Wind power cluster dispatch optimization method and system driven by digital twin

Abstract

The present invention discloses a wind power cluster dispatch optimization method and system driven by digital twins, and relates to the field of data processing technology. The method includes: performing topological simulation according to a power plant group, constructing a first-level digital twin model, performing spatial division according to power plants and power generation units, obtaining second-level and third-level digital twin models, embedding a power generation capacity evaluation function, connecting with a power generation capacity fusion layer, generating a global power generation twin mapping model, and performing power generation forecasting and power dispatching. The present invention solves the technical problem that the traditional power generation capacity forecasting method focuses on the power generation capacity of the overall plant cluster, ignoring the differences between individual power generation units, resulting in inaccurate power generation capacity forecasting, and affecting the flexibility and accuracy of power dispatching. It achieves the technical effect of gradually improving the accuracy of power generation forecasting by constructing a multi-level digital twin model from power plant groups to power generation units, thereby improving the flexibility and accuracy of power dispatching.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a wind power cluster scheduling optimization method and system driven by digital twins. Background Art

[0002] To achieve intelligent and refined wind farm cluster power dispatch, it is often necessary to combine power generation capacity forecasts to provide data support for power dispatch. However, traditional power generation capacity forecasting methods typically consider the overall power generation capacity of a site cluster, ignoring the impact of individual differences. While this method offers the advantage of fast calculation speed, the same meteorological factors can lead to different power generation capacities. This is especially true when the service life and model of power generation units in a site cluster vary significantly. This results in poor power generation capacity forecast accuracy, impacting the accuracy and flexibility of power dispatch. Summary of the Invention

[0003] This application provides a digital twin-driven wind power cluster scheduling optimization method and system to solve the technical problem that the traditional power generation capacity prediction method in the existing technology focuses on the power generation capacity of the entire site cluster, ignores the differences between individual power generation units, resulting in inaccurate power generation capacity prediction, and affects the flexibility and accuracy of power scheduling.

[0004] The first aspect of the present application provides a digital twin-driven wind power cluster scheduling optimization method, which includes: performing topological simulation based on a power plant group to construct a first-level digital twin model; spatially dividing the first-level digital twin model according to power plants to obtain a second-level digital twin model; spatially dividing the second-level digital twin model according to power generation units to obtain a third-level digital twin model; embedding a power generation capacity evaluation function in the power generation unit twin model of the third-level digital twin model to obtain a scattered power generation capacity twin mapping model; constructing a power generation capacity fusion layer, connecting it to the output layer of the scattered power generation capacity twin mapping model, and generating a global power generation twin mapping model; collecting meteorological factor prediction information, and generating predicted power generation through the global power generation twin mapping model; and performing power scheduling based on the predicted power generation in response to the power consumption request of the substation.

[0005] The second aspect of the present application provides a digital twin-driven wind power cluster scheduling optimization system, which includes: a first-level digital twin model construction module, which is used to perform topological simulation based on the power plant group to construct a first-level digital twin model; a second-level digital twin model construction module, which is used to spatially divide the first-level digital twin model according to the power plant station to obtain a second-level digital twin model; a third-level digital twin model construction module, which is used to spatially divide the second-level digital twin model according to the power generation unit to obtain a third-level digital twin model; a power generation capacity twin mapping model construction module, which The power generation capacity twin mapping model construction module is used to embed the power generation capacity evaluation function in the power generation unit twin model of the three-level digital twin model to obtain the scattered power generation capacity twin mapping model; the power generation twin mapping model generation module, the power generation twin mapping model generation module is used to construct a power generation capacity fusion layer, connected to the output layer of the scattered power generation capacity twin mapping model, to generate a global power generation twin mapping model; the power generation prediction module, the power generation prediction module is used to collect meteorological factor prediction information, and generate predicted power generation through the global power generation twin mapping model; the power dispatching module, the power dispatching module is used to respond to the power consumption request of the substation area and perform power dispatch based on the predicted power generation.

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

[0007] The digital twin-driven wind power cluster scheduling optimization method and system provided in the present application relate to the field of data processing technology. By performing topological simulation based on the power plant group, a first-level digital twin model is constructed, and spatial division is performed according to the power plant and power generation unit to obtain the second-level and third-level digital twin models. After the power generation capacity evaluation function is embedded, it is connected with the power generation capacity fusion layer to generate a global power generation twin mapping model to perform power generation prediction and power scheduling. This solves the technical problem that the traditional power generation capacity prediction method in the existing technology focuses on the power generation capacity of the entire plant cluster and ignores the differences between individual power generation units, resulting in inaccurate power generation capacity prediction and affecting the flexibility and accuracy of power scheduling. It realizes the technical effect of gradually improving the accuracy of power generation prediction by constructing a multi-level digital twin model from the power plant group to the power generation unit, providing more accurate data support for power scheduling, and thus improving the flexibility and accuracy of power scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0009] Figure 1 A schematic flow chart of a wind power cluster scheduling optimization method driven by a digital twin provided in an embodiment of the present application;

[0010] Figure 2 A schematic diagram of the process of constructing a twin mapping model of scattered power generation capacity in the digital twin-driven wind power cluster scheduling optimization method provided in an embodiment of the present application;

[0011] Figure 3 Schematic diagram of the structure of the digital twin-driven wind power cluster scheduling optimization system provided in the embodiment of the present application.

[0012] Explanation of the accompanying drawings: first-level digital twin model construction module 11, second-level digital twin model construction module 12, third-level digital twin model construction module 13, power generation capacity twin mapping model construction module 14, power generation twin mapping model generation module 15, power generation prediction module 16, power dispatching module 17. DETAILED DESCRIPTION

[0013] This application provides a digital twin-driven wind power cluster scheduling optimization method and system to solve the technical problem that the traditional power generation capacity prediction method in the existing technology focuses on the power generation capacity of the entire site cluster, ignores the differences between individual power generation units, resulting in inaccurate power generation capacity prediction, and affects the flexibility and accuracy of power scheduling.

[0014] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only some of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0015] It should be noted that the terms "first", "second", etc. in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices. Example 1

[0016] like Figure 1 As shown, the present application provides a digital twin-driven wind power cluster scheduling optimization method, the method comprising:

[0017] P10: Perform topological simulation based on the power plant group and build a first-level digital twin model.

[0018] Specifically, topology simulation is performed based on the layout and connectivity of a cluster of power plants. These clusters, which include wind farms and photovoltaic power plants, are essentially multiple geographically dispersed power stations connected via a power transmission network. These typically include power plants using renewable energy sources such as wind and solar power. Topology simulation involves simulating the electrical connections, geographic distribution, and power transmission characteristics between these power plants to construct the topology of the power grid. This allows the simulation of the flow of power between different power plants and nodes.

[0019] This process utilizes 3D modeling software, a geographic information system (GIS), and a specialized simulation platform to construct a highly detailed and dynamic digital model—also known as a Level 1 digital twin—based on the measured data and design drawings of the power plant cluster. Digital twin technology involves creating a virtual model that reflects the state, characteristics, and operational status of the physical world in real time. Here, the Level 1 digital twin model is a holistic model of the entire power plant cluster, encompassing the overall operational status, power generation capacity, and power transmission of each power station. By integrating the cluster's physical data (such as equipment status and geographic location) with real-time operational data (such as power output, current, and voltage), this model also reflects current power production and transmission conditions in real time, providing the foundational data for subsequent, more refined models.

[0020] P20: Spatially divide the first-level digital twin model according to the power generation station to obtain a second-level digital twin model.

[0021] Optionally, the first-level digital twin model constructed in the previous step can be spatially partitioned based on the specific distribution and characteristics of power stations (including wind farms and photovoltaic power plants). This allows the complex power plant cluster to be further decomposed into specific power station levels, namely wind farms and photovoltaic power plants. A power station is the basic unit within a power plant cluster. For example, a wind farm consists of multiple wind turbines, while a photovoltaic power plant consists of a large number of photovoltaic modules, which together convert wind and solar energy into electricity. Each power station may vary in location, installed capacity, equipment type, and operating status, necessitating further refined management and modeling.

[0022] The spatial partitioning process utilizes spatial analysis techniques and geographic information systems (GIS) to decompose the monolithic first-level digital twin model into several sub-models based on the power plant's physical location, power generation capacity, and operating parameters. Each sub-model corresponds to a specific power plant. This partitioning enables the dispatching system to manage and optimize each plant based on its unique characteristics. For example, certain wind farms may experience significant variations in wind speed and weather conditions due to their geographical location, directly impacting their power generation capacity. Meanwhile, the power generation efficiency of photovoltaic power plants is affected by factors such as sunlight intensity and component aging. This refined spatial partitioning allows the operational status and power generation capacity of each power plant to be individually monitored and evaluated within the second-level digital twin model.

[0023] The secondary digital twin model, based on the division of power plants, not only includes real-time data from each plant (such as power generation and equipment status), but also leverages technologies such as big data analytics, cloud computing, and the Internet of Things to acquire and process this data in real time. This model provides a personalized management perspective for each plant, helping to optimize its power generation efficiency and supporting subsequent, more granular analysis at the power generation unit level. This refined model further improves the responsiveness and flexibility of the dispatch system, enabling dynamic adjustments to power dispatch based on the characteristics of each plant.

[0024] P30: The second-level digital twin model is spatially divided according to the power generation unit to obtain a third-level digital twin model.

[0025] Optionally, the accuracy of the second-level digital twin model can be further improved to the power generation unit level. This involves performing a refined spatial partitioning of wind turbines in a wind farm and photovoltaic modules in a photovoltaic power plant to construct a third-level digital twin model. A power generation unit refers to a single wind turbine in a wind farm or a single photovoltaic module in a photovoltaic power plant. Each unit may have different power generation performance and operating status, necessitating independent modeling and management.

[0026] To achieve this goal, high-precision spatial positioning technologies (such as GPS or RTK) are first used to determine the precise location of each power generation unit (wind turbine or photovoltaic panel). Subsequently, using 3D modeling and simulation, these units are mapped one by one into a secondary digital twin model based on their actual layout and physical characteristics, creating a more detailed spatial division.

[0027] When building a three-level digital twin model, special attention must be paid to accurately simulating the operating status of each power generation unit. For example, for wind turbines, priority consideration is given to parameters such as blade length, speed, power curve, and control system, as well as the impact of environmental factors such as wind speed, direction, and temperature on power generation efficiency. Using complex physical models and algorithms, the output power and operating status of wind turbines under different operating conditions are simulated. Similarly, for photovoltaic modules, priority consideration is given to factors such as photoelectric conversion efficiency, temperature coefficient, shading, and the impact of environmental factors such as light intensity and temperature on power generation performance.

[0028] To ensure the accuracy and real-time performance of the three-level digital twin model, Internet of Things (IoT) technology and intelligent sensor networks can be integrated. These sensors, located at key locations within the power generation units, collect real-time operational data (such as wind speed, direction, light intensity, temperature, current, and voltage) and transmit it to a data processing center via the IoT platform. The data processing center then processes and analyzes this data using big data analytics and machine learning, extracting key operational characteristics and updating them in real time within the three-level digital twin model. This provides a solid data foundation and technical support for efficient scheduling and optimized management of wind farm clusters.

[0029] P40: A power generation capacity evaluation function is embedded in the power generation unit twin model of the three-level digital twin model to obtain a scattered power generation capacity twin mapping model.

[0030] Further, such as Figure 2 As shown, step P40 in this embodiment of the application also includes:

[0031] P41: Obtain the first power generation unit twin model of the power generation unit twin model of the three-level digital twin model, wherein the first power generation unit twin model has a power generation unit model label and a power generation unit service time label; P42: According to the power generation unit model label and the power generation unit service time label, collect the meteorological factor record data set and the power generation record data set; P43: According to the meteorological factor record data set and the power generation record data set, construct the first power generation unit power generation capacity evaluation function; P44: Embed the first power generation unit power generation capacity evaluation function into the first power generation unit twin model; P45: When each power generation unit twin model has a power generation capacity evaluation function, obtain the scattered power generation capacity twin mapping model.

[0032] It should be understood that in order to enhance the practicality of the three-level digital twin model, by embedding the power generation capacity evaluation function within the twin model of each power generation unit, a scattered power generation capacity twin mapping model is constructed that can reflect the power generation potential of each power generation unit in real time.

[0033] First, a specific power generation unit twin model from the three-level digital twin model is selected as the starting point, namely the first power generation unit twin model. This model not only includes the physical structure and spatial location information of the power generation unit, but also includes the power generation unit model label and service life label. The power generation unit model label is used to identify different types of wind turbines or photovoltaic modules, as the power generation performance of different equipment models can vary significantly. The power generation unit service life label records the equipment's usage time and reflects the impact of equipment aging on power generation capacity. These two labels are important foundations for the subsequent construction of the power generation capacity evaluation function, as they can reflect the technical characteristics and aging of the power generation unit.

[0034] Next, using the power generation unit model tag and the power generation unit service life tag as indexes, the corresponding meteorological factor record dataset and power generation record dataset are retrieved from the database. The meteorological factor dataset includes records of environmental factors that may affect power generation efficiency, such as wind speed, wind direction, temperature, and humidity. The power generation record dataset records the actual power generation of the power generation unit at different time points. These datasets form the basis for constructing the power generation capacity evaluation function.

[0035] Then, using statistical, machine learning, or deep learning methods, the meteorological factor and power generation record datasets are analyzed and modeled. Specifically, these datasets serve as training data, and supervised training is performed using algorithms such as regression analysis, classification, or neural networks. This results in the construction of an evaluation function capable of predicting the power generation capacity of a power generation unit under different meteorological conditions. This function predicts the power generation capacity of the equipment under different meteorological conditions by analyzing the relationship between meteorological factors and power generation.

[0036] Furthermore, the constructed first power generation unit's power generation capacity evaluation function is embedded in the first power generation unit's twin model. This way, whenever the power generation unit's operating status or external environment changes, the twin model can quickly calculate its current power generation capacity through the evaluation function and update it in real time.

[0037] Finally, repeat the above process to embed the corresponding power generation capacity assessment function into each power generation unit twin model in the three-level digital twin model. When all power generation unit twin models have this function, a complete distributed power generation capacity twin mapping model is obtained. This model can intuitively display the power generation capacity distribution of each power generation unit in the entire wind power cluster or photovoltaic power station, providing strong support for power dispatch and operation and maintenance management.

[0038] Furthermore, step P43 of the embodiment of the present application further includes:

[0039] P43-1: Based on the meteorological factor record data set and the power generation record data set, the initial neural network architecture is trained to obtain a first-level power generation capacity evaluation function; P43-2: The first-level power generation capacity evaluation function is verified to obtain a verification deviation vector set; P43-3: System vector extraction is performed based on the verification deviation vector set to obtain a first-level system error vector modulus and a first-level random error modulus; P43-4: When the first-level system error vector modulus is less than the system error vector modulus threshold, and the first-level random error modulus is less than the random error modulus threshold, the first power generation unit power generation capacity evaluation function is generated; P43-5: When the first-level system error vector modulus is less than the system error vector modulus threshold, and the first-level random error modulus is greater than or equal to the random error modulus threshold, the initial neural network architecture continues to be trained based on the meteorological factor record data set and the power generation record data set.

[0040] Optionally, the specific process for constructing the first power generation unit's power generation capacity assessment function may include first training an initially selected neural network architecture using the collected meteorological factor and power generation record datasets. These neural networks are widely used in modeling and predicting complex systems due to their powerful nonlinear fitting capabilities. Through forward and backpropagation algorithms, the network's weights and bias parameters are continuously adjusted, allowing the network's output to gradually approximate actual power generation data. After multiple iterations of training, a first-level power generation capacity assessment function is obtained, capable of predicting the power generation capacity of a power generation unit based on the input meteorological factors.

[0041] Next, to evaluate the performance of the first-level power generation capacity evaluation function, we applied it to an independent validation dataset and validated the function to obtain a set of validation deviation vectors. The goal of the validation process is to identify errors and deviations in the predictions by comparing actual power generation data with the model's predictions. This set of deviation vectors reflects the range of error in the model's predictions.

[0042] Next, the verification deviation vector set is subjected to system vector extraction. System vectors refer to deviation components caused by inherent system characteristics (such as aging of power generation units and design flaws), while random errors are deviation components caused by random factors (such as measurement noise and unpredictable weather changes). The verification deviation vector is decomposed into a first-level systematic error vector modulus and a first-level random error modulus. These two indicators reflect the level of systematic error and random error of the evaluation function, respectively.

[0043] Furthermore, the first-level system error vector modulus and the first-level random error modulus are compared with preset first-level system error vector modulus thresholds and first-level random error modulus thresholds, respectively. The first-level system error vector modulus thresholds and the first-level random error modulus thresholds can be set by professionals based on analysis of historical error records. If the first-level system error vector modulus is less than the system error vector modulus threshold, and the first-level random error modulus is less than the random error modulus threshold, then the current first-level power generation capability assessment function meets the accuracy requirements and can be generated as the first power generation unit power generation capability assessment function.

[0044] However, if the first-level system error vector modulus is less than the system error vector modulus threshold, but the first-level random error modulus is greater than or equal to the random error modulus threshold, this indicates that the random error of the evaluation function is large and requires further optimization. In this case, the initial neural network architecture needs to be retrained using the meteorological factor record dataset and the power generation record dataset. By adjusting training parameters (such as the learning rate, batch size, and number of training rounds) or optimizing the neural network structure (such as adding hidden layers and changing the activation function), the random error modulus can be reduced and the overall performance of the evaluation function improved. This process may require multiple iterations until all accuracy requirements are met, thereby improving the model's predictive capabilities.

[0045] Furthermore, the embodiment of the present application further includes step P43-5a, and step P43-5a further includes:

[0046] P43-51a: When the modulus of the first-level system error vector is greater than or equal to the system error vector modulus threshold, a first-level residual fitting block is constructed according to the first-level system error vector; P43-52a: The input layer of the first-level residual fitting block is integrated with the output layer of the first-level power generation capacity evaluation function to obtain a second-level neural network architecture; P43-53a: Based on the meteorological factor record data set and the power generation record data set, iterative training is performed on the second-level neural network architecture, wherein the training data of the second-level neural network architecture is different from the training data of the initial neural network architecture.

[0047] Specifically, if the modulus of the first-level system error vector is greater than or equal to the set threshold, it means that the system error in the model is too large, the prediction accuracy is insufficient, and further optimization and adjustment are required. At this time, first construct a first-level residual fitting block based on the first-level system error vector. The first-level residual fitting block is a specially designed neural network layer or module, whose goal is to learn and compensate for the system error in the first-level power generation capacity evaluation function. The role of the residual fitting block is to capture the part that is not accurately expressed in the primary model by fitting the system error, and further optimize the output of the model. Residual fitting is a commonly used technical means that helps improve the overall prediction accuracy by modeling the error part of the model.

[0048] Next, the input layer of the first-level residual fitting block is fused with the output layer of the first-level power generation capacity assessment function. This fusion approach allows the first-level residual fitting block to directly receive the prediction results of the first-level assessment function and make further corrections based on them. By combining the outputs of the first-level residual fitting block and the original model, a more accurate prediction model is formed, namely the second-level neural network architecture. This fusion approach preserves the predictive power of the original model while improving the overall prediction accuracy by introducing corrections for systematic errors.

[0049] Furthermore, the secondary neural network architecture was iteratively trained using a dataset of meteorological factor and power generation records. Unlike the training data for the initial neural network architecture, this training data focused on optimizing the system error component. By introducing new training data, the secondary neural network architecture was able to better learn and adapt to changes in the system error, optimizing its prediction performance at a higher level and gradually reducing the residual error.

[0050] Throughout this process, residual fitting techniques are combined with neural network architecture optimization to improve the model's forecasting accuracy through multi-level error correction. Iterative training ensures the model's continuous self-adjustment and optimization, enabling it to better adapt to power generation capacity forecasting needs under varying meteorological conditions. This approach technically leverages the high flexibility of neural networks and, combined with error analysis, ensures greater reliability and accuracy in the final output of the power generation capacity assessment function.

[0051] Furthermore, step P43-3 of the embodiment of the present application also includes:

[0052] P43-31: Perform outlier cleaning on the verification bias vector set to obtain a selected verification bias vector set; P43-32: Extract the minimum value of the selected verification bias vector set and set it as the first verification bias vector; P43-33: Extract the maximum value of the selected verification bias vector set and set it as the second verification bias vector; P43-34: Calculate the mean of the sum of the first verification bias vector and the second verification bias vector to obtain the first-level system error vector; P43-35: Use the first verification bias vector to subtract the first-level system error vector and take the modulus value to obtain the first random error modulus value; P43-36: Use the second verification bias vector to subtract the first-level system error vector and take the modulus value to obtain the second random error modulus value; P43-37: Calculate the mean of the first random error modulus value and the second random error modulus value and set it as the first-level random error modulus value.

[0053] Optionally, the specific process for extracting a system vector based on the validation deviation vector set can include first performing outlier cleaning on the validation deviation vector set. Outliers (or abnormal values) are values ​​that significantly deviate from other data points, potentially due to measurement errors, data entry errors, or extreme events. The purpose of outlier cleaning is to remove these outliers to reduce their impact on subsequent analysis. By applying statistical methods (such as Z-score, IQR, etc.) or model-based detection methods, outliers in the validation deviation vector set can be identified and removed, thereby obtaining a cleaner and more reliable set of selected validation deviation vectors.

[0054] Next, the minimum and maximum values ​​from the cleaned set of selected validation deviation vectors are extracted, respectively, and set as the first and second validation deviation vectors. The first validation deviation vector represents the minimum deviation of the model error under different conditions, and the second validation deviation vector represents the maximum deviation of the model error under different conditions, which can help determine the boundary between systematic and random errors. Then, the mean of the sum of the first and second validation deviation vectors is calculated and used as the first-level systematic error vector. The first-level systematic error vector reflects the overall error trend of the model and represents the error components that are always present.

[0055] Subsequently, the first and second verification bias vectors are subtracted from the first systematic error vector and the second verification bias vector, and their modulos (i.e., absolute values) are taken to obtain a first random error modulus and a second random error modulus. These two modulos respectively reflect the magnitude of the random error remaining in the two verification bias vectors after the systematic error is removed.

[0056] Finally, the mean of the first and second random error moduli is calculated and set as the first-level random error modulus. This value represents the average level of random error in the entire set of validation deviation vectors and is an important indicator for evaluating the random error performance of the power generation capacity assessment function. It helps further understand the model's non-systematic errors under different circumstances and provides a more comprehensive assessment of the model's prediction accuracy.

[0057] P50: Construct a power generation capacity fusion layer, connect it with the output layer of the scattered power generation capacity twin mapping model, and generate a global power generation twin mapping model.

[0058] Furthermore, step P50 of the embodiment of the present application further includes:

[0059] P51: Obtain the first node power generation mapping model of the scattered power generation capacity twin mapping model; P52: Obtain the power generation section data of the first node in the preset time zone, wherein the power generation section data of the first node in the preset time zone includes the first node power generation record data and the actual power generation record data of the field group, and the preset time zone refers to the historical time zone six months before the current moment; P53: Perform power generation efficiency analysis based on the first node power generation record data and the actual power generation record data of the field group to obtain the temporary power generation efficiency of the first node; P54: Repeat the analysis to obtain the temporary power generation efficiency of the second node until the temporary power generation efficiency of the Qth node; P55: Generate the power generation capacity fusion layer based on the first node temporary power generation efficiency, the second node temporary power generation efficiency until the Qth node temporary power generation efficiency, perform weighted addition and full connection on each node of the scattered power generation capacity twin mapping model, and obtain the global power generation twin mapping model.

[0060] It should be understood that by constructing the power generation capacity fusion layer and connecting it with the output layer of the scattered power generation capacity twin mapping model generated in the previous step, a global power generation twin mapping model is generated. The power generation capacity fusion layer is a calculation layer that integrates the power generation capacity of each node. By combining the power generation efficiency of each node, it can generate an overall power generation capacity prediction model for the entire farm group.

[0061] Specifically, the first node's power generation capacity mapping model is extracted from the scattered point power generation capacity twin mapping model. This model is data-driven and can reflect the changes in the node's power generation capacity under different conditions.

[0062] Subsequently, the power generation section data for the first node's preset time zone is extracted. The preset time zone refers to a historical time zone six months prior to the current time. The purpose of this historical data is to analyze and verify the node's power generation capacity. This power generation section data for the first node's preset time zone includes the node's recorded power generation data and the actual power generation data of the farm group. This data provides information about the node's power generation performance across different historical time zones.

[0063] Next, we analyze the power generation efficiency of the first node based on its recorded power generation data and the actual power generation data of the cluster. By comparing the actual power generation with the theoretical power generation (or reference power generation), we calculate the temporary power generation efficiency of the first node. This efficiency value reflects the actual power generation performance of the node during the current time window.

[0064] To obtain a more comprehensive view of the system's power generation efficiency, the above analysis process is repeated, calculating the temporary power generation efficiency of the second node, the third node, and finally the Qth node. The efficiency analysis of each node is independent, but together they constitute part of the overall power generation efficiency of the system.

[0065] Finally, based on the temporary power generation efficiency of all nodes—namely, the temporary power generation efficiency of the first node, the temporary power generation efficiency of the second node, and so on to the Qth node—the power generation capacity of each node is weighted. The weights can be determined based on factors such as the node's power generation contribution and importance. Through this fusion process, a twin mapping model that reflects the system's global power generation capacity is obtained, namely the global power generation twin mapping model. This model not only includes detailed information about each node but also, through weighted fusion, accurately simulates and predicts the system's overall power generation capacity.

[0066] P60: Collect meteorological factor prediction information and generate predicted power generation through the global power generation twin mapping model.

[0067] Specifically, meteorological factor forecast information is combined with the global power generation twin mapping model to generate accurate power generation forecasts. First, meteorological factor forecast information is collected. This information typically includes key meteorological data such as wind speed, wind direction, solar radiation intensity, temperature, and humidity, and is one of the key factors affecting power generation. Especially for renewable energy generation such as wind and solar power, changes in meteorological conditions directly affect their power generation efficiency. Therefore, obtaining accurate meteorological factor forecast information from meteorological departments or professional meteorological forecasting agencies is the foundation for power generation forecasts.

[0068] Next, the collected meteorological factor forecast information is fed into the global power generation twin mapping model to predict future power generation. This model integrates the temporary power generation efficiency of each generating unit and the overall power generation capacity of the cluster, comprehensively considering the power generation capacity of each node and the actual power generation demand within the cluster. By inputting meteorological factors into this model, the system can simulate power generation under different weather conditions and generate predicted power generation.

[0069] Within the model, meteorological factor forecast information is used to adjust the model's input parameters to reflect the potential impact of future weather conditions on power generation. The model performs calculations and inferences based on the input meteorological factor data, combined with historical power generation data and system operating status data, ultimately generating a predicted power generation output. To ensure the accuracy of the forecast results, the global power generation twin mapping model can use machine learning algorithms or deep learning technologies to automatically learn complex patterns and relationships in the data. Through training on large amounts of historical data, the model's predictive capabilities are continuously improved. Simultaneously, online learning and optimization are performed based on real-time feedback data to adapt to the dynamic environment of the power generation system.

[0070] Ultimately, through the calculation and reasoning of the global power generation twin mapping model, a forecast of power generation for a period of time in the future is obtained. This forecast will provide an important reference for the scheduling and operation of the power system, helping to achieve the optimal allocation of power resources and balance supply and demand.

[0071] P70: In response to the power demand request of the substation, power dispatch is performed based on the predicted power generation.

[0072] Furthermore, step P70 of the embodiment of the present application further includes:

[0073] P71: The substation electricity consumption request includes the first substation electricity demand demand up to the Nth substation electricity demand demand, and the first substation power supply distance up to the Nth substation power supply distance; P72: Sort the first substation electricity demand up to the Nth substation electricity demand from large to small to obtain the first substation number sorting result, wherein the first substation number sorting result has a first set weight; P73: Sort the first substation power supply distance up to the Nth substation power supply distance from small to large to obtain the second substation number sorting result, wherein the second substation number sorting result has a second set weight; P74: According to the first set weight and the second set weight, the first substation number sorting result and the second substation number sorting result are weightedly added with the substation number sequence to obtain a substation demand score set; P75: Sort the substations from small to large according to the substation demand score set to obtain a third substation number sorting result; P76: According to the predicted power generation, power dispatch is performed according to the third substation number sorting result.

[0074] Optionally, the system responds to power demand requests from different substations and performs precise power dispatch based on the previously generated predicted power generation. The substation power demand requests refer to the power demand of each area in the power grid, including key data such as the required power consumption and power supply distance of multiple substations. These requests are subdivided into the power demand of the first substation to the required power consumption of the Nth substation, and the power supply distance of the first substation to the power supply distance of the Nth substation. The power demand is the power load required by each substation, and the power supply distance refers to the physical distance from the power source to the substation.

[0075] First, the first substation's power demand is sorted from the first substation to the Nth substation, from highest to lowest, to obtain a first substation number sorting result. This sorting result reflects the priority of power demand and is assigned a first set weight to indicate the relative importance of the demand. This ensures that substations with higher power demand are prioritized during power dispatch.

[0076] Next, the power supply distances from the first substation to the Nth substation are sorted in ascending order, yielding a second substation numbering ranking. Power supply distance is primarily considered to reduce losses during power transmission and improve transmission efficiency. Shorter power supply distances mean less energy loss and lower costs. This ranking result is assigned a second set weight to reflect its relative importance in power dispatch.

[0077] Furthermore, based on the first and second preset weights, the weighted sum of the first and second substation number sorting results is performed, with the substation number sequence number being the same. This process comprehensively considers both power demand and power supply distance to produce a more comprehensive and objective set of substation demand scores. Each score in this set corresponds to a substation, reflecting its priority in power dispatch and balancing demand and cost.

[0078] The third substation number ranking is then obtained by sorting the substation demand score set from smallest to largest. This ranking will serve as the basis for power dispatch, ensuring that the power demand of each substation is met while achieving the optimal allocation of power resources.

[0079] Finally, based on the predicted power generation, power dispatch is performed according to the third substation number sorting results. Actual power generation is matched with the substation's power demand, ensuring the stability and reliability of power supply through a reasonable dispatch strategy. This ensures the rational allocation of power while meeting priority needs, optimizes power utilization efficiency, and achieves intelligent and refined power dispatch.

[0080] In summary, the embodiments of the present application have at least the following technical effects:

[0081] This application constructs a first-level digital twin model by performing topological simulation based on power plant clusters. It then divides the space based on power stations and power generation units to obtain second- and third-level digital twin models. After embedding the power generation capacity evaluation function, it connects with the power generation capacity fusion layer to generate a global power generation twin mapping model for power generation forecasting and power dispatching. This achieves the technical effect of gradually improving the accuracy of power generation forecasting by constructing multi-level digital twin models from power plant clusters to power generation units, providing more accurate data support for power dispatching, and thus enhancing the flexibility and accuracy of power dispatching. Example 2

[0082] Based on the same inventive concept as the wind power cluster scheduling optimization method driven by digital twins in the aforementioned embodiment, Figure 3 As shown, this application provides a wind power cluster scheduling optimization system driven by digital twins. The system and method embodiments in the embodiments of this application are based on the same inventive concept. The system includes:

[0083] The first-level digital twin model construction module 11 is used to perform topological simulation according to the power plant group and construct a first-level digital twin model.

[0084] The secondary digital twin model construction module 12 is used to spatially divide the primary digital twin model according to the power generation station to obtain the secondary digital twin model.

[0085] The three-level digital twin model construction module 13 is used to spatially divide the two-level digital twin model according to the power generation unit to obtain a three-level digital twin model.

[0086] The power generation capacity twin mapping model construction module 14 is used to embed a power generation capacity evaluation function in the power generation unit twin model of the three-level digital twin model to obtain a scattered power generation capacity twin mapping model.

[0087] The power generation twin mapping model generation module 15 is used to construct a power generation capacity fusion layer, which is connected to the output layer of the scattered power generation capacity twin mapping model to generate a global power generation twin mapping model.

[0088] The power generation prediction module 16 is used to collect meteorological factor prediction information and generate predicted power generation through the global power generation twin mapping model.

[0089] The power dispatching module 17 is used to respond to the power demand request of the substation and perform power dispatch based on the predicted power generation.

[0090] Furthermore, the power generation capability twin mapping model construction module 14 is further configured to perform the following steps:

[0091] Obtain the first power generation unit twin model of the power generation unit twin model of the three-level digital twin model, wherein the first power generation unit twin model has a power generation unit model label and a power generation unit service time label; according to the power generation unit model label and the power generation unit service time label, collect a meteorological factor record data set and a power generation record data set; according to the meteorological factor record data set and the power generation record data set, construct a first power generation unit power generation capacity evaluation function; embed the first power generation unit power generation capacity evaluation function into the first power generation unit twin model; when each power generation unit twin model has a power generation capacity evaluation function, obtain the scattered power generation capacity twin mapping model.

[0092] Furthermore, the power generation capability twin mapping model construction module 14 is further configured to perform the following steps:

[0093] According to the meteorological factor record data set and the power generation record data set, the initial neural network architecture is trained to obtain a first-level power generation capacity evaluation function; the first-level power generation capacity evaluation function is verified to obtain a verification deviation vector set; the system vector is extracted according to the verification deviation vector set to obtain a first-level system error vector modulus and a first-level random error modulus; when the first-level system error vector modulus is less than the system error vector modulus threshold, and the first-level random error modulus is less than the random error modulus threshold, the first power generation unit power generation capacity evaluation function is generated; when the first-level system error vector modulus is less than the system error vector modulus threshold, and the first-level random error modulus is greater than or equal to the random error modulus threshold, the initial neural network architecture continues to be trained according to the meteorological factor record data set and the power generation record data set.

[0094] Furthermore, the power generation capability twin mapping model construction module 14 is further configured to perform the following steps:

[0095] When the modulus of the first-level system error vector is greater than or equal to the system error vector modulus threshold, a first-level residual fitting block is constructed according to the first-level system error vector; the input layer of the first-level residual fitting block is fused with the output layer of the first-level power generation capacity evaluation function to obtain a second-level neural network architecture; according to the meteorological factor record data set and the power generation record data set, iterative training is performed on the second-level neural network architecture, wherein the training data of the second-level neural network architecture is different from the training data of the initial neural network architecture.

[0096] Furthermore, the power generation capability twin mapping model construction module 14 is further configured to perform the following steps:

[0097] Perform outlier cleaning on the verification bias vector set to obtain a selected verification bias vector set; extract the minimum value of the selected verification bias vector set and set it as the first verification bias vector; extract the maximum value of the selected verification bias vector set and set it as the second verification bias vector; calculate the mean of the sum of the first verification bias vector and the second verification bias vector to obtain the first-level system error vector; use the first verification bias vector to subtract the first-level system error vector and take the modulus value to obtain a first random error modulus value; use the second verification bias vector to subtract the first-level system error vector and take the modulus value to obtain a second random error modulus value; calculate the mean of the first random error modulus value and the second random error modulus value and set it as the first-level random error modulus value.

[0098] Furthermore, the power generation twin mapping model generation module 15 is further configured to perform the following steps:

[0099] Obtain a first-node power generation mapping model of the scattered power generation capacity twin mapping model; obtain the power generation section data of the first node in the preset time zone, wherein the power generation section data of the first node in the preset time zone includes the first-node power generation record data and the actual power generation record data of the field group, and the preset time zone refers to the historical time zone six months before the current moment; perform power generation efficiency analysis based on the first-node power generation record data and the actual power generation record data of the field group to obtain the temporary power generation efficiency of the first node; repeat the analysis to obtain the temporary power generation efficiency of the second node until the temporary power generation efficiency of the Q-th node; generate the power generation capacity fusion layer based on the first node temporary power generation efficiency, the second node temporary power generation efficiency, and the Q-node temporary power generation efficiency, perform weighted addition and full connection on each node of the scattered power generation capacity twin mapping model, and obtain the global power generation twin mapping model.

[0100] Furthermore, the power dispatching module 17 is further configured to perform the following steps:

[0101] The substation electricity request includes the power consumption demand of the first substation to the power consumption demand of the Nth substation, and the power supply distance of the first substation to the power supply distance of the Nth substation; sorting from large to small according to the power consumption demand of the first substation to the power consumption demand of the Nth substation, and obtaining a first substation number sorting result, wherein the first substation number sorting result has a first set weight; sorting from small to large according to the power supply distance of the first substation to the power supply distance of the Nth substation, and obtaining a second substation number sorting result, wherein the second substation number sorting result has a second set weight; according to the first set weight and the second set weight, performing weighted addition of the first substation number sorting result and the second substation number sorting result with the same substation number sequence number to obtain a substation demand score set; sorting the substations from small to large according to the substation demand score set to obtain a third substation number sorting result; according to the predicted power generation, power dispatching is performed according to the third substation number sorting result.

[0102] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0103] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

[0104] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. Wind power cluster dispatch optimization method driven by digital twins, It is characterized in that include: Conduct topological simulation based on the power plant group and build a first-level digital twin model; spatially dividing the first-level digital twin model according to the power generation station to obtain a second-level digital twin model; Spatially dividing the secondary digital twin model according to the power generation unit to obtain a third-level digital twin model; Embedding a power generation capacity evaluation function in the power generation unit twin model of the three-level digital twin model to obtain a scattered point power generation capacity twin mapping model; Constructing a power generation capacity fusion layer, connecting it with the output layer of the scattered power generation capacity twin mapping model, and generating a global power generation twin mapping model; Collecting meteorological factor prediction information, and generating predicted power generation through the global power generation twin mapping model; In response to a power demand request from a substation, performing power dispatch based on the predicted power generation; The power generation capacity evaluation function is embedded in the power generation unit twin model of the three-level digital twin model to obtain a scattered power generation capacity twin mapping model, including: Obtaining a first power generation unit twin model of the power generation unit twin model of the three-level digital twin model, wherein the first power generation unit twin model has a power generation unit model label and a power generation unit service time label; According to the power generation unit model label and the power generation unit service time label, a meteorological factor record data set and a power generation record data set are collected; Constructing a power generation capacity evaluation function of a first power generation unit according to the meteorological factor record data set and the power generation record data set; Embedding the power generation capacity evaluation function of the first power generation unit into the twin model of the first power generation unit; When each power generation unit twin model has a power generation capacity evaluation function, the scattered point power generation capacity twin mapping model is obtained; According to the meteorological factor record data set and the power generation record data set, constructing a power generation capacity evaluation function of the first power generation unit includes: According to the meteorological factor record data set and the power generation record data set, training the initial neural network architecture to obtain a first-level power generation capacity evaluation function; Verifying the primary power generation capacity evaluation function to obtain a verification deviation vector set; Perform system vector extraction according to the verification deviation vector set to obtain a first-level system error vector modulus and a first-level random error modulus; When the first-level system error vector modulus is less than the system error vector modulus threshold, and the first-level random error modulus is less than the random error modulus threshold, generating the first power generation unit power generation capacity evaluation function; When the first-level system error vector modulus is less than the system error vector modulus threshold, and the first-level random error modulus is greater than or equal to the random error modulus threshold, continuing to train the initial neural network architecture according to the meteorological factor record data set and the power generation record data set; Among them, it also includes: When the first-level system error vector modulus is greater than or equal to the system error vector modulus threshold, constructing a first-level residual fitting block according to the first-level system error vector; Fusing the input layer of the primary residual fitting block with the output layer of the primary power generation capacity evaluation function to obtain a secondary neural network architecture; Performing iterative training on a secondary neural network architecture according to the meteorological factor record data set and the power generation record data set, wherein the training data of the secondary neural network architecture is different from the training data of the initial neural network architecture; Among them, performing system vector extraction according to the verification deviation vector set to obtain a first-level system error vector modulus and a first-level random error modulus includes: Performing outlier cleaning on the verification deviation vector set to obtain a selected verification deviation vector set; Extracting a minimum value of the selected verification deviation vector set and setting it as a first verification deviation vector; Extracting the maximum value of the selected verification deviation vector set and setting it as the second verification deviation vector; Calculating a mean of the sum of the first verification deviation vector and the second verification deviation vector to obtain the primary system error vector; Subtract the first-level system error vector from the first verification deviation vector and take the modulus value to obtain a first random error modulus value; Subtract the first-level system error vector from the second verification deviation vector and take the modulus value to obtain a second random error modulus value; Calculate the mean of the first random error modulus value and the second random error modulus value, and set it as the primary random error modulus value; Among them, constructing a power generation capacity fusion layer, connecting it with the output layer of the scattered power generation capacity twin mapping model, and generating a global power generation twin mapping model, including: Obtaining a first node power generation mapping model of the scattered point power generation capacity twin mapping model; Obtaining power generation section data of a first node preset time zone, wherein the power generation section data of the first node preset time zone includes the first node power generation record data and the field group actual power generation record data, and the preset time zone refers to the historical time zone six months before the current time; Perform power generation efficiency analysis based on the first node power generation record data and the field group actual power generation record data to obtain the first node temporary power generation efficiency; Repeat the analysis to obtain the temporary power generation efficiency of the second node until the temporary power generation efficiency of the Qth node; According to the temporary power generation efficiency of the first node, the temporary power generation efficiency of the second node, and up to the temporary power generation efficiency of the Qth node, the power generation capacity fusion layer is generated, and each node of the scattered power generation capacity twin mapping model is weightedly added and fully connected to obtain the global power generation twin mapping model; In response to the power demand request of the substation, power dispatch is performed based on the predicted power generation, including: The power consumption request of the substation area includes the power consumption required by the first substation area to the power consumption required by the Nth substation area, and the power supply distance of the first substation area to the power supply distance of the Nth substation area; Sorting the power consumption demand of the first substation to the power consumption demand of the Nth substation from large to small to obtain a first substation number sorting result, wherein the first substation number sorting result has a first set weight; Sorting the first area power supply distance to the Nth area power supply distance from small to large to obtain a second area number sorting result, wherein the second area number sorting result has a second set weight; According to the first set weight and the second set weight, performing weighted summation of the first substation number sorting result and the second substation number sorting result with the same substation number sequence number to obtain a substation demand score set; Sort the stations from small to large according to the station demand score set to obtain a third station number sorting result; According to the predicted power generation, power dispatch is performed according to the sorting result of the third substation number.

2. Digital twin-driven wind power cluster dispatch optimization system, It is characterized in that The system comprises: A first-level digital twin model construction module, wherein the first-level digital twin model construction module is used to perform topological simulation according to the power plant group to construct a first-level digital twin model; A secondary digital twin model construction module, wherein the secondary digital twin model construction module is used to spatially divide the primary digital twin model according to the power generation station to obtain a secondary digital twin model; A three-level digital twin model construction module, wherein the three-level digital twin model construction module is used to spatially divide the two-level digital twin model according to the power generation unit to obtain a three-level digital twin model; A power generation capacity twin mapping model construction module, wherein the power generation capacity twin mapping model construction module is used to embed a power generation capacity evaluation function in the power generation unit twin model of the three-level digital twin model to obtain a scattered power generation capacity twin mapping model; A power generation twin mapping model generation module, wherein the power generation twin mapping model generation module is used to construct a power generation capacity fusion layer, connected to the output layer of the scattered power generation capacity twin mapping model, to generate a global power generation twin mapping model; A power generation prediction module, the power generation prediction module is used to collect meteorological factor prediction information and generate predicted power generation through the global power generation twin mapping model; A power dispatching module, the power dispatching module is used to respond to the power demand of the substation area and perform power dispatching based on the predicted power generation; The power generation capacity twin mapping model building module is further used to perform the following steps: Obtaining a first power generation unit twin model of the power generation unit twin model of the three-level digital twin model, wherein the first power generation unit twin model has a power generation unit model label and a power generation unit service time label; According to the power generation unit model label and the power generation unit service time label, a meteorological factor record data set and a power generation record data set are collected; Constructing a power generation capacity evaluation function of a first power generation unit according to the meteorological factor record data set and the power generation record data set; Embedding the power generation capacity evaluation function of the first power generation unit into the twin model of the first power generation unit; When each power generation unit twin model has a power generation capacity evaluation function, the scattered point power generation capacity twin mapping model is obtained; The power generation capacity twin mapping model building module is further used to perform the following steps: According to the meteorological factor record data set and the power generation record data set, training the initial neural network architecture to obtain a first-level power generation capacity evaluation function; Verifying the primary power generation capacity evaluation function to obtain a verification deviation vector set; Perform system vector extraction according to the verification deviation vector set to obtain a first-level system error vector modulus and a first-level random error modulus; When the first-level system error vector modulus is less than the system error vector modulus threshold, and the first-level random error modulus is less than the random error modulus threshold, generating the first power generation unit power generation capacity evaluation function; When the first-level system error vector modulus is less than the system error vector modulus threshold, and the first-level random error modulus is greater than or equal to the random error modulus threshold, continuing to train the initial neural network architecture according to the meteorological factor record data set and the power generation record data set; The power generation capacity twin mapping model building module is further used to perform the following steps: When the first-level system error vector modulus is greater than or equal to the system error vector modulus threshold, constructing a first-level residual fitting block according to the first-level system error vector; Fusing the input layer of the primary residual fitting block with the output layer of the primary power generation capacity evaluation function to obtain a secondary neural network architecture; Performing iterative training on a secondary neural network architecture according to the meteorological factor record data set and the power generation record data set, wherein the training data of the secondary neural network architecture is different from the training data of the initial neural network architecture; The power generation capacity twin mapping model building module is further used to perform the following steps: Performing outlier cleaning on the verification deviation vector set to obtain a selected verification deviation vector set; Extracting a minimum value of the selected verification deviation vector set and setting it as a first verification deviation vector; Extracting the maximum value of the selected verification deviation vector set and setting it as the second verification deviation vector; Calculating a mean of the sum of the first verification deviation vector and the second verification deviation vector to obtain the primary system error vector; Subtract the first-level system error vector from the first verification deviation vector and take the modulus value to obtain a first random error modulus value; Subtract the first-level system error vector from the second verification deviation vector and take the modulus value to obtain a second random error modulus value; Calculate the mean of the first random error modulus value and the second random error modulus value, and set it as the primary random error modulus value; The power generation twin mapping model generation module is further used to perform the following steps: Obtaining a first node power generation mapping model of the scattered point power generation capacity twin mapping model; Obtaining power generation section data of a first node preset time zone, wherein the power generation section data of the first node preset time zone includes the first node power generation record data and the field group actual power generation record data, and the preset time zone refers to the historical time zone six months before the current time; Perform power generation efficiency analysis based on the first node power generation record data and the field group actual power generation record data to obtain the first node temporary power generation efficiency; Repeat the analysis to obtain the temporary power generation efficiency of the second node until the temporary power generation efficiency of the Qth node; According to the temporary power generation efficiency of the first node, the temporary power generation efficiency of the second node, and up to the temporary power generation efficiency of the Qth node, the power generation capacity fusion layer is generated, and each node of the scattered power generation capacity twin mapping model is weightedly added and fully connected to obtain the global power generation twin mapping model; The power dispatching module is further used to perform the following steps: The power consumption request of the substation area includes the power consumption required by the first substation area to the power consumption required by the Nth substation area, and the power supply distance of the first substation area to the power supply distance of the Nth substation area; Sorting the power consumption demand of the first substation to the power consumption demand of the Nth substation from large to small to obtain a first substation number sorting result, wherein the first substation number sorting result has a first set weight; Sorting the first area power supply distance to the Nth area power supply distance from small to large to obtain a second area number sorting result, wherein the second area number sorting result has a second set weight; According to the first set weight and the second set weight, performing weighted summation of the first substation number sorting result and the second substation number sorting result with the same substation number sequence number to obtain a substation demand score set; Sort the stations from small to large according to the station demand score set to obtain a third station number sorting result; According to the predicted power generation, power dispatch is performed according to the sorting result of the third substation number.

Images