Smart grid status monitoring method and system based on digital twin

Abstract

The present invention proposes a smart grid state monitoring method and system based on digital twins. The method includes: acquiring multi-source data from the grid operating environment to form a scenario feature vector and a corresponding operating scenario label; selecting a basic twin model template based on the operating scenario label, mapping the scenario feature vector to node parameters within the basic twin model template, and obtaining a corresponding parameter set and digital twin model instance; constructing a causal graph structure of the operating status between devices based on the digital twin model instance to reflect the causal dependency paths between the current grid state variables; performing trend risk judgment based on the abnormal causal path set, and outputting structured early warning response information. The present invention effectively solves the problem of insufficient generalization capability caused by the static nature of traditional twin structures.

Description

Technical Field

[0001] The present invention belongs to the field of digital twins, and in particular relates to a smart grid status monitoring method and system based on digital twins. Background Art

[0002] With the continued expansion of smart grids and the highly dynamic nature of their operating environments, traditional power status monitoring technologies, based on static rules and fixed structures, face significant challenges. In multi-source, heterogeneous environments, grid operating conditions are often affected by a combination of complex factors, such as weather changes, load fluctuations, and operational strategy switching. This leads to highly nonlinear and time-varying device operating conditions.

[0003] Although digital twin technology has been gradually introduced into power grid state modeling and monitoring in recent years, used to build virtual mirror systems of the power grid, most twin models employ fixed structures and parameter settings, making them difficult to dynamically adjust for different operating scenarios. This is particularly true when dealing with non-stationary operating conditions such as extreme heat, high-load disturbances, and sudden maintenance. These models exhibit significant prediction bias and response lag. Furthermore, existing state analysis methods, which are mostly based on statistical features or pattern recognition, lack the ability to model the inter-device linkage mechanisms. This often prevents effective explanations for abnormal conditions and makes it impossible to trace the critical paths that lead to them. Furthermore, the degradation of power grid equipment often exhibits trend-like evolutionary characteristics, such as slowly rising temperatures and approaching current boundaries. These slowly changing risks are difficult to detect in a timely manner through short-term monitoring. Current systems generally lack a modeling mechanism to track and model equipment state evolution trends, making it difficult to identify potential chronic hazards and structural risks.

[0004] Therefore, there is an urgent need for an intelligent monitoring method that can realize the closed loop of operation scenario perception, structural adaptive modeling, abnormal causal analysis and trend risk identification, so as to enhance the stability, interpretability and predictability of the digital twin system in the actual complex power grid environment. Summary of the Invention

[0005] The purpose of this invention is to propose a smart grid status monitoring method and system based on digital twins to address the above problems.

[0006] In order to achieve the above object, a first aspect of the present invention provides a smart grid status monitoring method based on digital twins, the method comprising the following steps:

[0007] S1. Acquire multi-source data in the power grid operation environment to form a scenario feature vector and a corresponding operation scenario label. The scenario feature vector includes a meteorological data subvector, a load data subvector, and an initial equipment operation status subvector.

[0008] S2. Select a basic twin model template according to the running scenario label, map the scenario feature vector to the node parameters in the basic twin model template, and obtain the corresponding parameter set and digital twin model instance;

[0009] S3. Construct a causal graph structure of the operating status between devices based on the digital twin model instance to reflect the causal dependency paths between the current power grid state variables; based on the causal graph structure, generate a set of abnormal causal paths activated in the current state through abnormal path identification, and attach an abnormal strength index for each abnormal causal path;

[0010] S4. Perform trend risk judgment based on the abnormal causal path set and output structured early warning response information.

[0011] Furthermore, the multi-source data includes external meteorological data, load-side operation data and equipment initial operation status data;

[0012] The external meteorological data comes from the regional meteorological station connected to the power grid and is received through the SCADA system interface; the load-side operation data is generated by the load acquisition module in the trunk substation, including active power, reactive power, three-phase imbalance and voltage fluctuation coefficient; the initial operating status data of the equipment is the basic status parameters of the transformer, busbar and capacitor, which are collected using temperature sensors, infrared thermal imagers, voltage transformers and current transformers.

[0013] Furthermore, the S1 specifically includes:

[0014] Performing a splicing operation on the meteorological data sub-vector, the load data sub-vector, and the initial equipment operation status sub-vector to form a scene feature vector;

[0015] The scene feature vector is input into a shallow neural network structure for scene classification, and the corresponding running scene label is output; wherein the shallow neural network structure includes an input layer, two fully connected layers with ReLU activation and a softmax output layer.

[0016] Furthermore, the basic twin model template is modeled by expert experience;

[0017] Among them, the basic twin model template is a graph structure , where the node set Represents equipment unit, edge set Indicates a physical connection or operational constraint.

[0018] Furthermore, the S2 specifically includes:

[0019] Obtaining the scene feature vector;

[0020] Mapping the scene feature vector into node parameters in the basic twin model template; wherein a parameter adaptation function and an exponential compression model are introduced into the mapping to dynamically adjust the node parameters;

[0021] Output digital twin model instance and parameter set.

[0022] Furthermore, the parameter adaptation function is used to dynamically adjust the operating threshold of each device node in the digital twin model instance by combining the basic rated current capacity, the dimension representing the load enhancement degree in the characteristic vector of the device node, and the operating complexity index of the device node in the digital twin model instance to design the dynamic load upper limit; the exponential compression model designs the dynamic temperature rise threshold by combining the factory-set maximum temperature rise value of the device node and the dimension representing the degree of high temperature environment risk in the characteristic vector.

[0023] Furthermore, the S3 specifically includes:

[0024] Get the digital twin model instance and build the state vector set ,in Representation node Running state variables;

[0025] Node The state of the adjacent nodes is taken as the explained variable. The state of is used as the explanatory variable to construct the causal scoring function:

[0026] ;

[0027] in, Representation node right the strength of the causal influence; Represents the mutual information between node state variables and is used to measure nonlinear dependencies; Indicates the ratio of the current device temperature rise state to its upper limit, which is used to characterize the criticality of the node to abnormal conditions; Indicates the difference in node status Distance, which indicates the degree of deviation from the stable coupling relationship; Representation node The historical average operating status value; Represents a constant, used to avoid the denominator being zero; 、 、 Represents the weight factor of the three items of mutual information strength, temperature rise risk sensitivity, and state deviation penalty;

[0028] Generate a new causal graph structure in the digital twin model instance according to the causal scoring function , where each edge All come with ratings ;

[0029] Identify the node set that deviates from the historical statistical interval from the current observation data , combined with the causal graph, perform maximum path tracing and calculate the abnormal impact strength of the current path. Several paths with the largest abnormal impact strength are output as abnormal causal paths to generate the abnormal causal path set activated in the current state; is an abnormal node, is the observed value of device node j, is the observed average value of device node j, is the standard deviation of the observed value of device node j.

[0030] Furthermore, in the causal graph structure In the ,edges whose weights are lower than the set threshold will be pruned to maintain the sparsity and interpretability of the causal graph structure;

[0031] The abnormal impact intensity is calculated as follows:

[0032] ;

[0033] in, For path The intensity of abnormal impact; is the weighted product of all causal edges in the path, indicating the efficiency of information transmission; It is the degree of abnormality of the endpoint, expressed in standardized form.

[0034] Furthermore, the S4 specifically includes:

[0035] For the abnormal causal path set, a short-term trend estimation is performed on the terminal node state value of each abnormal path to generate a mean change rate;

[0036] A trend risk scoring function is introduced into the mean change rate to obtain the path activation strength; wherein the trend risk scoring function is obtained by combining the activation strength of the abnormal causal path, the growth rate of the endpoint trend, and the system's sensitivity to the trend;

[0037] Rank the activation intensities of all the paths, set thresholds based on system experience or statistical quantiles, and classify them into different risk levels, including green / yellow / orange / red;

[0038] Based on the risk level, a recommended action strategy is generated through a built-in response rule base, combined with the device type, risk level and topological location of the endpoint device node.

[0039] In a second aspect of the present invention, a smart grid status monitoring system based on digital twins is provided, the system comprising:

[0040] A multi-source data acquisition unit is used to acquire multi-source data in the power grid operation environment to form a scenario feature vector and a corresponding operation scenario label. The scenario feature vector includes a meteorological data sub-vector, a load data sub-vector, and an initial equipment operation status sub-vector;

[0041] A digital twin initialization unit is used to select a basic twin model template according to the running scenario label, map the scenario feature vector to the node parameters in the basic twin model template, and obtain the corresponding parameter set and digital twin model instance;

[0042] An anomaly analysis unit is configured to construct a causal graph structure of the operating status between devices based on the digital twin model instance to reflect the causal dependency paths between the current power grid state variables; based on the causal graph structure, generate a set of abnormal causal paths activated in the current state through abnormal path identification, and attach an abnormal strength index for each abnormal causal path;

[0043] The abnormal warning unit is used to make trend risk judgments based on the abnormal causal path set and output structured warning response information.

[0044] The beneficial technical effects of the present invention are at least as follows:

[0045] This paper focuses on the core issues of weak adaptability of state monitoring in smart grids, poor causal traceability, and difficulty in identifying trend risks, and constructs a dynamic digital twin state monitoring system driven by operation scenarios.

[0046] First, the system perceives the current operating scenario through multi-source information fusion, and accordingly drives the structural template selection and fine-grained parameter injection of the digital twin model, forming a dynamic twin instance that can be adjusted as the scene switches, effectively solving the problem of insufficient generalization capability caused by the static nature of traditional twin structures.

[0047] Secondly, the present invention introduces a state causal graph construction mechanism, combining the physical connection relationships between devices with the linkage characteristics of operating states to form a directed graph structure with causal weights and state information injection capabilities. This enables chain identification and path-based attribution of abnormal states, improving the interpretability and accuracy of state judgments. On this basis, the system further analyzes the short-term state trends of the terminal nodes of the abnormal paths and constructs a multi-level risk index based on the upper limits of operating parameters and the risk amplification trend, achieving a fusion judgment and grading of abnormal state and trend risks.

[0048] The system ultimately outputs structured warning response information, providing direct support for control and maintenance strategies, completing a closed-loop monitoring process from real-time scenario identification to response recommendation generation. The synergy of these mechanisms forms a complete digital twin monitoring chain, encompassing scenario-driven, structural adaptation, causal identification, and trend warning capabilities. This represents a significant breakthrough in current technologies for complex power grid state modeling and risk management. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The present invention is further described with reference to the accompanying drawings. However, the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative effort.

[0050] Figure 1 This is a flow chart of the smart grid status monitoring method based on digital twins disclosed in an embodiment of the present invention.

[0051] Figure 2 This is a framework diagram of the smart grid status monitoring system based on digital twins disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The following describes embodiments of the present invention in detail. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended only to explain the present invention and are not to be construed as limiting the present invention.

[0053] Example 1

[0054] like Figure 1 As shown, the embodiment of the present invention provides a smart grid status monitoring method based on digital twins, the method comprising:

[0055] S1. Acquire multi-source data in the power grid operation environment to form a scenario feature vector and a corresponding operation scenario label. The scenario feature vector includes a meteorological data subvector, a load data subvector, and an initial equipment operation status subvector.

[0056] Specifically, this step aims to acquire multi-source data closely related to the grid's operating state in real time from the grid's operating environment, including meteorological information, load characteristics, and initial operating state parameters of key equipment. Based on this data, a scenario feature vector is constructed. A classification model is used to identify the current operating scenario label, providing a decision-making basis for the subsequent adaptive adjustment of the twin model's parameters. This step does not construct a general data preprocessing process, but rather models and classifies scenarios for typical grid operating states (such as high temperature and high load, pre-typhoon maintenance, and seasonal low-load fluctuations). This serves as a clear preparation for subsequent model adjustments rather than analysis, which is the core innovation of the design.

[0057] Furthermore, this step serves as the starting module of the system, and its input is provided by different types of data collection devices deployed in the smart grid:

[0058] External weather data Data from regional weather stations connected to the power grid is received through SCADA system interfaces. For example, a municipal control center can access data from the China Meteorological Administration or a local micrometeorological network. Parameters typically include ambient temperature, humidity, wind speed, and rainfall level. This data is transmitted in a JSON structure updated every five minutes. After preliminary cleaning by edge devices, it is sent to the central computing module. For example, wind speed can be used to determine the risk of high wind speeds impacting overhead power lines.

[0059] Load side operating data Generated by the load collection module in the trunk substation, this data includes active power, reactive power, three-phase imbalance, and voltage fluctuation coefficient, typically sampled once a minute. For example, at a 35kV distribution station, the system automatically extracts the average load and fluctuation rate over the past 10 minutes, reflecting the current level of power shortage.

[0060] Equipment initial operating status data This system primarily collects basic status parameters for core node equipment, such as transformers, busbars, and capacitors. These parameters include winding hotspot temperatures, bushing surface infrared image temperature rise, busbar voltage, and current fluctuations. Data sources include temperature sensors, infrared thermal imagers, voltage transformers, and current transformers. These data are accessed by intelligent terminals within the station and uploaded via an edge computing gateway. The data update cycle varies depending on the device type. For example, winding temperature may be collected every 30 seconds, while infrared thermal image data may be collected every 5 minutes. All data undergoes Min-Max normalization and is then integrated into the model.

[0061] Furthermore, after obtaining the above three types of input data, they are first spliced ​​to form a scene feature vector ,as follows:

[0062] ;

[0063] in: Represents the comprehensive feature vector of the current running scenario, with a dimension of , input into the subsequent scene classification model; Represents a vector concatenation operation, combining input data by dimension; Represents the meteorological data sub-vector, which comes from the weather station interface; Represents the load data sub-vector, which comes from the EMS load acquisition module; Represents the initial device operating status sub-vector, which comes from the sensor acquisition module.

[0064] Furthermore, the concatenated feature vectors are fed into a shallow neural network structure for scene classification. This network consists of an input layer, two fully connected layers with ReLU activation, and a softmax output layer, which outputs the category label of the current scene. The model structure is as follows:

[0065] Input layer: dimension equal to length;

[0066] First hidden layer: fully connected, 64 neurons, activation function is ReLU;

[0067] Second hidden layer: fully connected, 32 neurons, activation function is ReLU;

[0068] Output layer: nodes, corresponding to Class scene, output classification probability vector .

[0069] Final output scene label Determined by the following formula:

[0070] ;

[0071] in: Indicates the first The probability value of the class; The label number of the current running scenario is used as a discrete output variable to input the next step; Indicates the total number of scenario types defined in the system. Usually, it is defined as 5 to 8 categories according to the scheduling specifications, such as high temperature and high load, thunderstorm warning, and planned maintenance.

[0072] Furthermore, the model is deployed at the master station or regional edge computing nodes, supporting status refreshes every five minutes. Its training data consists of historical meteorological load data and manually annotated operation records, ensuring that its output corresponds to actual grid dispatch behavior.

[0073] This step outputs two core variables: : Current operating scenario label, discrete value, representing the current macro operating environment of the system; : Scene feature vector, structured expression of multi-source state, input to the next step for dynamic adjustment of twin parameters.

[0074] This step uses the meteorological, load, and equipment status data available to the project as input to construct a unified scenario feature vector, and uses a shallow neural network for classification and identification to form a structured operation scenario label.

[0075] S2. Select a basic twin model template according to the running scenario label, map the scenario feature vector to the node parameters in the basic twin model template, and obtain the corresponding parameter set and digital twin model instance.

[0076] Specifically, this step has been identified in the running scene and the corresponding eigenvector Based on the Structural selection and dynamic configuration of operating parameters. Traditional twin systems mostly rely on fixed modeling methods, which make it difficult to cope with frequently changing environmental factors and load fluctuation characteristics in the power grid. Especially when faced with complex operating scenarios such as extreme weather (such as high temperature, high humidity, thunderstorms) and seasonal peak loads, they often exhibit problems such as response lag and large model deviation. To this end, this step introduces a dual-channel model adaptation mechanism of scenario-driven + parameter injection, which not only enables the model to have structural selection capabilities, but also can inject fine-grained operating conditions into the structure, thereby improving the dynamic adaptability and operational consistency of the overall digital twin system.

[0077] First, according to the scene label Select a basic twin model template from the twin structure library, denoted as The structure library is modeled by expert experience and includes typical topological structures under common operating conditions, such as high temperature and heavy load in summer, maintenance status or weather warning. Each template is essentially a graph structure. , where the node set Represents equipment units (such as transformers, feeders, busbars), edge sets Represents physical connections or operational constraints (such as current transmission, temperature rise coupling of adjacent equipment, etc.). The structural level is not changed in this step, but personalized parameters are injected into the structure to form an adaptive instance model. .

[0078] Furthermore, to complete the parameter injection, Some key dimensions are mapped to node parameters in the model. The innovation here is that a parameter adaptation function with an operation complexity regularization term is introduced to adjust the operation threshold of each device node, taking into account not only its local working conditions but also its relative operation complexity in the network. For example, for a node Load limit , the adjustment function is designed as follows:

[0079] ;

[0080] in: Representation node The basic rated current capability is obtained from the equipment nameplate parameters or historical data; The dimension representing the load enhancement degree in the feature vector, such as current fluctuation rate, regional load growth rate, etc. It represents the operation complexity index of a node in the network, which can be calculated by the degree, edge weight, and access load number in its graph structure (e.g. ); 、 It represents the adjustment coefficient, which represents the weight of the positive load pressure and operation complexity on the threshold adjustment. The value is as follows , .

[0081] Furthermore, this function embodies a key innovation: it not only responds to the current external load state of the system but also incorporates the relative complexity of nodes in the network topology as a regularization term to prevent the risk of excessive load migration in highly connected nodes. This joint modeling of structural and state information, unlike the existing approach of adjusting device parameters individually, is a key innovative improvement of the present invention.

[0082] Furthermore, for equipment that is sensitive to weather influences (such as outdoor bare conductors and cooling-dependent transformers), the temperature rise threshold should also be set. Adapt to special scenarios and use an exponential compression model to enhance the model's responsiveness to equipment tolerance under extreme weather conditions:

[0083] ;

[0084] in: Representation device The factory-set maximum temperature rise value; The dimension representing the risk level of high temperature environment in the feature vector, such as the number of hours when the current temperature exceeds 30°C or the heat index; It represents the risk adjustment factor, which indicates the degree of suppression of temperature rise tolerance by high temperature. ~ Through the above two parameter adaptation mechanisms, on the one hand, the dynamic response capability of the model to load fluctuations can be enhanced, and on the other hand, the fault tolerance analysis capability of the model under extreme weather conditions can be improved. After the parameter injection is completed, the twin model is formed. It will be used in subsequent causal modeling and anomaly identification steps.

[0085] This step outputs the following two key variables, both of which directly serve the next step of causal graph construction and abnormal path tracing: : The digital twin model instance after scene perception and operation parameter adjustment has the structure and operation constraints under the current working conditions; : parameter set, including all nodes in the model, 、 Dynamic parameters such as the state constraint judgment and risk assessment in the reasoning stage.

[0086] S3. Construct a causal graph structure of the operating status between devices based on the digital twin model instance to reflect the causal dependency path between the current power grid state variables; based on the causal graph structure, generate a set of abnormal causal paths activated in the current state through abnormal path identification, and attach an abnormal strength index for each abnormal causal path.

[0087] Specifically, this step is based on the digital twin model instance built in step 2. and run parameter sets , further establish the causal graph structure of the operating status between devices , to achieve structured identification of abnormal states and explainable tracking of risk evolution paths. Compared with traditional power grid anomaly identification methods (such as fixed threshold-based alarms and sliding window detection), this step introduces a fusion modeling framework of causality + structural state for the first time. It no longer relies on abnormal offset judgments of isolated variables, but instead traces and attributes abnormal states based on the operational impact mechanism between device variables in the entire twin structure network. This method is particularly suitable for complex operating scenarios (such as sudden load changes combined with high temperatures, local equipment degradation, and network chain responses), and significantly improves the sensitivity and credibility of anomaly detection in a multi-source coupling environment.

[0088] Furthermore, the core of this step is to establish a causal diagram that can capture the variable linkage mechanism , and combined with this structure to achieve multi-node collaborative recognition of abnormal states. First, based on the twin model Graph structure , construct the state vector set ,in Representation node Operating state variables such as current, voltage, winding temperature, etc.

[0089] Then, with the node The state of the adjacent nodes is taken as the explained variable. The state of is used as the explanatory variable and the following causal scoring function is constructed:

[0090] ;

[0091] in: Representation node right the strength of the causal influence; The table represents the mutual information between node state variables, which is used to measure nonlinear dependencies and comes from the multi-time state sequence in the historical sliding window; Indicates the ratio of the current device temperature rise state to its upper limit, which is used to characterize the criticality of the node to abnormal conditions; Indicates the difference in node status Distance, which indicates the degree of deviation from the stable coupling relationship; Representation node The historical average operating status value (such as average temperature) is provided by the model playback module; Represents a constant, used to avoid the denominator being zero; 、 、 It represents the weight factor of the three items of mutual information strength, temperature rise risk sensitivity, and state deviation penalty. The actual setting is as follows , , .

[0092] It is important to note that this scoring function embodies three innovations: it introduces nonlinear state dependence (mutual information) into causal graph modeling, which makes up for the shortcomings of traditional linear correlation graph construction;

[0093] The risk status of the equipment itself (i.e., whether it is close to the operating upper limit) is included as part of the causal graph, so that the graph structure can dynamically reflect the impact of critical nodes on the system; a state difference penalty term is introduced to emphasize the consistency of changes and prevent accidental linkage from causing false causal chains.

[0094] Through the above formula, in the entire twin graph structure Generate a new directed causal graph on , where each edge All come with ratings . The edge weight is lower than the set threshold (such as ) will be pruned to maintain the sparsity and interpretability of the causal graph structure.

[0095] Then, abnormal path identification is performed. Observe data at the current moment of the system In the above example, we identify the set of nodes that deviate from the historical statistical interval. , combined with the causal graph to perform maximum path tracing, that is, to find the path from any node Departure, can pass Affects abnormal nodes Path , and calculate the abnormal impact intensity of the path:

[0096] ;

[0097] in: Indicates the path The intensity of abnormal impact; represents the weight product of all causal edges in the path, indicating the efficiency of information transmission; Indicates the degree of abnormality of the endpoint, expressed in a standardized manner.

[0098] Will The largest number of paths are output as abnormal chains, indicating the system's potential fault propagation chain or implicit risk chain.

[0099] This step outputs the following two variables, which provide input for the next step of the early warning generation and proactive intervention module: : Causal graph structure, reflecting the causal dependency path between the current grid state variables; : The set of abnormal causal paths activated in the current state, with the abnormal strength index of each path .

[0100] Understandably, this step maps the existing physical structure in the twin model into a causal graph structure with state logic. A new edge weight function is designed by combining mutual information, state criticality, and coupling stability, and a path activation mechanism is used to identify abnormal evolution chains. This structure-driven + state-injected causal modeling mechanism effectively overcomes the shortcomings of traditional time series methods in reflecting the logic of device linkage. It is particularly suitable for power grid operation scenarios in multi-source dynamic environments (such as sudden high-temperature high loads and localized substation degradation).

[0101] S4. Perform trend risk judgment based on the abnormal causal path set and output structured early warning response information.

[0102] Specifically, the goal of this step is to identify the abnormal causal path in the previous step. Conduct trend risk assessments and output structured early warning response information. Its core task is not to build a new model or pre-process the data again, but to evaluate the potential direction of risk evolution based on the existing abnormal causal path, using the historical evolution trend of state variables, boundary parameters, and path structure information. This assessment then divides the risk levels accordingly and outputs response strategies that can be used by the control system and the operation and maintenance system. This step is in the closed-loop implementation stage of the logic of the present invention and must focus on two aspects: the explicit trend of risk and the generation of intervention recommendations to ensure that the results have decision-making guidance significance.

[0103] Furthermore, this step first estimates the short-term trend of the terminal node state value of each abnormal path. in the past The state value sequence within a sampling period (such as winding temperature and current), and use simple linear difference to calculate the mean rate of change (trend), which is defined as follows:

[0104] ;

[0105] This value indicates the device The current state variable change rate. Sampling data Real-time simulation state playback data from SCADA system or twin system, time interval Usually 1 minute, Can be set to 5~10.

[0106] Next, we introduce a trend risk scoring function , which comprehensively considers the activation intensity of the causal path, the growth rate of the endpoint trend, and the system's sensitivity to the trend:

[0107] ;

[0108] in: represents the path activation intensity, which indicates the comprehensive impact of the path on the abnormal state, and comes from step 3; Representation node The trend growth rate of the state variable indicates whether the risk is amplifying; Indicates the upper limit of the node's allowed temperature rise, from the parameter set in step 2 Obtained in; Indicates the adjustment coefficient, it is recommended to be set to 0.5~1.5;

[0109] It is understandable that this function has two clear innovations: one is to explicitly introduce trend risks into the scoring model, rather than relying solely on the degree of abnormality of the current state; the other is to use the operating limits (such as the upper temperature rise limit) injected into the twin model for normalization to achieve risk comparability between devices and avoid misjudgment.

[0110] Then for all Sort by setting thresholds based on system experience or statistical quantiles to divide risk levels into different levels (green / yellow / orange / red). For example:

[0111] :normal; : Mild warning; : Moderate warning; : Serious warning.

[0112] After the path level is determined, the system uses the built-in response rule library to generate recommended operation strategies based on the device type, risk level and topological location of the terminal node, such as load adjustment, switching to backup equipment, and manual review. For example:

[0113] If the terminal device is an outdoor high-voltage switch, and , it is recommended to arrange emergency load transfer immediately; if the equipment is an indoor main transformer and the risk level is medium, it is recommended that the system prompt the operation and maintenance personnel to check the cooling system; these strategies are defined by the expert library, and the system automatically maps response suggestions based on the input conditions.

[0114] This step outputs the following two variables: : Warning information set, each item contains path number, risk index , risk level, endpoint device ID, and recommended response measures; : The maximum trend risk value in the current system, used to determine whether it is necessary to enter a global emergency state or trigger a control strategy linkage.

[0115] Example 2

[0116] like Figure 2 As shown, an embodiment of the present invention further provides a smart grid status monitoring system based on digital twins, the system comprising:

[0117] The multi-source data acquisition unit 101 is used to acquire multi-source data in the power grid operation environment to form a scenario feature vector and a corresponding operation scenario label. The scenario feature vector includes a meteorological data sub-vector, a load data sub-vector, and an initial equipment operation status sub-vector.

[0118] The digital twin initialization unit 102 is used to select a basic twin model template according to the operating scenario label, map the scenario feature vector to the node parameters in the basic twin model template, and obtain the corresponding parameter set and digital twin model instance;

[0119] The abnormality analysis unit 103 is configured to construct a causal graph structure of the operating status between devices based on the digital twin model instance to reflect the causal dependency paths between the current power grid state variables; based on the causal graph structure, generate a set of abnormal causal paths activated in the current state through abnormal path identification, and attach an abnormality strength index for each abnormal causal path;

[0120] The abnormal warning unit 104 is used to make trend risk judgment based on the abnormal causal path set and output structured warning response information.

[0121] The foregoing description of specific embodiments of the present disclosure is intended to illustrate a method for performing a multi-tasking process. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0122] The systems, devices, modules, or units described in the above embodiments may be implemented by computer chips or entities, or by products having certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0123] For the convenience of description, the above devices are described as being divided into various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0124] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the embodiments of this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0125] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0126] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0127] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0128] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0129] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0130] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0131] It should also be noted that the terms include, comprise, or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "includes a..." does not preclude the presence of additional identical elements in the process, method, commodity, or apparatus that includes the element.

[0132] This specification may be described in the general context of computer-executable instructions, such as program modules, executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media, including storage devices.

[0133] The various embodiments in this specification are described in a progressive manner. Similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the system embodiments are generally similar to the method embodiments, so the description is relatively simple. For relevant parts, refer to the description of the method embodiments.

[0134] Finally, it should be noted that the digital twin-based smart grid status monitoring platform disclosed in the embodiment of the present invention is only a preferred embodiment of the present invention, which is only used to illustrate the technical solution of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, ordinary technicians in this field should understand that it is still possible to modify the technical solutions recorded in the aforementioned embodiments, or to replace some of the technical features therein with equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A smart grid status monitoring method based on digital twins, characterized in that: The method comprises the following steps: S1. Acquire multi-source data in the power grid operation environment to form a scenario feature vector and a corresponding operation scenario label. The scenario feature vector includes a meteorological data subvector, a load data subvector, and an initial equipment operation status subvector. S2. Select a basic twin model template according to the running scenario label, map the scenario feature vector to the node parameters in the basic twin model template, and obtain the corresponding parameter set and digital twin model instance; S3. Construct a causal graph structure of the operating status between devices based on the digital twin model instance to reflect the causal dependency paths between the current power grid state variables; based on the causal graph structure, generate a set of abnormal causal paths activated in the current state through abnormal path identification, and attach an abnormal strength index for each abnormal causal path; S4. Perform trend risk judgment based on the abnormal causal path set and output structured early warning response information; Said S3 specifically includes: Get the digital twin model instance and build the state vector set ,in Representation node Running state variables; Node Status As the explained variable, the adjacent nodes Status As explanatory variables, construct the causal scoring function: ； in, Representation node right the strength of the causal influence; Represents the mutual information between node state variables and is used to measure nonlinear dependencies; Indicates the ratio of the current device temperature rise state to its upper limit, which is used to characterize the criticality of the node to abnormal conditions; Indicates the difference in node status Distance, which indicates the degree of deviation from the stable coupling relationship; Representation node The historical average operating status value; Represents a constant, used to avoid the denominator being zero; 、 、 Represents the weight factor of the three items of mutual information strength, temperature rise risk sensitivity, and state deviation penalty; Generate a new causal graph structure in the digital twin model instance according to the causal scoring function , where each edge All come with ratings ; Identify the node set that deviates from the historical statistical interval from the current observation data , combined with the causal graph, perform maximum path tracing and calculate the abnormal impact strength of the current path. Several paths with the largest abnormal impact strength are output as abnormal causal paths to generate the abnormal causal path set activated in the current state; is an abnormal node, is the observed average value of device node j, is the standard deviation of the observation value of device node j; In the causal graph structure In the ,edges whose weights are lower than the set threshold will be pruned to maintain the sparsity and interpretability of the causal graph structure; The abnormal impact intensity is calculated as follows: ； in, For path The intensity of abnormal impact; is the weighted product of all causal edges in the path, indicating the efficiency of information transmission; It is a standardized expression of the degree of abnormality of the endpoint.

2. The smart grid status monitoring method based on digital twin according to claim 1 is characterized in that: The multi-source data includes external meteorological data, load side operation data and equipment initial operation status data; The external meteorological data comes from the regional meteorological station connected to the power grid; the load-side operation data is generated by the load acquisition module in the trunk substation, including active power, reactive power, three-phase imbalance and voltage fluctuation coefficient; the initial operating status data of the equipment are the basic status parameters of the transformer, busbar and capacitor, which are collected using temperature sensors, infrared thermal imagers, voltage transformers and current transformers.

3. The smart grid status monitoring method based on digital twin according to claim 1 is characterized in that: Said S1 specifically includes: Performing a splicing operation on the meteorological data sub-vector, the load data sub-vector, and the initial equipment operation status sub-vector to form a scene feature vector; The scene feature vector is input into a shallow neural network structure for scene classification, and the corresponding running scene label is output; wherein the shallow neural network structure includes an input layer, two fully connected layers with ReLU activation and a softmax output layer.

4. The smart grid status monitoring method based on digital twin according to claim 1 is characterized in that: The basic twin model template is modeled based on expert experience; Among them, the basic twin model template is a graph structure , where the node set Represents equipment unit, edge set Indicates a physical connection or operational constraint.

5. The smart grid status monitoring method based on digital twin according to claim 1 is characterized in that: Said S2 specifically includes: Obtaining the scene feature vector; Mapping the scene feature vector into node parameters in the basic twin model template; wherein a parameter adaptation function and an exponential compression model are introduced into the mapping to dynamically adjust the node parameters; Output digital twin model instance and parameter set.

6. The smart grid status monitoring method based on digital twin according to claim 5 is characterized in that: The parameter adaptation function is used to dynamically adjust the operating threshold of each device node in the digital twin model instance by combining the basic rated current capacity, the dimension representing the load enhancement degree in the feature vector of the device node, and the operation complexity index of the device node in the digital twin model instance to design a dynamic load upper limit; The exponential compression model designs a dynamic temperature rise threshold by combining the factory-set maximum temperature rise value of the device node and the dimension representing the degree of high temperature environment risk in the feature vector.

7. The smart grid status monitoring method based on digital twin according to claim 1 is characterized in that: Said S4 specifically includes: For the abnormal causal path set, a short-term trend estimation is performed on the terminal node state value of each abnormal path to generate a mean change rate; A trend risk scoring function is introduced into the mean change rate to obtain the path activation strength; wherein the trend risk scoring function is obtained by combining the activation strength of the abnormal causal path, the growth rate of the endpoint trend, and the system's sensitivity to the trend; Rank the activation intensities of all the paths, set thresholds based on system experience or statistical quantiles, and classify them into different risk levels, including green / yellow / orange / red; Based on the risk level, an operation strategy is generated through a built-in response rule base, combined with the device type, risk level and topological location of the endpoint device node.

8. A system for executing the smart grid status monitoring method based on digital twins according to claim 1, characterized in that: The system comprises: A multi-source data acquisition unit is used to acquire multi-source data in the power grid operation environment to form a scenario feature vector and a corresponding operation scenario label. The scenario feature vector includes a meteorological data sub-vector, a load data sub-vector, and an initial equipment operation status sub-vector; A digital twin initialization unit is used to select a basic twin model template according to the running scenario label, map the scenario feature vector to the node parameters in the basic twin model template, and obtain the corresponding parameter set and digital twin model instance; An anomaly analysis unit is configured to construct a causal graph structure of the operating status between devices based on the digital twin model instance to reflect the causal dependency paths between the current power grid state variables; based on the causal graph structure, generate a set of abnormal causal paths activated in the current state through abnormal path identification, and attach an abnormal strength index for each abnormal causal path; The abnormal warning unit is used to make trend risk judgments based on the abnormal causal path set and output structured warning response information.

Images