Intelligent scheduling method and system based on efficient coordination of source network load storage

Abstract

The invention provides an intelligent scheduling method and system based on source network load storage efficient collaboration, and relates to the technical field of source network load storage systems.The method comprises the steps that operation data of all nodes in a power grid are collected in real time through Internet of Things equipment, integrity verification and preliminary evaluation are conducted on the operation data, and an initial operation data set is formed; performing data cleaning and integration on the initial operation data set to form a unified operation data set; performing feature extraction and state analysis based on the operation data set to obtain a multi-dimensional feature vector; according to the multi-dimensional feature vector and an artificial intelligence technology, constructing a source network load storage collaborative energy management model; and obtaining a power grid operation state and load demand prediction result based on the model. According to the method, a full-process closed loop from data acquisition, processing, strategy generation to execution feedback is constructed, so that source-grid-load-storage efficient collaborative scheduling under data driving is realized, and finally, the goals of safe, stable and economic optimized operation of the power grid are achieved.

Description

Technical Field

[0001] This invention relates to the field of source-grid-load-storage system technology, and in particular to a smart scheduling method and system based on efficient coordination of source, grid, load and storage. Background Technology

[0002] With the accelerated transformation of the energy structure and the advancement of dual-carbon goals, a new type of power system based on new energy sources is developing rapidly, and the coupling between power generation, grid, load, and storage is deepening. Against this backdrop, achieving efficient and coordinated operation of power generation, grid, load, and storage has become crucial for improving the flexibility, economy, and security of the power system. As a core technological means supporting this goal, the effectiveness of intelligent dispatching systems largely depends on their ability to process and analyze multi-source operational data from the power grid.

[0003] However, existing scheduling methods still have significant shortcomings at the data processing level, which restricts further improvement in the effectiveness of coordinated scheduling. First, existing methods lack effective integrity verification and quality assessment mechanisms for operational data collected from multiple nodes of the power grid, and their ability to handle abnormal and missing data is insufficient, which may lead to low data reliability and affect the accuracy of subsequent analysis. Second, some traditional methods are unable to deeply integrate and extract features from massive heterogeneous data from multiple links such as power sources, grids, loads, and storage. The comprehensiveness and representativeness of feature construction are insufficient, and the traditional prediction and modeling techniques they rely on are difficult to accurately capture the complex nonlinear characteristics of the power grid's operating state, which may result in limited load forecast accuracy and weak adaptability of energy management models. Finally, some existing methods generate scheduling instructions based on static or preset models, lacking a mechanism for dynamic feedback and closed-loop correction based on real-time power grid status. This results in insufficient adaptability of scheduling strategies when facing actual fluctuations in power grid operation, making it difficult to achieve precise coordinated control of power sources, grids, loads, and storage. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide a smart scheduling method and system based on efficient coordination of power generation, grid, load and storage. By constructing a closed loop of the entire process from data acquisition, processing, strategy generation to execution feedback, it realizes efficient coordinated scheduling of power generation, grid, load and storage under data drive, and ultimately achieves the goal of power grid safety, stability and economic optimization.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: Firstly, a smart scheduling method based on efficient source-grid-load-storage coordination is provided, the method comprising: The operation data of each node in the power grid is collected in real time by IoT devices, and the integrity of the operation data is verified and preliminarily evaluated to form an initial operation dataset. The initial operational dataset is cleaned and integrated to form a unified operational dataset; feature extraction and state analysis are performed on the operational dataset to obtain multidimensional feature vectors; based on the multidimensional feature vectors and artificial intelligence technology, a source-grid-load-storage coordinated energy management model is constructed; based on this model, the power grid operation status and load demand prediction results are obtained. Based on the power grid operating status and load demand forecast results, and combined with the preset optimization objectives, an initial dispatch strategy is generated. Based on the initial dispatch strategy, a set of monitoring nodes for strategy feedback is determined in the key areas of the power grid. The operating data of the monitoring nodes is collected to construct the power grid operating status assessment area. The assessment area is divided into multiple sub-areas, and the electrical characteristic parameters of each sub-area are extracted. Based on the electrical characteristic parameters, the corresponding correction coefficients are calculated. The initial dispatch strategy is corrected using the correction coefficients to obtain the corrected dispatch strategy. The revised scheduling strategy is parsed into specific equipment control commands, which are then sent to the corresponding smart terminals. Based on the smart terminals, the access and exit of distributed energy resources, the charging and discharging status of energy storage devices, and the load management strategy are controlled by automated actuators, thereby achieving real-time scheduling and control of power grid resources.

[0006] Secondly, a smart scheduling system based on efficient coordination of source, grid, load, and storage includes: The data acquisition module is used to collect operational data from each node in the power grid in real time through IoT devices, perform integrity verification and preliminary evaluation on the operational data, and form an initial operational dataset. The data analysis module is used to clean and integrate the initial operating dataset to form a unified operating dataset; based on the operating dataset, feature extraction and state analysis are performed to obtain multi-dimensional feature vectors; based on the multi-dimensional feature vectors and artificial intelligence technology, a source-grid-load-storage coordinated energy management model is constructed; based on this model, the power grid operating status and load demand prediction results are obtained. The strategy generation module is used to generate an initial scheduling strategy based on the power grid operating status and load demand forecast results, combined with preset optimization objectives; based on the initial scheduling strategy, a set of monitoring nodes for strategy feedback is determined in the key areas of the power grid; the operating data of the monitoring nodes is collected to construct a power grid operating status assessment area; the assessment area is divided into multiple sub-regions, and the electrical characteristic parameters of each sub-region are extracted; the corresponding correction coefficients are calculated based on the electrical characteristic parameters; the initial scheduling strategy is corrected using the correction coefficients to obtain the corrected scheduling strategy. The execution control module is used to parse the modified scheduling strategy into specific equipment control commands and send the equipment control commands to the corresponding smart terminals. Based on the smart terminals, the module controls the access and disconnection of distributed energy sources, adjusts the charging and discharging status of energy storage devices, and adjusts the load management strategy through automated actuators, thereby realizing real-time scheduling and control of power grid resources.

[0007] Thirdly, a computing device, comprising: One or more processors; A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0008] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0009] The above-described solution of the present invention has at least the following beneficial effects: Real-time data collection of power grid node operations via IoT devices ensures timely and comprehensive data acquisition. Data integrity verification and preliminary assessment filter out invalid and incomplete data, ensuring the initial operational dataset possesses basic reliability. The initial operational dataset is cleaned and integrated to eliminate data format differences, redundancy, and errors, forming a unified and standardized operational dataset, improving the consistency of subsequent data analysis. Based on this unified dataset, features are extracted and models are built to deeply mine key data information, ensuring the model closely reflects actual power grid operation and providing high-quality data support for power grid status assessment and load forecasting. Combining power grid status, load forecasting results, and... The initial scheduling strategy is optimized to ensure it has clear data support. Data is collected from monitoring nodes to focus on key areas of the power grid, which are then divided into sub-regions and electrical parameters are extracted to make the correction coefficient calculation more accurate. The initial strategy is then revised using the correction coefficients to better adapt it to the actual operating characteristics of each sub-region of the power grid. The revised scheduling strategy is parsed into specific equipment control commands, transforming the abstract strategy into executable operations. The commands are encapsulated into standardized messages based on the power grid communication protocol to ensure the accuracy and stability of command transmission. The status feedback signals of the actuators are collected and compared with the control commands in a closed loop. The execution effect is confirmed through data comparison to ensure the accuracy of power grid scheduling and control. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating a smart scheduling method based on efficient coordination between source, grid, load, and storage, provided by an embodiment of the present invention.

[0011] Figure 2 This is a schematic diagram of an intelligent scheduling system based on efficient coordination of source, grid, load and storage provided by an embodiment of the present invention. Detailed Implementation

[0012] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0013] like Figure 1 As shown, an embodiment of the present invention proposes a smart scheduling method based on efficient source-grid-load-storage coordination, the method comprising the following steps: Step 100: Real-time collection of operational data from each node in the power grid via IoT devices; integrity verification and preliminary evaluation of the operational data to form an initial operational dataset. Step 200: Clean and integrate the initial operating dataset to form a unified operating dataset; perform feature extraction and state analysis based on the operating dataset to obtain multi-dimensional feature vectors; construct a source-grid-load-storage coordinated energy management model based on the multi-dimensional feature vectors and artificial intelligence technology; and obtain the power grid operating status and load demand prediction results based on the model. Step 300: Based on the power grid operating status and load demand forecast results, and combined with preset optimization objectives, an initial dispatch strategy is generated; based on the initial dispatch strategy, a set of monitoring nodes for strategy feedback is determined in the key areas of the power grid; the operating data of the monitoring nodes is collected to construct a power grid operating status assessment area; the assessment area is divided into multiple sub-areas, and the electrical characteristic parameters of each sub-area are extracted; based on the electrical characteristic parameters, the corresponding correction coefficients are calculated; the initial dispatch strategy is corrected using the correction coefficients to obtain the corrected dispatch strategy. Step 400: The revised scheduling strategy is parsed into specific equipment control commands, and the equipment control commands are sent to the corresponding smart terminals. Based on the smart terminals, the access and exit of distributed energy sources, the charging and discharging status of energy storage devices, and the load management strategy are controlled by automated actuators, thereby realizing real-time scheduling and control of power grid resources.

[0014] In this embodiment of the invention, real-time acquisition of operational data from various nodes within the power grid via IoT devices ensures the timeliness and comprehensiveness of data acquisition. Integrity verification and preliminary assessment of the operational data can filter out invalid and incomplete data in advance, ensuring the initial operational dataset has basic reliability. Cleaning and integrating the initial operational dataset eliminates data format differences, redundant content, and errors, forming a unified and standardized operational dataset, improving the efficiency and consistency of subsequent data analysis. Feature extraction and state analysis based on the unified operational dataset yield multi-dimensional feature vectors, enabling in-depth mining of key information in the data and comprehensively reflecting the multi-dimensional state of power grid operation. Combining multi-dimensional feature vectors with artificial intelligence technology to construct a source-grid-load-storage collaborative energy management model, supported by high-quality, multi-dimensional data, allows the model to better reflect the actual operation of the power grid, thereby clearly understanding the current operating state of the power grid and future load demand. Based on the power grid operating state, load demand prediction results, and preset optimization objectives, a generation of... The initial scheduling strategy ensures that it initially meets the overall operational needs of the power grid. By identifying monitoring nodes to collect data and constructing a power grid operation status assessment area, targeted data can be obtained from key areas of the power grid, improving the accuracy of the assessment. Dividing the assessment area into multiple sub-regions and extracting electrical characteristic parameters for each sub-region refines the operational characteristics of different regions. Calculating correction coefficients based on electrical characteristic parameters and revising the initial scheduling strategy allows the strategy to fully adapt to the actual operational characteristics of different sub-regions of the power grid, enhancing its relevance and feasibility, and ensuring its effective implementation in different regions. Parsing the revised scheduling strategy into specific equipment control commands and sending them to smart terminals ensures the clarity and accuracy of command transmission. Relying on smart terminals and automated actuators to control the access and exit of distributed energy sources, adjust the charging and discharging status of energy storage devices, and adjust load management strategies, real-time scheduling and control of power grid resources can be achieved based on accurate strategies supported by high-quality initial data, ensuring stable power grid operation.

[0015] In a preferred embodiment of the present invention, step 100 above involves collecting operational data from each node within the power grid in real time using IoT devices, performing integrity verification and preliminary evaluation on the operational data, and forming an initial operational dataset, including: Step 101: Perform data integrity verification on the operational data collected in real time through IoT devices and calculate its data integrity rate. If the data integrity rate is higher than a preset threshold, it is determined to be valid data; otherwise, a data supplementation mechanism is triggered. Specifically, this includes: first, determining the specific range of operational data collected by IoT devices for each node in the power grid. This range must at least cover the real-time current data, real-time voltage data, real-time load data, and real-time power generation data of distributed energy sources for each node, ensuring that the collected data can cover the basic parameters required for power grid operation status assessment; next, performing data integrity verification, first preset the total number of data items that need to be collected for each power grid node within a unit collection cycle, where the unit... The data acquisition cycle, such as 5 minutes or 10 minutes, can be set according to the accuracy requirements of power grid dispatching. For example, if a hub node needs to acquire four data items—current, voltage, load, and power generation—within a preset unit cycle, then the total number of data items for that node is set to 4. Then, the number of valid data items actually acquired by the node within the current unit acquisition cycle and verified by preliminary format is counted. This preliminary format verification excludes outliers that clearly exceed the physical range, such as data where the current value exceeds the node's rated current by more than 30%. Based on this, the data integrity rate is calculated by dividing the number of valid data items in the current cycle of that node by the preset total number of data items to obtain the real-time data integrity rate of that node in the current cycle.

[0016] Furthermore, a preset data integrity rate threshold is established. This threshold can be set according to the importance level of the node in the power grid, such as setting the integrity rate threshold of the main power grid node to 95% and the integrity rate threshold of the branch node to 90%. The calculated real-time data integrity rate of the node is compared with the corresponding threshold. If the integrity rate is higher than the preset threshold, the data collected by the node in the current period is determined to be valid data. If the integrity rate is lower than the preset threshold, a data supplementation mechanism is immediately triggered. The specific operation of the supplementation mechanism is to first send a data retransmission instruction to the IoT acquisition device of the node. The instruction clearly indicates the missing data items and the retransmission time limit. If the retransmitted data is received within the retransmission time limit and the recalculated integrity rate meets the standard, the data is included in the category of valid data. If the integrity rate still does not meet the standard after retransmission, the backup acquisition device of the node is activated to supplement the missing data items. The backup acquisition device is such as redundant sensors deployed in adjacent nodes, until the integrity rate of the collected data meets the preset threshold or reaches the preset upper limit of the number of supplementation times, such as 3 times, to ensure the integrity and availability of valid data.

[0017] Step 102: Based on the valid data, extract the fluctuation characteristics of current and voltage; calculate the fluctuation rate index based on the fluctuation characteristics; perform correlation analysis between the fluctuation rate index and load demand data to calculate the real-time power balance of each node. Specifically, this includes: performing subsequent calculations based on the valid data, extracting the fluctuation characteristics of current and voltage, first setting a time window for feature extraction, the length of which can be set according to the grid fluctuation characteristics, such as selecting 10 consecutive unit acquisition cycles as a feature extraction window; for each grid node, extracting all real-time current data and real-time voltage data within the time window from the valid data, calculating the maximum and minimum values ​​of the current data within the window respectively, the difference between the two is the current fluctuation amplitude within the window, similarly calculating the difference between the maximum and minimum values ​​of the voltage data to obtain the voltage fluctuation amplitude; simultaneously calculating the standard deviation of the current data and the standard deviation of the voltage data within the time window, further introducing a standard deviation ellipse. The circular algorithm optimizes the extraction of fluctuation features. Specifically, it first constructs a two-dimensional data point set using the current data at each moment within the time window as the horizontal axis coordinate and the voltage data at the corresponding moment as the vertical axis coordinate. Then, it calculates the mean center of this data point set, which is the coordinate point formed by the arithmetic mean of the horizontal axis coordinates and the arithmetic mean of the vertical axis coordinates of all data points, and uses it as the center of the standard deviation ellipse. Next, it calculates the standard deviation of this data point set in the horizontal and vertical directions, which correspond to the lengths of the major and minor semi-axes of the standard deviation ellipse, respectively, to reflect the primary and secondary strengths of the joint fluctuations of current and voltage. Then, by calculating the covariance matrix of the data point set, it determines the rotation angle of the standard deviation ellipse. This rotation angle can characterize the main direction of the joint fluctuations of current and voltage. The current fluctuation amplitude, voltage fluctuation amplitude, current standard deviation, voltage standard deviation, and the center coordinates, major and minor semi-axes, and rotation angle of the standard deviation ellipse are used together as the current and voltage fluctuation features within the current time window of this node.

[0018] Further calculations of volatility indicators are performed. For current volatility, the current fluctuation amplitude obtained above is divided by the arithmetic mean of all current data within the time window to obtain the relative current volatility. For voltage volatility, the voltage fluctuation amplitude is divided by the arithmetic mean of all voltage data within the time window to obtain the relative voltage volatility. The relative current volatility and the relative voltage volatility are used together as the volatility indicators for the current time window of this node.

[0019] Based on this, a correlation analysis is performed between volatility indicators and load demand data to calculate the real-time power balance of each node. A vector projection algorithm is introduced to enhance the accuracy of the correlation analysis. The specific operation is as follows: First, extract the real-time load demand data of the node within the same time window from the valid data. This data includes the node's own electricity load data and the load data transmitted to adjacent nodes. The relative current volatility and relative voltage volatility are combined to form a volatility feature vector, and the node's own electricity load data and the load data transmitted to adjacent nodes are combined to form a load demand vector. Then, a projection space for the correlation analysis is set. This projection space uses the power transmission efficiency of the grid node as the benchmark dimension. The first projection coefficient of the volatility feature vector in this projection space, as well as the load... The second projection coefficient of the demand vector in the projection space is used to determine the correlation weight between the volatility index and the load demand data by the ratio of the first projection coefficient to the second projection coefficient. At the same time, the real-time power generation data of the distributed energy connected to the node in the same period is extracted. Based on the power balance relationship and combined with the above correlation weight, the real-time supply power of the node is first calculated, which is the sum of the real-time power generation of the distributed energy and the real-time power input to the node by the grid. Then, the real-time power consumption of the node is calculated, which is the sum of the node's own power load and the real-time power output to adjacent nodes. Finally, the real-time power balance degree is calculated by dividing the absolute value of the difference between the real-time supply power and the real-time consumption power by the real-time supply power to obtain the real-time power balance degree of the node in the current time window.

[0020] Step 103: Based on the real-time power balance, perform a preliminary assessment of the power grid operating status to obtain status assessment parameters; identify power imbalance nodes based on the status assessment parameters to obtain a power imbalance node list; integrate the power imbalance node list with the operating data to form an initial operating dataset, specifically including: firstly, perform a preliminary assessment of the power grid operating status based on the real-time power balance to obtain status assessment parameters; firstly, set a reasonable range for the power balance based on the functional positioning and operating requirements of each node in the power grid. For example, for industrial load nodes with high power supply stability requirements, the reasonable range for their power balance is set to 0 to 5%, while for residential load nodes with relatively low power supply stability requirements... The nodes are assigned a reasonable range of 0% to 8%. The real-time power balance of each node is compared with the corresponding reasonable range. If the real-time power balance of the node is within the reasonable range, the operating status evaluation parameter of the node is determined to be normal. If the real-time power balance exceeds the reasonable range, and it is verified that the real-time power supply of the node is greater than the real-time power consumption, the status evaluation parameter is determined to be power surplus. If the real-time power balance exceeds the reasonable range and the real-time power supply is less than the real-time power consumption, the status evaluation parameter is determined to be power deficit. Through the above comparison process, the status evaluation parameters of all grid nodes are obtained, thereby completing the preliminary evaluation of the overall operating status of the grid.

[0021] Further, power imbalance nodes are identified based on state assessment parameters to obtain a power imbalance node list. The state assessment parameters of all grid nodes are traversed, and nodes with state assessment parameters indicating power surplus or power deficit are selected. For each selected node, its node number, the grid region it belongs to, and the current real-time power balance value are recorded. The grid region is, for example, the eastern power supply area or the western wind power aggregation area. The power surplus or power deficit is calculated by subtracting real-time power consumption from real-time power supply, and the power deficit is calculated by subtracting real-time power supply from real-time power consumption. This information is then organized according to a preset format to form a power imbalance node list, where the preset format includes, for example, node number, region, balance value, imbalance type, and imbalance amount.

[0022] First, collect all valid operating data of all nodes obtained in step 101. This data includes raw data such as real-time current, real-time voltage, real-time load demand, and real-time power generation of distributed energy sources for each node, as well as the volatility index calculated in step 102. Then, associate and match the information in the power imbalance node list with the valid operating data of the corresponding nodes. That is, for power imbalance nodes, add information such as imbalance status (surplus / deficit) and imbalance amount to their original valid operating data. For nodes whose status is assessed as normal, maintain the integrity of their valid operating data. Through the above association and integration operations, an initial operating dataset is formed that covers the basic operating data of each node in the power grid, key calculation indicators, and special information on imbalance nodes.

[0023] In this embodiment of the invention, by calculating the integrity rate of the operational data and performing integrity verification, the data quality can be quantitatively judged to ensure that the data entering subsequent stages has basic validity. When the data integrity rate does not reach a preset threshold, a data supplementation mechanism is triggered to reduce data loss and avoid deviations in subsequent analysis due to missing key data. Based on the effective data, the fluctuation characteristics of current and voltage can be extracted, and key information reflecting the dynamic changes of power grid electrical parameters can be mined from the original operational data. By calculating the volatility index through fluctuation characteristics, the degree of change of electrical parameters can be quantitatively presented. The volatility index is correlated with load demand data, and the real-time power level of each node is calculated. The power balance function can establish the correlation between changes in electrical parameters and load demand, clarify the power supply and demand matching status of each node, and provide specific data basis for power grid operation status assessment. Based on the real-time power balance, a preliminary assessment of the power grid operation status is conducted to obtain status assessment parameters, which can transform the abstract power grid operation status into specific parameters that can be referenced, clearly reflecting the overall operation status of the power grid. Based on the status assessment parameters, power imbalance nodes are identified and a list is generated, which can accurately locate key nodes in the power grid where power supply and demand are mismatched. The list of power imbalance nodes is integrated with the operation data to form an initial operation dataset, providing more comprehensive and focused data support for subsequent data cleaning, model building, and other stages.

[0024] In a preferred embodiment of the present invention, step 200 involves cleaning and integrating the initial operating dataset to form a unified operating dataset; performing feature extraction and state analysis based on the operating dataset to obtain a multi-dimensional feature vector; constructing a source-grid-load-storage coordinated energy management model based on the multi-dimensional feature vector and artificial intelligence technology; and obtaining the grid operating state and load demand prediction results based on this model, including: Step 201 involves cleaning the initial running dataset by using the interquartile range (IMR) method to identify and remove outlier data, and then using linear interpolation to fill in missing values, thus forming a unified running dataset. Specifically, this includes: first, performing outlier identification and removal using the IMR method. This requires determining the quartiles of each data item in the initial running dataset, such as current, voltage, load demand, and power generation. Specifically, all data for a given data item is sorted in ascending order. The value at the 25th percentile of the total sorted data is taken as the first quartile, and the value at the 75th percentile is taken as the third quartile. Then, the IMR of that data item is calculated, which is the difference between the third quartile and the first quartile. Next, the outlier judgment range is determined. The lower threshold is the first quartile minus 1.5 times the IMR, and the upper threshold is the third quartile plus 1.5 times the IMR. Data exceeding these thresholds is considered outlier and removed from the initial running dataset.

[0025] Furthermore, missing value imputation is performed on the dataset after removing outlier data. This operation is accomplished using linear interpolation. Specifically, the data items of each node in the dataset are traversed first to identify the time points where data is missing. For a missing data point, the two adjacent valid data points are identified, and their corresponding timestamps and data values ​​are recorded. Then, according to the principle of linear interpolation, the imputation value of the missing data point is calculated. That is, the time interval and data difference between the two valid data points are calculated first, and then the data difference is allocated proportionally according to the proportion of the time interval between the missing data point and the previous valid data point to the total time interval, thus obtaining the imputation value of the missing data point. After all missing data points are imputed in the above way, the format of the dataset is standardized. For example, the data of each node is arranged in a fixed field order of node number, timestamp, current, voltage, load, power generation, and power balance, ensuring that the units of each data item are consistent, such as current being uniformly converted to amperes and voltage to volts. Finally, a standardized, complete, and anomaly-free unified operating dataset is formed.

[0026] Step 202: Based on the unified operational dataset, extract the time-series variation features of the load data to obtain a load time-series feature set; simultaneously, calculate the fluctuation features of distributed energy generation power to obtain a power fluctuation feature set; fuse the load time-series feature set and the power fluctuation feature set to construct a multi-dimensional feature vector, specifically including: firstly, extracting the time-series variation features of the load data, requiring the setting of the time granularity for feature extraction, which can be set according to the time accuracy requirements of power grid dispatching, for example, dividing the time interval by hours; for each power grid node, extracting the node's time intervals from the unified operational dataset. The load demand data within a time interval is used to calculate the average load, peak load, trough load, and peak-to-trough difference for each time interval. The average load is the arithmetic mean of all load data within the interval, the peak load is the maximum value of the load data within the interval, the trough load is the minimum value of the load data within the interval, and the peak-to-trough difference is the difference between the peak value and the trough value. At the same time, the load change rate of adjacent time intervals is calculated, which is the difference between the average load of the later interval and the average load of the earlier interval divided by the average load of the earlier interval. The above average load, peak load, trough load, peak-to-trough difference, and change rate are arranged in chronological order to form the load time series feature set of that node.

[0027] Meanwhile, to calculate the fluctuation characteristics of distributed energy generation power, a time window for calculating the fluctuation characteristics is first set, for example, selecting four consecutive time intervals as a time window. For each node connected to distributed energy, the power generation data within each time window is extracted from the unified operation dataset, and the maximum power, minimum power, fluctuation amplitude, and power standard deviation within each time window are calculated. Among them, the fluctuation amplitude is the difference between the maximum and minimum values, and the power standard deviation is the arithmetic square root of the sum of squares of the deviations of each power data from the power mean within the window. These parameters are arranged in the order of the time windows to form the power fluctuation characteristic set of the node.

[0028] Based on this, the load time-series feature set and the power fluctuation feature set are fused to construct a multi-dimensional feature vector. Specifically, the two feature sets are first aligned according to the time dimension to ensure that the load features and power features in the same time interval or time window correspond one-to-one. Then, the aligned feature parameters are combined in a preset order. For example, the multi-dimensional feature vector of each time node includes the load mean, load peak, load valley, load change rate, maximum power, minimum power, power fluctuation amplitude, and power standard deviation. By combining the core features of the load side and the generation side in this way, a multi-dimensional feature vector that can comprehensively reflect the power grid operation status is finally formed.

[0029] Step 203: Input the multidimensional feature vector into the prediction model built on artificial intelligence technology. Through the forward propagation calculation of the prediction model, obtain the quantitative assessment result of the power grid operation status including stability indicators and the predicted load demand value for future periods. Specifically, this includes: firstly, preprocessing the multidimensional feature vector to normalize the values ​​of each feature parameter to the same range, such as the interval between 0 and 1, to avoid the influence of differences in the magnitude of feature parameters on the model calculation results; before inputting the preprocessed multidimensional feature vector into the model, the prediction model based on artificial intelligence technology needs to be constructed. The specific construction process is as follows: the prediction model adopts a deep learning model, such as a long short-term memory network model. The model structure includes an input layer, a hidden layer, and an output layer. The number of neurons in the input layer is consistent with the dimension of the multidimensional feature vector to accurately receive the preprocessed feature vector. The system employs a multidimensional feature vector and two to three long short-term memory (LSTM) network layers in the hidden layer. The number of neurons in each hidden layer is determined by the dimension of the multidimensional feature vector. For example, if the feature vector is 8-dimensional, the number of neurons in each layer is set to 32 or 64. The hidden layers are connected via fully connected layers, and a dropout mechanism is introduced to avoid overfitting during model training. The output layer has two types of output nodes based on the prediction target. One type outputs the quantitative evaluation results of the power grid operation status, with the number of nodes matching the types of stability indicators. For example, if the indicators include voltage deviation rate and frequency fluctuation range, two output nodes are set. The other type outputs the predicted load demand value for future periods, with the number of nodes matching the number of time nodes in the future preset period. For example, if the prediction is for the next 24 hours and one time node is set for each hour, 24 output nodes are set.

[0030] After the model is built, the model parameters need to be optimized through training to ensure that the model has accurate evaluation and prediction capabilities. The specific training process is as follows: Select a unified operation dataset of the power grid for 1 to 3 years in history, and construct a multi-dimensional feature vector from it according to the method in step 202 as training input data. At the same time, collect the actual power grid operation status data (such as actual voltage deviation rate, actual frequency fluctuation range) and actual load data for the corresponding historical periods as training label data. Use the mean square error function as the loss function of the model to calculate the error between the model prediction result and the training label data. Select the adaptive momentum estimation optimizer as the optimizer for model training to achieve efficient adjustment of the model weight matrix and bias term. Divide the training input data and label data into training set and validation set in a 7:3 ratio. Input the training set data into the model in batches for iterative training. After each iteration, use the validation set data to evaluate the model performance. If the decrease in validation error is less than the preset threshold in 5 to 10 consecutive iterations, the model error is considered to have converged, training is stopped, and the optimization of model parameters is completed.

[0031] The preprocessed multidimensional feature vectors are then input into the trained prediction model. Specifically, the multidimensional feature vectors are first input into the model's input layer, which performs an initial mapping, transforming them into a vector format that the model's hidden layers can process. Next, the model's forward propagation is executed. The hidden layers, based on the trained weight matrix and bias terms, perform linear transformations and nonlinear activation processing on the input vectors, such as using the ReLU activation function for nonlinear mapping. Through the progressive calculations of multiple hidden layers, the power grid operation patterns and load change trends contained in the feature vectors are extracted. The model's output layer outputs two types of results based on the calculation results of the hidden layers: one type includes power grid operation data containing stability indicators. The quantitative assessment results of grid operation status include stability indicators such as voltage deviation rate and frequency fluctuation range. The voltage deviation rate is the ratio of the difference between the actual voltage and the rated voltage to the rated voltage, and the frequency fluctuation range is the maximum difference in grid frequency within a certain period. These indicators are obtained by linear regression calculation in the output layer to obtain specific quantitative values. Another type is the load demand forecast value for future periods. This forecast value is calculated by the model learning the historical load time series characteristics and power fluctuation characteristics to obtain the estimated load demand value at each time node in the future preset period, such as the next 24 hours or the next 72 hours. Finally, the quantitative assessment of grid operation status and load demand forecast are completed.

[0032] Step 204: Based on the power grid operation status assessment results and load demand forecasts, establish the energy balance relationship among the power source, grid, load, and storage components to obtain a coordinated scheduling rule containing multi-objective constraints; based on the coordinated scheduling rule, construct a coordinated energy management model for the power source, grid, load, and storage components; through the coordinated energy management model, obtain the power grid operation status and load demand forecast results, specifically including: based on the power grid operation status assessment results and load demand forecasts, establish the energy balance relationship among the power source, grid, load, and storage components, first statistically analyzing the total power supplied by the power source, the total power consumed by the load, and the regulating power of the storage side, wherein the total power supplied by the power source includes the sum of the distributed energy generation power and the power input from the large power grid to the nodes. The total power consumption on the load side includes the sum of the node's own electrical load and the load output to adjacent nodes. The power regulation on the storage side is the charging and discharging power of the energy storage equipment, which is negative during charging and positive during discharging. Based on the principle of energy conservation, a balance equation is established: total power supplied on the source side = total power consumed on the load side + power regulation on the storage side. At the same time, combined with stability indicators, such as voltage deviation rate needing to be controlled within ±5% and frequency fluctuation range needing to be controlled within 50±0.2Hz, the power constraint range of each link is determined, thereby forming a collaborative scheduling rule containing multiple objective constraints. These multiple objective constraints include economic constraints: lowest operating cost; stability constraints: voltage and frequency within thresholds; and environmental constraints: new energy consumption rate not less than 90%.

[0033] Based on the aforementioned coordinated scheduling rules, a source-grid-load-storage coordinated energy management model is constructed. Specifically, the energy balance relationship is used as the core constraint of the model, and the multi-objective constraints in the coordinated scheduling rules are transformed into the model's objective function, such as the lowest operating cost and the highest renewable energy absorption rate. Historical operating data, such as historical load data and historical power generation data, are incorporated into the model training to ensure that the model can adapt to the actual operating characteristics of the power grid. Finally, through the calculation of this coordinated energy management model, two types of results are output: one is the power grid operating status result, including the power matching situation of each link of source, grid, load, and storage, such as the matching degree between source-side supply and load-side demand, and the compensation effect of storage-side regulation on power imbalance; the other is the load demand prediction result. This result is based on the prediction value in step 203, and is corrected by combining the energy balance relationship of each link of source, grid, load, and storage, so that the prediction result is more in line with the actual energy supply capacity of the power grid. Ultimately, the power grid operating status and load demand prediction results that can support the generation of subsequent scheduling strategies are obtained.

[0034] In this embodiment of the invention, the interquartile range method is used to remove abnormal data and the linear interpolation method is used to fill in missing values, which can reduce the impact of equipment failure or acquisition interference and ensure data continuity and integrity. The resulting unified operation dataset has a standardized format and reliable quality, providing a high-quality data foundation for subsequent feature extraction and model construction. Targeted extraction of load time-series features (such as mean, peak value, and rate of change) and distributed energy power fluctuation features (such as fluctuation amplitude and standard deviation) fully explores key information about power grid operation. The two feature sets are fused to construct a multi-dimensional feature vector, avoiding the limitations of a single dimension and providing comprehensive feature support for artificial intelligence prediction models. The multi-dimensional feature vector is then input into A. The I-prediction model, through forward propagation calculations, transforms characteristic information into quantitative assessment results of power grid operation containing stability indicators and future load demand forecasts. The former transforms the power grid status into specific reference indicators, while the latter clarifies energy allocation requirements, jointly providing accurate data support for establishing the energy balance relationship between power sources, grids, loads, and storage. Combining power grid operation assessments and load forecasts, the established energy balance relationship between power sources, grids, loads, and storage is more realistic. The multi-objective collaborative scheduling rules formed accordingly can take into account the needs of each link. The collaborative energy management model built based on these rules outputs power grid status and load forecast results that can comprehensively reflect the collaborative operation status, providing a reliable basis for the generation of subsequent scheduling strategies.

[0035] In a preferred embodiment of the present invention, step 300 involves generating an initial scheduling strategy based on the power grid operating status and load demand forecast results, combined with a preset optimization objective; determining a set of monitoring nodes for strategy feedback within key areas of the power grid based on the initial scheduling strategy; collecting operating data from the monitoring nodes to construct a power grid operating status assessment area; dividing the assessment area into multiple sub-areas and extracting electrical characteristic parameters for each sub-area; calculating corresponding correction coefficients based on the electrical characteristic parameters; and using the correction coefficients to correct the initial scheduling strategy to obtain a corrected scheduling strategy, including: Step 301: Based on the power grid operation status and load demand forecast results, the power grid operation stability index, power generation cost economic index, and new energy absorption rate index are quantified into multi-objective optimization constraints. Specifically, this includes: based on the power grid operation status and load demand forecast results, for the power grid operation stability index, extracting the predicted voltage deviation rate and frequency fluctuation range values ​​from the forecast results, setting the voltage deviation rate constraint threshold to ±5%, meaning that the voltage deviation rate needs to be controlled within this range in actual operation, and setting the frequency fluctuation range constraint threshold to 50±0.2Hz, using these as quantitative constraints for the stability index; secondly, for the power generation cost economic index, statistically analyzing the unit power generation cost of each type of generator unit during the forecast period, where each type of generator unit, such as coal-fired units... For coal-fired power plants and photovoltaic power plants, the unit power generation cost is calculated based on fuel prices and maintenance costs for coal-fired power plants, and on initial investment depreciation and operation and maintenance costs for photovoltaic power plants. The constraint threshold for the total power generation cost during the forecast period is set to be within 105% of the historical best cost for the same period, thus achieving quantitative constraints on economic indicators. Finally, for the renewable energy absorption rate indicator, based on the predicted total renewable energy generation and the maximum renewable energy that the grid can accept, where the total renewable energy generation is the sum of the predicted photovoltaic and wind power generation, the constraint threshold for the renewable energy absorption rate is set to be no less than 90%, meaning that the actual absorption must reach more than 90% of the predicted total power generation. Through the above operations, the three core indicators are transformed into clear and quantifiable multi-objective optimization constraints, providing a clear boundary basis for subsequent optimization calculations.

[0036] Step 302: The multi-objective constraints are transformed into a single-objective optimization problem using a linear weighted method, and an initial population containing multiple stochastic scheduling schemes is generated. Based on the initial population, the fitness value of each scheduling scheme is calculated, and high-quality individuals are selected through a selection mechanism. Specifically, this includes: First, the single-objective optimization problem is transformed using a linear weighted method. According to the power grid's operational needs during the forecast period, such as prioritizing stability during peak electricity consumption and prioritizing economic efficiency during off-peak periods, weight coefficients are assigned to three quantitative constraint indicators. For example, the weight of the power grid operation stability indicator is set to 0.4, the weight of the power generation cost economic indicator is set to 0.3, and the weight of the new energy absorption rate indicator is set to 0.3. The actual calculated values ​​of each indicator are multiplied by their corresponding weights and then summed to obtain the single-objective optimization function, i.e., the total weighted score. Second, the initial population is generated, with each individual in the population corresponding to a stochastic scheduling scheme. The scheduling scheme includes the output plan of each generator unit, the charging and discharging time and power value of energy storage equipment, and the transfer time of adjustable load on the load side. The output plan of each generator unit is such as the output time and power value of coal-fired units and the priority output strategy of photovoltaic units. Based on the rated parameters of the grid equipment, 50 to 100 scheduling schemes are randomly generated to form an initial population. The rated parameters of the grid equipment are such as the maximum and minimum output of the generator units and the maximum charging and discharging power of the energy storage. On this basis, the fitness value of each scheduling scheme is calculated. Each scheduling scheme is substituted into a single-objective optimization function, and the total weighted score obtained is the fitness value of the scheme. The higher the score, the better the scheme. Finally, the elite retention selection mechanism is used to select high-quality individuals. The scheduling schemes with the top 30% of the fitness values ​​are selected as high-quality individuals to provide a basis for subsequent population iterative optimization.

[0037] Step 303: Perform single-point crossover and uniform mutation operations sequentially on the high-quality individuals to generate a new population; based on the new population, determine whether the maximum number of iterations or fitness convergence condition has been reached; if the condition is met, select the individual with the highest fitness from the current population as the initial scheduling strategy; if the condition is not met, return to perform the fitness-based selection operation and continue the iterative optimization process to finally obtain the initial scheduling strategy, which specifically includes: first, performing a single-point crossover operation on the high-quality individuals, randomly selecting two high-quality individuals as parents, randomly determining a crossover point in the parameter sequence of the scheduling scheme, such as using the 5th time period of the energy storage charging and discharging plan as the crossover point, swapping the parameter fragments after the crossover point of the two parents to generate two child scheduling schemes; then, performing a uniform mutation operation on the crossover-generated child schemes, for each child scheduling scheme's parameters, such as unit output... For the power value and energy storage charging and discharging power, one or two parameter points are randomly selected and slightly adjusted within the allowable range of these parameters (e.g., the unit output is between the maximum and minimum output), such as a fluctuation of ±5%, to generate new offspring individuals. The offspring individuals generated by crossover and mutation are integrated to form a new population. Then, the iteration termination condition is determined. The maximum number of iterations is preset to 50, and the fitness convergence condition is set to the change of the optimal fitness value of the population in 5 consecutive iterations being less than 0.5%. If the new population reaches the maximum number of iterations or meets the fitness convergence condition, the individual with the highest fitness value in the current population is selected as the initial scheduling strategy. If the condition is not met, the new population is used as the initial population for a new round, and the selection operation based on fitness in step 302 is returned to continue the iterative optimization process until the termination condition is met, and the initial scheduling strategy is finally obtained.

[0038] Step 304: Based on the initial scheduling strategy, calculate the power flow distribution of each branch of the power grid and calculate the voltage stability coefficient of each node; identify key power transmission paths based on the power flow distribution, and identify voltage stability weak nodes based on the voltage stability coefficient; merge and deduplicate the nodes on the key power transmission paths and the voltage stability weak nodes to determine a set of monitoring nodes for strategy feedback, specifically including: firstly, calculating the power flow distribution of each branch of the power grid, using the Newton-Raphson power flow calculation method, taking the unit output plan and load demand forecast values ​​in the initial scheduling strategy as input, substituting them into the node admittance matrix of the power grid, solving for the voltage amplitude and phase angle of each node, and then calculating the active and reactive power transmission direction and magnitude of each branch (such as lines and transformers) to form the power flow distribution of each branch; simultaneously calculating the voltage stability coefficient of each node. The voltage stability coefficient is calculated using sensitivity analysis to determine the sensitivity of the node voltage to the injected power, i.e., the voltage change caused by the change in injected power. The reciprocal of the sensitivity value is used as the voltage stability coefficient. The smaller the coefficient, the more sensitive the node voltage is to power changes, and the weaker the voltage stability. Critical power transmission paths are identified based on the power flow distribution, and the top 20% of branches in terms of active power transmission are selected. The path formed by these branches connected in series from the power source side to the load side is the critical power transmission path, and all nodes on this path are extracted. Vulnerable nodes are identified based on the voltage stability coefficient, and nodes with a voltage stability coefficient less than a preset threshold, such as 0.8, are selected. Finally, the nodes on the critical power transmission path and the weak voltage stability nodes are merged, and duplicate nodes are removed by comparing node numbers to determine a set of monitoring nodes for strategy feedback.

[0039] Step 305: Collect voltage, current, and power operating data of the monitoring nodes. Based on the geographical location and electrical connection of the monitoring nodes, construct a power grid operation status assessment area. Specifically, this includes: first, determining the monitoring data acquisition scheme, setting the data acquisition cycle to 5 minutes, and collecting real-time voltage data, real-time current data, and real-time power data of each node in the monitoring node set. Real-time voltage data refers to the effective value of line voltage, real-time current data refers to the effective value of phase current, and real-time power data refers to total active power and total reactive power. The data is uploaded to the data processing center in real time through IoT acquisition devices. Subsequently, based on the monitoring nodes... The assessment area is constructed based on geographical location and electrical connection relationships. First, the actual geographical coordinates of each monitoring node are obtained, such as latitude and longitude. Nodes with adjacent geographical locations (e.g., a distance of less than 5 kilometers) are grouped together. At the same time, the geographical location grouping is adjusted in conjunction with electrical connection relationships, such as whether the nodes belong to the same substation power supply range or whether they are directly connected by lines, to ensure that the nodes in the same group have direct or indirect electrical connections. Each adjusted group of nodes and the power grid equipment it covers, such as lines and loads, are jointly delineated as a power grid operation status assessment area to ensure that the assessment area can reflect the overall operation status of a certain local area of ​​the power grid.

[0040] Step 306: Based on the monitoring node data in the power grid operation status assessment area, calculate the electrical distance between each monitoring node to form an electrical distance matrix; based on the electrical distance matrix, aggregate monitoring nodes with similar electrical distances into the same cluster; according to the cluster division result, divide the assessment area into multiple sub-regions with relatively uniform electrical characteristics, specifically including: firstly, calculating the electrical distance between each monitoring node, using the equivalent impedance value between two nodes as the quantitative index of electrical distance, obtaining the resistance and reactance of the line between two nodes through the power grid parameter manual, calculating the modulus of the equivalent impedance, that is, the square root of the sum of the squares of the resistance and the squares of the reactance, which is the electrical distance between two nodes; if there is no direct line connection between two nodes, calculate the total equivalent impedance through the shortest path between them, as the electrical distance between the two nodes. The shortest path is the path with the fewest nodes. The electrical distances between all monitoring nodes are arranged in order of node number to form an electrical distance matrix, where rows and columns are node numbers and elements represent the electrical distance between corresponding nodes. Clustering is performed based on this electrical distance matrix using the K-means clustering algorithm. The number of clusters is preset, such as 3 to 5 depending on the size of the evaluation area. Electrical distance is used as the cluster similarity index, aggregating monitoring nodes with similar electrical distances into the same cluster. Similar electrical distances mean the minimum average distance between nodes within the same cluster. Based on the cluster division results, each cluster and its corresponding power grid equipment (lines, loads) are divided into a sub-region to ensure relatively uniform electrical characteristics within each sub-region, such as impedance and power transmission characteristics.

[0041] Step 307: Based on the sub-region division results, obtain the real-time operating data of all monitoring nodes in each sub-region; according to the real-time operating data, calculate the voltage deviation rate by the percentage deviation between the voltage of each node and the rated voltage, calculate the line load rate by the ratio of the actual power to the rated capacity of each line, and calculate the power factor by the ratio of the active power to the apparent power of each node. Specifically, this includes: first, obtaining the real-time operating data of all monitoring nodes in each sub-region; retrieving the voltage, current, and power data of each monitoring node in the current acquisition cycle from the data processing center; calculating the voltage deviation rate of each node by subtracting the rated voltage value of the power grid from the real-time voltage value of the node, such as 10kV, 3kV, 4kV, 5kV, 6 ... For 5kV, the voltage deviation is obtained, then the voltage deviation is divided by the rated voltage and multiplied by 100% to obtain the voltage deviation rate. The line load rate of each line is calculated to obtain the rated capacity of the line, which is determined according to the line type. For example, the rated capacity of a 10kV line is 2MVA. The line load rate is obtained by dividing the real-time actual power (total active power) of the line by the rated capacity and multiplying by 100%. The power factor of each node is calculated by dividing the real-time total active power of the node by the real-time apparent power. Apparent power is the product of the effective voltage value and the effective current value. Through the above calculations, three key electrical parameters of each monitoring node are obtained, providing data support for subsequent sub-regional characteristic analysis.

[0042] Step 308 involves statistically analyzing the voltage deviation rate, line load rate, and power factor of all nodes within the same sub-region, calculating their maximum, minimum, and average values ​​to form the electrical characteristic parameters of that sub-region. Specifically, this includes: for each sub-region, collecting voltage deviation rate data from all monitoring nodes within that region; iterating through all data to find the maximum and minimum values, where the maximum value is the highest voltage deviation rate within the sub-region, and the minimum value is the lowest voltage deviation rate within the sub-region; summing all voltage deviation rate data and dividing by the number of nodes to obtain the average voltage deviation rate; and using the same method to analyze all nodes within the sub-region... The line load rate data is statistically analyzed to calculate the maximum, minimum, and average values ​​of the line load rate. Simultaneously, the power factor data of all nodes within the sub-region is statistically analyzed to calculate the maximum, minimum, and average values ​​of the power factor. The maximum, minimum, and average voltage deviation rates, the maximum, minimum, and average line load rates, the maximum, minimum, and average power factors of each sub-region are compiled into a set of parameters. This set of parameters represents the electrical characteristic parameters of the sub-region, comprehensively reflecting its electrical operating status.

[0043] Step 309: Based on the electrical characteristic parameters, extract the voltage deviation rate data of each sub-region and calculate its average value as a voltage stability index; simultaneously, extract the line load rate data of each sub-region and calculate its standard deviation as a load balance index. Specifically, this includes: first, extracting the voltage deviation rate data of each sub-region, obtaining the average voltage deviation rate from the electrical characteristic parameters of the sub-region, and using this average value directly as the voltage stability index of the sub-region; the closer the average value is to 0, the more stable the voltage of the sub-region. Second, extracting the line load rate data of each sub-region, obtaining the line load rate values ​​of all lines from the electrical characteristic parameters of the sub-region, first calculating the average of these values, i.e., the average line load rate, then calculating the deviation of each line load rate value from the average value, summing the squares of all deviations and dividing by the number of lines to obtain the average of the sum of squares of deviations, taking the arithmetic square root of this average value to obtain the standard deviation of the line load rate, and using this standard deviation as the load balance index of the sub-region; the smaller the standard deviation, the more balanced the load distribution of each line in the sub-region. Through the above operations, the multi-dimensional electrical characteristic parameters of the sub-region are transformed into two core evaluation indicators, providing a key basis for subsequent correction coefficient calculations.

[0044] Step 310: Multiply the voltage stability index by the corresponding weighting coefficient to obtain the weighted voltage stability index; multiply the load balancing index by the corresponding weighting coefficient to obtain the weighted load balancing index; add the weighted voltage stability index and the weighted load balancing index to obtain a weighted sum; normalize the weighted sum to obtain a correction coefficient, specifically including: first, determining the weighting coefficients based on the priority requirements of power grid dispatch, such as prioritizing voltage stability during peak periods and prioritizing load balancing during off-peak periods, assigning weights to the voltage stability index and the load balancing index, for example, setting the weight of the voltage stability index to 0.6 and the weight of the load balancing index to 0.4; calculating the weighted voltage stability index by multiplying the voltage stability index of the sub-region by the corresponding weighting coefficient, i.e., 0. 0.6, obtain the weighted voltage stability index; simultaneously calculate the weighted load balance index, multiply the load balance index of the sub-region by the corresponding weight coefficient, i.e., 0.4, to obtain the weighted load balance index; add the weighted voltage stability index and the weighted load balance index to obtain the weighted sum; normalize the weighted sum, with the preset correction coefficient range of 0.8 to 1.2, first calculate the maximum and minimum values ​​of the weighted sums of all sub-regions, subtract the minimum value from the weighted sum of a certain sub-region to obtain the difference, then divide the difference by (maximum value - minimum value) to obtain the normalization ratio, multiply the normalization ratio by 1.2 - 0.8 = 0.4 and add 0.8 to obtain the correction coefficient of that sub-region, ensuring that the correction coefficient is within a reasonable adjustment range and avoiding over-correction.

[0045] Step 311: Based on the correction coefficient, perform proportional scaling on the power generation plan, energy storage charging and discharging plan, and load control command in the initial scheduling strategy, including: multiplying the unit output value in the power generation plan by the correction coefficient to obtain the scaled power generation plan; multiplying the power value in the energy storage charging and discharging plan by the correction coefficient to obtain the scaled energy storage charging and discharging plan; multiplying the power setpoint in the load control command by the correction coefficient to obtain the scaled load control command; and integrating the scaled power generation plan, energy storage charging and discharging plan, and load control command to obtain the corrected scheduling strategy. Specifically, this includes: first, scaling the power generation plan in the initial scheduling strategy, extracting the output value of each generator unit in the initial power generation plan, such as the output value of coal-fired unit A in period 9 being 150MW, and the output value of photovoltaic unit B in period 12 being 80MW, multiplying each output value by the correction coefficient of the corresponding sub-region, such as a sub-region correction coefficient of 1.05, to obtain the scaled output value, that is, the output value of coal-fired unit A becomes 150 × 1.05 = 157.5MW, and the output value of photovoltaic unit B becomes 80MW. The output value of generator unit B becomes 80 × 1.05 = 84MW. All scaled output values ​​are integrated to form a scaled power generation plan. Next, the energy storage charging and discharging plan is scaled proportionally. The charging and discharging power values ​​of each energy storage device in the initial plan are extracted. For example, the discharging power of energy storage C in time period 14 is 50MW. Each power value is multiplied by a correction factor to obtain the scaled charging and discharging power value: 50 × 1.05 = 52.5MW, forming the scaled energy storage charging and discharging plan. Simultaneously, the load control command is scaled proportionally. The power setpoints of adjustable loads in the initial command are extracted. For example, the power reduction setpoint for industrial load D is 20MW. Each setpoint is multiplied by a correction factor to obtain the scaled power setpoint: 20 × 1.05 = 21MW, forming the scaled load control command. Finally, the scaled power generation plan, energy storage charging and discharging plan, and load control command are collaboratively verified to ensure that the total power generation matches the total load and energy storage adjustment, i.e., total power generation = total load + energy storage charging - energy storage discharging. After verification, these are integrated into a corrected dispatch strategy.

[0046] In this embodiment of the invention, the indicators of grid operation stability, power generation cost economy, and new energy absorption rate are quantified into multi-objective optimization constraints, so that the optimization objectives have specific data basis and the constraints are clear and explicit, laying a data foundation for subsequent optimization calculations. A linear weighted method is used to transform the multi-objective constraints into a single-objective problem, simplifying the optimization complexity. By generating an initial population and calculating fitness to select high-quality individuals, high-quality solutions are focused on, improving the efficiency of subsequent optimization. Single-point crossover and uniform mutation are used to enrich the diversity of the new population, and combined with iterative convergence judgment, continuous optimization of the solution is ensured. Finally, the initial scheduling strategy with the highest fitness is selected to ensure the optimality of the strategy. Based on power flow direction and voltage stability coefficient data, key transmission paths and voltage weak nodes are accurately identified, and the set of monitoring nodes is determined by merging and deduplication, providing a highly targeted data source for subsequent feedback.

[0047] The system collects operational data from monitoring nodes and constructs an assessment area by combining geographical location and electrical connection relationships. This ensures comprehensive data coverage and adherence to the actual power grid structure, guaranteeing a reliable assessment foundation. An electrical distance matrix is ​​calculated, and nearby nodes are aggregated to divide the area into sub-regions. This ensures relatively uniform electrical characteristics within the sub-regions, facilitating accurate extraction and analysis of sub-region electrical parameters. Voltage deviation rate, line load rate, and power factor are calculated to quantify the electrical status of the sub-regions, providing specific and referable data support for sub-region characteristic analysis. The maximum and average values ​​of electrical parameters within each sub-region are statistically analyzed to form characteristic parameters, comprehensively reflecting the overall electrical condition of the sub-region and avoiding the limitations of single data sources.

[0048] Voltage stability and load balance indicators are extracted by means and standard deviation, respectively, focusing on the core electrical characteristics of sub-regions to provide key data for the calculation of correction coefficients. The weighted indicators are calculated by combining weights and normalized to obtain the correction coefficients, so that the correction coefficients integrate the needs of different indicators and ensure that the correction direction is adapted to the actual dispatching needs. The execution ratios of power generation, energy storage, and load plans are scaled and integrated, and the strategy is adapted to the characteristics of sub-regions through data adjustment, thereby improving the pertinence and practical applicability of the dispatching strategy.

[0049] In a preferred embodiment of the present invention, step 400 above involves parsing the modified scheduling strategy into specific equipment control commands and sending the equipment control commands to the corresponding smart terminals. Based on the smart terminals, automated actuators are used to control the access and disconnection of distributed energy resources, adjust the charging and discharging status of energy storage devices, and adjust load management strategies, thereby achieving real-time scheduling and control of power grid resources, including: Step 401: Parse the revised scheduling strategy into specific equipment control instructions, including converting the power generation plan into generator start / stop instructions and output adjustment instructions, converting the energy storage charging / discharging plan into energy storage system charging / discharging power instructions, and converting the load control instructions into load switching instructions and power adjustment instructions. Specifically, this includes: First, for the conversion of the power generation plan, traverse each record in the power generation plan module of the revised scheduling strategy, extracting the unique generator unit number, planned execution period, and planned output value within that period for each record. The unique generator unit number is, for example, #1 coal-fired unit and #2 photovoltaic unit; the planned execution period is, for example, 08:00 to 12:00 on the same day. :00; Next, compare the current operating status of the generator set, such as shutdown, standby, or operation. If the planned output value is greater than the minimum starting output of the unit and the current status is shutdown, where the minimum starting output of the unit is 50MW for a coal-fired unit, then a unit start / stop command is generated. The command content includes the unit number, start time, and initial output value after start. If there is a difference between the planned output value and the current actual output value of the unit, such as the current output being 100MW and the planned output being 120MW, then the output adjustment amount is calculated, which is the planned output value minus the current actual output value. An output adjustment command is generated, and the command content includes the unit number, adjustment period, adjustment amount, and target output value after adjustment.

[0050] Secondly, regarding the conversion of energy storage charging and discharging plans, the system extracts the device number, planned charging and discharging time period, and planned charging and discharging power value for each energy storage device within the energy storage plan module of the scheduling strategy. For example, the device number is #1 (energy storage converter) and #2 (battery energy storage system). The planned charging and discharging time period is from 12:00 to 15:00 on the same day, and the planned charging and discharging power value during this period is negative for charging and positive for discharging. Based on the rated charging and discharging power of the energy storage device (e.g., #1 energy storage device has a rated discharge power of 20MW), the system checks whether the planned power value is within the rated range. If it is within the range, a charging and discharging power command for the energy storage system is directly generated. The command includes the energy storage device number, charging and discharging time period, target power value, and power adjustment step size. If the planned power value exceeds the rated range, the commanded power value is corrected to the rated power value. For example, if the planned discharge is 25MW and the rated power is 20MW, it is corrected to 20MW, and the command is then generated again.

[0051] Finally, regarding the conversion of load control commands, the adjustable load numbers, planned switching periods, and planned power adjustment values ​​are extracted from the load planning module of the scheduling strategy. For example, the adjustable load numbers are like #1 industrial load and #2 commercial building load, and the planned switching period is like 18:00 to 20:00 on the current day. If the planned power adjustment value is 0, meaning the current load operation is maintained, no switching command needs to be generated. If the planned power adjustment value is greater than the current load power (e.g., current 10MW, planned 15MW), a load switching command is generated, containing the load number, switching time, number of load branches to be connected, and power per branch. If the planned power adjustment value is less than the current load power (e.g., current 10MW, planned 8MW), the load reduction is calculated (current power minus planned power), and a power adjustment command is generated, containing the load number, adjustment period, reduction amount, and target power value after reduction. Through the above classification and conversion, specific control commands covering power generation, energy storage, and load are formed.

[0052] Step 402: Based on the power grid communication protocol, the device control command is encapsulated into a standardized communication message and sent to the corresponding smart terminal through the power dedicated communication network. Specifically, this includes: first, determining the power grid communication protocol type, using the IEC61850 standard protocol commonly used in the power industry. This protocol is suitable for data interaction between smart grid devices. According to the message structure specified in this protocol, the device control command generated in step 401 is encapsulated. The message structure consists of three parts: the first part is the message header, occupying 8 bytes, containing the command sending timestamp, the sending device number, and the command type identifier. The command sending timestamp is accurate to milliseconds. The first part is the terminal device number, such as the dispatch center server number; the second part is the instruction type identifier, such as 01 representing a power generation instruction and 02 representing an energy storage instruction. The third part is the control parameter area, the length of which is dynamically adjusted according to the instruction content. The key parameters in the device control instruction are converted into binary data according to the encoding format specified in the protocol and filled into the parameter area in sequence. Key parameters include unit number, time period, and power value, and the encoding format is ASCII code. The fourth part is the check code, which occupies 4 bytes. The cyclic redundancy check (CRC) algorithm is used to calculate the CRC value between the message header and the binary data in the control parameter area and fill it into the check code area to complete the encapsulation of the standardized communication message.

[0053] Next, the command issuance operation is executed. First, a communication link is established between the dispatch center and the intelligent terminal, which is achieved through a dedicated power communication network, such as a fiber optic private network from a 220kV substation to the dispatch center or a 230MHz wireless private network on the distribution network side. Based on the device number in the device control command, a preset device-terminal address mapping table is queried. This table stores the IP address or physical address of the intelligent terminal corresponding to each device, determining the target intelligent terminal address to which each command needs to be issued. The encapsulated standardized communication messages are classified according to the target address and issued one by one to the corresponding intelligent terminals through the communication link. At the same time, the issuance time, target terminal address, and message identifier of each command are recorded for subsequent command transmission status traceability.

[0054] Step 403: The smart terminal receives and parses the standardized communication message, extracts the control parameters therein, and obtains the specific execution signal. Specifically, this includes: First, after receiving the standardized communication message, the smart terminal performs a message integrity check, extracts the CRC checksum at the end of the message, and recalculates the CRC value for the binary data in the message header and control parameter area. The two calculated CRC values ​​are then compared. If the comparison matches, the message is determined to be undamaged, and the next step of parsing is initiated. If the comparison does not match, a message retransmission request is sent to the dispatch center, including the identifier of the message that failed the check, until a complete message is received and verified.

[0055] Subsequently, the data is disassembled according to the message structure specified in the IEC 61850 protocol. First, the instruction sending timestamp, sender number, and instruction type identifier are extracted from the message header to confirm whether the instruction source and type match the range of equipment managed by the terminal. For example, the energy storage terminal only processes type 02 instructions. Then, binary data is extracted from the control parameter area and converted back to ASCII code according to the protocol encoding format to reconstruct the key parameters of the equipment control instruction, such as unit number #1, adjustment period from 08:00 to 12:00, and target output of 120MW. Based on the instruction type identifier... The system identifies and organizes the restored parameters into execution signals according to a preset format. For example, the execution signal for the power generation output adjustment command is: Equipment number: #1 unit; Execution period: 08:00 to 12:00; Adjustment direction: capacity increase; Adjustment amount: 20MW; Target output: 120MW. The execution signal for the energy storage charging and discharging command is: Equipment number: #1 energy storage; Execution period: 12:00 to 15:00; Operating mode: discharging; Target power: 20MW. This ensures that the parameters of the execution signals are clear and the format is uniform, so that they can directly drive the automatic actuators.

[0056] Step 404: Based on the execution signal, the grid connection switch status of the distributed energy source is controlled by the automated actuator to realize access and exit, the power conversion unit of the energy storage system is adjusted to realize charging and discharging status control, and the load control switch and regulating device are operated to realize the adjustment of the load management strategy, so as to obtain the action execution result of the automated actuator. Specifically, it includes: First, for the control of the access and exit of the distributed energy source, the automated actuator (such as a grid-connected circuit breaker or disconnecting switch) receives the grid connection switch status instruction in the execution signal, such as closing the #1 photovoltaic unit grid-connected circuit breaker at 08:00. The built-in control module reads the action time and switch status in the signal, where the switch status is closed or open. When the specified action time is reached, the control module outputs a drive signal, such as a 220V DC drive voltage, to drive the operating mechanism of the circuit breaker to act. At the same time, the current value at both ends of the circuit breaker is collected by the current sensor. If the current value changes from 0 to the preset grid-connected current, such as the rated grid-connected current of the #1 photovoltaic unit of 100A, it is determined that the distributed energy source has been successfully connected; if the current value changes from the rated value to 0, it is determined that the exit has been successfully completed. The action completion time and actual status are recorded to form the action execution result.

[0057] Secondly, regarding the adjustment of the charging and discharging state of energy storage devices, the automated actuator (such as the power conversion system PCS) receives the charging and discharging power command in the execution signal. For example, if the #1 energy storage discharges 20MW from 12:00 to 15:00, the PCS control module first converts the target power value into the corresponding DC-side voltage and current command. Based on the power conversion efficiency of the PCS, such as 95%, the DC-side power needs to be set to 20 / 0.95≈21.05MW. By adjusting the switching frequency of the IGBT (Insulated Gate Bipolar Transistor), the power flow direction and magnitude on the DC and AC sides are controlled. At the same time, the output power of the PCS is collected in real time by the power sensor. If the actual output power is stable within ±2% of the target power value, such as 19.6 to 20.4MW, the charging and discharging state adjustment is determined to be complete. The average power and fluctuation range during the adjustment period are recorded to form the action execution result.

[0058] Finally, regarding the adjustment of the load management strategy, automated actuators (such as load control switches and variable frequency drives) receive load switching / power adjustment commands in the execution signals, such as reducing the #1 industrial load by 2MW at 18:00. If it is a switching command, the control switch drive module outputs a signal to close or open the specified load branch, such as opening two 1MW load branches. The total load power is collected through the power meter to confirm that the power change is consistent with the command. If it is a power adjustment command, the variable frequency drive adjusts the output frequency according to the target power value in the command, such as reducing the frequency from 50Hz to 48Hz. The actual load power is collected in real time until it stabilizes near the target value. The power values ​​before and after the adjustment and the adjustment time are recorded to form the action execution result. The action execution results of the three types of equipment are stored in association with the equipment number and command identifier to provide a data foundation for subsequent closed-loop verification.

[0059] Step 405: Based on the execution results of the automated actuator, the generated status feedback signal is collected in real time. The status feedback signal is compared and verified with the equipment control command in a closed loop to confirm the accurate execution of the power grid resource scheduling control and complete the real-time scheduling and control of power grid resources. Specifically, this includes: First, collecting status feedback signals. After completing the action, the automated actuator collects the actual operating status parameters of the equipment in real time through built-in status monitoring sensors, such as position sensors, power sensors, and voltage sensors. For example, the actual position (closed / open) and grid current of the distributed energy grid-connected switch, the actual output power and operating mode (charging / discharging) of the energy storage PCS, and the actual power of the load and the status of the branch switch are collected. These actual parameters are organized into a status feedback signal according to a preset format. The preset format includes the equipment number, command identifier, feedback time, and actual parameter value, and is transmitted back to the dispatch center through the power dedicated communication network.

[0060] Next, a closed-loop comparison verification is performed. After receiving the status feedback signal, the dispatch center retrieves the corresponding equipment control command based on the command identifier in the signal, extracts the target parameters from the command, such as target grid connection status closed, target output 120MW, and target discharge power 20MW; the actual parameters in the feedback signal are compared with the command target parameters one by one, for example, whether the actual position of the grid connection switch is consistent with the command target status, and whether the deviation between the actual output value and the target output value is within the allowable range, such as allowable deviation of ±2% for power generation output, ±3% for energy storage power, and ±5% for load power; if all parameter comparisons meet the requirements, the command is determined to have been executed accurately, and the dispatch center... The log records the execution success and verification time. If there is a parameter deviation exceeding the allowable range, such as an actual output of 115MW and a target of 120MW with a deviation of 4.2%, it is determined that there is an execution deviation. An instruction retry signal is immediately sent to the corresponding intelligent terminal, requiring the automated actuator to re-execute the action and collect feedback signals for comparison again. If the requirements are still not met after three consecutive retries, an alarm mechanism is triggered, such as sending an alarm message to the dispatcher, while recording the deviation parameters and the number of retries. After all instructions have been compared and verified or the execution deviation has been resolved, the overall accuracy of the power grid resource scheduling and control is confirmed, and the real-time scheduling and control process of the power grid resources is completed.

[0061] In this embodiment of the invention, the modified scheduling strategy is parsed into specific equipment control commands such as generator start-up and shutdown, output regulation, energy storage charging and discharging power, load switching, and power regulation. This transforms the macro-scheduling strategy into operational commands that equipment can directly execute, clarifying the specific operational requirements of each device and providing clear data for subsequent precise equipment execution. Based on the power grid communication protocol, the equipment control commands are encapsulated into standardized communication messages to ensure that the command transmission format conforms to power grid communication standards, avoiding errors caused by format differences. Commands are issued through a dedicated power communication network, improving the stability and timeliness of command transmission and ensuring efficient delivery of control commands to the intelligent terminal. After receiving the standardized communication messages, the intelligent terminal parses them, accurately extracts the control parameters, and converts them into specific execution signals. The system converts message data into operational signals that can be recognized by the terminal, ensuring that the intelligent terminal can accurately understand the core requirements of the control instructions and provide correct data guidance for the actions of the actuators. Based on the execution signals, the system drives the automated actuators to control the distributed energy grid connection switches, energy storage power conversion units, and load control devices, respectively. The system converts the execution signals into actual equipment operation actions, realizing precise control over the access and disconnection of distributed energy, energy storage charging and discharging, and load management, and completing the implementation of instructions into actual dispatch actions. The system collects the status feedback signals of the automated actuators in real time and performs closed-loop comparison and verification with the issued equipment control instructions. Through data comparison, it confirms whether the equipment actions are consistent with the instruction requirements, promptly detects and corrects possible execution deviations, and ensures the accuracy and effectiveness of power grid resource dispatch control.

[0062] like Figure 2 As shown, embodiments of the present invention also provide a smart scheduling system based on efficient coordination of source, grid, load, and storage, comprising: The data acquisition module is used to collect operational data from each node in the power grid in real time through IoT devices, perform integrity verification and preliminary evaluation on the operational data, and form an initial operational dataset. The data analysis module is used to clean and integrate the initial operating dataset to form a unified operating dataset; based on the operating dataset, feature extraction and state analysis are performed to obtain multi-dimensional feature vectors; based on the multi-dimensional feature vectors and artificial intelligence technology, a source-grid-load-storage coordinated energy management model is constructed; based on this model, the power grid operating status and load demand prediction results are obtained. The strategy generation module is used to generate an initial scheduling strategy based on the power grid operating status and load demand forecast results, combined with preset optimization objectives; based on the initial scheduling strategy, a set of monitoring nodes for strategy feedback is determined in the key areas of the power grid; the operating data of the monitoring nodes is collected to construct a power grid operating status assessment area; the assessment area is divided into multiple sub-regions, and the electrical characteristic parameters of each sub-region are extracted; the corresponding correction coefficients are calculated based on the electrical characteristic parameters; the initial scheduling strategy is corrected using the correction coefficients to obtain the corrected scheduling strategy. The execution control module is used to parse the modified scheduling strategy into specific equipment control commands and send the equipment control commands to the corresponding smart terminals. Based on the smart terminals, the module controls the access and disconnection of distributed energy sources, adjusts the charging and discharging status of energy storage devices, and adjusts the load management strategy through automated actuators, thereby realizing real-time scheduling and control of power grid resources.

[0063] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0064] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0065] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0066] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A smart scheduling method based on efficient source-grid-load-storage coordination, characterized in that, The method includes: Step 100: Real-time collection of operational data from each node in the power grid via IoT devices; integrity verification and preliminary evaluation of the operational data to form an initial operational dataset. Step 200: Clean and integrate the initial operating dataset to form a unified operating dataset; perform feature extraction and state analysis based on the operating dataset to obtain multi-dimensional feature vectors; construct a source-grid-load-storage coordinated energy management model based on the multi-dimensional feature vectors and artificial intelligence technology; and obtain the power grid operating status and load demand prediction results based on the model. Step 300: Based on the power grid operating status and load demand forecast results, and combined with preset optimization objectives, an initial dispatch strategy is generated; based on the initial dispatch strategy, a set of monitoring nodes for strategy feedback is determined in the key areas of the power grid; the operating data of the monitoring nodes is collected to construct a power grid operating status assessment area; the assessment area is divided into multiple sub-areas, and the electrical characteristic parameters of each sub-area are extracted; based on the electrical characteristic parameters, the corresponding correction coefficients are calculated; the initial dispatch strategy is corrected using the correction coefficients to obtain the corrected dispatch strategy. Step 400: The revised scheduling strategy is parsed into specific equipment control commands, and the equipment control commands are sent to the corresponding smart terminals. Based on the smart terminals, the access and exit of distributed energy sources, the charging and discharging status of energy storage devices, and the load management strategy are controlled by automated actuators, thereby realizing real-time scheduling and control of power grid resources.

2. The intelligent scheduling method based on efficient source-grid-load-storage coordination according to claim 1, characterized in that, Step 100: Real-time collection of operational data from each node within the power grid via IoT devices; integrity verification and preliminary assessment of the operational data to form an initial operational dataset, including: Perform data integrity verification on the operational data collected in real time through IoT devices and calculate its data integrity rate; if the data integrity rate is higher than the preset threshold, it is determined to be valid data; otherwise, a data supplementation mechanism is triggered. Based on the effective data, the fluctuation characteristics of current and voltage are extracted; the fluctuation rate index is calculated based on the fluctuation characteristics; the fluctuation rate index is correlated with the load demand data to calculate the real-time power balance of each node. Based on the real-time power balance, a preliminary assessment of the power grid's operating status is performed to obtain status assessment parameters; power imbalance nodes are identified according to the status assessment parameters to obtain a power imbalance node list; the power imbalance node list is integrated with the operating data to form an initial operating dataset.

3. The intelligent scheduling method based on efficient source-grid-load-storage coordination according to claim 2, characterized in that, Step 200: Clean and integrate the initial running dataset to form a unified running dataset; perform feature extraction and state analysis based on the running dataset to obtain a multi-dimensional feature vector; Based on multidimensional feature vectors and artificial intelligence technology, a source-grid-load-storage collaborative energy management model is constructed. Based on this model, the power grid operating status and load demand forecast results are obtained, including: The initial running dataset is cleaned by using the interquartile range method to identify and remove outlier data, and missing values ​​are filled in by linear interpolation to form a unified running dataset. Based on the unified operational dataset, the time-series variation features of the load data are extracted to obtain a load time-series feature set; simultaneously, the fluctuation features of distributed energy generation power are calculated to obtain a power fluctuation feature set; the load time-series feature set and the power fluctuation feature set are fused to construct a multi-dimensional feature vector. The multidimensional feature vector is input into a prediction model built on artificial intelligence technology. Through the forward propagation calculation of the prediction model, a quantitative assessment result of the power grid operation status including stability indicators and a predicted load demand value for future periods are obtained. Based on the power grid operation status assessment results and load demand forecasts, energy balance relationships among power sources, grids, loads, and storage are established, resulting in collaborative scheduling rules with multi-objective constraints. Based on these collaborative scheduling rules, a collaborative energy management model for power sources, grids, loads, and storage is constructed. Through this collaborative energy management model, the power grid operation status and load demand forecasts are obtained.

4. The intelligent scheduling method based on efficient source-grid-load-storage coordination according to claim 3, characterized in that, Step 300: Based on the power grid operation status and load demand forecast results, and combined with the preset optimization objectives, generate an initial scheduling strategy; Based on the initial scheduling strategy, a set of monitoring nodes for strategy feedback is identified in the critical areas of the power grid, including: Based on the power grid operation status and load demand forecast results, the power grid operation stability index, power generation cost economy index and new energy absorption rate index are quantified into multi-objective optimization constraints. The multi-objective constraints are transformed into a single-objective optimization problem using a linear weighting method, and an initial population containing multiple random scheduling schemes is generated. Based on the initial population, the fitness value of each scheduling scheme is calculated, and a selection mechanism is used to select high-quality individuals. Single-point crossover and uniform mutation operations are sequentially performed on the high-quality individuals to generate a new population. Based on the new population, it is determined whether the maximum number of iterations or the fitness convergence condition has been reached. If the condition is met, the individual with the highest fitness in the current population is selected as the initial scheduling strategy. If the condition is not met, the selection operation based on fitness is returned to continue the iterative optimization process, and finally the initial scheduling strategy is obtained. Based on the initial scheduling strategy, the power flow distribution of each branch of the power grid is calculated, and the voltage stability coefficient of each node is calculated. Critical power transmission paths are identified based on the power flow distribution, and weak voltage stability nodes are identified based on the voltage stability coefficient. The nodes on the critical power transmission paths and the weak voltage stability nodes are merged and deduplicated to determine a set of monitoring nodes for strategy feedback.

5. The intelligent scheduling method based on efficient source-grid-load-storage coordination according to claim 4, characterized in that, Operational data from monitoring nodes are collected to construct a power grid operation status assessment area; this assessment area is divided into multiple sub-regions, and the electrical characteristic parameters of each sub-region are extracted, including: The voltage, current, and power operation data of the monitoring nodes are collected, and a power grid operation status assessment area is constructed based on the geographical location and electrical connection relationship of the monitoring nodes. Based on the monitoring node data in the power grid operation status assessment area, the electrical distance between each monitoring node is calculated to form an electrical distance matrix; based on the electrical distance matrix, monitoring nodes with similar electrical distances are aggregated into the same cluster; according to the cluster division results, the assessment area is divided into multiple sub-regions with relatively uniform electrical characteristics. Based on the sub-region division results, real-time operating data of all monitoring nodes in each sub-region are obtained; based on the real-time operating data, the voltage deviation rate is obtained by calculating the percentage deviation between the voltage of each node and the rated voltage, the line load rate is obtained by calculating the ratio of the actual power of each line to the rated capacity, and the power factor is obtained by calculating the ratio of the active power to the apparent power of each node. The voltage deviation rate, line load rate and power factor of all nodes in the same sub-region are statistically analyzed, and their maximum, minimum and average values ​​are calculated to form the electrical characteristic parameters of the sub-region.

6. The intelligent scheduling method based on efficient source-grid-load-storage coordination according to claim 5, characterized in that, Calculate the corresponding correction factor based on the electrical characteristic parameters; The initial scheduling policy is modified using correction factors to obtain the modified scheduling policy, including: Based on the electrical characteristic parameters, voltage deviation rate data for each sub-region is extracted, and its average value is calculated as a voltage stability index; at the same time, line load rate data for each sub-region is extracted, and its standard deviation is calculated as a load balance index. Multiply the voltage stability index by the corresponding weighting coefficient to obtain the weighted voltage stability index; multiply the load balance index by the corresponding weighting coefficient to obtain the weighted load balance index; add the weighted voltage stability index and the weighted load balance index to obtain the weighted sum; normalize the weighted sum to obtain the correction coefficient. Based on the correction coefficient, the generation plan, energy storage charging and discharging plan, and load control command in the initial scheduling strategy are scaled proportionally, including: multiplying the unit output value in the generation plan by the correction coefficient to obtain the scaled generation plan; multiplying the power value in the energy storage charging and discharging plan by the correction coefficient to obtain the scaled energy storage charging and discharging plan; multiplying the power setpoint in the load control command by the correction coefficient to obtain the scaled load control command; and integrating the scaled generation plan, energy storage charging and discharging plan, and load control command to obtain the corrected scheduling strategy.

7. The intelligent scheduling method based on efficient source-grid-load-storage coordination according to claim 6, characterized in that, Step 400: The revised scheduling strategy is parsed into specific equipment control commands, and these commands are sent to the corresponding smart terminals. Based on the smart terminals, automated actuators control the access and disconnection of distributed energy resources, adjust the charging and discharging status of energy storage devices, and adjust load management strategies, thereby achieving real-time scheduling and control of grid resources, including: The revised scheduling strategy is parsed into specific equipment control instructions, including converting the power generation plan into generator start-stop instructions and output adjustment instructions, converting the energy storage charging and discharging plan into energy storage system charging and discharging power instructions, and converting the load control instructions into load switching instructions and power adjustment instructions. Based on the power grid communication protocol, the equipment control commands are encapsulated into standardized communication messages and sent to the corresponding smart terminals through the power dedicated communication network. The intelligent terminal receives and parses the standardized communication message, extracts the control parameters therein, and obtains the specific execution signal; Based on the execution signal, the grid connection switch status of the distributed energy is controlled by the automated actuator to realize access and disconnection, the power conversion unit of the energy storage system is adjusted to realize charging and discharging status control, and the load control switch and regulating device are operated to realize the adjustment of the load management strategy, so as to obtain the action execution result of the automated actuator. Based on the execution results of the automated actuators, the generated status feedback signals are collected in real time. The status feedback signals are compared and verified with the equipment control commands in a closed loop to confirm the accurate execution of power grid resource scheduling and control, and to complete the real-time scheduling and control of power grid resources.

8. A smart scheduling system based on efficient coordination of source, grid, load, and storage, wherein the system implements the method as described in any one of claims 1 to 7, characterized in that, include: The data acquisition module is used to collect operational data from each node in the power grid in real time through IoT devices, perform integrity verification and preliminary evaluation on the operational data, and form an initial operational dataset. The data analysis module is used to clean and integrate the initial running dataset to form a unified running dataset; Based on the aforementioned running dataset, feature extraction and state analysis are performed to obtain a multidimensional feature vector; Based on multidimensional feature vectors and artificial intelligence technology, a source-grid-load-storage coordinated energy management model is constructed; based on this model, the power grid operation status and load demand prediction results are obtained. The strategy generation module is used to generate an initial scheduling strategy based on the power grid operating status and load demand forecast results, combined with preset optimization objectives. Based on the initial scheduling strategy, a set of monitoring nodes for strategy feedback is determined in the key areas of the power grid; The operation data of the monitoring nodes are collected to construct the power grid operation status assessment area; the assessment area is divided into multiple sub-areas, and the electrical characteristic parameters of each sub-area are extracted; the corresponding correction coefficients are calculated based on the electrical characteristic parameters; the initial scheduling strategy is corrected using the correction coefficients to obtain the corrected scheduling strategy. The execution control module is used to parse the modified scheduling strategy into specific equipment control commands and send the equipment control commands to the corresponding smart terminals. Based on the smart terminals, the module controls the access and disconnection of distributed energy sources, adjusts the charging and discharging status of energy storage devices, and adjusts the load management strategy through automated actuators, thereby realizing real-time scheduling and control of power grid resources.

9. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 7.

Images