Power grid infrastructure project scene construction model method and system

Abstract

The invention relates to the technical field of smart power grids, and discloses a power grid infrastructure project scene construction model method and system, and the method comprises the steps: S101, collecting the geographic coordinates of power grid facilities, and delimiting N service areas according to the geographic coordinates; performing layered scene construction according to the delimited service area to obtain a scene source set; collecting monitoring data, and adjusting the scene source set by using an adaptive adjustment algorithm to obtain a scene basic set; and establishing a scene model according to the scene basic set. According to the method, a dynamic and flexible power grid scene construction model is established, the challenge of a modern power grid infrastructure project on flexibility is effectively solved, through a data driving method and technology, a power grid can rapidly adapt to changes, online monitoring and decision optimization can be achieved, and therefore efficient and reliable power supply service is ensured.

Description

Technical Field

[0001] The present application relates to the field of smart grid technology, and in particular to a method and system for constructing a model for a power grid infrastructure project scenario. Background Art

[0002] In today's rapidly developing society, electricity, as the cornerstone of modern life and economic development, faces growing demand and an increasingly complex power supply environment. With the introduction of renewable energy, the popularization of smart grids and the large-scale application of electric vehicles, power grid infrastructure projects must have higher flexibility and adaptability to meet the changing power demand and unpredictable system changes.

[0003] The power grid not only needs to respond to fluctuating power demand in real time, but also needs to adapt to equipment failures, environmental changes, and the diversity of user power consumption patterns. Traditional scenarios modeling methods for power grid infrastructure projects often lack flexibility and cannot quickly adjust power supply strategies and resource allocation. However, over-emphasizing flexibility may lead to inefficient construction of scenario models for power grid infrastructure projects. Summary of the invention

[0004] In order to solve the above problems, an embodiment of the present invention provides a method for constructing a model of a power grid infrastructure project scenario, the method comprising:

[0005] Collect the geographic coordinates of power grid facilities and define N service areas based on the geographic coordinates;

[0006] Construct hierarchical scenarios according to the designated service areas to obtain scenario source sets;

[0007] Collect monitoring data, use an adaptive adjustment algorithm to adjust the scene source set, and obtain the scene base set;

[0008] Build a scene model based on the scene basis set.

[0009] Furthermore, the method for defining the service area includes:

[0010] The geographical coordinates of the power grid facilities are determined as seed points, and the Voronoi diagram is generated based on the seed points using the Voronoi diagram generation technology to obtain the preliminary service area;

[0011] Collecting load data and user distribution data in the preliminary service area, so as to obtain the total load and average load of each preliminary service area;

[0012] A total load threshold and an average load threshold are set, and a high load area and a low load area are defined according to the total load threshold and the average load threshold.

[0013] The high-load area and the low-load area are merged according to the merging strategy to obtain N service areas.

[0014] Furthermore, the method for defining the high load area and the low load area includes:

[0015] High load area:

[0016] Condition 1: The preliminary service area with a total load greater than the total load threshold;

[0017] Condition 2: The preliminary service area with an average load greater than the average load threshold;

[0018] If any of conditions 1 and 2 are met, it is a high load area;

[0019] Low load area:

[0020] Condition 3: The preliminary service area with a total load less than or equal to the total load threshold;

[0021] Condition 4: The average load of the preliminary service area is less than or equal to the load threshold;

[0022] The low load zone is defined as the zone that meets any of conditions 3 and 4.

[0023] Furthermore, the method for constructing a hierarchical scene includes:

[0024] A three-layer structural framework is set up, and the three-layer structural framework includes the first layer, the second layer and the third layer;

[0025] The first layer is a combination of N service areas, which together with the three-dimensional visual construction blueprint within the first layer constitute the first layer;

[0026] The second layer covers a single service area, which together with the three-dimensional visual construction blueprint within the second layer constitutes the second layer;

[0027] The third layer covers the power grid facilities within a single service area, and together with the three-dimensional visual construction blueprint within the third layer, they constitute the third layer;

[0028] Once the three-layer structure framework is filled, it becomes a layered scene.

[0029] Furthermore, the method for obtaining the scene basic set is:

[0030] A clustering algorithm is used to analyze the overall operation status of the first layer for the sensor data within the first layer, identify the high-efficiency area and the low-efficiency area, and obtain the labels of the high-efficiency area and the low-efficiency area;

[0031] In the second layer, the sensor data within the second layer is input into the pre-built load demand assessment model to output the load demand;

[0032] In the third layer, the sensor data within the second layer is input into the pre-built state recognition model to output the classification results; the scene source set is obtained, which includes hierarchical scenes, labels of efficient areas, labels of inefficient areas, load demands and classification results.

[0033] Further, the method of identifying the efficient and inefficient areas includes:

[0034] S101: Select and mark the cluster to be formed, and randomly select two data points as the initial cluster centers c = {c1, c2};

[0035] S102: Calculate each data point x i With each cluster center c j The Euclidean distance of:

[0036] The data point x i Assign to the cluster center c with the closest Euclidean distance j ;

[0037] S103: Recalculate the central mean c of each cluster j ;

[0038] S104: Iterate S102 to S103 until the cluster center no longer changes or the maximum number of iterations is reached.

[0039] Furthermore, the training method of the load demand assessment model includes:

[0040] All sensor data within the second layer are used as input of the load demand model. The load demand model takes the load demand corresponding to each set of sensor data as output, takes the actual load demand corresponding to each set of sensor data as the prediction target, and takes minimizing the sum of the first prediction accuracies of all predicted load demands as the training target; the load demand model is trained until the sum of the first prediction accuracies converges; the load demand model is a linear regression model.

[0041] Furthermore, the training method of the state recognition model includes:

[0042] All sensor data within the third layer are used as input of the state recognition model, the state recognition model takes the risk probability corresponding to each group of sensor data as output, takes the actual risk probability corresponding to each group of sensor data as the prediction target, and takes minimizing the sum of the second prediction accuracies of all predicted risk probabilities as the training target; the state recognition model is trained until the sum of the second prediction accuracies reaches convergence and the training is stopped; the state recognition model is a support vector machine model.

[0043] Furthermore, the method for obtaining the scene basic set includes:

[0044] Determine the set point, input monitoring data to the PID controller, calculate the error, proportional term, integral term, differential term and total control output:

[0045] The scenario source set is adjusted according to the total control output of the PID controller to obtain the scenario base set.

[0046] In another aspect, a grid infrastructure project scenario construction model system includes:

[0047] A service area delineation module, the service area delineation module is used to collect the geographical coordinates of the power grid facilities and delineate N service areas according to the geographical coordinates;

[0048] A layered architecture module, wherein the layered architecture module is used to construct a layered scene according to the defined service area to obtain a scene source set;

[0049] An adaptive control module, wherein the adaptive control module is used to collect monitoring data, adjust the scene source set using an adaptive adjustment algorithm, and obtain a scene basic set;

[0050] A visualization building module is used to establish a scene model according to a scene basic set.

[0051] The power grid infrastructure project scenario modeling method and system provided by the present invention have the following effects:

[0052] The present invention effectively solves the challenge of flexibility in modern power grid infrastructure projects by establishing a dynamic and flexible power grid scenario construction model. Through data-driven methods and technologies, the power grid can not only quickly adapt to changes, but also monitor and optimize decisions online, thereby ensuring efficient and reliable power supply services;

[0053] The present invention utilizes Voronoi diagram generation technology to divide the service area according to the geographic coordinates of the power grid facilities, ensuring that each substation reasonably covers the surrounding area. This method provides a more accurate geographical distribution perspective for power grid resource allocation; the three-layer structural framework (global layer, regional layer, scenario layer) provides a clear hierarchical perspective for power grid infrastructure projects, enabling effective integration and analysis of information at different levels. From the global to the local, the operation status of the power grid is deeply understood layer by layer, and the traceability and consistency of data are ensured; through the adaptive adjustment algorithm, using the PID controller and monitoring data, the power grid can dynamically adjust the resource configuration according to the real-time feedback, which enhances the response and flexibility of the power grid when facing emergencies and adapting to seasonal load fluctuations; the linear regression model and the support vector machine model are used for load demand forecasting and equipment status identification, realizing data-driven decision-making, combined with sensor data analysis, correctly predicting user load and equipment working status, and improving the reliability and safety of power grid operation; through the construction of real-time data monitoring and visualization interface, the power grid managers can intuitively understand the operation status and potential problems of the power grid, so as to make decisions quickly, and the application of data integration platform and front-end visualization framework enhances the accessibility of information and the real-time nature of decision support; the identification and merging strategy of high-load area and low-load area ensures the reasonable allocation of resources in the power grid, and through effective management, reduces potential safety risks and improves the stability of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 A flow chart of a method for constructing a model for a power grid infrastructure project scenario in Example 1;

[0055] Figure 2 This is a conceptual diagram of the three-layer structure framework in Example 1;

[0056] Figure 3 A schematic diagram of the connection of a model system for a power grid infrastructure project scenario in Example 2. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] Embodiment 1:

[0059] See also Figure 1 As shown, this embodiment provides a method for constructing a model for a power grid infrastructure project scenario, the method comprising:

[0060] Collect the geographic coordinates of power grid facilities and define N service areas based on the geographic coordinates;

[0061] Construct hierarchical scenarios according to the designated service areas to obtain scenario source sets;

[0062] Collect monitoring data, use an adaptive adjustment algorithm to adjust the scene source set, and obtain the scene base set;

[0063] Build a scene model based on the scene basis set.

[0064] Grid facilities include substations, transmission lines and distribution stations. The geographical coordinates of grid facilities are generally in longitude and latitude, and are located using software such as GPS.

[0065] It is particularly noted that, since infrastructure projects already have specific three-dimensional visual construction blueprints such as design drawings during construction, the present invention will not elaborate on the source of obtaining the three-dimensional visual construction blueprint.

[0066] The methods for defining service areas include:

[0067] The geographic coordinates of the power grid facilities are determined as seed points. The Voronoi diagram is generated based on the seed points using the Voronoi diagram generation technology. That is, the Voronoi diagram is generated using Python libraries (such as SciPy, Matplotlib, and GeoPandas) or GIS software (such as QGIS and ArcGIS) to obtain the preliminary service area (the polygonal area in the Voronoi diagram is the preliminary service area).

[0068] The Voronoi diagram generation technology is used to determine the service area based on the geographic coordinates of the power grid facilities, ensuring that each substation covers the surrounding area geographically. This location-based division method is more accurate and can effectively reflect the distribution of resources.

[0069] Generate Voronoi diagram code example:

[0070]

[0071] Collect load data and user distribution data within the preliminary service area to obtain the total load and average load of each preliminary service area; load data and user distribution data are generally available on the official website of the local power grid project.

[0072] Total load: The sum of all user loads within the preliminary service area.

[0073] Average load: The average load of each user in the preliminary service area.

[0074] A total load threshold and an average load threshold are set, and a high load area and a low load area are defined according to the total load threshold and the average load threshold.

[0075] High load area:

[0076] Condition 1: The total load of the preliminary service area is greater than the total load threshold.

[0077] Condition 2: The average load of the preliminary service area is greater than the average load threshold.

[0078] The high load zone is when either condition 1 or condition 2 is met.

[0079] Low load area:

[0080] Condition 3: The total load of the preliminary service area is less than or equal to the total load threshold.

[0081] Condition 4: The average load of the preliminary service area is less than or equal to the load threshold.

[0082] The low load zone is defined as the zone that meets any of conditions 3 and 4.

[0083] The high-load area and the low-load area are merged according to the merging strategy to obtain N service areas (the polygonal area in the Voronoi diagram after the merging is completed is the service area).

[0084] Merger strategies include geographic proximity and merger conditions;

[0085] Geographic proximity: Only merge directly adjacent low-load areas. Use boundary information in the Voronoi diagram to determine proximity between areas.

[0086] Merge condition: If the total load of the high-load area and the total load of the target low-load area are still below the total load threshold, they can be merged.

[0087] Code example:

[0088]

[0089] During the merging operation, it is necessary to traverse all high-load areas and try to merge them with each adjacent low-load area.

[0090] Code example:

[0091]

[0092] It is important to note that the merging process may render some areas invalid (e.g. a substation no longer serves a certain part), so the list needs to be updated to avoid repeated merging.

[0093] The strategy of merging high-load areas and low-load areas can be dynamically adjusted according to actual needs to effectively maintain the load balance of the power grid. Geographic proximity ensures the feasibility of merging, thereby avoiding resource waste and management complexity. This merging strategy is highly flexible, enabling the power grid to respond quickly to load changes and emergencies, improving the reliability and stability of the overall system.

[0094] The methods for obtaining scene source sets include:

[0095] like Figure 2 As shown, a three-layer structure framework is set, which includes the first layer (global layer), the second layer (regional layer) and the third layer (scene layer);

[0096] First layer (global layer):

[0097] Purpose: To outline the structure, layout and operation strategy of the entire power grid and provide a global perspective.

[0098] Content: Overall architecture diagram of power grid infrastructure, introduction to main areas and their functions.

[0099] Second layer (regional layer):

[0100] Purpose: To deeply analyze the functions, equipment distribution and interrelationships of N service areas.

[0101] Content: Power supply, load conditions and equipment status of N service areas.

[0102] The third layer (scene layer):

[0103] Purpose: To demonstrate the implementation and execution of individual devices and specific operations.

[0104] Content: equipment operating parameters, monitoring data, operating procedures, etc.

[0105] The first layer is a combination of N service areas, which together with the three-dimensional visual construction blueprint within the first layer constitute the first layer;

[0106] The second layer covers a single service area, which together with the three-dimensional visual construction blueprint within the second layer constitutes the second layer;

[0107] The third layer covers the power grid facilities within a single service area, and together with the three-dimensional visual construction blueprint within the third layer, they constitute the third layer;

[0108] Once the three-layer structure framework is filled, it becomes a layered scene.

[0109] The sensor data within each layer is collected into the corresponding hierarchical database, which includes the first-layer database, the second-layer database and the third-layer database; the first-layer database contains the second-layer database, and the second-layer database contains the third-layer database.

[0110] Sensor data includes data that can be identified by sensors, such as equipment operating parameters (current, voltage, frequency, etc.), environmental parameters (temperature, humidity, etc.), and load data.

[0111] A clustering algorithm is used to analyze the overall operating status of the first layer using the sensor data within the first layer, identify high-efficiency areas and low-efficiency areas, obtain labels for high-efficiency areas and low-efficiency areas, and provide guidance for global optimization.

[0112] In the second layer, the sensor data within the second layer are input into the pre-built load demand assessment model, the load demand is output, and the load changes in each service area are predicted, thereby optimizing the allocation of power grid resources.

[0113] In the third layer, the sensor data within the second layer is input into the pre-built state recognition model, and the classification results are output. The classification results include the equipment name and status, and the operating characteristics and status are judged to provide a basis for the optimization of the operation process.

[0114] The scenario source set includes hierarchical scenarios, labels of efficient areas, labels of inefficient areas, load demands and classification results.

[0115] The method for acquiring scene source sets provides a clear hierarchical perspective for power grid infrastructure projects through the construction of a three-layer structural framework. It also lays the foundation for subsequent data analysis and resource optimization. In the event of failures or emergencies, the detailed information provided by the three-layer structure can quickly identify affected areas and equipment and optimize emergency response plans.

[0116] Methods for identifying efficient and inefficient areas include:

[0117] Use Z-Score to standardize the sensor data within the first layer, and input the standardized sensor data matrix x, where the rows of the sensor data matrix represent the service area and the columns represent the sensor data;

[0118] S101: Select and mark the clusters to be formed, i.e., the high-efficiency area and the low-efficiency area, and randomly select two data points as the initial clustering centers c = {c1, c2}; the data points in the high-efficiency area show higher efficiency, lower failure rate and stable power load pattern, which means that the power system operates well in these areas and resources are used reasonably; the data points in the low-efficiency area show low efficiency, high failure rate and unstable power load pattern, which means that the power system is under pressure or potential problems in these places, and optimization measures may need to be taken.

[0119] S102: Calculate each data point x i With each cluster center c j The Euclidean distance of:

[0120]

[0121] In the formula, d(c j ,x i ) represents the i To cluster center c j The distance For data point x i The mth eigenvalue of is the cluster center c j The mth eigenvalue of , where n is the feature dimension of the data point.

[0122] The data point x i Assign to the cluster center c with the closest Euclidean distance j .

[0123] S103: Recalculate the central mean c of each cluster j ; In particular, the cluster center c j It not only indicates the representative characteristics of the cluster, but also serves as a data carrier.

[0124] S104: Iterate S102 to S103 until the cluster center no longer changes (converges) or the maximum number of iterations (preset) is reached.

[0125] The training method of the load demand assessment model includes:

[0126] All sensor data within the second layer are used as the input of the load demand model. The load demand model takes the load demand corresponding to each set of sensor data as the output, the actual load demand corresponding to each set of sensor data as the prediction target, and minimizing the sum of the first prediction accuracies of all predicted load demands as the training target.

[0127] Among them, the calculation formula for the first prediction accuracy is: JE g =(JG g -JT g ) 2 , where g is the number of each set of sensor data, JE g is the first prediction accuracy, JG g is the predicted load demand corresponding to the g-th group of sensor data, JT g is the actual load demand corresponding to the g-th group of sensor data; the load demand model is trained until the sum of the first prediction accuracy reaches convergence and the training is stopped; the load demand model is a linear regression model.

[0128] The training methods of the state recognition model include:

[0129] All sensor data within the third layer are used as the input of the state recognition model. The state recognition model takes the risk probability corresponding to each set of sensor data as the output, the actual risk probability corresponding to each set of sensor data as the prediction target, and the sum of the second prediction accuracies of minimizing all predicted risk probabilities as the training target.

[0130] Among them, the calculation formula for the second prediction accuracy is: QR b =(QG b -QT b ) 2 , where b is the number of each set of sensor data, QR b is the second prediction accuracy, QG b is the predicted risk probability corresponding to the sensor data of group b, QT b is the actual risk probability corresponding to the bth group of sensor data; the state recognition model is trained until the sum of the second prediction accuracy reaches convergence and the training is stopped; the state recognition model is a support vector machine model.

[0131] Monitoring data includes sensor data, equipment anomaly and fault alarm data. Monitoring data can be collected through SCADA (supervisory control and data acquisition system), smart meters, sensor networks and other methods.

[0132] Methods for obtaining scene base sets include:

[0133] Determine the set point (target value), which is the ideal target value preset by the sensor data when the power grid is running;

[0134] Input monitoring data to the PID controller and calculate the error e(t):

[0135] e(t) = set point - actual monitoring data;

[0136] Calculate the proportional term P(t):

[0137] P(t)=K p ·e(t);

[0138] In the formula, K p is the proportional gain, which is a constant that adjusts the response strength of the proportional control. The larger the proportional gain, the more sensitive the system is to errors and the faster it can correct deviations. If it is set too high, it may cause the system to oscillate excessively near the set point. If it is set too low, the system response may be too slow. t represents the current time.

[0139] Calculate the integral term I(t):

[0140]

[0141] In the formula, τ represents the integral time constant; e(τ) represents the error at the past moment; K v Represents the integral gain, which is a constant that adjusts the response strength of the integral control; the integral gain is used to eliminate static deviations and deal with long-term errors; a higher integral gain will make the system respond faster to sustained deviations, but it may also cause system oscillation or slow response.

[0142] Calculate the differential term D(t):

[0143]

[0144] In the formula, K d is the differential gain: this is a constant that adjusts the response strength of the differential control; the differential gain is used to predict future error changes, mainly used to reduce overshoot and improve system stability; a higher differential gain will reduce the overshoot of the system and slow down the response. If it is set improperly, it may cause the system to respond too slowly.

[0145] Calculate the total control output:

[0146] U(t)=P(t)+I(t)+D(t);

[0147] Adjust the scenario source set according to the total control output of the PID controller to obtain the scenario base set;

[0148] Exemplary:

[0149] The set point is the target load, according to the total control output of the PID controller:

[0150] If U(t) is greater than 0, it means that the current power generation is insufficient to meet the load demand, and the power output of the generator is adjusted to the required level in real time.

[0151] If U(t) is less than 0, it means that the current power generation exceeds the demand, and the power output of the generator is adjusted in real time to reduce to the required level.

[0152] Through real-time data monitoring and the application of PID controllers, the flexibility of the power grid is improved, enabling it to better adapt to seasonal and temporal load fluctuations, while effectively responding to unexpected events (such as surges in power demand or equipment failures).

[0153] Use data integration platforms (QGIS, ArcGIS, Hadoop and Spark) and front-end visualization frameworks (such as D3.js, Plotly, Dash, etc.) to build scenario models based on the scenario base set.

[0154] Embodiment 2:

[0155] like Figure 3As shown, based on the same inventive concept as the method for building a model for a power grid infrastructure project scenario in the aforementioned embodiment, the present application provides a power grid infrastructure project scenario building model system, and the system and method embodiments in the present application embodiment are based on the same inventive concept. The system includes:

[0156] A service area delineation module is used to collect the geographical coordinates of power grid facilities and delineate N service areas according to the geographical coordinates;

[0157] A hierarchical architecture module is used to construct hierarchical scenarios according to the defined service areas and obtain a scenario source set;

[0158] The adaptive control module is used to collect monitoring data, adjust the scene source set using an adaptive adjustment algorithm, and obtain the scene base set;

[0159] A visual building block for building scene models based on a scene basis set.

[0160] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0161] What has been described above is only a preferred specific implementation manner of the embodiments of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can make equivalent substitutions or changes according to the technical scheme and concept of the present application within the technical scope disclosed in the present application, which should be covered by the protection scope of the present application.

Claims

1. A method for constructing a model for a power grid infrastructure project scenario, characterized in that: Methods include: Collect the geographic coordinates of power grid facilities and define N service areas based on the geographic coordinates; Construct hierarchical scenarios according to the designated service areas to obtain scenario source sets; Collect monitoring data, use an adaptive adjustment algorithm to adjust the scene source set, and obtain the scene base set; Build a scene model based on the scene basis set.

2. The method according to claim 1, characterized in that The method for defining the service area includes: The geographical coordinates of the power grid facilities are determined as seed points, and the Voronoi diagram is generated based on the seed points using the Voronoi diagram generation technology to obtain the preliminary service area; Collecting load data and user distribution data in the preliminary service area, so as to obtain the total load and average load of each preliminary service area; A total load threshold and an average load threshold are set, and a high load area and a low load area are defined according to the total load threshold and the average load threshold. The high-load area and the low-load area are merged according to the merging strategy to obtain N service areas.

3. The method according to claim 2, characterized in that The method for defining the high load area and the low load area includes: High load area: Condition 1: The preliminary service area with a total load greater than the total load threshold; Condition 2: The preliminary service area with an average load greater than the average load threshold; If any of conditions 1 and 2 are met, it is a high load area; Low load area: Condition 3: The preliminary service area with a total load less than or equal to the total load threshold; Condition 4: The average load of the preliminary service area is less than or equal to the load threshold; The low load zone is defined as the zone that meets any of conditions 3 and 4.

4. The method according to claim 1, characterized in that The methods for constructing a layered scene include: A three-layer structural framework is set up, and the three-layer structural framework includes the first layer, the second layer and the third layer; The first layer is a combination of N service areas, which together with the three-dimensional visual construction blueprint within the first layer constitute the first layer; The second layer covers a single service area, which together with the three-dimensional visual construction blueprint within the second layer constitutes the second layer; The third layer covers the power grid facilities within a single service area, and together with the three-dimensional visual construction blueprint within the third layer, they constitute the third layer; Once the three-layer structure framework is filled, it becomes a layered scene.

5. The method according to claim 4, characterized in that The method for obtaining the scene basic set includes: A clustering algorithm is used to analyze the overall operation status of the first layer for the sensor data within the first layer, identify the high-efficiency area and the low-efficiency area, and obtain the labels of the high-efficiency area and the low-efficiency area; In the second layer, the sensor data within the second layer is input into the pre-built load demand assessment model to output the load demand; In the third layer, the sensor data within the second layer is input into the pre-built state recognition model to output the classification results; the scene source set is obtained, which includes hierarchical scenes, labels of efficient areas, labels of inefficient areas, load demands and classification results.

6. The method according to claim 5, characterized in that Methods for identifying efficient and inefficient areas include: S101: Select and mark the cluster to be formed, and randomly select two data points as the initial cluster centers c = {c1, c2}; S102: Calculate each data point x i With each cluster center c j The Euclidean distance of: The data point x i Assign to the cluster center c with the closest Euclidean distance j ; S103: Recalculate the central mean c of each cluster j ; S104: Iterate S102 to S103 until the cluster center no longer changes or the maximum number of iterations is reached.

7. The method according to claim 5, characterized in that The training method of the load demand assessment model includes: All sensor data within the second layer are used as input of the load demand model. The load demand model takes the load demand corresponding to each set of sensor data as output, takes the actual load demand corresponding to each set of sensor data as the prediction target, and takes minimizing the sum of the first prediction accuracies of all predicted load demands as the training target; the load demand model is trained until the sum of the first prediction accuracies converges; the load demand model is a linear regression model.

8. The method according to claim 5, characterized in that The training method of the state recognition model includes: All sensor data within the third layer are used as input of the state recognition model, the state recognition model takes the risk probability corresponding to each group of sensor data as output, takes the actual risk probability corresponding to each group of sensor data as the prediction target, and takes minimizing the sum of the second prediction accuracies of all predicted risk probabilities as the training target; the state recognition model is trained until the sum of the second prediction accuracies reaches convergence and the training is stopped; the state recognition model is a support vector machine model.

9. The method according to claim 1, characterized in that Methods for obtaining scene base sets include: Determine the set point, input monitoring data to the PID controller, calculate the error, proportional term, integral term, differential term and total control output: The scenario source set is adjusted according to the total control output of the PID controller to obtain the scenario base set.

10. A power grid infrastructure project scenario construction model system, characterized in that: include: A service area demarcation module, the service area demarcation module is used to collect the geographical coordinates of the power grid facilities and demarcate N service areas according to the geographical coordinates; A layered architecture module, wherein the layered architecture module is used to construct a layered scene according to the defined service area to obtain a scene source set; An adaptive control module, wherein the adaptive control module is used to collect monitoring data, adjust the scene source set using an adaptive adjustment algorithm, and obtain a scene basic set; A visualization building module is used to establish a scene model according to a scene basic set.

Images