Power grid safety guiding type multi-strategy collaborative energy storage regulation and control method

Abstract

The embodiment of the invention provides a power grid safety guiding type multi-strategy collaborative energy storage regulation and control method. The method comprises the steps that photovoltaic output prediction data and user load prediction data are acquired; dynamically correcting the photovoltaic output prediction data and the user load prediction data based on equipment operation monitoring information; taking the corrected prediction data as input, combining with a power grid operation constraint condition, and generating an energy storage regulation and control instruction set through an artificial intelligence optimization algorithm; and performing cooperative charging and discharging control on each energy storage unit according to the regulation and control instruction set so as to realize a safe operation target of the power grid. According to the method, optimization balance of energy storage among multiple targets of improving photovoltaic absorption, stabilizing voltage, preventing overload and countercurrent and the like is realized, the energy storage is converted into an active safety defense line of the power grid through closed-loop execution and feedback, and the robustness and the operation efficiency of the high-proportion photovoltaic power distribution network are remarkably enhanced.

Description

TECHNICAL FIELD

[0001] The present document relates to the technical field of power grid regulation, and in particular to a power grid safety-oriented multi-strategy coordinated energy storage regulation method. BACKGROUND

[0002] With the promotion of the double carbon target, the penetration rate of distributed renewable energy represented by photovoltaics in distribution networks is rapidly increasing. However, photovoltaic power generation has the characteristics of intermittency and volatility, and large-scale access brings two core challenges to power grid operation: first, voltage quality problems, photovoltaic output fluctuations cause frequent voltage overruns in distribution areas, especially at the end of the feeder, which may cause household appliances to burn out and other accidents; second, the risk of reverse overload, when photovoltaic power is large and local load is low, reverse power flow may cause distribution transformers and lines to overload. Traditional solutions mainly rely on reactive power compensation devices and grid expansion and reconstruction, but have the problems of slow response speed and high investment cost.

[0003] In the operation of the distribution network, energy storage systems are widely recognized as an effective means to smooth photovoltaic fluctuations, but existing energy storage scheduling techniques have the following limitations: first, the target orientation is misplaced, existing strategies mainly optimize economic efficiency, and power grid safety is taken as a boundary condition rather than a core target, resulting in frequent voltage overrun events in high photovoltaic penetration scenarios; second, the lack of device coordination, the dispatching system does not access device operation monitoring data, which may misdispatch fault energy storage units that have been warned; third, multi-dimensional regulation requirements, power grid companies have introduced photovoltaic preferential consumption, orderly charging and other power grid safety operation requirements based on their own operation safety, and energy storage regulation strategies should fully consider such requirements to ensure power grid operation safety and power quality. SUMMARY

[0004] One or more embodiments of the present specification provide a power grid safety-oriented multi-strategy coordinated energy storage regulation method, comprising: obtaining photovoltaic output prediction data and user load prediction data; based on device operation monitoring information, dynamically correcting the photovoltaic output prediction data and user load prediction data; using the corrected prediction data as input, combining power grid operation constraints, and generating a set of energy storage regulation instructions through an artificial intelligence optimization algorithm; coordinately charging and discharging each energy storage unit according to the regulation instruction set to achieve the power grid safety operation target.

[0005] Further, the obtaining photovoltaic output prediction data specifically comprises: Based on meteorological data, micro-meteorological data, actual irradiance, component temperature and historical photovoltaic output data, a deep neural network is used to construct a spatial feature extraction module and a time feature extraction module, and a future 24-hour photovoltaic output prediction curve is generated through an adaptive weight fusion layer; When the prediction deviation exceeds 15% of the rated capacity, a dynamic error compensation mechanism is triggered, and the prediction result is corrected in combination with the device abnormal state and health data.

[0006] Further, the acquisition of user load prediction data specifically includes: Based on customer profile data, historical load curve, real-time electricity consumption data, meteorological data, date type and special event marking, a graph neural network is used to construct a feature embedding module and a time series prediction module, and a future 24-hour hourly load prediction curve is dynamically fused and output. When the prediction deviation exceeds the set threshold, a real-time feedback calibration mechanism is started, and the prediction result is corrected in combination with the device abnormal state and health data.

[0007] Further, the power grid operation constraint conditions include: Energy storage constraint: the charging and discharging power does not exceed the rated capacity, the SOC change rate does not exceed 5% / minute, and the SOC change gradient does not exceed 3% / minute; Voltage constraint: voltage fluctuation rate does not exceed 1% / second, voltage imbalance does not exceed 2%; Load constraint: line load rate does not exceed 85%, transformer load rate does not exceed 90%, and single load adjustment amount does not exceed 25% of the current load.

[0008] Further, the generation of the regulation instruction set includes the following strategies: Photovoltaic consumption strategy: start energy storage charging during the period of large photovoltaic generation, the charging power is the minimum of 1.2 times the photovoltaic excess power and the available capacity of the energy storage, and the energy storage unit with high marginal benefit coefficient is preferentially called based on the voltage sensitivity matrix; Ordered charging and discharging strategy: the charging period is divided into prohibited, limited and encouraged intervals, the charging power quota is allocated according to the transformer load rate, and the substation load rate is ensured to be not more than 85%.

[0009] Further, the objective function of the optimization algorithm includes: Maximize voltage qualification rate, control substation voltage in the range of 207-235V, and terminal user voltage deviation not more than ±5%; Minimize reverse power, control each node reverse power flow not more than 10% of line capacity; Minimize reverse flow impact, control reverse flow duration ratio less than 0.001%, and maximum single reverse flow energy less than 0.1kWh; Minimize energy storage losses, control the cycle efficiency to no less than 92%, and maintain the SOC between 40% and 80%.

[0010] Furthermore, the optimization algorithm employs an improved deep reinforcement learning algorithm, including a hybrid architecture of PPO and DDPG, which generates charging and discharging power and duration instructions for each energy storage unit within the next 24 hours with a time granularity of 15 minutes.

[0011] Furthermore, the energy storage control instruction set is output in granularity of 96 points / day, and the instruction format includes energy storage unit identifier, charging and discharging power, start time and duration.

[0012] This specification provides one or more embodiments of an electronic device, including: Processor; and, A memory is configured to store computer-executable instructions, which, when executed, cause the processor to implement the steps of the above-described grid security-oriented multi-strategy coordinated energy storage control method.

[0013] This specification provides one or more embodiments of a storage medium for storing computer-executable instructions that, when executed, implement the steps of the above-described grid security-oriented multi-strategy coordinated energy storage control method.

[0014] By adopting the embodiments of the present invention, with the goal of safe grid operation, an optimized system is constructed with voltage stability, prevention of reverse overload, and prevention of energy storage backflow as the core. This system can reduce voltage over-limit accidents, improve the voltage qualification rate of users, strengthen the safety of energy storage systems, and support the safe operation of the power grid. A multi-strategy collaborative framework for photovoltaic consumption and orderly charging is constructed, realizing the requirements for safe grid operation such as the linkage between photovoltaic consumption and voltage regulation.

[0015] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in one or more embodiments of this specification or in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1A flowchart of a power grid safety-oriented multi-strategy collaborative energy storage regulation method is provided for one or more embodiments of the present specification. Figure 2 A structural schematic diagram of an electronic device is provided for one or more embodiments of the present specification. DETAILED DESCRIPTION In order for those skilled in the art to better understand the technical solutions in one or more embodiments of the present specification, the technical solutions in one or more embodiments of the present specification will be described clearly and completely below in conjunction with the accompanying drawings of one or more embodiments of the present specification. Obviously, the described embodiments are only a part of the embodiments of the present specification, rather than all the embodiments. Based on one or more embodiments of the present specification, all other embodiments obtained by those skilled in the art without creative labor should belong to the protection scope of the present document.

[0018] Method embodiments According to an embodiment of the present application, a power grid safety-oriented multi-strategy collaborative energy storage regulation method is provided, Figure 1 A flowchart of a power grid safety-oriented multi-strategy collaborative energy storage regulation method is provided for one or more embodiments of the present specification, as Figure 1 shown, the power grid safety-oriented multi-strategy collaborative energy storage regulation method according to the embodiment of the present application specifically includes: S1. Obtain photovoltaic output prediction data and user load prediction data.

[0019] In terms of photovoltaic output prediction, multiple source data are comprehensively utilized, including macro weather forecast, micro weather data reflecting local weather changes, real-time actual irradiance and photovoltaic component temperature, and historical photovoltaic output records. These data are input into a deep neural network model to predict future photovoltaic output. The deep neural network integrates two independent modules, including a spatial feature extraction module and a temporal feature extraction module. The spatial feature extraction module is used to extract the distribution trend features of illumination intensity and temperature changes from weather and geographical data. The temporal feature extraction module is used to analyze the time regularity of historical output data and mine its internal regularity and dependence relationship at different time scales. The outputs of the two modules are integrated through an adaptive weight fusion layer. According to different weather types and time periods, the contribution weights of spatial and temporal features are dynamically adjusted, and finally a 24-hour photovoltaic output prediction curve is generated. In terms of user load prediction, multi-dimensional data analysis is also adopted, based on customer profile data (power consumption category, installed capacity), historical load curve, real-time power consumption data, meteorological data (temperature, humidity), date type (weekday / holiday) and special event markers, a hybrid prediction architecture of graph neural network is used for prediction. The graph neural network includes a feature embedding module and a time series prediction module. The feature embedding module is used to extract static features of user attributes and device types, and to convert static information such as user attributes and device types into feature vectors that can be deeply learned by computers. The time series prediction module is used to analyze the time dependence of load data and the influence of external factors, capture the dynamic rules of load changes over time, and analyze the impact of external factors on load. The two modules are dynamically fused to output a 24-hour hourly load prediction curve. To ensure the continuous accuracy of load prediction, a real-time feedback calibration mechanism is designed. When the deviation between the actual load and the predicted value exceeds the set threshold, the real-time feedback calibration mechanism is started, and the prediction curve is corrected in combination with the device abnormal state and health data. In this embodiment, the threshold is set to 20%.

[0020] S2. Based on the device operation monitoring information, dynamically correct the photovoltaic output prediction data and user load prediction data.

[0021] In terms of photovoltaic output prediction, a dynamic error compensation mechanism is introduced. When the deviation between the actual measurement value and the predicted value of the photovoltaic output exceeds 15% of the rated capacity, the dynamic error compensation mechanism is triggered, the incremental learning process is started, the latest operation data in the sliding time window is used to quickly update the model parameters, and the prediction result is corrected in real time in combination with the device abnormal state information and the device health data, wherein the device abnormal state information includes communication interruption and inverter failure, and the device health data includes component attenuation rate and connector aging degree.

[0022] In terms of user load prediction, to ensure the continuous accuracy of load prediction, a real-time feedback calibration mechanism is designed. When the deviation between the actual load and the predicted value exceeds the set threshold, the real-time feedback calibration mechanism is started, and the prediction curve is corrected in combination with the device abnormal state and health data. In this embodiment, the threshold is set to 20%.

[0023] S3. With the corrected prediction data as input, combined with the grid operation constraint conditions, the energy storage control instruction set is generated through an artificial intelligence optimization algorithm.

[0024] Fusion of high proportion of photovoltaic consumption demand, load dynamic response characteristics and grid safety constraints, construction of multi-time and space scale collaborative optimization model: The data input layer is used to input photovoltaic data, composite data, power grid data and energy storage data, the photovoltaic data including active / reactive power of each grid-connected point, power generation prediction curve (15 minutes); the load data including total load curve of a transformer area, adjustable load capacity, meter voltage monitoring value, load prediction curve (15 minutes); the power grid data including impedance parameters of a distribution line, rated capacity of a transformer, topological connection relationship; the energy storage data including SOC state of each energy storage unit, charging / discharging efficiency, health degree index of energy storage equipment.

[0025] Comprehensive power grid operation constraints are constructed, including: Energy storage constraints: the charging / discharging power does not exceed the rated capacity, the SOC change rate does not exceed 5% / minute, and the SOC change gradient does not exceed 3% / minute; Voltage constraints: the voltage fluctuation rate does not exceed 1% / second, and the voltage unbalance degree does not exceed 2%; Load constraints: the line load rate does not exceed 85%, the transformer load rate does not exceed 90%, and the single load adjustment amount does not exceed 25% of the current load.

[0026] Within the framework of the above rigid constraints, a multi-objective function is constructed for the purpose of achieving multiple optimal targets, including: Maximizing voltage qualification rate, controlling the transformer area voltage in the range of 207-235V, and the end-user voltage deviation not exceeding ±5%; Minimizing reverse power, controlling the reverse power flow of each node not exceeding 10% of the line capacity; Minimizing reverse flow impact, controlling the reverse flow duration ratio to be less than 0.001%, and the maximum single reverse flow energy to be less than 0.1kWh; Minimizing energy storage loss, controlling the cycle efficiency to be not less than 92%, and the SOC to be maintained between 40%-80%.

[0027] In order to find the optimal solution of the objective function, the embodiment adopts an improved deep reinforcement learning algorithm, specifically a hybrid architecture of proximal policy optimization (PPO) and deep deterministic policy gradient (DDPG), which takes the corrected prediction curve for the next 24 hours, the real-time power grid state and the real-time state of all energy storage units as input. At each 15-minute decision point, various possible energy storage action strategies are simulated, and based on the multi-objective function and the constraints, the long-term cumulative benefits of these strategies in terms of power grid safety, photovoltaic accommodation, equipment loss, etc. in the next few hours are evaluated, and finally an optimal energy storage control instruction set is generated as the charging / discharging power and duration instructions for each energy storage unit in the next 24 hours.

[0028] The generation of the control instruction set includes the following strategies: Photovoltaic consumption strategy: adopt sliding time window analysis method, identify photovoltaic large time period, that is, output >=80% rated capacity, when photovoltaic real-time output >=75% rated capacity and lasts for more than 30 minutes, start the energy storage charging mode, the charging power takes the minimum value of 1.2 times of the excess power of photovoltaic and the available capacity of energy storage, dynamically adjusts the charging curve, ensures that the voltage fluctuation <=1% / minute;Based on the voltage sensitivity matrix, the marginal benefit coefficient of each node energy storage consumption is calculated, the energy storage unit with high marginal benefit coefficient and closer to the photovoltaic grid connection point is preferentially called, and the photovoltaic power generation is maximized; Ordered charging and discharging strategy: adopt charging load optimization model, divide the charging period into three intervals of prohibition, restriction and encouragement, distribute charging power quota according to transformer load rate, and ensure that the transformer load rate does not exceed 85%.

[0029] The energy storage control instruction set is output with a granularity of 96 points / day, and the instruction format includes energy storage unit identification, charging and discharging power, start time and duration.

[0030] S4. According to the control instruction set, the energy storage units are controlled in coordination to realize the safe operation target of the power grid.

[0031] The present application has the following advantages: By adopting the embodiments of the present application, an optimization system is constructed with the goal of safe operation of the power grid, voltage stability, prevention of reverse overload and prevention of energy storage reverse flow as the core, which can reduce voltage out-of-limit accidents, improve user voltage qualification rate, strengthen energy storage system safety and support safe operation of the power grid;A photovoltaic consumption and ordered charging multi-strategy coordination framework is constructed to realize the requirements of power grid safe operation such as photovoltaic consumption and voltage regulation linkage.

[0032] Device embodiment one The embodiments of the present application provide an electronic device, as shown in the figure, comprising a memory 20, a processor 22 and a computer program stored on the memory 20 and executable on the processor 22, the computer program is executed by the processor 22 to realize the following method steps: Figure 2 S1. Obtain photovoltaic output prediction data and user load prediction data; S2. Based on the device operation monitoring information, dynamically correct the photovoltaic output prediction data and user load prediction data; S3. With the corrected prediction data as input, combine the power grid operation constraint condition, and generate an energy storage control instruction set through an artificial intelligence optimization algorithm; S4. According to the control instruction set, the energy storage units are controlled in coordination to realize the safe operation target of the power grid.

[0033] Device embodiment two ​The embodiment of the present application provides a computer readable storage medium, and the computer readable storage medium stores an implementation program of information transmission, and the program is executed by a processor 22 to implement the following method steps. S1. obtaining photovoltaic output prediction data and user load prediction data; S2. dynamically correcting the photovoltaic output prediction data and the user load prediction data based on equipment operation monitoring information; S3. taking the corrected prediction data as input, combining power grid operation constraint conditions, and generating a set of energy storage regulation instructions through an artificial intelligence optimization algorithm; S4. performing collaborative charging and discharging control on each energy storage unit according to the set of regulation instructions, so as to achieve a power grid safe operation target.

[0034] The computer readable storage medium in the embodiment includes but is not limited to ROM, RAM, a magnetic disk or an optical disk, and the like.

[0035] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A grid security-oriented multi-strategy coordinated energy storage control method, characterized in that, include: Obtain photovoltaic power output forecast data and user load forecast data; Based on equipment operation monitoring information, the photovoltaic power output prediction data and user load prediction data are dynamically corrected. Using the corrected forecast data as input and combined with power grid operation constraints, an energy storage control instruction set is generated through artificial intelligence optimization algorithms. The energy storage units are coordinated to charge and discharge according to the control instruction set, so as to achieve the goal of safe operation of the power grid.

2. The method according to claim 1, characterized in that, The acquisition of photovoltaic power output prediction data specifically includes: Based on meteorological data, micro-meteorological data, actual irradiance, module temperature and historical photovoltaic power output data, a spatial feature extraction module and a temporal feature extraction module are constructed using a deep neural network, and a photovoltaic power output prediction curve for the next 24 hours is generated through an adaptive weight fusion layer. When the prediction deviation exceeds 15% of the rated capacity, a dynamic error compensation mechanism is triggered, which corrects the prediction result by combining the abnormal status and health data of the equipment.

3. The method according to claim 1, characterized in that, The acquisition of user load prediction data specifically includes: Based on customer profile data, historical load curves, real-time electricity consumption data, meteorological data, date types, and special event markers, a graph neural network is used to construct a feature embedding module and a time series prediction module. After dynamic fusion, the hourly load prediction curve for the next 24 hours is output. When the prediction deviation exceeds the set threshold, a real-time feedback calibration mechanism is activated to correct the prediction results by combining the abnormal status and health data of the equipment.

4. The method according to claim 1, characterized in that, The power grid operation constraints include: Energy storage constraints: charging and discharging power shall not exceed the rated capacity, the SOC change rate shall not exceed 5% / minute, and the SOC change gradient shall not exceed 3% / minute; Voltage constraints: voltage fluctuation rate not exceeding 1% / second, voltage imbalance not exceeding 2%; Load constraints: Line load rate shall not exceed 85%, transformer load rate shall not exceed 90%, and single load adjustment shall not exceed 25% of the current load.

5. The method according to claim 1, characterized in that, The generation of the control instruction set includes the following strategies: Photovoltaic consumption strategy: Start energy storage charging during peak photovoltaic power generation periods. The charging power is the minimum of 1.2 times the excess photovoltaic power and the available energy storage capacity. Based on the voltage sensitivity matrix, priority is given to calling energy storage units with high marginal benefit coefficients. Orderly charging and discharging strategy: Divide the charging period into prohibited, restricted and encouraged zones, allocate charging power quotas according to transformer load rate, and ensure that the load rate of the transformer area does not exceed 85%.

6. The method according to claim 1, characterized in that, The objective function of the optimization algorithm includes: To maximize voltage compliance, the voltage in the control area should be within the range of 207-235V, and the voltage deviation at the end user should not exceed ±5%. Minimize reverse power and control the reverse power flow at each node to not exceed 10% of the line capacity; To minimize the impact of backflow, the duration of backflow should be controlled to be less than 0.001%, and the maximum energy of a single backflow should be less than 0.1 kWh. Minimize energy storage losses, control the cycle efficiency to no less than 92%, and maintain the SOC between 40% and 80%.

7. The method according to claim 1, characterized in that, The optimization algorithm adopts an improved deep reinforcement learning algorithm, including a hybrid architecture of PPO and DDPG, and generates charging and discharging power and duration instructions for each energy storage unit in the next 24 hours with a time granularity of 15 minutes.

8. The method according to claim 1, characterized in that, The energy storage control instruction set is output in granularity of 96 points / day, and the instruction format includes energy storage unit identifier, charging and discharging power, start time and duration.

9. An electronic device, characterized in that, include: processor; as well as, A memory configured to store computer-executable instructions, which, when executed, cause the processor to implement the steps of the grid security-oriented multi-strategy coordinated energy storage control method as described in any one of claims 1 to 8.

10. A storage medium, characterized in that, Used to store computer-executable instructions, which, when executed, implement the steps of the grid security-oriented multi-strategy coordinated energy storage control method as described in any one of claims 1 to 8.

Images