 1 is used as the timestamp t. 21 .

[0116] 3) The inspection system sends a request to other nodes i to inquire about the current time. When sending the request, the inspection system records the current time as a timestamp t. 1i .

[0117] 4) Upon receiving the current time from another node i, the inspection system records the current time as a timestamp t. 3i Use the current time sent back by other points i as the timestamp t. 2i .

[0118] 5) The function expression for calculating the time difference between the clock source and other nodes i is:

[0119] , (1)

[0120] 6) If Δt 1i If the threshold is exceeded, it is recommended to shorten the inspection cycle and provide corresponding solutions.

[0121] 7) Repeat steps 1) to 6), where other nodes 1 become other nodes 2, until the time difference of all online nodes in the power grid control system has been calculated.

[0122] In this embodiment, a calculation method based on dual nodes and multiple timestamps is used to monitor clock deviation. Compared with traditional single timestamp or simple time synchronization algorithms, it can more accurately calculate the time difference between the clock source and other nodes, promptly detect nodes whose clock deviation exceeds the threshold, and provide corresponding handling suggestions. This helps to improve the clock synchronization accuracy of the power grid control system and ensure the stable operation of the system.

[0123] This embodiment can also combine big data analytics and machine learning techniques to mine and analyze historical clock deviation data, establishing a clock deviation prediction model. By comprehensively considering multi-dimensional data such as power grid operating status, environmental factors, and equipment performance, the changing trend of clock deviation can be predicted in advance, enabling an adaptive clock adjustment strategy. That is, before deviation occurs, the system can automatically adjust clock parameters to avoid system performance degradation or increased failure risk caused by clock deviation, thereby improving the reliability and stability of the power grid.

[0124] Step 5: Resource prediction and optimization;

[0125] Based on the acquired node data, a periodic wavelet-like function is constructed to predict CPU utilization and a linear and exponential compound growth model is constructed to predict storage space occupancy. Resource allocation is then optimized based on the prediction results.

[0126] In this embodiment, the step of constructing a periodic wavelet-like function for predicting CPU utilization includes:

[0127] Step 501: Define a periodic wavelet function to describe the variation pattern of CPU utilization:

[0128] Step 502: Based on the operating characteristics of the power grid control system, estimate the range of values ​​for the parameters of the periodic wavelet function;

[0129] Step 503: Determine the parameter calculation step size by combining the required accuracy of the prediction results and the computing power of the inspection system;

[0130] Step 504: Fit the actual historical CPU usage data by combining multiple periodic wavelet functions, and obtain the optimal parameters of the periodic wavelet functions by minimizing the sum of squared errors;

[0131] Step 505: Predict CPU utilization using the sum of multiple periodic wavelet functions containing optimal parameters.

[0132] In specific application embodiments, the method for predicting the CPU utilization and memory usage of nodes in a power grid control system based on periodic wavelet functions includes:

[0133] 1) For CPU utilization, construct a periodic wavelet-like function:

[0134] (2)

[0135] in, The percentage of resources used by a single application (service) to run. The start time of the operation. For runtime, It is a repeating cycle.

[0136] 2) Based on the operating characteristics of the power grid control system, estimate... 、 、 、 The value range is as follows: [Amp] Min Amp Max ]、[t pMin , t pMax ]、[t RMin , t RMax ]、[t TMin , t TMax ];

[0137] 3) Based on the required accuracy of the forecast results and the computing power of the inspection system, determine... 、 、 、 The calculation step size is Amp. Step t pStep t RStep t TStep ;

[0138] 4) Historical CPU utilization data (Ratio) obtained from power grid control system nodes. CPU (t), denoted as A, is based on the maximum number of applications (services) running in the nodes of the power grid control system. Max Using A Max The sum of formulas (2) is used to approximate Ratio. CPU (t), and use formula (3) to calculate the approximation error. ,when At the minimum value, the corresponding Amp, t p 、t R 、t T The value matrix Mat is the optimal parameter that describes the historical data change pattern of CPU utilization, as shown in formula (5).

[0139] (3)

[0140] in, 、 These are the start and end times of the historical data used for analysis. As shown in formula (4):

[0141] (4)

[0142] in, [Amp Min :Amp Step :Amp Max ], [t pMin :t pStep :t pMax ], [t RMin :t RStep :t RMax ], [t TMin :t TStep :t TMax ]

[0143] (5)

[0144] 5) At some point in the future t f Substituting into equation (6), the node CPU utilization can be predicted.

[0145] (6)

[0146] in,

[0147] (7)

[0148] (8)

[0149] ...

[0150] (9)

[0151] 6) If the predicted CPU utilization exceeds the set threshold, it is recommended to shorten the inspection cycle and recommend corresponding solutions.

[0152] 7) If the latest CPU usage exceeds the set threshold, an alarm will be issued and a corresponding handling solution will be recommended.

[0153] In this embodiment, the method for constructing a linear and exponential compound growth model for predicting storage space occupancy includes:

[0154] Step 511: Construct the linear and exponential compound growth model function;

[0155] Step 512: Obtain historical data on node storage space occupancy, and process the historical data to obtain the dataset;

[0156] Step 513: Fit the dataset using the linear and exponential compound growth model function to obtain the parameter values ​​of the linear and exponential compound growth model function;

[0157] Step 514: Use the constructed linear and exponential compound growth model function to predict the node storage space occupancy rate at a future time. When the prediction result exceeds the threshold, shorten the inspection cycle and recommend corresponding handling solutions.

[0158] In specific application embodiments, the method for predicting the node storage space occupancy rate of a power grid control system based on a compound growth model includes:

[0159] 1) Constructing linear and exponential compound growth model functions

[0160] (10)

[0161] in, Indicates linear growth. This represents exponential growth, where a, b, c, and d are undetermined coefficients, and t is time. This represents the storage space occupancy rate as a function of time t.

[0162] 2) Remove historical data on node storage space occupancy rate S H The data Sd decreases in the middle, thus obtaining the dataset S. H1 .

[0163] 3) Use formula (10) to process dataset S H1 By performing a fitting operation, the values ​​of a, b, c, and d are obtained, in order as follows: ass b ass c ass d ass .

[0164] 4) If the node's storage space usage is reduced manually, including removing files or replacing hard drives, then the latest storage space usage will be set to 'a'. ass .

[0165] 5) At some point in the future t f Substituting into formula (11), we obtain the predicted node storage space occupancy rate.

[0166] (11)

[0167] 6) If the predicted storage space occupancy rate exceeds the set threshold, it is recommended to shorten the inspection cycle and recommend corresponding handling solutions.

[0168] 7) If the latest storage space occupancy rate exceeds the set threshold, an alarm will be issued and a corresponding handling solution will be recommended.

[0169] In this embodiment, periodic wavelet-like functions are used to predict CPU utilization and memory usage, and a compound growth model is used to predict storage space utilization. This approach fully uncovers the inherent patterns and trends in the data, resulting in more accurate and scientific predictions compared to traditional methods. It helps maintenance personnel plan resource allocation in advance, optimize the performance of the power grid control system, and avoid system failures caused by insufficient resources.

[0170] It is understood that this embodiment can integrate deep learning and reinforcement learning techniques on the basis of existing periodic wavelet-like and compound growth models to further improve the accuracy and intelligence of resource prediction. Deep learning can automatically mine complex features and nonlinear relationships in data, while reinforcement learning can dynamically adjust and optimize based on prediction results and actual feedback, enabling the prediction model to better adapt to the dynamic changes and uncertainties in resource use in the power grid control system. For example, in response to sudden load increases or equipment failures, it can more accurately predict resource demand and take corresponding measures in advance.

[0171] In this embodiment, in addition to the operational data of the power grid control system itself, more external data can be integrated for resource forecasting, such as economic development data, social activity data, and climate change data. These data have potential correlations with power grid load and resource usage. Through the fusion analysis of multi-source data, a more comprehensive understanding of the changing trends in resource demand can be achieved, improving the accuracy and reliability of forecasts. For example, based on information about large-scale events held in the city, the increase in power grid load can be predicted in advance, allowing for the rational allocation of power generation and transmission resources.

[0172] This embodiment also provides a power grid control intelligent inspection system based on digital twins, including:

[0173] The digital twin model construction and management module is used to divide the physical entities of the power grid control system into multiple granularities, create digital twin model components corresponding to each granularity, and load the components to build the digital twin model.

[0174] The data communication and protocol adaptation module is used to acquire node data using a full range of secure communication protocols.

[0175] The clock skew monitoring and early warning module is used to calculate clock skew based on dual-node multi-timestamp.

[0176] The node resource prediction and analysis module is used to classify and predict node resources based on periodic wavelet-like models and compound growth models.

[0177] In this embodiment, the digital twin model construction and management module includes:

[0178] The component library management submodule stores and manages a massive number of digital twin model components, covering regional components and node components created based on physical entities at various granularities. The component library features a categorized indexing function, storing component files according to multi-dimensional tags such as administrative region, equipment room name, and node name, ensuring that component files can be quickly retrieved and accessed. Simultaneously, a component update mechanism is implemented. When the power grid control system undergoes equipment upgrades, additions, or layout adjustments, the assembly parameters, geometric parameters, size parameters, and color parameters of the corresponding components are updated promptly based on the new physical entity parameters and geometric dimensions, ensuring consistency between the digital twin model and the real-world entities.

[0179] The model assembly and rendering engine leverages graphics processing and modeling algorithms to efficiently construct 3D models after reading component data from corresponding folders via an access interface. On one hand, it precisely outlines the components based on geometric and dimensional parameters, and uses color parameters to give them a realistic appearance. On the other hand, it accurately positions the spatial relative locations of each component based on assembly parameters, assembling scattered regional and node components into a complete and hierarchical digital twin model of the power grid control system. This engine supports real-time rendering, dynamically displaying model status changes as data is updated, providing a clear and intuitive visual interface for power grid control system personnel.

[0180] In this embodiment, the data communication and protocol adaptation module has a built-in library of secure communication protocols covering a full range of versions, centrally managing and storing files for various SSH protocol versions. The secure communication protocol library is continuously updated and maintained to ensure it covers both older and newer protocol types that may be used by various nodes in the power grid control system. When the inspection system establishes communication with a node, it quickly locates and loads the compatible protocol in the secure communication protocol library based on the protocol version number returned by the node, seamlessly connecting the communication link and ensuring smooth and stable data interaction.

[0181] In this embodiment, by centrally managing and storing a full range of secure communication protocol libraries, and quickly adapting and loading the corresponding protocols based on the protocol version numbers fed back by the nodes, smooth and stable communication between the inspection system and each node of the power grid control system can be effectively guaranteed. This mechanism not only improves the system's compatibility but also enhances its reliability and security.

[0182] In this embodiment, the clock deviation monitoring and early warning module includes:

[0183] The timestamp recording unit is used to record the timestamps of time query requests sent to the clock source node and other nodes;

[0184] The clock skew calculation unit is used to calculate the time difference between the clock source and other nodes based on the timestamp;

[0185] The threshold comparison and early warning unit is used to compare with preset thresholds and mark and suggest actions for nodes that exceed the thresholds.

[0186] In this embodiment, the clock deviation monitoring and early warning module automatically and periodically sends time query requests to the clock source node and other online nodes according to the method flow of calculating clock deviation based on dual nodes and multiple timestamps. It accurately records a series of timestamps and calculates the time difference. By comparing with a preset clock deviation threshold, nodes that exceed the threshold are marked, and the results are fed back to the visual operation and maintenance interface in real time to alert the power grid control system staff, enabling them to adjust the node clocks according to the deviation.

[0187] In this embodiment, the node resource prediction and analysis module includes:

[0188] The periodic wavelet analysis engine unit is used to analyze CPU utilization and memory usage data. By utilizing a constructed periodic wavelet function, it delves into the inherent periodic fluctuation characteristics of these data. Historical monitoring data is imported into the engine in time series format. Through complex mathematical transformations and parameter fitting, the engine extracts frequency components and amplitude information associated with the operating cycles of different applications (services), predicts future utilization trends, and assesses the strain on node computing resources in advance, providing a quantitative basis for load balancing and allocation.

[0189] The compound growth model computation unit is used for predicting node storage space occupancy. It employs a linear and exponential compound growth model function, combined with historical node storage growth data, to fit the optimal model parameters. Considering that data growth is influenced by factors such as explosive growth in business data volume and the accumulation of log files, the model dynamically adjusts the weights of linear and exponential growth to accurately predict when storage space will be exhausted. This alerts power grid control system personnel to promptly clean up redundant data and expand storage capacity, ensuring the safe and stable storage of data in the power grid control system.

[0190] In this embodiment, the power grid control intelligent inspection system based on digital twins also includes a human-machine interface module. The human-machine interface includes an operating condition switching viewing area, a function bar, a company location display area, a storage space usage area, a node display area, a digital twin model display area, a CPU usage area, a memory usage area, and an alarm area, and has good human-machine interaction performance.

[0191] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A smart inspection method for power grid control based on digital twins, characterized in that, Includes the following steps: Step 1: Construct a full-node, multi-granularity digital twin model; The physical entities of the power grid control system are divided into multiple granularities, and digital twin model components corresponding to each granularity are created and loaded to construct a digital twin model. Step 2: Establish the mapping relationship between the physical entities of the power grid control system and the digital twin model; Information on all nodes of the physical entity of the power grid control system is stored in a configuration file in a hierarchical manner. The information in the configuration file is associated with the component information of the digital twin model through the node name, so as to realize the mapping at different granularities. Step 3: Obtaining node data; By integrating all versions of the SSH protocol, the inspection system can connect with different nodes and obtain node data information. Step 4: Clock Deviation Detection and Early Warning; The time difference between the clock source node and other nodes is calculated based on the acquired node data information. It is then determined whether the time difference exceeds the set threshold. If the time difference exceeds the threshold, the inspection cycle is adjusted. Step 5: Resource prediction and optimization; Based on the obtained node data, a periodic wavelet-like function is constructed to predict CPU utilization and a linear and exponential compound growth model is constructed to predict storage space occupancy. Resource allocation is then optimized based on the prediction results. In step 2, the methods for implementing mapping at different granularities include: Step 201: Organize the information of all physical nodes of the power grid control system according to the hierarchy of the node's region, work location, computer room, and cabinet, and store the collected information in the configuration file; Step 202: Name the component files using the node names, so that the component file name of each node is consistent with its name in the configuration file; Step 203: Read and parse the node information in the configuration file. Based on the node name, read the corresponding component file, parse the parameters in the component file, and associate the information in the configuration file with the information in the component file through the node name. Step 204: Load the component files in the corresponding folder through the access interface to realize the mapping between regional physical entities and digital twin models; Step 205: Associate the information in the configuration file with the information in the component file through the node name to realize the mapping between node-level physical entities and digital twin models; In step 5, the method for constructing a linear and exponential compound growth model for predicting storage space occupancy includes: Step 511: Construct the linear and exponential compound growth model function; Step 512: Obtain historical data on node storage space occupancy, and process the historical data to obtain the dataset; Step 513: Fit the dataset using the linear and exponential compound growth model function to obtain the parameter values ​​of the linear and exponential compound growth model function; Step 514: Use the constructed linear and exponential compound growth model function to predict the node storage space occupancy rate at a future time. When the prediction result exceeds the threshold, shorten the inspection cycle and recommend corresponding handling solutions.

2. The intelligent power grid control and inspection method based on digital twins according to claim 1, characterized in that, Step 2 also includes node state mapping, which sets different component colors for different node-level components and region-level components to display the overall node status. The node-level component color setting is based on the inspection results of the application, clock, CPU, memory, and hard disk. The region-level component color setting is based on the inspection results and the node affiliation stored in the configuration file, and then performs a Boolean operation to set the region-level component color.

3. The intelligent power grid control and inspection method based on digital twins according to claim 1, characterized in that, In step 3, the method for connecting the inspection system with different nodes includes: Step 301: Use a socket to establish a Transmission Control Protocol (TCP) connection between the inspection system and the node using the node's IP address and port number. The socket is a network communication method. Step 302: Send a request to the node to inquire about the supported SSH protocol version number through the established Transmission Control Protocol (TCP) connection; Step 303: After finding the SSH protocol with the same version number, load the protocol and establish an SSH connection.

4. The intelligent power grid control and inspection method based on digital twins according to claim 1, characterized in that, In step 4, the method for calculating the clock offset includes: Step 401: The inspection system sends a request to the clock source node to inquire about the current time, and records the current time as the first timestamp; Step 402: Use the current time of the information sent back by the clock source as the second timestamp, and use the current time of the information received by the inspection system from the clock source as the third timestamp; Step 403: The inspection system sends a request to other nodes to inquire about the current time, and records the current time as the fourth timestamp; Step 404: Use the current time of the information sent back by the other nodes as the fifth timestamp, and use the current time of the inspection system receiving the information sent back by the other nodes as the sixth timestamp; Step 405: Calculate the time deviation between the clock source and the node based on the first, second, third, fourth, fifth, and sixth timestamps.

5. The intelligent power grid control and inspection method based on digital twins according to claim 4, characterized in that, The time deviation The function expression is: ， Among them, t 11 t represents the current time when the inspection system sends a time request to clock source node 1. 21 t represents the current time sent back by clock source node 1. 31 t represents the current time when the inspection system receives the time response sent back by clock source node 1. 1i t represents the current time when the inspection system sends a time request to other node i. 2i t represents the current time sent back by other node i. 3i This represents the current time when the inspection system receives a time response from another node i.

6. The intelligent power grid control and inspection method based on digital twins according to claim 1, characterized in that, In step 5, the step of constructing a periodic wavelet function for predicting CPU utilization includes: Step 501: Define a periodic wavelet function to describe the variation pattern of CPU utilization: Step 502: Based on the operating characteristics of the power grid control system, estimate the range of values ​​for the parameters of the periodic wavelet function; Step 503: Determine the parameter calculation step size by combining the required accuracy of the prediction results and the computing power of the inspection system; Step 504: Fit the actual historical CPU usage data by combining multiple periodic wavelet functions, and obtain the optimal parameters of the periodic wavelet functions by minimizing the sum of squared errors; Step 505: Predict CPU utilization using the sum of multiple periodic wavelet functions containing optimal parameters.

7. A power grid control intelligent inspection system based on digital twins, used to implement the power grid control intelligent inspection method based on digital twins as described in claim 1, characterized in that, include The digital twin model construction and management module is used to divide the physical entities of the power grid control system into multiple granularities, create digital twin model components corresponding to each granularity, and load the components to construct the digital twin model. The data communication and protocol adaptation module is used to acquire node data using a full range of secure communication protocols. The clock skew monitoring and early warning module is used to calculate clock skew based on dual-node multi-timestamp. The node resource prediction and analysis module is used to classify and predict node resources based on periodic wavelet-like and compound growth models. The clock deviation monitoring and early warning module includes: The timestamp recording unit is used to record the timestamps of time query requests sent to the clock source node and other nodes; The clock skew calculation unit is used to calculate the time difference between the clock source and other nodes based on the timestamp; The threshold comparison and early warning unit is used to compare with preset thresholds and mark and suggest actions for nodes that exceed the thresholds.

Images