Distribution network distributed photovoltaic centralized scheduling method and distribution and storage system

Abstract

The invention discloses a distribution network distributed photovoltaic centralized scheduling method and a distribution and storage system, and relates to the technical field of optical storage intelligent scheduling, and the method comprises the steps: obtaining regional photovoltaic output prediction data and load demand prediction data, and generating a global scheduling index through combining the topological data of a power distribution network; determining a short-term optimal scheduling index of the region according to the global scheduling index; and generating a regional control instruction based on the short-term optimal scheduling index, and executing the instruction to realize optimal scheduling of cross-regional photovoltaic and energy storage. According to the method, photovoltaic output and load uncertainty can be flexibly dealt with, the problems of good power grid power imbalance and voltage fluctuation are effectively avoided, meanwhile, cross-regional scheduling is achieved through global scheduling indexes, photovoltaic consumption can be maximized, light abandoning can be reduced, energy storage utilization can be optimized, finally, cross-regional reliable scheduling is achieved on the premise that the energy storage utilization rate is low, and the method is suitable for large-scale popularization and application. And the cross-regional scheduling efficiency and reliability are improved.

Description

Technical Field

[0001] This application relates to the field of intelligent photovoltaic and energy storage scheduling technology, specifically a centralized scheduling method and energy storage system for distributed photovoltaic power grids. Background Technology

[0002] With the energy transition driving the large-scale integration of distributed photovoltaic (PV) and other renewable energy sources into the distribution network, PV output is highly volatile due to sunlight exposure, and load is becoming increasingly complex due to user behavior and the widespread adoption of new electrical equipment. The superposition of uncertainties at both the source and load ends poses challenges to the safe and stable operation and dispatch of the distribution network. Traditional PV-storage dispatch relies on deterministic optimization and accurate prediction, but these strategies are prone to failure when faced with rapid fluctuations and abrupt changes in source and load. Furthermore, the low prediction accuracy makes it difficult to accurately capture the fluctuating characteristics of PV output and load demand. Conservative dispatch limits PV utilization or may lead to curtailment due to improper dispatch, reducing energy efficiency. Some solutions introduce technologies such as reinforcement learning, but their ability to handle rapid fluctuations and abrupt changes in source and load in the distribution network is limited, failing to ensure that dispatch strategies maintain good performance in the face of diverse and extreme uncertainties. Summary of the Invention

[0003] The purpose of this application is to address the problem that conventional photovoltaic-storage scheduling methods struggle to balance power output and load demand fluctuations, as well as the poor inter-regional scheduling effect caused by low energy storage utilization efficiency, in cross-regional scenarios with diverse power sources and extremely uncertain loads. A centralized scheduling method for distributed photovoltaic power in distribution networks and a distribution-storage system are proposed. This method determines global scheduling indicators based on photovoltaic power output and load demand forecast data within a region, and then obtains short-term optimal scheduling indicators based on these global indicators. This achieves reliable cross-regional scheduling while improving energy storage utilization efficiency, thereby enhancing the efficiency and reliability of cross-regional scheduling.

[0004] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows: In a first aspect, embodiments of this application provide a method for centralized dispatching of distributed photovoltaic power in a distribution network. The method includes: acquiring regional photovoltaic power output forecast data and load demand forecast data, and generating a global dispatch index in combination with distribution network topology data; determining a short-term optimal dispatch index for the region based on the global dispatch index; generating a regional control command based on the short-term optimal dispatch index, and then executing the command to achieve optimal dispatching of photovoltaic power and energy storage across regions.

[0005] In this scheme, global dispatch indicators are formulated based on the photovoltaic output forecast data and load demand forecast data of each region. These global dispatch indicators are then combined with the actual data of the distribution network to generate the global dispatch indicators. Furthermore, the short-term optimal dispatch indicators are obtained, which are the optimal regional dispatch plans. This approach can flexibly respond to uncertainties in photovoltaic output and load, effectively avoid problems such as power imbalance and voltage fluctuations in the power grid, and ensure the stable operation of the distribution network. At the same time, the global dispatch indicators enable cross-regional dispatch, which can maximize photovoltaic absorption, reduce curtailment, and optimize energy storage utilization. This achieves full automation and intelligence of the dispatch process, meeting the rapid dispatch needs of distributed photovoltaic centralized energy storage in the distribution network.

[0006] Preferably, the step of acquiring regional photovoltaic power output forecast data and load demand forecast data, and generating global scheduling indicators by combining them with distribution network topology data, includes: acquiring power distributed resource monitoring data in each region, and making short-term forecasts of regional photovoltaic power output and load demand by combining historical monitoring data and meteorological data to obtain regional photovoltaic power output forecast data and load demand forecast data; and generating global scheduling indicators based on the photovoltaic power output forecast data and load demand forecast data combined with distribution network topology data.

[0007] Preferably, the power distributed resource monitoring data includes at least photovoltaic power plant monitoring data, energy storage device monitoring data, load-side operating parameters, and power data from distribution network line monitoring points; by integrating the photovoltaic power plant monitoring data, energy storage device monitoring data, load-side operating parameters, power data from distribution network line monitoring points, and corresponding historical monitoring data and meteorological data from each region, photovoltaic output characteristic sets and load demand characteristic sets for each region are extracted; the regional photovoltaic output characteristic sets are used as input to the output prediction model to obtain short-term photovoltaic output prediction data for the region; and the load demand characteristic sets are used as input to the load demand prediction model to obtain short-term load demand prediction data for the region.

[0008] Preferably, the step of generating global scheduling indicators based on the photovoltaic output forecast data and load demand forecast data combined with distribution network topology data includes: integrating short-term photovoltaic output forecast data and short-term load demand forecast data from various regions, performing data preprocessing, and combining the data with distribution network topology data to form a global dataset; constructing a multi-regional collaborative optimization model based on global scheduling objectives and constraints, solving the model based on the global dataset to obtain the global optimization result; and converting the global optimization result into global scheduling indicators.

[0009] Preferably, the step of constructing a multi-regional collaborative optimization model based on global scheduling objectives and constraints, and obtaining the global optimization result by solving the model based on the global dataset, includes: constructing a regional collaborative objective function to minimize the regional collaborative operation cost, and establishing constraints based on the internal operation characteristics of the regions and the operational constraints between regions; determining the decision variables of the multi-regional collaborative optimization model, including continuous variables and binary variables, and configuring the objective function, constraints, and solver parameters of the model; and solving the multi-regional collaborative optimization model by running the solver based on the global dataset to obtain the global optimization result of the decision variables.

[0010] Preferably, determining the short-term optimal scheduling index for a region based on the global scheduling index includes: constructing multi-layered constraints based on short-term forecast data of photovoltaic output and short-term forecast data of load demand for each region, combined with the global scheduling index; constructing regional scheduling targets based on global electricity purchase cost, curtailment of solar power, power fluctuations, and energy storage losses, and constructing a regional scheduling optimization model by integrating the multi-layered constraints; and solving the regional scheduling optimization model to obtain the short-term optimal scheduling index for the region.

[0011] Preferably, the step of constructing multi-layered constraints based on short-term photovoltaic power output forecast data and short-term load demand forecast data for each region, combined with the global scheduling indicators, includes: decomposing the global scheduling indicators to obtain global decision constraints; determining regional scheduling constraints based on the short-term photovoltaic power output forecast data and short-term load demand forecast data for each region; and establishing multi-layered constraints based on the global decision constraints and the regional scheduling constraints.

[0012] Preferably, the global scheduling indicators include at least the cross-regional power exchange power indicator, the regional energy storage plan indicator, and the regional photovoltaic absorption rate indicator; the cross-regional power exchange power, the regional energy storage plan, and the regional photovoltaic absorption rate are constructed in a regional constraint manner based on boundary values ​​to obtain the corresponding cross-regional power exchange constraints, energy storage global trajectory constraints, and regional photovoltaic absorption constraints; the cross-regional power exchange constraints, energy storage global trajectory constraints, and regional photovoltaic absorption constraints serve as global decision constraints.

[0013] Preferably, the step of generating regional control instructions based on the short-term optimal scheduling index includes: comparing the regional power distributed resource monitoring data with the short-term optimal scheduling index to obtain the scheduling deviation; using a rule base to determine the adjustment plan based on the magnitude and type of the scheduling deviation, thereby obtaining the regional control instructions.

[0014] The beneficial effects of this application are: Global dispatch indicators are formulated by combining regional photovoltaic output forecast data and load demand forecast data. These global dispatch indicators are then combined with actual distribution network data to generate short-term optimal dispatch indicators, i.e., the optimal regional dispatch plan. This approach can flexibly respond to uncertainties in photovoltaic output and load, effectively avoid problems such as power imbalance and voltage fluctuations in the power grid, and ensure the stable operation of the distribution network. At the same time, the global dispatch indicators enable cross-regional dispatch, which can maximize photovoltaic absorption, reduce curtailment, and optimize energy storage utilization, thereby achieving global resource utilization and realizing full automation and intelligence of the dispatch process. This meets the rapid dispatch needs of distributed photovoltaic centralized energy storage in the distribution network. Attached Figure Description

[0015] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings. The drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings.

[0016] Figure 1 A flowchart of a centralized dispatching method for distributed photovoltaic power generation in a distribution network is provided for an embodiment of this application.

[0017] Figure 2 This is a flowchart illustrating the modeling and solving of a mixed-integer linear programming problem, as provided in an embodiment of this application.

[0018] Figure 3 This is a schematic diagram of a distribution and storage system module for centralized dispatching of distributed photovoltaic power in a distribution network, provided as an embodiment of this application.

[0019] Figure 4 This is a flowchart illustrating the periodic dynamic adjustment of area control commands, as provided in an embodiment of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely one preferred embodiment of this application and are only used to explain this application. They do not limit the scope of protection of this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] Example 1: As Figure 1 As shown, a method for centralized dispatching of distributed photovoltaic power in a distribution network includes the following steps: S1. Obtain regional photovoltaic power output forecast data and load demand forecast data, and generate global scheduling indicators by combining them with distribution network topology data.

[0022] As an optional implementation, step S1 includes: S11. Obtain monitoring data of distributed power resources in each region, and combine historical monitoring data and meteorological data to make short-term forecasts of regional photovoltaic output and load demand, thereby obtaining regional photovoltaic output forecast data and load demand forecast data. S12. Generate global scheduling indicators based on the photovoltaic output forecast data and load demand forecast data combined with the distribution network topology data.

[0023] Furthermore, in this embodiment, the regions are divided based on feeders. Each region includes a photovoltaic power station, energy storage equipment, load-side users, and distribution network line monitoring points. The boundaries between regions are defined by distribution network line connection points.

[0024] Specifically, the regional division should closely follow the physical characteristics of the distribution network and the practical needs of dispatching. Based on the feeder division, the power supply range of the 10kV feeder is used as the dividing boundary. Each feeder corresponds to a fixed power supply area (such as a 10kV feeder covering a residential area, a surrounding industrial park, several distributed photovoltaic power stations and a centralized energy storage device). The feeders are separated by the distribution network line connection point. This can ensure that the area includes photovoltaic power stations, energy storage devices, load-side users and distribution network line monitoring points, while avoiding the data collection dispersion and the chaotic execution of dispatching instructions caused by cross-feeder division.

[0025] In some examples, acquiring real-time and reliable data from photovoltaic power plants, energy storage devices, load-side equipment, and distribution network monitoring points in various regions forms the basis for upper-level decision-making. Specifically, on the photovoltaic side, multiple sensors and irradiance meters collect data on the output power, DC voltage / current, module temperature, and irradiance of the photovoltaic array. On the energy storage side, the BMS (Battery Management System) collects data on the SOC (State of Charge), individual cell voltage, total voltage, charging / discharging current, and battery temperature of the energy storage batteries; the PCS (Bidirectional Converter System) collects data on the charging / discharging power and reactive power output of the energy storage. On the load side, smart meters and load controllers collect real-time power, voltage, current, and electricity consumption periods of important loads. On the distribution network side, distribution network monitoring terminals collect data on the voltage, frequency, line current, and power factor of distribution network points. This real-time data effectively reflects the operating status of the distributed photovoltaic centralized system in the distribution network. By monitoring and comparing real-time data, feedback on dispatch execution and improvement can be obtained, thereby dynamically adjusting dispatch plans and control commands.

[0026] As an optional implementation, the power distributed resource monitoring data includes at least photovoltaic power plant monitoring data, energy storage device monitoring data, load-side operating parameters, and power data from distribution network line monitoring points; By integrating the photovoltaic power plant monitoring data, energy storage device monitoring data, load-side operating parameters, power data from distribution network line monitoring points, and corresponding historical monitoring data and meteorological data from various regions, the photovoltaic output characteristic set and load demand characteristic set of each region are extracted. Using the regional photovoltaic power output feature set as input to the power output prediction model, short-term photovoltaic power output prediction data for the region is obtained; Using the load demand feature set as input to the load demand forecasting model, short-term load demand forecast data for the region is obtained.

[0027] In some embodiments, photovoltaic (PV) output is affected by environmental factors, requiring the capture of the nonlinear relationship between irradiance and PV output. Besides collecting real-time data on PV, energy storage, load, and distribution network within the region, historical PV output and meteorological data (such as temperature, wind speed, and cloud cover) from the previous 24 hours are also needed as core data. This data is cleaned and normalized, representing it as a feature dataset X, where X = [PV output from the previous 24 hours, irradiance from the previous 24 hours, temperature from the previous 24 hours, and cloud cover from the previous 24 hours]. Subsequently, a regional PV output prediction model is constructed based on Seq2Seq and TimeDistributed. Seq2Seq (sequence-to-sequence model) automatically learns the mapping relationship from input to output through end-to-end training; TimeDistributed (time distribution layer) allows the reuse of the same network layer at different time steps.

[0028] Specifically, for the input feature dataset, a time-series sample format is constructed with a time granularity of 15 minutes (time steps: 24 hours / 15 minutes = 96, feature count: 4) to obtain the input features required for multi-step prediction. The encoder of the model extracts the time-series features of the input features, focusing on capturing the source-load patterns and cloud influences of the previous 24 hours. Then, the decoder generates a photovoltaic power output sequence for the next 16 steps (16 * 15 minutes = 4 hours). The output layer then performs a fully connected layer on each of the 16 steps to output the photovoltaic power output for each step, ultimately obtaining a short-term prediction of regional photovoltaic power output. During model training, the error loss between the predicted output and the true value is calculated to optimize the overall prediction accuracy. In particular, to enhance the model's predictive ability under extreme weather conditions, the following strategies are adopted: actively selecting extreme weather samples based on historical data, or constructing such samples through data augmentation techniques, to expand the training set and improve the model's ability to learn patterns in extreme scenarios. Meanwhile, during the inference process, extreme weather conditions are identified based on irradiance and cloud cover in the input features, and higher weights are assigned to the corresponding prediction errors in the loss function, thereby effectively improving the model's predictive adaptability and robustness to extreme weather.

[0029] Understandably, load demand forecasting is dominated by user behavior and social factors, requiring a focus on capturing the impact of time cycles and external events. Therefore, the collected real-time data also needs to include historical load data and time characteristics to reflect periodic patterns such as weekday evening peak hours and weekend load troughs, as well as other influencing factors like temperature. By organizing this core data, a regional load demand characteristic dataset is obtained, which can be represented as Y = [previous week's load, hours, day of the week, whether it's a holiday, previous week's temperature]. Similarly, to construct a regional load demand forecasting model, the characteristic dataset needs to be split by hour, indicating the hour (0-23), day of the week (0-6), whether it's a weekend (1=yes, 0=no), and the corresponding temperature. A time-series sample format is then constructed (time step 7 * 24 hours = 168, feature number 5) to obtain the input features.

[0030] Specifically, the regional load demand forecasting model uses a three-layer LSTM core network structure with an attention mechanism. The first layer uses a bidirectional LSTM to learn periodic patterns; the second layer uses LSTM and an attention mechanism to automatically increase the weights of "holidays" and "temperature," enhancing the influence of event features; the third layer achieves nonlinear mapping through a fully connected layer and an output layer, transforming time-series features into specific load values; ultimately, short-term forecast data for regional load demand is obtained. In particular, the input features employ a sliding window method to implement a rolling forecasting mechanism, i.e., features from hours 1-168 predict the load demand for hour 169, and features from hours 2-169 predict the load demand for hour 170, addressing the issue of accuracy decay in long-term forecasts.

[0031] As an optional implementation, step S12 includes: S121. Integrate short-term forecast data of photovoltaic output and short-term forecast data of load demand in various regions, perform data preprocessing, and combine them with distribution network topology data to form a global dataset. S122. Construct a multi-regional collaborative optimization model based on global scheduling objectives and constraints, and solve the model using the global dataset to obtain the global optimization result; S123. Convert the global optimization result into a global scheduling index.

[0032] As an optional implementation, step S122 includes: The objective function for regional collaboration is constructed to minimize the operational cost of regional collaboration, and constraints are established based on the operational characteristics within the region and the operational constraints between regions. Determine the decision variables for the multi-regional collaborative optimization model, including continuous and binary variables, and configure the model's objective function, constraints, and solver parameters; Based on the global dataset, the multi-regional collaborative optimization model is solved by running a solver to obtain the global optimization results of the decision variables.

[0033] Specifically, short-term forecast data of photovoltaic power output and short-term forecast data of load demand from various regions are integrated and preprocessed to ensure that the time scale and time range of the forecast data are completely consistent across all regions. Distribution network topology data is obtained based on the topology of the distribution network, which includes: the location of each region, the interconnections between regions and their transmission capacity limits, and their connection points to the main grid.

[0034] In this embodiment, considering that inter-regional scheduling enables regions with surplus photovoltaic power to supply electricity to regions with high loads, thereby achieving global resource utilization, the objective function for regional coordination needs to minimize the total operating cost. The objective function is as follows: ; ; ; ; ; ; in, This represents the total operating cost. This represents the total cost of electricity purchase. This represents the total revenue from selling electricity. This indicates a total penalty for abandoning light. Indicates total energy storage loss. Indicates the unit price of electricity. This represents the total power purchased from the main grid. Indicates time interval, Indicates the unit price of electricity sold. This represents the total power sold to the main grid. This indicates a penalty for abandoning light by the unit. Indicates the region The power of abandoned light, Indicates the total number of regions. This represents the cost per unit of charge / discharge power lost. Indicates the area The charging / discharging power.

[0035] Furthermore, the constraints must include both intra-region and inter-region constraints. Specifically, the intra-region constraints include power balance constraints, which can be expressed as: ; in, , Indicates the region Predicted photovoltaic power output Indicates the area Switching power with the main network Indicates the area With all other regions The switching power, Indicates the area The predicted load demand power; energy storage operation constraints can be expressed as: ; ; ; in, Indicates the area exist Charge state during a time period Indicates the area exist Charge state during a time period Indicates the area The rated capacity of energy storage, Indicates the area Maximum charging power of energy storage Indicates the area Maximum discharge power of energy storage.

[0036] The transmission capacity constraints of the tie lines between regions must be met, which can be expressed as: ; in, For the region With the region The maximum power allowed to be transmitted on the interconnecting lines; the power constraints for switching between the area and the main network can be expressed as: ; in, For the region Maximum power at the connection point with the main network.

[0037] It should be noted that before constructing a multi-regional collaborative optimization model, it is necessary to define the model's decision variables, which include continuous variables (such as...). , , (etc.) and binary variables (such as variables that identify the charge and discharge states of energy storage). ), where binary variables Indicates the area exist Energy storage charging and discharging status during a given period of time. Indicates the charging status of energy storage. This indicates the discharge state of the energy storage, which is to ensure that the storage cannot be in a charging and discharging state at the same time.

[0038] Furthermore, such as Figure 2 As shown, a multi-regional collaborative optimization model is constructed based on decision variables, objective function, and constraints. Then, a mixed-integer linear programming solver is selected, and solver parameters (such as solution time limits, tolerance gaps, etc.) are configured. During the solution process, the solver uses the branch and bound method to solve the model. After the solution is completed, the solution status needs to be determined: if the solution fails or the result is infeasible, the variable definitions, parameter settings, and constraints of the model need to be further diagnosed. Common problems include overly tight constraints, logical conflicts between constraints, and incorrect initial values ​​of variables. The model needs to be adjusted for the specific problem and the solution needs to be restarted. If the solution is successful, a global optimization result that satisfies the objective function and all constraints can be obtained.

[0039] Furthermore, the global optimization result is the optimal value of all decision variables, which needs to be transformed into a global scheduling index, such as from... The variables are extracted to determine the planned power transmission and reception between any two regions in each future time period, yielding the cross-regional power exchange power index, which is the core of inter-regional dispatching; from Extract the energy storage plan for each region from the variables. The trajectory is used to obtain regional energy storage plan targets. Furthermore, to maximize the utilization efficiency of regional photovoltaic resources while reducing dependence on the main grid and dispatch pressure, regional photovoltaic absorption rate targets can be preset to limit the proportion of photovoltaic power generation effectively absorbed within the region. These global dispatch targets will be distributed to each region to guide dispatching within the region.

[0040] Understandably, a multi-regional collaborative optimization model is constructed based on mixed integer linear programming (MILP). This model is a single-objective optimization model, and the global optimization result is obtained through the solver, which intuitively displays the power flow situation in each region and at each time point.

[0041] In other embodiments, the global scheduling metrics are dynamically updated by calculating operational deviations. These global scheduling metrics for multi-regional collaboration are not fixed or unique in the long term; the obtained global scheduling metrics need to be dynamically adjusted through real-time monitoring to avoid scheduling failures due to prediction errors.

[0042] As one implementation, after the global scheduling indicators are issued to the regions, the actual operating data of the scheduled regions is obtained at specific intervals and compared with the target values ​​of the scheduling indicators to identify the operational deviations. For example, if the actual photovoltaic output in region A is much lower than the predicted photovoltaic output, the energy storage in that region cannot complete the expected charging and deliver the required amount of electricity to region DB. A dynamic adjustment strategy is implemented based on the range of deviation. When the deviation is small (e.g., ≤5%), an optimal scheduling indicator is formulated within the region, and fine-tuned using dynamically adjusted regional control commands to ensure that the region's operating status is as close as possible to the expected scheduling target. When the deviation is large (e.g., >5%), a multi-regional collaborative optimization model is triggered for re-optimization, resulting in new global scheduling indicators that are issued to the corresponding regional scheduling layer. By monitoring and closed-loop adjusting each region, the feasibility and effectiveness of the global scheduling indicators are ensured, achieving optimal resource allocation.

[0043] S2. Determine the short-term optimal scheduling index for the region based on the global scheduling index.

[0044] As an optional implementation, step S2 includes: S21. Based on the short-term forecast data of photovoltaic output and short-term forecast data of load demand in each region, and in conjunction with the global scheduling index, construct multi-layer constraint conditions. S22. Construct regional dispatch targets based on global power purchase costs, curtailment of solar power, power fluctuations, and energy storage losses, and integrate the aforementioned multi-layered constraints to construct a regional dispatch optimization model; S23. Solve the regional scheduling optimization model to obtain the short-term optimal scheduling index of the region.

[0045] In this embodiment, after each region obtains the global scheduling indicators, it needs to be converted into regional scheduling constraints. At the same time, regional scheduling also ensures the safe operation of equipment within the region. Therefore, the constraints of regional scheduling include multi-level constraints at both the global and regional levels, ensuring that the regional scheduling plan is both in line with global coordination and adapted to local realities.

[0046] As an optional implementation, step S21 includes: S211. Decompose the global scheduling index to obtain global decision constraints; S212. Determine regional dispatch constraints based on short-term forecast data of regional photovoltaic output and short-term forecast data of load demand. S213. Establish multi-layered constraint conditions based on the global decision constraints and the regional scheduling constraints.

[0047] As an optional implementation, the global scheduling indicators include at least the cross-regional power exchange power indicator, the regional energy storage plan indicator, and the regional photovoltaic absorption rate indicator; The cross-regional power exchange power, regional energy storage plan and regional photovoltaic absorption rate are constructed in a regional constraint manner based on the boundary values ​​to obtain the corresponding cross-regional power exchange constraints, energy storage global trajectory constraints and regional photovoltaic absorption constraints. Cross-regional power exchange constraints, global energy storage trajectory constraints, and regional photovoltaic consumption constraints serve as global decision constraints.

[0048] In this embodiment, the global scheduling indicators may include indicators such as inter-regional power exchange power, regional energy storage plans, and regional photovoltaic absorption rates. These global scheduling indicators are broken down into quantifiable regional constraints to ensure no deviation from the global objective. Specifically, based on the inter-regional power exchange power, if a region is a power-transmitting region in the global scheduling indicators, since the global scheduling indicators already limit the inter-regional power exchange power to not exceed the maximum power allowed for transmission via inter-regional tie lines, it is only necessary to limit the actual regional interaction power to not exceed the inter-regional power exchange power. Similarly, if a region is a power-receiving region, then the actual inter-regional interaction power must not be lower than the inter-regional power exchange power. This allows regional decisions to be fine-tuned based on predicted data within the region, while ensuring that the fine-tuning results do not affect the safe operation of the power-transmitting or power-receiving regions.

[0049] Understandably, based on the regional energy storage plan, regional energy storage constraints are derived. The region must ensure that its actual regional energy storage charging and discharging power and actual energy storage SOC are as consistent as possible with the global dispatch targets, allowing for fluctuations of ±5%, to avoid excessive deviation from the global objectives. Furthermore, since the regional photovoltaic (PV) grid integration rate can be transformed into the relationship between regional curtailment and short-term forecasts of regional PV output, for example, if the global dispatch targets require a regional PV grid integration rate ≥ 95%, then the regional PV grid integration constraint is: curtailment ≤ short-term forecast of regional PV output × 5%. These global decision constraints, extracted from the global dispatch targets, effectively ensure that the regional dispatch plan does not deviate from the global objectives.

[0050] At the same time, by combining short-term forecast data of regional photovoltaic output and load demand with the parameters of local equipment operation, constraints are set to ensure the safe operation within the region, which complement the global decision-making constraints.

[0051] Specifically, the step of establishing multi-level constraint conditions based on the global decision constraints and the regional scheduling constraints includes: The regional dispatch constraints include at least energy storage operation constraints, power balance constraints, and distribution network security constraints. The global decision constraints and the regional scheduling constraints are merged, and multi-layered constraint conditions are formed according to the priority of each constraint.

[0052] Specifically, distribution network safety constraints are designed to prevent line overload and voltage exceeding limits, and are expressed as follows: ; ; in, Indicates in Time period line transmission power, Indicates the maximum rated power transmitted by the line. Indicates that the node is Voltage during time period Indicates the minimum voltage at the node. This represents the maximum voltage at the node. The energy storage operation constraints and power balance constraints within the regional dispatch constraints have already been used when setting global dispatch indicators; their reuse here is to adapt to the real-time operational status of the region. The constraint boundaries can be adjusted based on regional operation. Furthermore, other regional dispatch constraints, such as photovoltaic ramp-up constraints and grid purchase cost constraints, can be determined to further limit equipment safety and economic efficiency.

[0053] Furthermore, global decision-making constraints and regional scheduling constraints are merged, with clear priorities to avoid constraint conflicts. In this embodiment, regional energy storage operation constraints and distribution network security constraints ensure regional safe operation and are given the first priority; global decision-making constraints are given the second priority, while cross-regional power exchange constraints, energy storage global trajectory constraints, and regional photovoltaic consumption constraints must be strictly followed under the premise of meeting security constraints; other regional scheduling constraints are given the third priority and can be flexibly adjusted when the first two priorities are met, thus obtaining multi-layered constraint conditions.

[0054] As an optional implementation, in step S22, the regional scheduling target is represented as follows: ; ; ; in, and The objective function used is the same as that used when obtaining the global scheduling metrics, so it will not be described again. For the predicted photovoltaic power output of the region, To make actual contributions to regional photovoltaic power, To obtain a positive function, This represents the actual inter-regional interaction power. This refers to cross-regional power interaction. Furthermore, weighting coefficients can be assigned to the four sub-objective functions according to different scenarios. For example, in scenarios emphasizing photovoltaic power consumption, weighting coefficients can be assigned to... Assign a higher weight to the sub-objective function than to other sub-objective functions to highlight its importance.

[0055] Furthermore, based on the multi-layered constraints and optimization objectives, a regional scheduling optimization model is constructed and solved using mixed-integer linear programming (MILP) to obtain the optimal scheduling indices for short-term photovoltaic (PV) and energy storage in the region. Specifically, scheduling optimization variables are set, such as PV output, energy storage charging / discharging power, energy storage SOC, inter-regional exchange power, and grid exchange power. The multi-layered constraints and optimization objectives are transformed into a mixed-integer linear programming form, and the solver is called to obtain the optimal values ​​corresponding to the scheduling optimization variables. The results are then converted into the optimal scheduling indices for short-term PV and energy storage in the region.

[0056] In this embodiment, a regional scheduling target is constructed based on the global electricity purchase cost, curtailment of solar power, power fluctuations, and energy storage losses. The regional scheduling optimization model is constructed by integrating multi-layer constraints. The aim is to achieve both the goal of low global energy consumption and low curtailment of solar power, and the need for low regional costs and stable operation.

[0057] S3. Generate regional control instructions based on the short-term optimal scheduling index, and then execute the instructions to achieve optimal scheduling of photovoltaic and energy storage across regions.

[0058] As an optional implementation, generating area control instructions based on the short-term optimal scheduling index includes: The scheduling deviation is obtained by comparing the regional power distributed resource monitoring data with the short-term optimal scheduling index; Based on the magnitude and type of the scheduling deviation, a rule base is used to determine and adjust the plan, thereby obtaining regional control instructions.

[0059] Specifically, the optimal scheduling indicators for short-term photovoltaic and energy storage in a region are formulated based on their corresponding short-term forecast data. In actual operation, forecast errors and unexpected interferences (such as the sudden start of a large motor) are inevitable. Therefore, in this embodiment, the optimal scheduling indicators for short-term photovoltaic and energy storage in a region are fine-tuned based on the latest real-time situation to generate a safe and reliable regional control command that can be directly executed.

[0060] In some examples, specific planned values ​​are extracted from the optimal scheduling indicators of regional short-term photovoltaic (PV) and energy storage, such as energy storage charging / discharging power and PV output within a certain time period. Then, based on data from monitoring points of PV power plants, energy storage devices, load side, and distribution network lines within the region, these values ​​are compared with the planned values ​​corresponding to the current time point to calculate the deviation. The deviation can be categorized into PV output deviation, energy storage power deviation, and energy storage SOC deviation. Depending on the type and magnitude of the deviation, a rule base or lightweight real-time optimization is used to determine how to adjust the plan. When photovoltaic (PV) output deviation occurs, the primary task is to maintain the real-time power balance of the region. Priority is given to adjusting energy storage output to compensate for the deviation. The energy storage discharge power increases the corresponding PV output deviation, and it is verified that the increased energy storage discharge power does not exceed the maximum charge / discharge power range of the energy storage, thus obtaining a regional energy storage control command. If energy storage power deviation occurs, the corresponding energy storage power deviation can be increased or decreased by adjusting the energy storage charge / discharge power, while ensuring that the line voltage remains within a safe range. Simultaneously, the voltage of the PV inverter can be controlled, thus obtaining a regional control command. If energy storage SOC deviation occurs, under the premise of satisfying power balance and voltage constraints, the energy storage power command is fine-tuned to gradually return it to the planned trajectory over a period of time, thus obtaining a regional control command. All regional control commands are integrated and uniformly scheduled and operated through the execution module.

[0061] Example 2, as Figure 3 As shown in the embodiments of this application, a distribution and storage system corresponding to a distribution network distributed photovoltaic centralized dispatching method is also provided, which includes: The forecasting module is used to acquire regional photovoltaic power output forecast data and load demand forecast data, and combine them with distribution network topology data to generate global scheduling indicators. The control module is used to determine the short-term optimal scheduling index for the region based on the global scheduling index. The execution module is used to generate regional control instructions based on the short-term optimal scheduling index, and then execute the instructions to achieve optimal scheduling of photovoltaic and energy storage across regions.

[0062] In some possible embodiments, the prediction module includes a sensing execution unit and a short-term prediction unit. The sensing execution unit is deployed at the locations of physical equipment such as photovoltaic power plants, energy storage devices, load-side devices (such as industrial plants, residential distribution boxes), and distribution network line monitoring points. It achieves real-time data collection through hardware such as sensors, smart meters, and data acquisition terminals. The real-time data includes data from photovoltaic power plants, energy storage devices, load-side devices, and distribution network line monitoring points.

[0063] Specifically, the sensing and execution unit collects data from photovoltaic power stations, energy storage devices, load side and distribution network line monitoring points in the area through physical devices, and transmits it to the short-term forecasting unit. Based on the collected real-time data, combined with historical data and meteorological conditions in the area, the short-term forecasting unit predicts the photovoltaic output and load conditions in the area in the near future, and obtains short-term forecast data of photovoltaic output and load demand in the area.

[0064] In some possible embodiments, the distribution and storage system further includes a cloud-based decision-making module; The cloud-based decision-making module collects short-term forecasts from multiple regions, formulates multi-regional collaborative scheduling strategies, and obtains global scheduling indicators which are then distributed to the corresponding regions. The regional scheduling layer receives the global scheduling indicators from the cloud-based decision-making module, transforms them into corresponding constraints, and then optimizes the scheduling plan based on the short-term forecast data and actual operation data within the region. This ensures that the region achieves optimal scheduling under global collaborative optimization, and the obtained short-term optimal scheduling indicators are then sent to the execution module.

[0065] Furthermore, the execution module compares the short-term optimal scheduling index of the region with the actual data of the region, generates the corresponding regional control command, and executes the command to complete the scheduling, realizing full automation and intelligence of the scheduling process, and meeting the rapid scheduling needs of distributed photovoltaic centralized energy storage in the distribution network.

[0066] Furthermore, the area control commands are repeatedly issued according to a preset cycle, and dynamic adjustments are made by receiving execution feedback from the area control commands.

[0067] Specifically, such as Figure 4 As shown, the execution module has a preset control cycle (usually on the minute level) to ensure rapid response and effective execution of regional control commands. Within each control cycle, after the command is issued, the module receives real-time feedback signals from the sensing and execution unit, monitors the actual execution status of the command, and collects real-time data on photovoltaic output, energy storage status, load demand, and distribution network operation within the region after the command's action. This data is then used to assess whether the expected scheduling objectives have been achieved. If the objectives are not met, the module compares the data from monitoring points of photovoltaic power plants, energy storage devices, load sides, and distribution network lines within the region after command execution with the planned values ​​of the optimal scheduling indicators. Based on the deviation, the command is adjusted, and a new regional control command is issued. The next cycle continues to receive command execution feedback, thereby continuously optimizing the control strategy to form a closed-loop adjustment, achieving dynamic adjustment of regional control commands and continuously improving the real-time performance and effectiveness of the control commands.

[0068] In some examples, the short-term forecast data for regions A and B are as follows: The predicted total over-generation of photovoltaic power in region A during time period a is... The predicted total load gap for area B during time period b is Based on short-term forecast data for regions A and B, the cloud-based decision-making module formulates a global decision to transmit power from region A to region B. Energy storage in Zone B needs to be charged to 80% of its capacity before peak load. Therefore, based on local short-term photovoltaic forecasts and cloud-based decision-making, each region formulates a regional dispatch plan as follows: During the over-generation period 'a', Zone A transmits electricity to Zone B per unit time. The remaining excess power will be used for local energy storage charging; during period c, when area B receives power from area A, it will be used for... Charge local energy storage before peak load, and during peak load period b, energy storage will... Discharge fills the gap, where c lies between a and b. Based on the scheduling plan, control commands are generated according to scheduling instructions and real-time data. Simultaneously, adjustments are made autonomously based on real-time data. Region A generates commands for photovoltaic output, inter-regional interconnection power, and energy storage charging power; Region B generates commands for inter-regional interconnection power and energy storage charging / discharging power. Finally, Regions A and B execute the regional control commands respectively to complete this scheduling. Therefore, through this scheduling, curtailment in Region A is zero, and Region B reduces its grid power purchases. This reduces overall losses, effectively maximizes regional photovoltaic absorption, reduces curtailment, and optimizes energy storage utilization.

[0069] The above-described embodiments are preferred embodiments of this application and are not intended to limit the specific scope of this application. The scope of this application includes but is not limited to the specific embodiments described above. All equivalent changes made in accordance with the shape, structure, and method of this application are within the protection scope of this application.

Claims

1. A method for centralized dispatching of distributed photovoltaic power in a distribution network, characterized in that: Includes the following steps: Acquire regional photovoltaic power output forecast data and load demand forecast data, and combine them with distribution network topology data to generate global dispatch indicators; Determine the short-term optimal scheduling index for the region based on the global scheduling index; Based on the aforementioned short-term optimal scheduling index, regional control commands are generated, and then the commands are executed to achieve optimal scheduling of photovoltaic and energy storage across regions.

2. The method for centralized dispatching of distributed photovoltaic power in a distribution network according to claim 1, characterized in that: The process of acquiring regional photovoltaic power output forecast data and load demand forecast data, and combining them with distribution network topology data to generate global scheduling indicators, includes: Acquire monitoring data of distributed power resources in each region, and combine historical monitoring data with meteorological data to make short-term forecasts of regional photovoltaic output and load demand, thereby obtaining regional photovoltaic output forecast data and load demand forecast data. A global scheduling index is generated based on the photovoltaic output forecast data, load demand forecast data, and distribution network topology data.

3. The method for centralized dispatching of distributed photovoltaic power in a distribution network according to claim 2, characterized in that: The power distributed resource monitoring data includes at least photovoltaic power plant monitoring data, energy storage device monitoring data, load-side operating parameters, and power data from distribution network line monitoring points; By integrating the photovoltaic power plant monitoring data, energy storage device monitoring data, load-side operating parameters, power data from distribution network line monitoring points, and corresponding historical monitoring data and meteorological data from various regions, the photovoltaic output characteristic set and load demand characteristic set of each region are extracted. Using the regional photovoltaic power output feature set as input to the power output prediction model, short-term photovoltaic power output prediction data for the region is obtained; Using the load demand feature set as input to the load demand forecasting model, short-term load demand forecast data for the region is obtained.

4. The method for centralized dispatching of distributed photovoltaic power in a distribution network according to claim 2, characterized in that: The step of generating global scheduling indicators based on the photovoltaic power output forecast data, load demand forecast data, and distribution network topology data includes: Integrate short-term forecast data of photovoltaic output and short-term forecast data of load demand from various regions, and after data preprocessing, combine them with distribution network topology data to form a global dataset. A multi-regional collaborative optimization model is constructed based on global scheduling objectives and constraints, and the global optimization result is obtained by solving the model based on the global dataset. The global optimization result is then transformed into a global scheduling metric.

5. A method for centralized dispatching of distributed photovoltaic power in a distribution network according to claim 4, characterized in that: The construction of a multi-regional collaborative optimization model based on global scheduling objectives and constraints, and the obtaining of global optimization results by solving the model using a global dataset, includes: The objective function for regional collaboration is constructed to minimize the operational cost of regional collaboration, and constraints are established based on the operational characteristics within the region and the operational constraints between regions. Determine the decision variables for the multi-regional collaborative optimization model, including continuous and binary variables, and configure the model's objective function, constraints, and solver parameters; Based on the global dataset, the multi-regional collaborative optimization model is solved by running a solver to obtain the global optimization results of the decision variables.

6. The method for centralized dispatching of distributed photovoltaic power in a distribution network according to claim 1, characterized in that: The step of determining the short-term optimal scheduling index for a region based on the global scheduling index includes: Based on the short-term forecast data of photovoltaic output and short-term forecast data of load demand in each region, multi-layered constraint conditions are constructed in conjunction with the global scheduling indicators. Regional scheduling objectives are constructed based on global electricity purchase costs, curtailment of solar power, power fluctuations, and energy storage losses, and a regional scheduling optimization model is constructed by integrating the aforementioned multi-layered constraints. Solve the regional scheduling optimization model to obtain the short-term optimal scheduling index for the region.

7. A method for centralized dispatching of distributed photovoltaic power in a distribution network according to claim 6, characterized in that: The process involves constructing multi-layered constraints based on short-term forecasts of photovoltaic power output and load demand in each region, combined with the global scheduling indicators. These constraints include: The global scheduling index is decomposed to obtain global decision constraints; Regional dispatch constraints are determined based on short-term forecasts of photovoltaic output and load demand in the region. Multi-layered constraint conditions are established based on the global decision constraints and the regional scheduling constraints.

8. A method for centralized dispatching of distributed photovoltaic power in a distribution network according to claim 7, characterized in that: The global scheduling indicators include at least the cross-regional power exchange power indicator, the regional energy storage plan indicator, and the regional photovoltaic absorption rate indicator. The cross-regional power exchange power, regional energy storage plan and regional photovoltaic absorption rate are constructed in a regional constraint manner based on the boundary values ​​to obtain the corresponding cross-regional power exchange constraints, energy storage global trajectory constraints and regional photovoltaic absorption constraints. Cross-regional power exchange constraints, global energy storage trajectory constraints, and regional photovoltaic consumption constraints serve as global decision constraints.

9. A method for centralized dispatching of distributed photovoltaic power in a distribution network according to claim 2, characterized in that: The generation of area control instructions based on the short-term optimal scheduling index includes: The scheduling deviation is obtained by comparing the regional power distributed resource monitoring data with the short-term optimal scheduling index; Based on the magnitude and type of the scheduling deviation, a rule base is used to determine and adjust the plan, thereby obtaining regional control instructions.

10. A distribution and storage system for centralized dispatch of distributed photovoltaic power in a distribution network, characterized in that: A method for centralized dispatching of distributed photovoltaic power in a distribution network as described in any one of claims 1-9, comprising: The forecasting module is used to acquire regional photovoltaic power output forecast data and load demand forecast data, and combine them with distribution network topology data to generate global scheduling indicators. The control module is used to determine the short-term optimal scheduling index for the region based on the global scheduling index. The execution module is used to generate regional control instructions based on the short-term optimal scheduling index, and then execute the instructions to achieve optimal scheduling of photovoltaic and energy storage across regions.

Images