Multi-objective optimization scheduling method for distributed photovoltaic energy storage system based on big data

Abstract

The application discloses a distributed photovoltaic energy storage system multi-objective optimization scheduling method based on big data, and relates to the technical field of photovoltaic energy storage scheduling. The distributed photovoltaic energy storage system multi-objective optimization scheduling method based on big data comprises energy storage data acquisition, energy storage region division and multi-objective optimization scheduling. The application constructs a source-side photovoltaic predicted power curve and a load-side power consumption predicted power curve through historical system operation data and meteorological monitoring data, realizes power grid region dynamic division in combination with a power grid topological structure and node net load characteristics, improves the pertinence of subsequent energy storage scheduling, matches different energy storage solving modes according to the identification result, guarantees the optimization effect of the scheduling result in a conventional operation state, improves the scheduling response capability in a power rapid fluctuation state, and ensures the adaptability to complex operation conditions and operation stability.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic energy storage scheduling technology, and in particular to a multi-objective optimization scheduling method for distributed photovoltaic energy storage systems based on big data. Background Technology

[0002] With the widespread application of distributed photovoltaic power generation and energy storage systems in distribution networks, the uncertainty of new energy output and the dynamic changes in user-side load demand are becoming increasingly apparent, making the grid operation state more time-varying and complex. Most existing photovoltaic energy storage dispatching technologies adopt fixed area division and fixed optimization strategies, which are difficult to fully reflect the differences in operating characteristics of different areas and the dynamic changes in operating status, resulting in the need to further improve the system dispatching flexibility and resource utilization efficiency.

[0003] Therefore, there is a need for a multi-objective optimization scheduling method and system for distributed photovoltaic energy storage systems based on big data, in order to improve the adaptability of distributed photovoltaic energy storage systems to complex operating scenarios and realize the optimized allocation and coordinated scheduling of photovoltaic energy storage resources within the power grid area. Summary of the Invention

[0004] This invention aims to provide a multi-objective optimization scheduling method for distributed photovoltaic energy storage systems based on big data, which improves the local consumption capacity of photovoltaic power and the utilization efficiency of energy storage resources, and enhances the overall supply and demand coordination capability of the power grid.

[0005] A multi-objective optimization scheduling method for distributed photovoltaic energy storage systems based on big data includes the following steps: Step S1: Energy storage data acquisition; acquire historical system operation data and meteorological monitoring data of the photovoltaic energy storage system in the current power grid area; there are several electrical nodes in the current power grid area, and several photovoltaic energy storage nodes exist in the photovoltaic energy storage system; Step S2: Energy storage area division; extract load power consumption characteristics of each node based on historical system operation data; construct the source-side photovoltaic predicted power curve and load-side predicted power curve of each electrical node based on the load power consumption characteristics of each node; divide the power grid area based on historical system operation data and meteorological monitoring data, and divide the current power grid area into multiple photovoltaic power supply sub-regions; Step S3: Multi-objective optimization scheduling; Construct different photovoltaic energy storage multi-objective optimization models for different photovoltaic energy supply sub-regions; The photovoltaic energy storage multi-objective optimization model includes an energy storage objective function and an energy storage solution mode; Within a preset energy storage optimization scheduling period, for any photovoltaic energy supply sub-region, optimize and solve based on the energy storage objective function and energy storage solution mode in the photovoltaic energy storage multi-objective optimization model to obtain the corresponding regional photovoltaic energy storage scheduling scheme; Complete the scheduling of the photovoltaic energy storage system in the current power grid area according to the regional photovoltaic energy storage scheduling scheme corresponding to each photovoltaic energy supply sub-region.

[0006] As a preferred embodiment of the present invention, the specific steps for dividing the power grid area based on historical system operation data and meteorological monitoring data include: Photovoltaic irradiance temperature characteristics are extracted based on meteorological monitoring data; the predicted photovoltaic power curve on the source side and the predicted power consumption curve on the load side of the same electrical node are calculated to obtain the node time-series net load curve; Using electrical nodes as graph nodes and the physical electrical lines between electrical nodes as edges, a basic physical connectivity graph of the power grid is constructed. For any two adjacent graph nodes with edges, the inverse correlation coefficient between the time-series net load curves of the adjacent graph nodes is calculated. The inverse correlation coefficient represents the degree of complementarity of the net load change trends between adjacent electrical nodes. The load asymmetry weight of the edge is assigned according to the degree of complementarity. Based on the physical connectivity graph of the power grid with asymmetric load weights, the graph segmentation constraint is to maximize the sum of weights within sub-regions and minimize the weights of cross-sub-region edges. A graph clustering algorithm is used to dynamically segment the current power grid region to obtain multiple photovoltaic power supply sub-regions.

[0007] As a preferred embodiment of the present invention, the specific steps for constructing different photovoltaic energy storage multi-objective optimization models according to different photovoltaic energy supply sub-regions include: For any photovoltaic power supply sub-region, the total predicted photovoltaic power on the source side and the total predicted electricity consumption on the load side of all electrical nodes in the photovoltaic power supply sub-region during the preset energy storage optimization scheduling period are calculated based on the predicted photovoltaic power curve on the source side and the predicted electricity consumption curve on the load side. The photovoltaic power supply characteristic coefficient and asymmetric power fluctuation rate are calculated based on the total predicted photovoltaic power on the source side and the total predicted electricity consumption on the load side. The photovoltaic energy supply characteristic coefficients are used to match the corresponding energy storage objective function for the photovoltaic energy supply sub-regions; the objective function of the energy storage objective function includes photovoltaic absorption, tie-line stability and energy storage lifetime. The solution mode for matching the energy storage to the photovoltaic power supply sub-region is based on the asymmetric power fluctuation rate; the energy storage solution mode includes conventional solution mode and special solution mode.

[0008] As a preferred embodiment of the present invention, the specific steps for matching the corresponding energy storage objective function for the photovoltaic power supply sub-region based on the photovoltaic power supply characteristic coefficient include: A preset source-load balance benchmark value is established; among which, the photovoltaic absorption sub-item, the tie-line stability sub-item, and the energy storage lifetime sub-item correspond to the photovoltaic absorption sub-item weight, the tie-line stability sub-item weight, and the energy storage lifetime sub-item weight, respectively; Calculate the directional offset between the photovoltaic power supply characteristic coefficient and the source-load balance benchmark value; determine the weights of the photovoltaic consumption component, the tie-line stability component, and the energy storage lifetime component based on the directional offset. An energy storage objective function is constructed based on the photovoltaic absorption component, the tie-line stability component, the energy storage lifetime component, and the corresponding weights of the photovoltaic absorption component, the tie-line stability component, and the energy storage lifetime component.

[0009] As a preferred embodiment of the present invention, the specific steps for matching the energy storage solution mode corresponding to the photovoltaic power supply sub-region based on the asymmetric power fluctuation rate include: Extract the node time-series net load curves of all electrical nodes in the photovoltaic power supply sub-region and spatially superimpose and fuse them to obtain the regional time-series net power curve; within a preset sliding time window, calculate the first time derivative of the regional time-series net power curve and take the effective value of the absolute value as the asymmetric power fluctuation rate; at the same time, calculate the second time derivative of the regional time-series net power curve to obtain the asymmetric power fluctuation rate. When the asymmetric power fluctuation rate exceeds a preset first threshold and the asymmetric power fluctuation change rate exceeds a preset second threshold, a special solution mode is matched; otherwise, a regular solution mode is matched.

[0010] As a preferred embodiment of the present invention, the specific steps for optimization based on the energy storage objective function and energy storage solution mode in the photovoltaic energy storage multi-objective optimization model include: Obtain operational status data for each photovoltaic energy storage node within the photovoltaic power supply sub-region; When the energy storage solution mode is the conventional solution mode, a comprehensive optimization objective is constructed by combining the operating status data and the energy storage objective function; the comprehensive optimization objective is iteratively solved to obtain the corresponding photovoltaic energy storage scheduling scheme. When the energy storage solution mode is a special solution mode, the power mutation state is obtained based on the operating status data; the tie-line stability enhancement factor is calculated based on the power mutation state; the tie-line stability enhancement factor is used to amplify and correct the tie-line stability component weight, and the photovoltaic absorption component weight and the energy storage lifetime component weight are compressed and corrected to obtain a new energy storage objective function; a transient optimization objective is constructed based on the new energy storage objective function; the transient optimization objective is iteratively solved to obtain the corresponding photovoltaic energy storage scheduling scheme.

[0011] As a preferred embodiment of the present invention, the specific steps for scheduling the photovoltaic energy storage system in the current power grid area according to the regional photovoltaic energy storage scheduling scheme corresponding to each photovoltaic energy supply sub-region include: Based on the regional photovoltaic energy storage scheduling scheme corresponding to each photovoltaic energy supply sub-region, calculate the regional power surplus or regional power deficit of each photovoltaic energy supply sub-region within the preset energy storage optimization scheduling cycle; construct a regional coordination relationship matrix between photovoltaic energy supply sub-regions based on the regional power surplus and regional power deficit; and determine the target photovoltaic energy supply sub-region combination with power complementarity based on the regional coordination relationship matrix. For any target photovoltaic power supply sub-region combination, priority is given to using regional power surplus to coordinate regional power deficit across regions, resulting in cross-regional power coordination results; corresponding regional collaborative dispatch instructions are generated based on the cross-regional power coordination results; regional photovoltaic energy storage dispatch schemes within each photovoltaic power supply sub-region are synchronously modified based on the regional collaborative dispatch instructions, resulting in a comprehensive photovoltaic energy storage dispatch scheme; and the optimized dispatch of the photovoltaic energy storage system in the current power grid region is completed based on the comprehensive photovoltaic energy storage dispatch scheme.

[0012] The present invention has the following advantages: 1. This invention constructs source-side photovoltaic power prediction curves and load-side electricity consumption prediction curves using historical system operation data and meteorological monitoring data. It also dynamically divides the power grid area by combining the grid topology and node net load characteristics, thereby improving the targeting of subsequent energy storage dispatch. By constructing an asymmetric power fluctuation rate evaluation mechanism, the operating status of photovoltaic energy supply sub-regions is identified, and different energy storage solution modes are matched according to the identification results. Under normal operating conditions, the optimization effect of dispatch results is guaranteed, and under rapid power fluctuation conditions, the dispatch response capability is improved, ensuring adaptability and operational stability to complex operating conditions.

[0013] 2. By introducing an inter-regional power complementarity coordination mechanism, this invention identifies regional power surplus and regional power deficit and performs cross-regional collaborative scheduling, realizing resource sharing and collaborative utilization among different photovoltaic energy supply sub-regions. This helps reduce the scheduling pressure caused by local regional supply and demand imbalances and improves the overall utilization level of photovoltaic energy storage resources in the entire power grid area. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating the multi-objective optimization scheduling method for distributed photovoltaic energy storage systems based on big data, as used in an embodiment of the present invention. Detailed Implementation

[0015] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this invention.

[0016] Example 1: A multi-objective optimization scheduling method for distributed photovoltaic energy storage systems based on big data (see [link]). Figure 1 As shown, it includes the following steps: Step S1: Energy storage data acquisition; acquire historical system operation data and meteorological monitoring data of the photovoltaic energy storage system in the current power grid area; there are several electrical nodes in the current power grid area, and several photovoltaic energy storage nodes exist in the photovoltaic energy storage system; Historical system operation data refers to the operation information continuously recorded by the power grid dispatching platform, energy management system, and energy storage monitoring system during historical operation. It mainly includes load power data, photovoltaic power generation data, energy storage charge and discharge power data, energy storage state of charge data, voltage data, and line power flow data of each electrical node. Meteorological monitoring data refers to environmental information related to photovoltaic power generation capacity, mainly including data such as light intensity, ambient temperature, module surface temperature, cloud cover, humidity, and wind speed.

[0017] An electrical node is a point in the power grid topology where electricity is collected or distributed. It can correspond to a transformer node, a user access node, or a distribution line connection node. A photovoltaic energy storage node is an energy node that is connected to both a photovoltaic power generation unit and an energy storage unit. It is used to perform subsequent power regulation and energy management.

[0018] During data acquisition, historical system operation data and meteorological monitoring data were first synchronized according to a unified sampling period, aligning data from different sources by timestamp. Then, missing values ​​were filled using adjacent time-time interpolation, and anomalous data were identified using sliding window statistical analysis, with the window mean replacing outliers. Next, data of different units were standardized, converting power, voltage, temperature, and other units to a unified numerical range to eliminate the impact of dimensional differences on subsequent analysis results. Finally, corresponding data indexes were established for electrical nodes and photovoltaic energy storage nodes, forming a multi-source historical dataset containing node operating status information, photovoltaic power generation information, energy storage operation information, and meteorological environment information. This provides fundamental data support for subsequent load consumption characteristic extraction, photovoltaic power prediction, and power grid area division.

[0019] Step S2: Energy storage area division; extract load power consumption characteristics of each node based on historical system operation data; construct the source-side photovoltaic predicted power curve and load-side predicted power curve of each electrical node based on the load power consumption characteristics of each node; divide the power grid area based on historical system operation data and meteorological monitoring data, and divide the current power grid area into multiple photovoltaic power supply sub-regions; Load power consumption characteristics refer to data features that reflect the power consumption patterns and trends of electrical nodes under different time conditions. They are used to describe the periodicity, volatility, and response characteristics of node loads. In practice, firstly, load power data of each electrical node within a preset time period is extracted from historical system operation data and classified and organized according to different time scales such as daily, weekly, and monthly. Then, the load power sequence is segmented using a fixed-length sliding time window. Within each time window, basic statistical features such as average load value, maximum load value, minimum load value, load fluctuation amplitude, and load change rate are calculated. The load fluctuation amplitude is the difference between the maximum and minimum load values ​​within the window, and the load change rate is the ratio of the load change at adjacent sampling times to the time interval. Further, the periodic patterns of the load power sequence are analyzed to statistically analyze the power consumption distribution at different times and extract typical load patterns corresponding to weekdays and non-weekdays, peak hours and off-peak hours. Then, the obtained statistical features and periodic features are normalized and node load feature vectors are constructed in chronological order, thereby forming load power consumption characteristics that can characterize the historical power consumption behavior and future power consumption trends of each electrical node.

[0020] The source-side photovoltaic power forecast curve refers to the curve formed by the change of output power of photovoltaic power generation units over time during the predicted future dispatch period, which is used to reflect the future power generation capacity that the photovoltaic side can provide; the load-side power consumption forecast curve refers to the curve formed by the change of load power over time during the predicted future dispatch period, which is used to reflect the future electricity demand of the user side.

[0021] In practice, the first step is to correlate the load power consumption characteristics of each electrical node with meteorological monitoring data over time to establish a mapping relationship between historical load changes and environmental factors. Then, a photovoltaic power prediction model is constructed using historical photovoltaic power generation data, irradiance data, and temperature data. During the prediction process, meteorological forecast information for future time periods is input in chronological order to calculate the photovoltaic power generation value at each moment, and the prediction results for consecutive moments are connected to form a source-side photovoltaic power prediction curve. Simultaneously, a load prediction model is constructed using historical load power data and extracted load power consumption characteristics. The moment information, periodic characteristics, and historical load change patterns corresponding to future time periods are used as input data to calculate the load demand power value at each future moment, and load-side power prediction curves are generated in chronological order. Finally, the source-side photovoltaic power prediction curve and load-side power prediction curve for the same electrical node within a preset future scheduling cycle are obtained, providing a predictive data foundation for subsequent source-load matching analysis, regional division, and energy storage optimization scheduling.

[0022] The specific steps for dividing the power grid into regions based on historical system operation data and meteorological monitoring data include: Photovoltaic irradiance temperature characteristics are extracted based on meteorological monitoring data; the predicted photovoltaic power curve on the source side and the predicted power consumption curve on the load side of the same electrical node are calculated to obtain the node time-series net load curve; Photovoltaic irradiance temperature characteristics are environmental characteristic parameters used to characterize the variation law of photovoltaic power generation capacity. They essentially reflect the power generation potential of photovoltaic modules under different meteorological conditions. Among them, irradiance intensity represents the solar radiation energy received per unit area and is the main factor affecting photovoltaic power generation output. Ambient temperature and module surface temperature affect the photoelectric conversion efficiency of photovoltaic cells.

[0023] In practice, the following steps are taken: First, data related to photovoltaic power generation, such as irradiance, ambient temperature, module surface temperature, humidity, and cloud cover, are extracted from meteorological monitoring data and synchronized at uniform time intervals. Then, a sliding time window is used to statistically analyze continuous meteorological data, calculating the average irradiance, irradiance fluctuation amplitude, average temperature, and temperature change rate within the window. The irradiance fluctuation amplitude is the difference between the maximum and minimum irradiance values ​​within the window, and the temperature change rate is the ratio of the temperature change at adjacent moments to the time interval. Furthermore, historical photovoltaic power generation data is combined to analyze the variation patterns of photovoltaic output power under different meteorological conditions, forming photovoltaic irradiance-temperature characteristics that can reflect future trends in power generation capacity, providing an environmental characteristic basis for subsequent net load calculation and regional division.

[0024] The node time-series net load curve refers to the curve showing the actual power change that an electrical node needs to obtain from or output to the external power grid during the future forecast period, and is used to reflect the source-load balance state of the node; where net load represents the difference between the load demand power and the local photovoltaic power generation power.

[0025] In practice, the predicted power curves of the source-side photovoltaic (PV) and load-side electricity consumption for the same electrical node are first extracted in chronological order to ensure that both sets of curves have the same time resolution and prediction duration. Then, at each prediction time, the difference between the predicted power consumption on the load side and the predicted power of the source-side PV is calculated. A positive difference indicates that the node has external power demand, while a negative difference indicates that the node has surplus PV power. Finally, the net load values ​​obtained at each prediction time are connected in chronological order to form a continuous node-time-series net load curve. This curve not only reflects the future power demand changes of the node but also demonstrates the dynamic matching relationship between PV power generation and load electricity consumption.

[0026] Using electrical nodes as graph nodes and the physical electrical lines between electrical nodes as edges, a basic physical connectivity graph of the power grid is constructed. For any two adjacent graph nodes with edges, the inverse correlation coefficient between the time-series net load curves of the adjacent graph nodes is calculated. The inverse correlation coefficient represents the degree of complementarity of the net load change trends between adjacent electrical nodes. The load asymmetry weight of the edge is assigned according to the degree of complementarity. A power grid basic physical connectivity graph is a data model that uses a graph structure to describe the topological relationships of a power grid; in this model, graph nodes correspond to electrical nodes in the actual power grid, and edges correspond to the actual transmission and distribution line connections between electrical nodes.

[0027] In practice, the power grid topology data is first read to identify all electrical nodes and their interconnections. Then, each electrical node is mapped to a graph node in the graph structure, and the physical transmission and distribution lines between two nodes are mapped to the corresponding graph edges. Furthermore, basic parameters such as line impedance, line capacity, and line length are recorded as edge attributes to construct a basic physical connectivity graph of the power grid that can truly reflect the physical connection relationships of the power grid. This graph not only preserves the original topology of the power grid but also provides a structural foundation for node association analysis and cluster analysis in the subsequent regional division process.

[0028] The inverse correlation coefficient is an evaluation index used to measure the degree of complementarity in the net load change trends of two adjacent nodes. When the net load of one node increases while the net load of the other node decreases, the two exhibit a strong complementary relationship. In practice, the node time-series net load curves corresponding to the two adjacent electrical nodes are first extracted and the curves are standardized to eliminate the influence of differences in load scale between different nodes. Then, the correlation coefficient between the two net load sequences is calculated to describe the consistency of the change trends of the two curves. Finally, the correlation results are reverse-converted so that the stronger the negative correlation, the larger the corresponding inverse correlation coefficient.

[0029] At this point, a larger inverse correlation coefficient indicates that the two nodes have strong source-load complementarity characteristics in the same period, that is, when one node has power surplus, the other node is more likely to have power demand. Further, the obtained inverse correlation coefficient is mapped to the load asymmetric weight of the corresponding connection, where the load asymmetric weight is used to describe the power complementarity value between the nodes at both ends of the connection. The larger the weight, the more suitable it is to be assigned to the same energy supply area for coordinated scheduling.

[0030] Based on the physical connectivity graph of the power grid with asymmetric load weights, the graph segmentation constraint is to maximize the sum of weights within sub-regions and minimize the weights of cross-sub-region edges. A graph clustering algorithm is used to dynamically segment the current power grid region to obtain multiple photovoltaic power supply sub-regions.

[0031] A photovoltaic (PV) power supply sub-region refers to a local area within the power grid that has strong internal source-load complementarity and high collaborative scheduling value. In practice, the basic physical connectivity graph of the power grid with asymmetric load weights is first used as the clustering input. Then, a graph partitioning objective function is established to maximize the sum of edge weights within the same sub-region and minimize the sum of edge weights between different sub-regions, thus ensuring close internal connections and weak coupling between regions after partitioning. Next, a graph clustering algorithm is used for iterative calculation. In each round of calculation, the impact of the node partitioning scheme on the objective function is evaluated, and the region to which the node belongs is continuously adjusted. When the region partitioning result stabilizes after multiple iterations and the objective function no longer significantly improves, the iteration stops, and the final partitioning result is output. This results in multiple PV power supply sub-regions. Because this partitioning process considers both the physical connection relationship of the power grid and the net load complementarity relationship of the nodes, it enables better matching of PV resources and load demand within the same region, providing a more reasonable scheduling unit for subsequent energy storage optimization scheduling and regional collaborative control.

[0032] Step S3: Multi-objective optimization scheduling; Construct different photovoltaic energy storage multi-objective optimization models for different photovoltaic energy supply sub-regions; The photovoltaic energy storage multi-objective optimization model includes an energy storage objective function and an energy storage solution mode; Within a preset energy storage optimization scheduling period, for any photovoltaic energy supply sub-region, optimize and solve based on the energy storage objective function and energy storage solution mode in the photovoltaic energy storage multi-objective optimization model to obtain the corresponding regional photovoltaic energy storage scheduling scheme; Complete the scheduling of the photovoltaic energy storage system in the current power grid area according to the regional photovoltaic energy storage scheduling scheme corresponding to each photovoltaic energy supply sub-region.

[0033] The specific steps for constructing different photovoltaic energy storage multi-objective optimization models based on different photovoltaic energy supply sub-regions include: For any photovoltaic power supply sub-region, the total predicted photovoltaic power on the source side and the total predicted electricity consumption on the load side of all electrical nodes in the photovoltaic power supply sub-region during the preset energy storage optimization scheduling period are calculated based on the predicted photovoltaic power curve on the source side and the predicted electricity consumption curve on the load side. The photovoltaic power supply characteristic coefficient and asymmetric power fluctuation rate are calculated based on the total predicted photovoltaic power on the source side and the total predicted electricity consumption on the load side. The predicted total energy of source-side photovoltaic power refers to the total expected power generation reflected by the predicted power curve of source-side photovoltaic power corresponding to all electrical nodes in the photovoltaic power supply sub-region within the preset energy storage optimization scheduling period; the predicted total energy of load-side electricity consumption refers to the total expected electricity consumption reflected by the predicted power curve of load-side electricity consumption corresponding to all electrical nodes in the photovoltaic power supply sub-region within the preset energy storage optimization scheduling period.

[0034] Among them, the source side refers to the power generation side, that is, the side where the photovoltaic power generation unit is located; the predicted total energy represents the cumulative power generation obtained by integrating the predicted photovoltaic power at each moment in the future scheduling cycle; its physical meaning is the total power that the photovoltaic power supply sub-region can theoretically provide in the future scheduling cycle, used to characterize the size of the new energy supply capacity in the region; the load side refers to the power consumption side, that is, the side where the user load is located; the predicted total energy represents the cumulative power consumption obtained by integrating the predicted load power at each moment in the future scheduling cycle; its physical meaning is the total power demand expected to be generated by the photovoltaic power supply sub-region in the future scheduling cycle, used to characterize the size of the load demand in the region.

[0035] The photovoltaic power supply characteristic coefficient is used to characterize the degree of matching between photovoltaic power supply capacity and regional load demand, while the asymmetric power fluctuation rate is used to characterize the severity and imbalance of regional net power changes. In specific implementation, the source-side photovoltaic predicted power curve and load-side electricity consumption predicted power curve corresponding to all electrical nodes in the photovoltaic power supply sub-region are first extracted, and time integration is performed according to the preset energy storage optimization scheduling cycle. The predicted power at each moment is multiplied by the corresponding time interval and then accumulated to obtain the photovoltaic predicted energy and load predicted energy of each electrical node within the scheduling cycle. Specifically, the predicted photovoltaic power value of each electrical node at each moment within the scheduling cycle can be multiplied with the corresponding time interval and then summed. The results of all electrical nodes in the region can then be summarized to obtain the total predicted photovoltaic energy on the source side. Similarly, the predicted load power value of each electrical node at each moment within the scheduling cycle can be multiplied with the corresponding time interval and then summed. The results of all electrical nodes in the region can then be summarized to obtain the total predicted load-side electricity consumption.

[0036] A photovoltaic power supply characteristic coefficient is constructed by further utilizing the ratio between the predicted total photovoltaic energy on the source side and the predicted total electricity consumption on the load side. When the coefficient is large, it indicates that the photovoltaic power supply capacity in the region is strong, and when the coefficient is small, it indicates that the load demand in the region is dominant.

[0037] Meanwhile, the node time-series net load curves corresponding to all electrical nodes in the region are extracted and spatially superimposed to form the regional net power sequence; the net power change at adjacent sampling times is calculated, and the average amplitude, maximum amplitude, and fluctuation dispersion of the net power change within the entire scheduling cycle are statistically analyzed. By comprehensively analyzing the intensity and frequency of net power change, the asymmetric power fluctuation rate is obtained, thereby reflecting the regional power fluctuation risk level and providing a basis for subsequent objective function construction and solution mode selection.

[0038] The photovoltaic energy supply characteristic coefficients are used to match the corresponding energy storage objective function for the photovoltaic energy supply sub-regions; the objective function of the energy storage objective function includes photovoltaic absorption, tie-line stability and energy storage lifetime. The energy storage objective function is a comprehensive evaluation function used to guide the charging and discharging decisions of the energy storage system. Its role is to establish a coordination relationship among multiple scheduling objectives, so that the energy storage system can automatically adjust the optimization direction according to the regional operating characteristics. Specifically, the steps for matching the corresponding energy storage objective function to the photovoltaic energy supply sub-region based on the photovoltaic energy supply characteristic coefficients include: A preset source-load balance benchmark value is established; among which, the photovoltaic absorption sub-item, the tie-line stability sub-item, and the energy storage lifetime sub-item correspond to the photovoltaic absorption sub-item weight, the tie-line stability sub-item weight, and the energy storage lifetime sub-item weight, respectively; The source-load balance benchmark value is a reference standard value used to measure whether the photovoltaic supply capacity and load demand capacity are in a balanced state. In essence, it represents the characteristic parameter corresponding to the relatively balanced generation capacity on the source side and the electricity demand on the load side in the photovoltaic power supply sub-region. Usually, the photovoltaic power supply characteristic coefficient corresponding to the equal total predicted photovoltaic energy on the source side and the total predicted electricity demand on the load side can be set as the source-load balance benchmark value.

[0039] The photovoltaic (PV) grid integration sub-item evaluates the extent to which PV power generated within a region is absorbed and utilized by local loads or energy storage systems. The higher the PV grid integration level, the less curtailment occurs. The tie-line stability sub-item evaluates the stability of the power exchange process between the region and the external power grid. The smaller the tie-line power fluctuation, the more stable the grid operation. The energy storage lifetime sub-item evaluates the degree of lifespan loss of energy storage equipment during operation. Its goal is to reduce the aging effects caused by frequent charging and discharging while meeting dispatch requirements. To achieve multi-objective collaborative optimization, corresponding weight parameters need to be configured for the above three objective sub-items. The weight parameters are used to characterize the importance of each objective in the current dispatch scenario, providing a basis for subsequent dynamic optimization.

[0040] Calculate the directional offset between the photovoltaic power supply characteristic coefficient and the source-load balance benchmark value; determine the weights of the photovoltaic consumption component, the tie-line stability component, and the energy storage lifetime component based on the directional offset. The directional offset is a characteristic quantity used to describe the degree and direction of deviation of the current photovoltaic (PV) power supply state from the ideal source-load balance state. The magnitude of the offset reflects the degree of deviation, while the direction reflects whether the region is currently in a PV supply-dominated or load demand-dominated state. In practice, the directional offset is first obtained by calculating the difference between the PV power supply characteristic coefficient and the source-load balance benchmark value. When the directional offset is positive, it indicates that the PV supply capacity on the source side in the region exceeds the load demand capacity, and the region is in a PV surplus state. When the directional offset is negative, it indicates that the load demand capacity in the region exceeds the PV supply capacity, and the region is in a load-dominated state. When the quantity is close to zero, it indicates that the region is in a relatively balanced state. Subsequently, weights are allocated according to the magnitude and direction of the directional offset. When the region is in a state of photovoltaic surplus, the weight of the photovoltaic absorption component is increased, so that the energy storage system can prioritize the absorption and utilization of surplus photovoltaic power. When the region is in a load-dominated state, the weight of the tie-line stability component is increased, so that the energy storage system can prioritize the suppression of power fluctuations between the region and the external power grid. When the region is close to a balanced state, the weight of the energy storage lifetime component is increased, so that the system pays more attention to the balanced operation and lifetime protection of energy storage devices. Through the above methods, the optimization focus of the objective function can be dynamically adjusted according to the regional operating characteristics, rather than using a fixed weight configuration.

[0041] It should be noted that in this embodiment, the directional offset is used to characterize the source-load imbalance degree of the current photovoltaic power supply sub-region; the weighting parameter is not a fixed value preset, but is adaptively generated based on the correspondence between the directional offset and the statistical characteristics of historical operating samples. Specifically, firstly, a sample set of directional offsets is constructed using historical system operating data, and the correlation between photovoltaic absorption level, tie-line power fluctuation level, and energy storage lifetime loss level under different directional offsets is statistically analyzed; then, photovoltaic absorption sensitivity curves, tie-line stability sensitivity curves, and energy storage lifetime sensitivity curves are established respectively.

[0042] Once the current directional offset is obtained during real-time operation, it is mapped to the three types of sensitivity curves mentioned above to obtain the corresponding photovoltaic (PV) absorption sensitivity value, tie-line stability sensitivity value, and energy storage lifetime sensitivity value. The sensitivity value reflects the strength of the objective function's response to changes in the current source-load imbalance state; the larger the sensitivity value, the more important the corresponding objective is to the current operating state. Using these three sensitivity values ​​as the basis for weight generation, they are normalized to obtain the PV absorption component weight, tie-line stability component weight, and energy storage lifetime component weight.

[0043] When the directional offset increases towards the direction of photovoltaic surplus, the response intensity corresponding to the photovoltaic absorption sensitivity curve automatically increases, thus the weight of the photovoltaic absorption component increases accordingly. When the directional offset increases towards the direction of load dominance, the response intensity corresponding to the tie-line stability sensitivity curve automatically increases, thus the weight of the tie-line stability component increases accordingly. When the directional offset approaches the source-load balance state, the response intensity corresponding to the energy storage lifetime sensitivity curve reaches a relatively high level, thus the weight of the energy storage lifetime component increases automatically.

[0044] An energy storage objective function is constructed based on the photovoltaic absorption component, the tie-line stability component, the energy storage lifetime component, and the corresponding weights of the photovoltaic absorption component, the tie-line stability component, and the energy storage lifetime component.

[0045] The energy storage objective function is a comprehensive evaluation model used to guide the charging and discharging decisions of energy storage systems. Its role is to establish a unified decision-making basis among multiple optimization objectives, enabling the energy storage system to automatically seek the optimal dispatch scheme based on the regional operating status. In specific implementation, the evaluation values ​​corresponding to the photovoltaic absorption component, the tie-line stability component, and the energy storage lifetime component are calculated separately. Among them, the photovoltaic absorption component can be characterized by the proportion of photovoltaic power effectively utilized in the region to the total photovoltaic power generation; the tie-line stability component can be characterized by the amplitude of tie-line power fluctuation or the rate of power change within the dispatch cycle; and the energy storage lifetime component can be characterized by the degree of change in the charging and discharging depth of energy storage units, the degree of change in the number of cycles, or the lifetime loss index. Subsequently, each objective component is weighted and combined with its corresponding weight so that the importance of each objective can reflect the current regional operating needs. Finally, a unified energy storage objective function is formed. This energy storage objective function can integrate photovoltaic absorption demand, grid stability demand, and energy storage lifetime demand into the same optimization framework, and automatically adjust the optimization direction according to the operating characteristics of different photovoltaic energy supply sub-regions, thereby improving the pertinence, adaptability, and overall operating efficiency of energy storage dispatch.

[0046] The solution mode for matching the energy storage to the photovoltaic power supply sub-region is based on the asymmetric power fluctuation rate; the energy storage solution mode includes conventional solution mode and special solution mode.

[0047] Specifically, the steps for matching the energy storage solution mode corresponding to the photovoltaic power supply sub-region based on the asymmetric power fluctuation rate include: Extract the node time-series net load curves of all electrical nodes in the photovoltaic power supply sub-region and spatially superimpose and fuse them to obtain the regional time-series net power curve; within a preset sliding time window, calculate the first time derivative of the regional time-series net power curve and take the effective value of the absolute value as the asymmetric power fluctuation rate; at the same time, calculate the second time derivative of the regional time-series net power curve to obtain the asymmetric power fluctuation rate. The node time-series net load curve refers to the net power change curve of a certain electrical node in a continuous time series. Its value is obtained by subtracting the load power consumption from the photovoltaic power generation at the corresponding time. When the net load is positive, it means that the power generation of the node is greater than the power consumption. When the net load is negative, it means that the power consumption of the node is greater than the power generation.

[0048] To reflect the overall operational status of the entire photovoltaic power supply sub-region, the time-series net load curves of all electrical nodes within the region are first time-aligned according to a unified time scale to eliminate data deviations caused by different sampling periods. Subsequently, the net load values ​​corresponding to all nodes are accumulated at each sampling time to obtain the regional time-series net power curve, which reflects the changes in power surplus and power deficit of the entire photovoltaic power supply sub-region. The regional time-series net power curve essentially reflects the real-time balance between photovoltaic power generation and load demand within the region and is an important basis for subsequent judgment of the regional operational status.

[0049] After obtaining the regional time-series net power curve, a sliding time window method is used to dynamically analyze the curve. The sliding time window refers to selecting a continuous time interval of fixed length and moving it forward continuously as time progresses to achieve real-time monitoring of the power change trend. Within each sliding time window, the change between adjacent sampling times of the regional time-series net power curve is calculated, and the first time derivative of the curve is approximately represented by the ratio of the change to the sampling time interval. The first time derivative reflects the rate of change of the regional net power, that is, the speed at which the net power increases or decreases per unit time.

[0050] Since the system needs to pay attention to both the power rise and power fall processes, the absolute value of the first-order time derivative is taken to eliminate the influence of the direction of change. The effective value is calculated for all absolute value results within the window. That is, the average level of the squared values ​​of each data point is calculated first, and then the square root is taken to obtain the asymmetric power volatility rate that can comprehensively reflect the overall fluctuation intensity. Compared with the simple average value, the effective value can better highlight the impact of large fluctuation events, and therefore is more suitable as an evaluation index of the degree of regional power fluctuation.

[0051] To further identify the development trend of power fluctuations, it is also necessary to calculate the second time derivative of the regional net power curve. The second time derivative reflects the rate of change of power itself, that is, whether the power fluctuation is intensifying or weakening. When the second time derivative is large, it indicates that the regional net power is rapidly deviating from the current state, and the system may soon experience violent fluctuations. When the second time derivative is small, it indicates that the power change trend is relatively gentle.

[0052] To improve computational stability, the regional time-series net power curve can be smoothed and filtered first to eliminate the interference of random noise on the derivative calculation results. Then, the second-order time derivative is calculated using the difference between adjacent first-order time derivatives, and the calculation results within the entire sliding time window are statistically analyzed to obtain the asymmetric power fluctuation rate, which can reflect the current regional power fluctuation development trend. The asymmetric power fluctuation rate is used to characterize the current fluctuation intensity, and the asymmetric power fluctuation rate is used to characterize the evolution trend of the fluctuation intensity. The combination of the two can simultaneously reflect the current state and the future direction of change.

[0053] When the asymmetric power fluctuation rate exceeds a preset first threshold and the asymmetric power fluctuation change rate exceeds a preset second threshold, a special solution mode is matched; otherwise, a regular solution mode is matched.

[0054] Energy storage solution modes refer to the different optimization solution strategies adopted by energy storage systems under different operating conditions. Among them, conventional solution modes are suitable for scenarios where the system operation is relatively stable and the power change is relatively gradual; special solution modes are suitable for scenarios where photovoltaic output fluctuates rapidly, load changes suddenly, or regional power imbalance expands rapidly.

[0055] To achieve automatic switching of the solution mode, the distribution of asymmetric power fluctuation rate and asymmetric power fluctuation change rate under different operating conditions is first statistically analyzed using historical system operation data. Then, a first preset threshold and a second preset threshold are determined in combination with the requirements for safe operation of the power grid. The first preset threshold is used to determine whether the current power fluctuation has reached a level that requires close attention. The second preset threshold is used to determine whether the current power fluctuation is in a state of rapid deterioration.

[0056] During real-time operation, the system continuously calculates the asymmetric power fluctuation rate and asymmetric power fluctuation change rate corresponding to the current photovoltaic power supply sub-region and compares them with the corresponding thresholds. When the asymmetric power fluctuation rate exceeds the preset first threshold, it indicates that the region has experienced significant power fluctuations. When the asymmetric power fluctuation change rate exceeds the preset second threshold, it indicates that the fluctuation is still expanding and the region's operating state may evolve from a stable state to an abnormal state. At this time, the system determines that the current region has entered a high-risk fluctuation condition and automatically matches a special solution mode, so that the subsequent optimization process prioritizes the stability of tie-line power and rapid response capability to improve the system's ability to cope with sudden power shocks.

[0057] Conversely, if at least one of the above two conditions is not met, it indicates that the regional operation status is still within a controllable range. In this case, the conventional solution mode is matched, and a better overall energy storage scheduling result is obtained by using a complete multi-objective optimization process. By examining both the fluctuation intensity and the fluctuation development trend at the same time, misjudgments caused by relying on a single indicator can be avoided, thereby improving the accuracy and reliability of energy storage scheduling strategy switching.

[0058] A preset first threshold is used to determine whether regional power fluctuations have reached an abnormal level. In this embodiment, the regional time-series net power curve within the historical operating cycle is first extracted, and the asymmetric power fluctuation rate at the corresponding time is calculated in the same way as in real-time operation. Then, the distribution characteristics of all historical asymmetric power fluctuation rate samples are statistically analyzed, and their long-term average level and fluctuation dispersion are calculated. Furthermore, based on the long-term average level and the fluctuation dispersion, a normal fluctuation range and an abnormal fluctuation range are constructed. When the asymmetric power fluctuation rate calculated in real-time exceeds the upper boundary of the normal range formed by historical statistics, the value corresponding to this boundary is determined as the preset first threshold. The first threshold obtained in this way can reflect the maximum fluctuation level allowed by the current power grid region under normal operating conditions.

[0059] A preset second threshold is used to determine whether power fluctuations are in a rapidly deteriorating state. In this embodiment, the asymmetric power fluctuation change rate at the corresponding moment is first calculated based on historical operating data, and the probability distribution characteristics of the fluctuation change rate under different operating conditions are statistically analyzed. Then, abnormal event samples that have occurred in the historical data, such as significant power surges, tie line fluctuations exceeding limits, or energy storage emergency responses, are screened, and the fluctuation change rate data corresponding to the abnormal event is extracted. By analyzing the change rate characteristics before the abnormal event is triggered, a change rate critical value that can effectively distinguish between normal and abnormal states is determined, and this critical value is determined as the preset second threshold. The second threshold obtained in this way can reflect the critical characteristics of the power fluctuation transitioning from a stable state to a rapidly deteriorating state.

[0060] The preset first threshold and the preset second threshold are not fixed parameters, but are dynamically updated based on historical operating data from the most recent consecutive scheduling cycles. The system periodically re-statistically re-analyzes the distribution characteristics of asymmetric power fluctuation rate and asymmetric power fluctuation change rate, and synchronously corrects the corresponding thresholds, so that the thresholds can be adaptively adjusted with seasonal changes, weather changes and load structure changes.

[0061] The specific steps for optimization based on the energy storage objective function and energy storage solution mode in the photovoltaic energy storage multi-objective optimization model include: The system acquires operational status data for each photovoltaic (PV) energy storage node within the PV power supply sub-region. This operational status data reflects the real-time operational and adjustability capabilities of each energy storage node. Specifically, real-time operational data for each PV energy storage node is first collected through the distribution automation system, energy storage management system, and on-site monitoring terminals. The acquired data includes at least the current PV power generation, current load power, remaining energy storage capacity, maximum rechargeable energy storage capacity, maximum dischargeable energy storage capacity, energy storage charging / discharging status, and inverter operating status. The collected data is then time-synchronized to ensure that data from each node corresponds to the same scheduling time. For data with missing values, outliers, or communication delays, corrections are made using interpolation of data from adjacent time points and outlier removal methods, resulting in a unified operational status dataset. This operational status dataset not only reflects the real-time availability of energy storage resources within the current region but also provides constraints and decision-making basis for subsequent optimization solutions.

[0062] When the energy storage solution mode is the conventional solution mode, a comprehensive optimization objective is constructed by combining the operating status data and the energy storage objective function; the comprehensive optimization objective is iteratively solved to obtain the corresponding photovoltaic energy storage scheduling scheme. The conventional solution mode is suitable for scenarios where the regional operating status is relatively stable. In this case, the system has sufficient computation time and can simultaneously consider multiple optimization objectives such as photovoltaic utilization, grid stability, and energy storage lifetime. In specific implementation, the real-time operating constraints of each energy storage node are first extracted based on the operating status data, including energy storage remaining capacity constraints, charging and discharging power constraints, state of charge safety constraints, and inverter rated power constraints. Then, the aforementioned constructed energy storage objective function is called, and the photovoltaic absorption component, tie-line stability component, and energy storage lifetime component are used as evaluation indicators in a unified optimization system.

[0063] Furthermore, the operational constraints are integrated with the energy storage objective function to form a comprehensive optimization objective. Among these, the photovoltaic (PV) absorption component aims to improve the local utilization level of PV power within the region; the tie-line stability component aims to reduce power exchange fluctuations between the region and the external power grid; and the energy storage lifetime component aims to reduce the lifespan loss of energy storage equipment caused by frequent charging and discharging. By unifying multiple objectives into a single optimization framework, the subsequent solution results can simultaneously meet the requirements of economy, stability, and equipment reliability.

[0064] After the comprehensive optimization objective is established, the optimal charging and discharging strategy of each energy storage node in the future scheduling cycle needs to be determined through iterative optimization. In specific implementation, an initial scheduling scheme that satisfies all operating constraints is first generated and used as the starting point for optimization. Then, the evaluation results corresponding to the comprehensive optimization objective are calculated based on the current scheduling scheme, and the contribution of each objective component to the overall performance is analyzed. Then, the charging and discharging power allocation of each energy storage node is gradually adjusted according to the direction of objective improvement, and the optimization results are recalculated.

[0065] As the iteration process continues, the system continuously compares the optimization gain between two adjacent calculations. When the optimization gain drops below the preset convergence criterion, the current scheduling result is considered to have stabilized. Finally, the system outputs the charging power, discharging power, and state of charge change plan for each energy storage node in each scheduling period, thus forming a corresponding photovoltaic energy storage scheduling scheme. This scheme can fully utilize photovoltaic power generation resources while ensuring stable regional power exchange and safe operation of energy storage equipment.

[0066] When the energy storage solution mode is a special solution mode, the power mutation state is obtained based on the operating status data; the tie-line stability enhancement factor is calculated based on the power mutation state; the tie-line stability enhancement factor is used to amplify and correct the tie-line stability component weight, and the photovoltaic absorption component weight and the energy storage lifetime component weight are compressed and corrected to obtain a new energy storage objective function; a transient optimization objective is constructed based on the new energy storage objective function; the transient optimization objective is iteratively solved to obtain the corresponding photovoltaic energy storage scheduling scheme.

[0067] A power surge state refers to an operational state in which the net power within a photovoltaic (PV) power supply sub-region changes significantly within a short period of time. In practice, firstly, a regional net power sequence is constructed based on real-time operational data, and the power change between consecutive sampling times is continuously monitored. Then, the net power deviation between the current and previous times, as well as the power change trend across multiple consecutive times, are calculated. When the power change exceeds the historical normal operating range and the trend continues to strengthen, the system is determined to have entered a power surge state. Simultaneously, cross-validation is performed using PV output change information, load change information, and energy storage response information to avoid misjudgments due to measurement noise or short-term communication anomalies. Through power surge state identification, abnormal fluctuation events that may affect the stable operation of the power grid can be detected in advance.

[0068] The tie-line stability enhancement factor is used to characterize the importance of maintaining tie-line power stability at the current moment. Essentially, it is a target enhancement parameter dynamically generated based on the risk of power surges. In practice, firstly, characteristic parameters such as the net power change amplitude, rate of change, and duration corresponding to power surge states are extracted. Then, the influence of these characteristic parameters on tie-line power fluctuations is analyzed, and an influence mapping relationship is established by combining tie-line response under similar operating conditions in historical operating data. Further, the tie-line stability enhancement factor is calculated based on the current power surge risk level. When the net power change amplitude is larger, the rate of change is faster, and the duration is longer, it indicates a higher risk of the tie-line being impacted, and the corresponding tie-line stability enhancement factor is larger. Conversely, when the power surge is small, the tie-line stability enhancement factor remains at a low level. This enhancement factor objectively reflects the importance of tie-line stability control under the current operating state, rather than relying on manual experience for setting.

[0069] In the special solution mode, the primary task of the system changes from comprehensive optimization to rapid suppression of power surges. Therefore, it is necessary to dynamically adjust the importance of each optimization objective. In specific implementation, the tie-line stability enhancement factor is first applied to the tie-line stability component, so that the contribution ratio of the tie-line stability component in the overall objective function increases synchronously with the risk level. Then, based on the enhanced tie-line stability component ratio, the photovoltaic absorption component and the energy storage lifetime component are adaptively compressed and adjusted to keep the total weight of the three objective components consistent.

[0070] It should be noted that the weight adjustment here is not a fixed ratio set manually, but is automatically generated based on the contribution of power mutation status to the improvement of overall dispatch performance. When the risk of power mutation is high, the tie line stability component is given higher priority. When the risk decreases, the weights of each objective component gradually return to normal levels, and finally a new energy storage objective function is formed, so that the system can prioritize the stability of grid operation under abnormal conditions.

[0071] The transient optimization objective is a dedicated optimization model established for scenarios with rapid power fluctuations. Its core task is to suppress tie-line power surges in the shortest possible time. In practice, a new energy storage objective function is first used as the evaluation basis, and the adjustable power range, remaining energy storage capacity, and real-time response capability of each energy storage node are extracted. Then, an optimization objective system with tie-line power stability as the core is established, while retaining photovoltaic utilization and energy storage lifetime constraints as auxiliary objectives. Furthermore, the transient optimization objective is constructed by combining the current operating status data. Compared with conventional solution modes, the transient optimization objective pays more attention to the rapid response capability under short time scales, thereby ensuring that the system can respond to sudden power fluctuations in a timely manner.

[0072] Once the transient optimization objective is established, the system immediately initiates a rapid optimization solution process. In practice, an initial response scheme is first generated based on the current adjustable capacity of each energy storage node. Then, the effectiveness of this scheme in suppressing tie-line power fluctuations is evaluated, and the charging and discharging power allocation of each energy storage node is adjusted based on the optimization results. In each round of calculation, priority is given to ensuring the stability of the tie-line power target, while also considering photovoltaic utilization and energy storage lifetime requirements. Once the optimization results meet the preset stability conditions or reach the specified response time, the scheduling results are immediately output without pursuing a globally optimal solution. Ultimately, a photovoltaic energy storage scheduling scheme is formed, containing charging and discharging commands for each energy storage node, power adjustment amplitude, and execution timing information, and is then distributed to the corresponding energy storage devices for execution. This achieves rapid suppression of sudden power fluctuations and improves the operational stability and scheduling response capability of the entire photovoltaic power supply sub-region.

[0073] The specific steps for scheduling the photovoltaic energy storage system in the current power grid area according to the regional photovoltaic energy storage scheduling plan corresponding to each photovoltaic energy supply sub-region include: Based on the regional photovoltaic energy storage scheduling scheme corresponding to each photovoltaic energy supply sub-region, calculate the regional power surplus or regional power deficit of each photovoltaic energy supply sub-region within the preset energy storage optimization scheduling cycle; construct a regional coordination relationship matrix between photovoltaic energy supply sub-regions based on the regional power surplus and regional power deficit; and determine the target photovoltaic energy supply sub-region combination with power complementarity based on the regional coordination relationship matrix. The regional photovoltaic energy storage dispatch scheme is the optimal charge and discharge control result generated for a single photovoltaic energy supply sub-region during the aforementioned optimization solution stage. It can reflect the power supply and demand status of the sub-region in the future dispatch cycle. In specific implementation, the regional photovoltaic energy storage dispatch scheme corresponding to each photovoltaic energy supply sub-region is first read, and the planned photovoltaic power generation, planned load power consumption, and planned energy storage charge and discharge power at each time in the corresponding dispatch cycle are statistically analyzed. Then, the regional net power balance result at each dispatch time is calculated in chronological order.

[0074] When the total available power within a region exceeds the region's load demand, the excess is defined as the region's power margin; when the total available power within a region is less than the region's load demand, the deficiency is defined as the region's power deficit. To avoid misjudgments caused by instantaneous fluctuations, the power margin and power deficit over the entire scheduling cycle can be further integrated over time to obtain a comprehensive power margin and comprehensive power deficit that reflect the overall supply and demand situation of the region. This process can identify which sub-regions have surplus regulation capacity and which sub-regions have power deficits, providing basic data for subsequent cross-regional coordinated scheduling.

[0075] The regional coordination relationship matrix is ​​a data structure used to describe the power complementarity between different photovoltaic power supply sub-regions. Essentially, it reflects whether the regions have the value of coordinated scheduling. In specific implementation, all photovoltaic power supply sub-regions are first numbered in a unified order and a corresponding matrix structure is established. Then, for any two photovoltaic power supply sub-regions, their regional power surplus and regional power deficit information are extracted respectively, and the supply and demand matching degree of the two in the same scheduling cycle is analyzed.

[0076] When one region has a power surplus and another region has a power deficit, it indicates that the two have potential complementary capabilities. Further analysis is conducted by considering the electrical connections between regions, line transmission capacity, and power exchange distances to calculate corresponding coordination matching indices. A higher coordination matching index indicates a better supply-demand balance achieved through power exchange between regions. Finally, all coordination matching indices between regions are filled into the corresponding positions in the matrix to form a regional coordination relationship matrix, thereby achieving a quantitative expression of the regional coordination potential of the entire power grid.

[0077] Power complementarity refers to the relationship where the power surplus of one photovoltaic (PV) sub-region can effectively compensate for the power deficit of another PV sub-region. In practice, the process begins by traversing all coordination matching indices in the regional coordination matrix and sorting them by index value. Then, regional combinations that meet preset coordination conditions are selected. These conditions include at least the existence of a power supply-demand complementarity relationship, sufficient line transmission capacity to meet exchange requirements, and the absence of new operational constraint conflicts during cross-regional power transmission. For multiple candidate regional combinations, combinations with higher coordination matching indices are prioritized to improve power exchange efficiency. Ultimately, several target PV sub-region combinations are formed, each containing at least one power surplus region and at least one power deficit region. This step connects multiple sub-regions that were originally operating independently, creating regional combinations with collaborative optimization capabilities.

[0078] For any target photovoltaic power supply sub-region combination, priority is given to using regional power surplus to coordinate regional power deficit across regions, resulting in cross-regional power coordination results; corresponding regional collaborative dispatch instructions are generated based on the cross-regional power coordination results; regional photovoltaic energy storage dispatch schemes within each photovoltaic power supply sub-region are synchronously modified based on the regional collaborative dispatch instructions, resulting in a comprehensive photovoltaic energy storage dispatch scheme; and the optimized dispatch of the photovoltaic energy storage system in the current power grid region is completed based on the comprehensive photovoltaic energy storage dispatch scheme.

[0079] Cross-regional power coordination refers to the redistribution of available power resources among different photovoltaic power supply sub-regions to improve the overall power grid's supply and demand balance. In practice, the power surplus and power deficit values ​​of each region in the target region combination are first extracted, and the total surplus and total deficit within the combination are calculated. Then, the direction and scale of power exchange are determined based on the principle of prioritizing the consumption of surplus photovoltaic power within the region.

[0080] When the regional power margin is greater than the regional power deficit, the margin is used first to compensate for the entire deficit; when the regional power margin is insufficient to compensate for the entire deficit, the power is allocated according to the deficit ratio. At the same time, the transmission capacity constraints and operational safety constraints of the inter-regional interconnection lines need to be checked to ensure that the power exchange process does not lead to line overload or voltage exceeding limits. Finally, the actual power exchange between each region is obtained, thus forming the inter-regional power coordination result, which reflects the actual collaborative energy supply relationship between regions.

[0081] Regional collaborative dispatch instructions are a set of data instructions used to guide each photovoltaic power supply sub-region to perform collaborative power exchange. In specific implementation, the power output that each region needs to increase or decrease is first determined based on the cross-regional power coordination results. Then, the coordination results are converted into executable dispatch control parameters, including the target power exchange value, execution time, power adjustment direction, and the adjustment requirements of the corresponding energy storage devices. Furthermore, based on the real-time operating status of each energy storage node in the region, it is determined which energy storage nodes will undertake the corresponding adjustment tasks. Finally, regional collaborative dispatch instructions are formed and sent to the corresponding photovoltaic power supply sub-regions through the dispatch communication network. These instructions not only include inter-regional power exchange requirements but also intra-regional resource adjustment requirements, thereby ensuring that the collaborative results can be accurately implemented.

[0082] Because regional coordinated dispatch introduces new power exchange relationships, the original regional photovoltaic energy storage dispatch scheme needs to be adjusted synchronously. In specific implementation, firstly, the power exchange tasks specified in the regional coordinated dispatch instructions are read and mapped to the corresponding photovoltaic energy supply sub-regions. Then, the charging and discharging plans of the energy storage units within each region are recalculated to meet the new power exchange requirements. For regions undertaking power output tasks, the discharge power of the corresponding energy storage units is appropriately increased or the charging power is reduced. For regions receiving power compensation, the operating status of the energy storage units is adjusted accordingly. During the adjustment process, the energy storage capacity constraints, charging and discharging power constraints, and equipment safety constraints need to be checked simultaneously to ensure that the revised dispatch scheme still meets the operating requirements. Finally, the revised dispatch results of each region are integrated to form a comprehensive photovoltaic energy storage dispatch scheme for the entire power grid region.

[0083] After the integrated photovoltaic energy storage dispatch scheme is formulated, it enters the actual implementation phase. In practice, the integrated photovoltaic energy storage dispatch scheme is first converted into control commands that each photovoltaic energy storage node can directly execute, including energy storage charging power commands, energy storage discharging power commands, power adjustment timing, and cross-regional power exchange control parameters. Then, control commands are issued to each energy storage inverter, power control device, and regional control terminal through the dispatch control platform. Each device executes corresponding charging / discharging operations and power adjustment operations according to the received control commands and provides real-time feedback on the execution status. The dispatch platform continuously monitors the actual operating results in each region. When a deviation is found between the actual operating status and the planned status, the optimized dispatch process can be re-triggered for dynamic correction. By combining intra-regional optimization with cross-regional collaborative optimization, not only can the local absorption level of photovoltaic power be improved, but the resource complementarity between different regions can also be enhanced, thereby improving the overall operating efficiency and power supply stability of the entire power grid's photovoltaic energy storage system.

[0084] In this embodiment, all data involved in the calculation undergoes uniform dimensional processing before entering the model; power data is uniformly converted to kilowatts or megawatts; electrical energy data is uniformly converted to kilowatt-hours or megawatt-hours; time data is uniformly converted to seconds, minutes, or hours; temperature data is uniformly converted to degrees Celsius; and light intensity data is uniformly converted to standard units of received light power per unit area.

[0085] In practice, the original units corresponding to each data field are first identified; then a unit conversion mapping table is established; all data are converted to a unified unit according to the mapping relationship; after the unit conversion is completed, outlier detection and data verification are performed; then, the maximum-minimum normalization or standardization method is used to eliminate numerical differences between different units; finally, the processed data is input into the corresponding model for calculation and optimization analysis.

[0086] When the data sampling periods of multiple data sources are inconsistent, the preset scheduling period is first used as a unified time benchmark; for high-frequency data, downsampling is performed by resampling and time aggregation; for low-frequency data, interpolation is used to complete the data; finally, a data sequence with a unified time scale is formed, thereby ensuring the data consistency and calculation accuracy of subsequent prediction, optimization and collaborative scheduling calculation processes.

[0087] Regarding the models or algorithms and implementation methods used in all calculation and simulation processes in this embodiment: The photovoltaic prediction process in this embodiment is based on a time series prediction model; in specific implementation, the original data is first cleaned and normalized; then, an input feature vector is constructed; the mapping relationship between input and output is established using training samples; the model parameters are continuously optimized through an error back-correction mechanism; finally, the trained model is used to predict the photovoltaic power generation curve in the future scheduling cycle.

[0088] The load forecasting process in this embodiment is implemented using the same time series analysis framework as photovoltaic forecasting. First, a historical load database is established; then, daily, weekly, and seasonal cycle features are extracted; a forecasting model is constructed using historical load patterns; and finally, the load-side power consumption forecast curve for the future scheduling cycle is output.

[0089] The power grid area division process in this embodiment is based on a graph clustering algorithm. In specific implementation, electrical nodes are abstracted as graph nodes, and electrical lines are abstracted as graph edges. An edge weight matrix is ​​constructed using the inverse correlation coefficient. Then, the connection tightness between nodes is calculated. The graph partitioning algorithm is used to continuously adjust the node affiliation relationship so that the weight within the same area is maximized and the weight between areas is minimized. Finally, multiple photovoltaic power supply sub-regions are formed.

[0090] In this embodiment, the calculation process of the reverse correlation coefficient is implemented using statistical correlation analysis. First, the time-series net load curves of the two nodes are standardized. Then, the correlation between the two time series is calculated. Finally, their reverse complementary characteristics are taken as the basis for regional division. The stronger the reverse correlation, the more complementary and regulatory value the two nodes have.

[0091] In this embodiment, the asymmetric power volatility calculation process is implemented using the derivative analysis method. First, the change between adjacent moments of the regional time-series net power curve is calculated; then, the rate of change is calculated; after taking the absolute value of the rate of change, the effective value is calculated; finally, the asymmetric power volatility that can reflect the intensity of power fluctuation is obtained.

[0092] In this embodiment, the calculation process of the asymmetric power fluctuation rate is implemented using the second-order difference analysis method. First, the first-order rate of change is calculated; then, the difference between adjacent rates of change is calculated; the difference is used to characterize the fluctuation trend; and finally, the asymmetric power fluctuation rate is obtained.

[0093] In this embodiment, both the comprehensive optimization objective and the transient optimization objective are implemented using a constrained optimization algorithm. In specific implementation, the objective function is first established; then, energy storage capacity constraints, charging and discharging power constraints, and tie-line operation constraints are established; the decision variables are continuously updated using an iterative optimization mechanism; when the objective function converges, the optimal solution is output; and finally, the corresponding photovoltaic energy storage scheduling scheme is formed.

[0094] In this embodiment, the regional coordination relationship matrix is ​​constructed using a matrix mapping algorithm. First, the matching degree between the power margin and power deficit between any two regions is calculated. Then, a coordination evaluation index is formed by combining the line capacity constraint. Finally, a complete regional coordination relationship matrix is ​​constructed, and the optimal power complementary region combination is determined using a matrix search method.

[0095] Regarding the determination and selection of all preset models and preset threshold parameters in this embodiment: The source-side photovoltaic power prediction curve construction model in this embodiment is formed by joint training using historical photovoltaic power output data and meteorological monitoring data. In specific implementation, firstly, historical photovoltaic power generation, irradiance, ambient temperature, weather type, and other data for at least one year are collected; then, time alignment and missing value repair are performed according to a uniform time interval; further, irradiance variation characteristics, temperature variation characteristics, and seasonal cycle characteristics are extracted as model inputs, and historical actual power generation is used as model outputs. The prediction model is trained through cross-validation. During the training process, the model parameters are continuously adjusted to minimize the prediction error; when multiple models meet the accuracy requirements, the model with lower computational complexity and better generalization ability is selected as the final prediction model.

[0096] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims. Parts not described in detail in this specification are prior art known to those skilled in the art.

Claims

1. A multi-objective optimization scheduling method for distributed photovoltaic energy storage systems based on big data, characterized in that, Includes the following steps: Step S1: Energy storage data acquisition; acquire historical system operation data and meteorological monitoring data of the photovoltaic energy storage system in the current power grid area; there are several electrical nodes in the current power grid area, and several photovoltaic energy storage nodes exist in the photovoltaic energy storage system; Step S2: Energy storage area division; extract load power consumption characteristics of each node based on historical system operation data; Based on the load power consumption characteristics of each node, the source-side photovoltaic power prediction curve and the load-side power consumption prediction curve of each electrical node are constructed respectively; based on historical system operation data and meteorological monitoring data, the power grid area is divided into multiple photovoltaic power supply sub-areas; Step S3: Multi-objective optimization scheduling; Construct different photovoltaic energy storage multi-objective optimization models for different photovoltaic energy supply sub-regions; The photovoltaic energy storage multi-objective optimization model includes an energy storage objective function and an energy storage solution mode; Within a preset energy storage optimization scheduling period, for any photovoltaic energy supply sub-region, optimize and solve based on the energy storage objective function and energy storage solution mode in the photovoltaic energy storage multi-objective optimization model to obtain the corresponding regional photovoltaic energy storage scheduling scheme; Complete the scheduling of the photovoltaic energy storage system in the current power grid area according to the regional photovoltaic energy storage scheduling scheme corresponding to each photovoltaic energy supply sub-region.

2. The multi-objective optimization scheduling method for distributed photovoltaic energy storage systems based on big data as described in claim 1, characterized in that, The specific steps for dividing the power grid into regions based on historical system operation data and meteorological monitoring data include: Photovoltaic irradiance temperature characteristics are extracted based on meteorological monitoring data; the predicted photovoltaic power curve on the source side and the predicted power consumption curve on the load side of the same electrical node are calculated to obtain the node time-series net load curve; Using electrical nodes as graph nodes and the physical electrical lines between electrical nodes as edges, a basic physical connectivity graph of the power grid is constructed. For any two adjacent graph nodes with edges, the inverse correlation coefficient between the time-series net load curves of the adjacent graph nodes is calculated. The inverse correlation coefficient represents the degree of complementarity of the net load change trends between adjacent electrical nodes. The load asymmetry weight of the edge is assigned according to the degree of complementarity. Based on the physical connectivity graph of the power grid with asymmetric load weights, the graph segmentation constraint is to maximize the sum of weights within sub-regions and minimize the weights of cross-sub-region edges. A graph clustering algorithm is used to dynamically segment the current power grid region to obtain multiple photovoltaic power supply sub-regions.

3. The multi-objective optimization scheduling method for distributed photovoltaic energy storage systems based on big data according to claim 2, characterized in that, The specific steps for constructing different photovoltaic energy storage multi-objective optimization models based on different photovoltaic energy supply sub-regions include: For any photovoltaic power supply sub-region, the total predicted photovoltaic power on the source side and the total predicted electricity consumption on the load side of all electrical nodes in the photovoltaic power supply sub-region during the preset energy storage optimization scheduling period are calculated based on the predicted photovoltaic power curve on the source side and the predicted electricity consumption curve on the load side. The photovoltaic power supply characteristic coefficient and asymmetric power fluctuation rate are calculated based on the total predicted photovoltaic power on the source side and the total predicted electricity consumption on the load side. The photovoltaic energy supply characteristic coefficients are used to match the corresponding energy storage objective function for the photovoltaic energy supply sub-regions; the objective function of the energy storage objective function includes photovoltaic absorption, tie-line stability and energy storage lifetime. The solution mode for matching the energy storage to the photovoltaic power supply sub-region is based on the asymmetric power fluctuation rate; the energy storage solution mode includes conventional solution mode and special solution mode.

4. The multi-objective optimization scheduling method for distributed photovoltaic energy storage systems based on big data according to claim 3, characterized in that, The specific steps for matching the corresponding energy storage objective function to the photovoltaic power supply sub-region based on the photovoltaic power supply characteristic coefficients include: A preset source-load balance benchmark value is established; among which, the photovoltaic absorption sub-item, the tie-line stability sub-item, and the energy storage lifetime sub-item correspond to the photovoltaic absorption sub-item weight, the tie-line stability sub-item weight, and the energy storage lifetime sub-item weight, respectively; Calculate the directional offset between the photovoltaic power supply characteristic coefficient and the source-load balance benchmark value; determine the weights of the photovoltaic consumption component, the tie-line stability component, and the energy storage lifetime component based on the directional offset. An energy storage objective function is constructed based on the photovoltaic absorption component, the tie-line stability component, the energy storage lifetime component, and the corresponding weights of the photovoltaic absorption component, the tie-line stability component, and the energy storage lifetime component.

5. The multi-objective optimization scheduling method for distributed photovoltaic energy storage systems based on big data according to claim 4, characterized in that, The specific steps for matching the energy storage solution mode corresponding to the photovoltaic power supply sub-region based on the asymmetric power fluctuation rate include: Extract the node time-series net load curves of all electrical nodes in the photovoltaic power supply sub-region and spatially superimpose and fuse them to obtain the regional time-series net power curve; within a preset sliding time window, calculate the first time derivative of the regional time-series net power curve and take the effective value of the absolute value as the asymmetric power fluctuation rate; at the same time, calculate the second time derivative of the regional time-series net power curve to obtain the asymmetric power fluctuation rate. When the asymmetric power fluctuation rate exceeds a preset first threshold and the asymmetric power fluctuation change rate exceeds a preset second threshold, a special solution mode is matched; otherwise, a regular solution mode is matched.

6. The multi-objective optimization scheduling method for distributed photovoltaic energy storage systems based on big data according to claim 5, characterized in that, The specific steps for optimization based on the energy storage objective function and energy storage solution mode in the photovoltaic energy storage multi-objective optimization model include: Obtain operational status data for each photovoltaic energy storage node within the photovoltaic power supply sub-region; When the energy storage solution mode is the conventional solution mode, a comprehensive optimization objective is constructed by combining the operating status data and the energy storage objective function; the comprehensive optimization objective is iteratively solved to obtain the corresponding photovoltaic energy storage scheduling scheme. When the energy storage solution mode is a special solution mode, the power mutation state is obtained based on the operating status data; the tie-line stability enhancement factor is calculated based on the power mutation state; the tie-line stability enhancement factor is used to amplify and correct the tie-line stability component weight, and the photovoltaic absorption component weight and the energy storage lifetime component weight are compressed and corrected to obtain a new energy storage objective function; a transient optimization objective is constructed based on the new energy storage objective function; the transient optimization objective is iteratively solved to obtain the corresponding photovoltaic energy storage scheduling scheme.

7. The multi-objective optimization scheduling method for distributed photovoltaic energy storage systems based on big data according to claim 6, characterized in that, The specific steps for scheduling the photovoltaic energy storage system in the current power grid area according to the regional photovoltaic energy storage scheduling plan corresponding to each photovoltaic energy supply sub-region include: Based on the regional photovoltaic energy storage scheduling scheme corresponding to each photovoltaic energy supply sub-region, calculate the regional power surplus or regional power deficit of each photovoltaic energy supply sub-region within the preset energy storage optimization scheduling cycle; construct a regional coordination relationship matrix between photovoltaic energy supply sub-regions based on the regional power surplus and regional power deficit; and determine the target photovoltaic energy supply sub-region combination with power complementarity based on the regional coordination relationship matrix. For any target photovoltaic power supply sub-region combination, priority is given to using regional power surplus to coordinate regional power deficit across regions, resulting in cross-regional power coordination results; corresponding regional collaborative dispatch instructions are generated based on the cross-regional power coordination results; regional photovoltaic energy storage dispatch schemes within each photovoltaic power supply sub-region are synchronously modified based on the regional collaborative dispatch instructions, resulting in a comprehensive photovoltaic energy storage dispatch scheme; and the optimized dispatch of the photovoltaic energy storage system in the current power grid region is completed based on the comprehensive photovoltaic energy storage dispatch scheme.

Images