Comprehensive energy station energy storage control method and device and storage medium

Abstract

The invention provides an integrated energy station energy storage control method and device and a storage medium, belongs to the technical field of integrated energy station energy storage control, and solves the technical problem that the photovoltaic utilization rate is difficult to improve in existing integrated energy station energy storage scheduling. The method comprises the following steps: collecting multi-source data of a comprehensive energy station system, constructing a data set, and dividing the data set into daily initial input data and real-time input data; constructing an energy storage flexible control model based on the multi-source data to generate a control strategy; the control strategy comprises an energy storage whole-day strategy and energy storage real-time control, the energy storage system is dynamically adjusted, and an energy storage control curve is output; converting the energy storage control curve into a fine adjustment control instruction, and sending the fine adjustment control instruction to an energy storage system; and constructing an energy storage strategy multi-objective optimization model, and solving to obtain a global optimal schedule charging and discharging strategy. The method is applied to energy storage strategy optimization, scheduling and control of the integrated energy station, and improves the photovoltaic consumption capability.

Description

Technical Field

[0001] This invention belongs to the field of integrated energy station energy storage control technology, and is specifically applied to the optimization, scheduling and control of energy storage strategies in integrated energy stations. Background Technology

[0002] As the proportion of renewable energy in the power structure continues to increase, traditional power systems face increasingly severe challenges in real-time source-load balance, significantly increasing the demand for flexible regulation capabilities. Energy storage technology, due to its ability to decouple energy production and consumption in time and space, is gradually becoming a key support for future high-proportion renewable energy power systems. Against this backdrop, integrated photovoltaic-storage-charging energy stations have emerged. This model, through the deep integration of photovoltaic power generation, energy storage systems, and charging facilities, achieves localized energy production, storage, and consumption, forming an integrated micro-energy system.

[0003] The following technical problems exist in the flexible control of industrial and commercial energy storage: Control objectives are difficult to achieve synergistically: Modern control strategies aim to optimize multiple objectives such as economy, stability, and sustainability. However, in practice, these objectives may conflict and are difficult to achieve simultaneously at their optimal levels. For example, pursuing economy may lead to frequent charging and discharging of energy storage systems, thereby affecting their stability and lifespan and reducing sustainability.

[0004] Data fusion and processing are challenging: Although the technical architecture is trending towards layering and collaboration, with the "cloud-edge-device" architecture becoming mainstream, photovoltaic, energy storage, and charging equipment are multi-source and heterogeneous, resulting in poor system compatibility. This can easily lead to data fragmentation, making it impossible to fully leverage the advantages of layering and collaboration architectures and affecting the effectiveness of global optimization and rapid control.

[0005] Insufficient integrated and coordinated control: Although the integration and coordination of "photovoltaic, energy storage, charging and load" is emphasized, in actual operation, due to the possible differences in the control strategies and parameter settings of each device, the forecast support for photovoltaic power generation and electricity load is insufficient. It is difficult to truly form a highly efficient and coordinated whole by using technologies such as flexible DC microgrids, resulting in low energy utilization efficiency.

[0006] The adaptability of intelligent decision-making algorithms is poor: Although algorithms based on Model Predictive Control (MPC), Machine Learning (ML), and Deep Reinforcement Learning (DRL) are gradually being applied to energy storage control, these algorithms have high data requirements. However, integrated photovoltaic-storage-charging energy stations, as a newly developing business scenario, have relatively small-scale photovoltaic power generation and energy storage, are greatly affected by the environment, and lack historical data accumulation, which limits the prediction accuracy and optimization effect of the algorithms. At the same time, these algorithms have high computational requirements, which are often difficult for small energy stations to meet, making it difficult for intelligent decision-making algorithms to be applied in practice.

[0007] Overall, integrated photovoltaic-storage-charging (PV-SGC) energy stations are a newly developing business scenario. The scale of PV power generation and energy storage is relatively small, making them highly susceptible to environmental influences and lacking historical data accumulation. Simultaneously, the rapid development of electric vehicles limits their responsiveness to electricity price fluctuations and changes in electricity load. Therefore, this presents significant challenges to the design, operation, and profitability of power stations. Existing methods require high data input and strong computing power, which are often insufficient for smaller integrated energy stations to implement such complex control strategies. Integrated energy stations require methods that are low-investment, have broad environmental requirements, are simple, and offer rapid response with strong adaptability. Summary of the Invention

[0008] In view of this, the present invention aims to propose an integrated energy station energy storage control method, device, computer product and storage medium based on a source-storage-load intelligent flexible control model, so as to solve the technical problem of difficulty in improving photovoltaic utilization in the energy dispatch of existing integrated energy stations.

[0009] To achieve the above objectives, the present invention adopts the following technical solution: This invention proposes an energy storage control method for integrated energy stations, the method comprising: S1. Collect multi-source data from the integrated energy station system, construct a dataset based on the multi-source data, and divide the dataset into initial daily input data and real-time input data; The multi-source data includes equipment operation data, environmental sensing data, and power grid interaction data; S2. Construct an energy storage flexible control model based on the multi-source data, and generate a control strategy using the energy storage flexible control model; The control strategy includes a daily energy storage strategy and real-time energy storage control; wherein... The energy storage all-day strategy is used to perform global planning of the energy storage system and generate a basic energy storage control plan for each hour of the day. The real-time energy storage control is based on the hourly basic energy storage control plan throughout the day, dynamically adjusting the energy storage system and outputting energy storage control curves; S3. Convert the energy storage control curve into a fine-tuning control command and send it to the energy storage system; S4. Construct a multi-objective optimization model for energy storage strategies and solve it using an improved particle swarm optimization algorithm to obtain the globally optimal scheduled charging and discharging strategy.

[0010] Furthermore, the equipment operation data is obtained by collecting data from the smart meters and energy storage system of the integrated energy station. The equipment operation data includes: real-time photovoltaic power generation, electricity load, energy storage system SOC, charging and discharging status and health status data.

[0011] Furthermore, the method for constructing the flexible control model for energy storage described in S2 includes: Based on the environmental perception data and power grid interaction data, photovoltaic power generation forecasting and electricity load forecasting are performed, and the forecast results include the daily photovoltaic power generation curve and the load forecast curve. Based on the prediction results and the equipment operation data, the energy storage flexible control model is constructed.

[0012] Furthermore, the construction process of the all-day energy storage strategy described in S2 is as follows: Enter the initial data for the day; The execution time of the control strategy is set to 0:00 every day; The strategy optimization objectives are set as system economy, load balance, and maximizing photovoltaic self-consumption; Set the constraints as power, capacity, and charge / discharge cycle limits; The particle swarm optimization algorithm is used to optimize the energy storage strategy and obtain the hourly energy storage charge and discharge power curves throughout the day, thereby generating a full-day charge and discharge strategy. Based on the energy storage equipment cluster, an energy storage equipment model library is established, energy storage equipment units are divided and strategy matching is performed to determine the distribution model of different energy storage equipment units; Based on the distribution models of different energy storage device units, the all-day charging and discharging strategy is flexibly decomposed and mapped to the output of the basic energy storage control plan for each hour of the day.

[0013] Furthermore, the flexible decomposition involves mapping the energy storage all-day strategy to the energy storage system, dividing the device operating state into charging, standby, and discharging, and outputting the control strategy to the devices in the energy storage system according to a timing control method.

[0014] Furthermore, the process of dynamically adjusting the energy storage system described in S2 includes: Input the real-time input data; The real-time fine-tuning interval for energy storage is set to 10 minutes. The adjusted control commands are obtained through real-time fine-tuning; The adjusted control command is sent to the energy storage system; Equipment scheduling is performed based on the priority of its stable operating status. It outputs a fine-tuning control command that is executed once every 10 minutes, generating an energy storage control curve.

[0015] Furthermore, the real-time fine-tuning includes re-acquiring multi-source data, evaluating the deviation from the currently executed control strategy, and making minor corrections to the charging and discharging power based on real-time photovoltaic data, load data, and electricity prices.

[0016] Furthermore, the S4 process includes: S41. Setting model optimization objectives includes: Photovoltaic-storage-transformation yields the greatest benefits: , Minimize transformer losses: , In the formula, Indicates the power consumption of the load; This represents the population mean; Indicates the number of consecutive low hours; This indicates the period from the start to the end of the consecutive off-peak electricity price period; Indicates the total variance; S42. Setting model constraints includes: Charging and discharging power is less than or equal to maximum power: , In the formula, t represents the charging and discharging time. Indicates discharge power. Indicates charging power; Indicates the maximum charging capacity. Indicates the maximum discharge power; Energy storage charging and discharging process capacity rules: In the formula, Indicates the capacity of the energy storage battery; Indicates the lower limit of the energy storage battery capacity; , indicating any time within 24 hours; Solar surplus energy storage: when hour, The amount of electricity used to charge the energy storage device; Prioritized discharge of stored energy: when hour, This indicates that the peak-valley price difference is greater than the energy storage cost; S43. An improved particle swarm optimization algorithm is used to solve the problem; Set the velocity vector for each particle and position vector Update particle velocity and position as follows: , Introduce a penalty function: , In the formula, Penalties for violating energy balance rules; Internet access penalties; This is a penalty for violating the rule that the remaining photovoltaic capacity is not stored. As a penalty factor; S44. Correct the particle velocity and position constraints on the particle motion range, and dynamically adjust the energy storage charging amount in conjunction with the real-time surplus of photovoltaic power generation. At that time, particles The amount of charge at any given time is: The energy storage SOC change curve is obtained; the process is iterated until the optimal all-day charging and discharging strategy is obtained.

[0017] The present invention also proposes an integrated energy station energy storage control device, the device comprising: Data input module: used to collect multi-source data from the integrated energy station system, construct a dataset based on the multi-source data, and divide the dataset into initial daily input data and real-time input data; the multi-source data includes equipment operation data, environmental sensing data, and power grid interaction data; Model and strategy construction module: used to construct an energy storage flexible control model and generate control strategies using the model; the control strategies include a daily energy storage strategy and a real-time energy storage control strategy; wherein, The energy storage all-day strategy is constructed based on the initial input data of the day and is used to perform global planning of the energy storage system and generate a basic energy storage control plan for each hour of the day. The real-time energy storage control is based on the real-time input data and combined with the basic energy storage control plan for each hour of the day to dynamically adjust the energy storage system and output the energy storage control curve. Control command generation module: used to convert the energy storage control curve into fine-tuning control commands and send them to the energy storage system; Strategy optimization module: Used to construct a multi-objective optimization model for energy storage strategies and solve it using an improved particle swarm optimization algorithm to obtain the globally optimal scheduled charging and discharging strategy.

[0018] The present invention also proposes a storage medium storing a computer program, which, when running, implements the method described in any one of the present invention.

[0019] Compared with the prior art, the beneficial effects of the present invention are: The energy storage flexible control model described in this invention is a model based on source-storage-load flexible control, used for energy storage control in integrated energy stations. This model is based on a multi-objective optimization model for energy storage strategies to achieve coordinated source-storage-load control of integrated energy stations. Among other things, By combining a full-day energy storage strategy with real-time control, when photovoltaic power generation exceeds the load and fluctuates, the surplus photovoltaic power can be prioritized for charging, avoiding curtailment and significantly improving the absorption capacity of photovoltaic power generation. By constructing a multi-objective optimization model for energy storage strategies, energy storage scheduling is combined with transformers and energy storage devices in microgrids, thereby reducing energy losses in the distribution network while achieving economical operation of energy storage. In the optimization algorithm, a single-objective optimization framework is adopted and a load balancing objective is embedded in the iteration process. This avoids the redundant calculation and slow convergence problems caused by Pareto multi-objective solutions, thereby improving the efficiency of policy solution.

[0020] The improved particle swarm optimization algorithm described in this invention, on the one hand, directly uses the photovoltaic surplus as charging power for the energy storage system when the photovoltaic power generation exceeds the load power, thereby fully utilizing photovoltaic power generation. On the other hand, it improves the iteration speed by prioritizing discharge based on load demand during periods of high electricity price differences, thus achieving load demand-based discharge priority control and improving system economy. Furthermore, through the above improvements, the convergence speed and optimal solution quality of the algorithm are improved under limited computing resources, thereby rapidly generating high-quality charging and discharging strategies.

[0021] The control strategy described in this invention combines a two-stage approach: daily strategy formulation and real-time fine-tuning control. This approach ensures both the integrity of the overall daily plan and allows for dynamic adjustments based on real-time input data during operation, balancing planning efficiency with flexibility and thus improving system efficiency and scheduling accuracy. Specifically, the initial daily input data constructs the daily energy storage strategy, which generates hourly energy storage control plans. The real-time input data and the daily charge / discharge strategy enable real-time energy storage control, generating energy storage control curves. This approach ensures a globally optimal operating plan at the beginning of the day and allows for rapid adjustments to control commands in real-time, guaranteeing both global planning and rapid response capabilities in the energy storage control process.

[0022] This invention belongs to the field of integrated energy station energy storage control technology, and is specifically applied to the optimization, scheduling and control of energy storage strategies in integrated energy stations, as well as improving photovoltaic absorption capacity. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or 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 embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0024] Figure 1 This is a flowchart of an integrated energy station energy storage control method according to the present invention.

[0025] Figure 2 This is a structural diagram of the energy storage flexible control model described in this invention.

[0026] Figure 3 This is a flowchart illustrating the construction of a flexible control model for energy storage based on the particle swarm optimization algorithm in this invention.

[0027] Figure 4 This is a flowchart illustrating how the improved particle swarm optimization algorithm is used to solve the flexible control model for energy storage in this invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] Specific implementation method one, such as Figure 1 As shown in this embodiment, a method for controlling energy storage in an integrated energy station includes: S1. Collect multi-source data from the integrated energy station system, construct a dataset based on the multi-source data, and divide the dataset into initial daily input data and real-time input data; The multi-source data includes equipment operation data, environmental sensing data, and power grid interaction data; S2. Construct an energy storage flexible control model based on the multi-source data, and generate a control strategy using the energy storage flexible control model; The control strategy includes a daily energy storage strategy and real-time energy storage control; wherein... The energy storage all-day strategy is used to perform global planning of the energy storage system and generate a basic energy storage control plan for each hour of the day. The real-time energy storage control is based on the hourly basic energy storage control plan throughout the day, dynamically adjusting the energy storage system and outputting energy storage control curves; S3. Convert the energy storage control curve into a fine-tuning control command and send it to the energy storage system; S4. Construct a multi-objective optimization model for energy storage strategies and solve it using an improved particle swarm optimization algorithm to obtain the globally optimal scheduled charging and discharging strategy.

[0030] In this embodiment, the method begins with step one, in which multi-source data of the integrated energy station system is collected, a dataset is constructed based on the multi-source data, and the dataset is divided into initial daily input data and real-time input data; the multi-source data includes equipment operation data, environmental sensing data, and grid interaction data; the equipment operation data is obtained by collecting data from the smart meters and energy storage system of the integrated energy station, and the equipment operation data includes: real-time photovoltaic power generation, electricity load, energy storage system SOC curve, charging and discharging status, and health status data.

[0031] Specifically, (1) Multi-source data fusion and access; The system integrates three types of real-time and near-real-time data streams, forming the data-driven foundation of the model and serving as its input: Environmental perception data includes weather forecasts (irradiance, temperature, humidity), geographic location information, and holiday and weekday markings. Equipment operation data: Real-time photovoltaic power generation, power load, energy storage system SOC, charge / discharge status and health status (SOH), obtained through smart meters (sampling interval 10 minutes) and BMS system; Grid interaction data includes time-of-use pricing signals (peak-valley-flat pricing), distribution area capacity limits, and possible demand response or ancillary service signals.

[0032] Then, step two is executed. In step two, an energy storage flexible control model is constructed based on the multi-source data, and a control strategy is generated using the energy storage flexible control model. The control strategy includes a daily energy storage strategy and real-time energy storage control; wherein... The energy storage all-day strategy is used to perform global planning of the energy storage system and generate a basic energy storage control plan for each hour of the day. The real-time energy storage control is based on the hourly basic energy storage control plan throughout the day, dynamically adjusting the energy storage system and outputting energy storage control curves; The method for constructing the flexible control model for energy storage includes: Based on the environmental perception data and power grid interaction data, photovoltaic power generation forecasting and electricity load forecasting are performed, and the forecast results include the daily photovoltaic power generation curve and the load forecast curve. Based on the prediction results and the equipment operation data, the energy storage flexible control model is constructed.

[0033] Specifically, the photovoltaic power generation forecast is based on the environmental perception data to predict future power generation and obtain a daily photovoltaic power generation curve; the electricity load forecast is based on the grid interaction data to predict future load changes and obtain a load forecast curve.

[0034] The construction process of the energy storage all-day strategy is as follows: Enter the initial data for the day; The execution time of the control strategy is set to 0:00 every day; The strategy optimization objectives are set as system economy, load balance, and maximizing photovoltaic self-consumption; Set the constraints as power, capacity, and charge / discharge cycle limits; The particle swarm optimization algorithm is used to optimize the energy storage strategy and obtain the hourly energy storage charge and discharge power curves throughout the day, thereby generating a full-day charge and discharge strategy. Based on the energy storage equipment cluster, an energy storage equipment model library is established, energy storage equipment units are divided and strategy matching is performed to determine the distribution model of different energy storage equipment units; Based on the distribution models of different energy storage device units, the all-day charging and discharging strategy is flexibly decomposed and mapped to the output of the basic energy storage control plan for each hour of the day.

[0035] The flexible decomposition involves mapping the energy storage all-day strategy to the energy storage system, dividing the device operating state into charging, standby, and discharging, and outputting the control strategy to the devices in the energy storage system according to a timing control method.

[0036] The process of dynamically adjusting the energy storage system includes: Input the real-time input data; The real-time fine-tuning interval for energy storage is set to 10 minutes. The adjusted control commands are obtained through real-time fine-tuning; The adjusted control command is sent to the energy storage system; Equipment scheduling is performed based on the priority of its stable operating status. It outputs a fine-tuning control command that is executed once every 10 minutes, generating an energy storage control curve.

[0037] The real-time fine-tuning includes re-acquiring multi-source data, evaluating the deviation from the currently executed control strategy, and making minor corrections to the charging and discharging power based on real-time photovoltaic data, load data, and electricity prices.

[0038] Specifically, decision-making and optimization include: The strategy optimization (day-ahead decision) aims to maximize overall benefits (including peak-valley arbitrage, reducing grid losses, and delaying investment). Considering energy storage SOC constraints, lifetime loss, and grid capacity limitations, a mixed integer programming model is constructed and solved using an improved particle swarm optimization (PSO) algorithm to generate the optimal 24-hour energy storage charging and discharging plan.

[0039] Online optimization (real-time control) receives the latest photovoltaic power generation, load and electricity price data every 10 minutes. Based on the model predictive control (MPC) framework, it performs closed-loop correction and real-time fine-tuning of the day-ahead plan to smooth out prediction errors and ensure that the system always approaches the optimal operating state.

[0040] Based on the operation of integrated energy stations, an energy storage control method with two stages, namely daily strategy formulation and real-time fine-tuning control, was constructed, which takes into account both planning and flexibility, and improves the system's operating efficiency and economy.

[0041] Figure 3 The architecture of the energy storage flexible control model, as well as its inputs and outputs, are shown. The input information is the multi-source data, and the output is the energy storage control strategy curve.

[0042] The control strategy is divided into two parts: daily planning of energy storage strategy and real-time control process of energy storage, which correspond to control tasks at different time scales.

[0043] Energy storage strategy scheduling, executed daily at 00:00; Input at the beginning of the day, at 00:00: 24-hour hourly load forecasting: based on modeling of historical load, weather, user behavior, etc. 24-hour hourly photovoltaic forecast: obtained from weather forecasts and historical irradiance data; Daily time-of-use electricity pricing information: divided into peak, flat, and valley periods.

[0044] Real-time input, every 10 minutes: Real-time load forecasting, short-term rolling forecasting; Real-time photovoltaic power generation forecast, updated every 10 minutes; Real-time electricity pricing, if a real-time electricity pricing mechanism exists; The original energy storage strategy was used to fine-tune the baseline.

[0045] This stage involves global optimization planning to generate a full-day charging and discharging strategy. The main steps include: Energy storage strategy optimization model: The system aims to maximize system economy (electricity price difference), load balance, and photovoltaic self-consumption; it satisfies energy storage constraints, including power, capacity, and charge / discharge cycles; and it employs algorithms such as particle swarm optimization to solve the problem.

[0046] Energy storage distribution model selected: The energy storage group within the station is divided into multiple subsets, such as by capacity / brand / power / lifespan, and an equipment model library is established; the appropriate strategy types for different energy storage units are determined, including charging, standby, and discharging types.

[0047] Flexible strategy decomposition: Based on the daily strategy, it is mapped to a device-level operation plan; It has three operating states: charging, standby, and discharging, and outputs them in a timing control manner. Output: A full-day, hourly energy storage control plan, detailed down to the status and power of each energy storage device.

[0048] The energy storage real-time control process is executed every 10 minutes. To cope with real-time environmental changes such as sudden load fluctuations and abnormal lighting, the strategy is fine-tuned every 10 minutes. Control task triggering: Re-collect prediction data every 10 minutes; evaluate the deviation from the current execution strategy.

[0049] Real-time strategy fine-tuning: The overall daily strategy framework is retained, and adjustments are made only for the current and next time steps; the charging and discharging power is finely adjusted based on real-time photovoltaic, load, and electricity price data; Avoid frequent state switching to increase device lifespan.

[0050] Flexible decomposition of control commands: Control commands that have been adjusted in real time are reissued to the energy storage cluster; priority is given to scheduling equipment with stable operating status.

[0051] Then, step three is executed, in which the energy storage control curve is converted into a fine-tuning control command and sent to the energy storage system; Specifically, output control commands; Control commands: Output fine-tuning control commands every 10 minutes, i.e., time resolution of 10 minutes, and combine them into the final energy storage control curve, and output the charging and discharging active power commands to the energy storage controller.

[0052] Forecast output includes 144 points of photovoltaic power generation curve for the next 24 hours, load forecast curve, and energy storage SOC change curve.

[0053] The model is designed based on the idea of ​​"data-algorithm" co-evolution. Its prediction method follows a three-stage dynamic migration path: probability-driven → hybrid-driven → data-driven. In the end, it forms a hybrid intelligent system that combines the high-precision prediction capability of machine learning with the strong robustness of traditional methods. It can continuously evolve to adapt to the ever-changing external environment and operational requirements.

[0054] Then, proceed to step four, in which a multi-objective optimization model for the energy storage strategy is constructed and solved using an improved particle swarm optimization algorithm to obtain the globally optimal scheduled charging and discharging strategy. S41. Setting model optimization objectives includes: Photovoltaic-storage-transformation yields the greatest benefits: , Minimize transformer losses: , In the formula, Indicates the power consumption of the load; This represents the population mean; Indicates the number of consecutive low hours; This indicates the period from the start to the end of the consecutive off-peak electricity price period; Indicates the total variance; S42. Setting model constraints includes: Charging and discharging power is less than or equal to maximum power: , In the formula, t represents the charging and discharging time. Indicates discharge power. Indicates charging power; Indicates the maximum charging capacity. Indicates the maximum discharge power; Energy storage charging and discharging process capacity rules: In the formula, Indicates the capacity of the energy storage battery; Indicates the lower limit of the energy storage battery capacity; , indicating any time within 24 hours; Solar surplus energy storage: when hour, The amount of electricity used to charge the energy storage device; Prioritized discharge of stored energy: when hour, This indicates that the peak-valley price difference is greater than the energy storage cost; S43. An improved particle swarm optimization algorithm is used to solve the problem; Set the velocity vector for each particle and position vector Update particle velocity and position as follows: , Introduce a penalty function: , In the formula, Penalties for violating energy balance rules; Internet access penalties; This is a penalty for violating the rule that the remaining photovoltaic capacity is not stored. As a penalty factor; S44. Correct the particle velocity and position constraints on the particle motion range, and dynamically adjust the energy storage charging amount in conjunction with the real-time surplus of photovoltaic power generation. At that time, particles The amount of charge at any given time is: The energy storage SOC change curve is obtained; the process is iterated until the optimal all-day charging and discharging strategy is obtained.

[0055] The improvements include introducing a heuristic strategy for time-of-use pricing, load, and photovoltaic surplus during particle initialization; and introducing a penalty function into the fitness function, which includes a penalty term for energy balance rules, a penalty term for grid connection, and a penalty term for unstored photovoltaic surplus.

[0056] The optimization objective described in the model is a multi-objective optimization, specifically, Optimal return: Photovoltaic-storage-transformer yields the highest returns.

[0057] , Minimizing transformer losses: Minimizing transformer losses is reflected in balanced electricity load, which means minimizing the overall variance during the longest consecutive off-peak period (the period with the longest off-peak). , in: Power consumption of the load; The overall mean; The number of consecutive off-peak hours, for example, if the off-peak electricity price starts at 0:00 and ends at 7:00. ; From the start to the end of the continuous off-peak electricity price period; Overall variance.

[0058] Model constraints: Charging and discharging power is less than or equal to maximum power. , , in: The charging and discharging time is generally 1 minute. Discharge power; Charging power; , , Maximum charging efficiency; Maximum discharge power; Energy storage charging and discharging process capacity rules: , in: Energy storage battery capacity; SoC with lower limit for energy storage battery capacity; 24-hour content, anytime; Solar surplus energy storage; when hour: The amount of electricity used to charge the energy storage.

[0059] , Prioritize the discharge of stored energy; when At that time, the peak-valley price difference is greater than the energy storage cost: , Other constraints: Peak-valley price difference exceeds energy storage cost. , Peak electricity price; Off-peak electricity prices; The price difference coefficient is generally taken as 0.3 yuan.

[0060] The optimization algorithm used in the model is Particle Swarm Optimization (PSO), which originates from the study of bird flock foraging behavior. The core principle of PSO is to leverage the information sharing among individuals within the swarm, causing the overall swarm's motion to evolve from disordered to ordered within the problem space, thus obtaining the optimal solution. In PSO, particle motion is achieved by maintaining a velocity vector. and position vector The formulas for updating the velocity and position of each particle, as described above, can be expressed as: ; Position update formula: ; Improvements include adding a penalty function and faster optimizations, including: Penalty function: Penalty = Penalty Factor × (Violation of Energy Balance Rules + Grid Connection + Violation of Failure to Store Surplus Solar Energy) , in: It is a penalty item for violating the energy balance rules; It is an internet-related penalty item; This is a violation of the penalty item for not storing surplus photovoltaic energy; It is a penalty factor, equivalent to a coefficient.

[0061] Quick optimization: The particle motion method is modified to ensure that particles move within a reasonable range, thereby improving the convergence and optimization effect of the algorithm.

[0062] Solar negative margin energy storage is a process in which solar power generation occurs when... At that time, particles The mandatory charging amount at any given time is set as follows: .

[0063] Algorithm Flow; The optimal algorithm for energy storage revenue targets employs particle swarm optimization, such as... Figure 4 As shown in the algorithm flowchart, the algorithm process revolves around calculating the benefits and losses related to photovoltaic energy storage, and combines the Particle Swarm Optimization (PSO) algorithm to optimize the energy storage strategy. The process is as follows: First, initialize the environment parameters, then construct the particles and the loss unit; In the loss calculation function stage, frequency conversion calculation, line loss calculation, and charging pile loss calculation are carried out respectively, covering no-load and load conditions. Then we proceed to the particle swarm optimization process: Initialize the particle swarm optimization energy storage strategy; Perform photovoltaic-storage load revenue calculation; Perform loss calculation function calculation; Determine the objective function, including the calculation of overall benefits and penalties; Then determine if the profit is optimal: If the profit is optimal, output the result. If the return is not optimal, the energy storage strategy is updated by particle motion, then the energy storage strategy boundary is checked and processed, and then the process is repeated back to the photovoltaic-storage load return calculation step.

[0064] Model building; the core of the source-storage-load intelligent flexible control model modeling flowchart is based on the particle swarm algorithm, where the particle dimension definition (dim=24) is the energy storage charging and discharging power for 24 hours a day, and the particle motion boundary is the maximum energy storage charging and discharging power.

[0065] Model input: Forecasted 24-hour electricity load; Forecasted 24-hour photovoltaic power generation; Time-of-use pricing for electricity; Current SoC; Energy storage parameters: capacity, power; Model output: 24-hour energy storage charging and discharging power.

[0066] The modeling process is as follows: First, initialize the energy storage strategy, then calculate the photovoltaic-storage revenue, and then calculate the penalty number. Check if the iteration count has been reached. If not, check if the penalty is 0. If the penalty is not 0, update the particle mechanics, perform particle energy storage strategy constraint checks and processing, then update the energy storage strategy with photovoltaic margin assistance, and then update the particle energy storage strategy with the goal of minimizing load balancing damage, before returning to the photovoltaic-storage revenue calculation step. If the penalty is 0, check if the revenue is optimal. If optimal, update the energy storage strategy and revenue, then return to the photovoltaic-storage revenue calculation step. If not optimal, return to the particle mechanics update step. If the iteration count has been reached, output the result.

[0067] Specific implementation method two: The integrated energy station energy storage control device described in this implementation method includes: Data input module: used to collect multi-source data from the integrated energy station system, construct a dataset based on the multi-source data, and divide the dataset into initial daily input data and real-time input data; the multi-source data includes equipment operation data, environmental sensing data, and power grid interaction data; Model and strategy construction module: used to construct an energy storage flexible control model and generate control strategies using the model; the control strategies include a daily energy storage strategy and a real-time energy storage control strategy; wherein, The energy storage all-day strategy is constructed based on the initial input data of the day and is used to perform global planning of the energy storage system and generate a basic energy storage control plan for each hour of the day. The real-time energy storage control is based on the real-time input data and combined with the basic energy storage control plan for each hour of the day to dynamically adjust the energy storage system and output the energy storage control curve. Control command generation module: used to convert the energy storage control curve into fine-tuning control commands and send them to the energy storage system; Strategy optimization module: Used to construct a multi-objective optimization model for energy storage strategies and solve it using an improved particle swarm optimization algorithm to obtain the globally optimal scheduled charging and discharging strategy.

[0068] Specific implementation method three: The storage medium described in this embodiment stores a computer program, which implements the method described in any one of the present invention when running.

Claims

1. An energy storage control method for an integrated energy station, characterized by, The method comprises: S1. Collecting multi-source data of the integrated energy station system, constructing a data set based on the multi-source data, and dividing the data set into initial input data and real-time input data; The multi-source data includes device operation data, environmental perception data, and grid interaction data; S2. Constructing a flexible control model for energy storage based on the multi-source data, and generating a control strategy using the flexible control model for energy storage; The control strategy includes an energy storage daily strategy and an energy storage real-time control; wherein, The energy storage daily strategy is used for global planning of the energy storage system to generate a basic energy storage control plan for each hour of the day; The energy storage real-time control is based on the basic energy storage control plan for each hour of the day to dynamically adjust the energy storage system and output an energy storage control curve; S3. Convert the energy storage control curve into fine-tuning control instructions and send them to the energy storage system; S4. Construct a multi-objective optimization model for energy storage strategy and solve it using an improved particle swarm optimization algorithm to obtain a globally optimal daily charging and discharging strategy.

2. The energy storage control method of claim 1, wherein, The device operation data is obtained by collecting intelligent electric meters and energy storage systems of the integrated energy station, and includes real-time photovoltaic power generation, power load, energy storage system SOC, charging and discharging state, and health status data.

3. The energy storage control method of claim 1, wherein, The method for constructing the flexible control model for energy storage in S2 comprises: Based on the environmental perception data and grid interaction data, perform photovoltaic power generation prediction and power load prediction to obtain prediction results including a daily photovoltaic power generation curve and a load prediction curve; Based on the prediction results and the device operation data, construct the flexible control model for energy storage.

4. The energy storage control method of claim 1, wherein, The construction process of the energy storage daily strategy in S2 is as follows: Input the initial input data; Set the execution time of the control strategy to 0 o'clock every day; Set the strategy optimization target to system economy, load balance, and maximum photovoltaic self-use; Set the constraint conditions to power, capacity, and charging and discharging frequency limits; Use the particle swarm algorithm to optimize the energy storage strategy and obtain a charging and discharging power curve for each hour of the day to generate a daily charging and discharging strategy; Based on the energy storage device cluster, establish an energy storage device model library, divide energy storage device units, and perform strategy matching to determine different energy storage device unit distribution models; According to different energy storage device unit distribution models, perform flexible decomposition of the daily charging and discharging strategy to map to an output basic energy storage control plan for each hour of the day.

5. The energy storage control method of claim 4, wherein, The flexible decomposition is to map the energy storage daily strategy to the energy storage system, divide the device operation state into charging, standby, and discharging, and output to the devices of the energy storage system in a time sequence control mode to execute the control strategy.

6. The energy storage control method of claim 1, wherein, The process of dynamically adjusting the energy storage system in S2 comprises: Input the real-time input data; Set the real-time fine-tuning interval time for energy storage to 10 minutes; Obtain adjusted control instructions through real-time fine-tuning; Send the adjusted control instructions to the energy storage system; Schedule the devices according to the priority of stable device operation state; Output fine-tuning control instructions executed every 10 minutes to generate an energy storage control curve.

7. The energy storage control method of claim 6, wherein, The real-time fine-tuning includes re-acquiring multi-source data, deviation evaluation with the currently executed control strategy, and power charging and discharging micro-correction according to real-time photovoltaic, load, and electricity price.

8. The energy storage control method of claim 1, wherein, The process of S4 includes: S41. Setting a model optimization target includes: Maximum variable revenue of light storage: ， Minimum transformer loss: ， In the formula, represents the load power consumption; represents the overall mean; represents the number of consecutive low valley hours; represents the start point to the end point of the consecutive low valley price; represents the total variance; S42. Setting a model constraint condition includes: The charging and discharging power is less than or equal to the maximum power: , ; wherein t represents the charging and discharging time, represents the discharging power, represents the charging power; represents the maximum charging power, represents the maximum discharging power; Energy storage charge and discharge process capacity rule: ; wherein, represents the energy storage battery capacity; represents the lower limit of the energy storage battery capacity; represents any time within 24 hours; Photovoltaic surplus energy storage: when the energy storage charging power is zero. Energy storage preferential discharge: when , , indicates that the peak-valley price difference is greater than the cost of energy storage; S43. Using an improved particle swarm optimization algorithm for solving; Set the velocity vector of each particle and the position vector Update the particle velocity and position to be: ， A penalty function is introduced: ， wherein is a violation of the energy balance rule penalty term; is an on-grid penalty term; is a violation of the PV surplus not stored energy penalty term; is a penalty factor; S44. Correct the particle velocity and position constraints to adjust the particle motion range, combine the real-time surplus of photovoltaic power generation to dynamically adjust the energy storage charging capacity, when the particle charging capacity at the moment is: , the energy storage SOC change curve is obtained; through iteration until the optimal all-day charging and discharging strategy is obtained.

9. An integrated energy station energy storage control device, characterized by, The device includes: A data input module: configured to acquire multi-source data of the integrated energy station system, construct a data set based on the multi-source data, and divide the data set into initial daily input data and real-time input data; the multi-source data includes equipment operation data, environmental perception data, and power grid interaction data; A model and strategy construction module: configured to construct an energy storage flexible control model, and generate a control strategy using the energy storage flexible control model; the control strategy includes an energy storage whole-day strategy and an energy storage real-time control; wherein, The energy storage whole-day strategy is constructed based on the initial daily input data, used for global planning of the energy storage system, and generating a whole-day hourly basic energy storage control plan; The energy storage real-time control is based on the real-time input data, combined with the whole-day hourly basic energy storage control plan, used for dynamic adjustment of the energy storage system, and outputting an energy storage control curve; A control instruction generation module: configured to convert the energy storage control curve into fine-tuning control instructions, and send the fine-tuning control instructions to the energy storage system; A strategy optimization module: configured to construct an energy storage strategy multi-objective optimization model, and use an improved particle swarm optimization algorithm for solving, to obtain a globally optimal daily charging and discharging strategy.

10. A storage medium, characterized by A computer program is stored in a storage medium, and the computer program realizes the method of any one of claims 1 to 8 when running. A computer program is stored in a storage medium, and the computer program realizes the method of any one of claims 1 to 8 when running.

Images