A generalized distributed energy storage participating in power distribution network operation aggregation regulation method

Abstract

The present application relates to a kind of generalized distributed energy storage involved in power distribution network operation aggregation regulation method, this method with resource layer-aggregation layer-scheduling layer three-layer architecture as bearing framework, multi-objective particle swarm optimization (MOPSO) is embedded in the day-ahead optimization link of scheduling layer, and coupling medium and long-term contract constraint and day-to-day rolling correction mechanism, form "contract constraint-day-ahead optimization-day-to-day correction" time sequence progressive closed loop, applicable to the application scenarios of power distribution network including 5G base station energy storage, electric vehicle energy storage and conventional independent energy storage, the distributed energy storage aggregation regulation of source network load storage collaborative benefit maximization as target is realized;The present application has the advantages of three-layer regulation architecture, constructing aggregation regulation model, realizing collaborative regulation, improving the stability and economy of distribution network.

Description

Technical Field

[0001] This invention belongs to the field of power system operation and control technology, specifically relating to a method for aggregated regulation of generalized distributed energy storage participating in distribution network operation. Background Technology

[0002] As the global energy structure accelerates towards decarbonization and cleaner energy, the installed capacity of new energy sources, represented by wind power and photovoltaics, continues to grow. According to statistics, the proportion of global renewable energy power generation exceeded 30% in 2022 and is expected to exceed 50% by 2030. However, the intermittency and volatility of new energy sources pose serious challenges to the safe operation of the power system, especially at the medium and low voltage distribution network level, where problems such as widening load peak-valley differences and insufficient new energy absorption capacity are becoming increasingly prominent. Distributed energy storage (DES), as a flexible regulation resource, is considered a key technology for improving the resilience of the distribution network and the capacity for new energy absorption due to its advantages such as proximity to the load side, fast response speed, and distributed deployment. Broadly speaking, distributed energy storage not only includes traditional electrochemical energy storage power stations but also encompasses diverse forms such as electric vehicles (EVs), 5G base station backup batteries, and residential energy storage. It is predicted that the global installed capacity of distributed energy storage will exceed 400GW by 2030, with electric vehicle energy storage accounting for more than 60% of the potential. Against this backdrop, how to achieve source-grid-load-storage synergistic optimization through efficient regulation of flexible resources has become one of the core issues in the construction of new power systems.

[0003] Due to the heterogeneity of distributed energy storage and the interests of multiple stakeholders, current distributed energy storage faces the following challenges in large-scale aggregation and coordinated control: The operating characteristics of distributed energy storage vary significantly, making it difficult for existing centralized control methods to adapt to their complex dynamic characteristics; In balancing multiple grid objectives such as peak shaving and renewable energy consumption with the economic interests of all parties, there are obvious coordination conflicts among these objectives, making it difficult to achieve Pareto optimality; The single-layer architecture of traditional virtual power plants lacks hierarchical collaborative design, failing to balance global optimization and local autonomy, and thus failing to meet the real-time response requirements of massive distributed resources; Existing distributed energy storage control strategies generally lack system modeling for the multi-timescale market of "medium-to-long-term contracts - day-ahead plans - intraday execution," and the hierarchical architecture design is mostly static functional definition, not deeply coupled with market timing, resulting in insufficient market adaptability and engineering practicality of the strategies; Therefore, it is essential to provide a generalized distributed energy storage aggregation control method for distribution network operation based on a three-layer control architecture, constructing an aggregation control model, achieving coordinated control, and improving distribution network stability and economy. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a generalized decentralized energy storage aggregation control method for distribution network operation, based on a three-layer control architecture, constructing an aggregation control model, realizing coordinated control, and improving the stability and economy of the distribution network.

[0005] The objective of this invention is achieved as follows: a generalized distributed energy storage aggregation and control method for participating in distribution network operation, using a three-layer architecture of "resource layer-aggregation layer-scheduling layer" as the physical carrier, setting hierarchical collaborative rules at three time scales: medium-to-long-term, day-ahead, and intraday, and employing a multi-objective particle swarm optimization algorithm as the core solution algorithm for day-ahead optimization of the scheduling layer, forming a strategy system that is architecture-supported, algorithm-driven, and time-series progressive. The method includes the following steps:

[0006] Step 1: Construct a three-layer collaborative hierarchical control architecture of "resource layer - aggregation layer - scheduling layer", complete the dynamic behavior modeling of generalized distributed energy storage, define the multi-objective optimization problem and build the algorithm framework, clarify the functional division of each layer in the three time scales of medium- and long-term, day-ahead and intraday, and establish the basic theoretical model for generalized distributed energy storage to participate in the operation of the distribution network.

[0007] Step 2: Based on the aforementioned fundamental theoretical model, and combining the multi-objective optimization framework with the multi-objective particle swarm optimization algorithm, construct a generalized decentralized energy storage aggregation control optimization model for participating in distribution network operation;

[0008] Step 3: Select a typical urban load area under a typical distribution network scenario, design a simulation case involving multiple entities and flexible resources, and conduct simulation verification and effect analysis of the control strategy.

[0009] The beneficial effects of this invention are as follows: This invention is a generalized decentralized energy storage aggregation and control method for participating in distribution network operation. In use, this invention uses a three-layer architecture of resource layer-aggregation layer-scheduling layer as the carrying framework, embedding the multi-objective particle swarm optimization (MOPSO) algorithm into the day-ahead optimization stage of the scheduling layer, and coupling medium- and long-term contract constraints with intraday rolling correction mechanism to form a time-series progressive closed loop of "contract constraints-day-ahead optimization-intraday correction". It is applicable to distribution network application scenarios including 5G base station energy storage, electric vehicle energy storage and conventional independent energy storage, and realizes decentralized energy storage aggregation and control with the goal of maximizing the synergistic benefits of source-grid-load-storage. This invention has the advantages of being based on a three-layer control architecture, constructing an aggregation and control model, realizing collaborative control, and improving the stability and economy of distribution networks. Attached Figure Description

[0010] Figure 1 This is a flowchart of the present invention.

[0011] Figure 2 This is a diagram of the layered architecture of the distributed energy storage of the present invention.

[0012] Figure 3 This is a framework diagram of the multi-objective particle swarm optimization algorithm of the present invention.

[0013] Figure 4 This is a line graph showing the 5G base station zone load power data of the present invention.

[0014] Figure 5 This is a graph showing the average number of power outages and the average annual power outage time according to the present invention.

[0015] Figure 6 This is a bar chart showing the photovoltaic output of the present invention.

[0016] Figure 7 This is a diagram of the time-of-use electricity pricing model of the present invention.

[0017] Figure 8 This is a PSO convergence curve diagram of the present invention.

[0018] Figure 9 This is a comparison chart of the total load before and after optimization according to the present invention.

[0019] Figure 10 This is an optimization diagram of the energy storage charging and discharging power in different regions according to the present invention.

[0020] Figure 11 This is a graph showing the SOC optimization curves for each region of this invention.

[0021] Figure 12 This is a comparison chart of SOC optimization in different regions of the present invention.

[0022] Figure 13 This is a graph showing the average SOC (State of Charge) curve of the electric vehicle charging station according to the present invention.

[0023] Figure 14 This is a graph showing the average SOC curve of the 5G base station according to the present invention.

[0024] Figure 15 This is a typical SOC curve for standalone energy storage according to the present invention. Detailed Implementation

[0025] This invention addresses the challenges of aggregation and control, multi-objective coordination conflicts, and the lack of a hierarchical architecture in distributed energy storage in power distribution networks due to their strong heterogeneity, dispersed spatiotemporal distribution, and conflicts of interest among multiple stakeholders. It proposes a hierarchical aggregation and control strategy for generalized distributed energy storage to maximize the synergistic benefits of source-grid-load-storage and improve the stability and economy of power distribution network operation.

[0026] The present invention will now be further described with reference to the accompanying drawings.

[0027] Example 1

[0028] like Figure 1-15 As shown, a method for aggregated control of generalized distributed energy storage participating in the operation of a distribution network includes the following steps:

[0029] Step 1: By constructing a three-layer collaborative hierarchical control architecture of "resource layer - aggregation layer - scheduling layer", dynamic behavior modeling, multi-objective optimization problem definition and algorithm framework, and defining the functional division of each layer under the three time scales of medium- and long-term, day-ahead and intraday, a theoretical model for the collaborative optimization of generalized distributed energy storage system in distribution network is established.

[0030] In this embodiment, the theoretical model for the collaborative optimization of a generalized distributed energy storage system in a distribution network employs the following control strategy under a multi-timescale market: The dispatch layer generates baseline charging and discharging capacities for each aggregate based on peak-shaving contracts in the medium to long term, and generates charging and discharging plans based on forecast data in the day-ahead period; the aggregation layer continuously corrects the deviation between the actual SOC and the day-ahead plan every 15-30 minutes during the day; and the resource layer satisfies local security constraints during execution. This achieves progressive control through medium- to long-term contract constraints, day-ahead multi-objective optimization, and intraday rolling correction.

[0031] The core hierarchy and functions of the model include:

[0032] ① Resource Layer: The resource layer consists of distributed energy storage units and local loads; distributed energy storage units include backup batteries for 5G base stations, energy storage in industrial and commercial parks, electric vehicles, etc.; the resource layer has local autonomy and can meet its own load needs such as base station communication and EV charging; it has dynamic adjustability and can adjust the charging and discharging power according to the instructions of the superior; there are backup constraints.

[0033] ② Aggregation Layer: The aggregation layer is implemented by regional agents, such as base station smart terminals and park energy management systems (EMS). Its core functions include state aggregation, which is to collect dynamic parameters such as energy storage SOC and power limits in real time; capability mapping, which unifies the frequency regulation capability of heterogeneous energy storage into dispatchable capacity; and instruction decomposition, which adapts global instructions into executable operations for each unit.

[0034] ③ Scheduling Layer: The scheduling layer consists of the distribution network scheduling center and the market trading platform. Its functional design includes global optimization based on the distribution network load curve and the market electricity price generation optimization target; coordinated control of the charging and discharging timing and power allocation of multiple types of energy storage; and market interaction in response to time-of-use electricity price signals and frequency regulation ancillary service demands. The advantage of this architecture is that it is compatible with the dynamic characteristics of different types of energy storage through layered design, such as the randomness of EV charging and base station backup constraints; it realizes edge-cloud collaboration, regional agents support local rapid response, and the scheduling layer realizes global optimization, taking into account both real-time performance and economy.

[0035] Specifically, the basic dynamic model and economic modeling of energy storage are as follows:

[0036] I. Basic Dynamic Model of Energy Storage

[0037] ① Evolution equation of state of charge (SOC): Energy storage unit The SOC evolves over time as follows: In the formula, For energy storage units During the period The state of charge; These are the energy storage charging and discharging efficiencies (unit: %). For time period Energy storage unit The charging and discharging power (kW); The time step (unit: hours).

[0038] ②SOC safety boundary constraints: The energy storage SOC must meet operational safety requirements: In the formula, For energy storage units The minimum state of charge (lower limit); For energy storage units Maximum state of charge (upper limit);

[0039] Lower limit dynamic adjustment: dynamically set based on regional power supply reliability requirements, combined with historical power outage data and load characteristics.

[0040] ③ Charge and discharge power limits: In the formula, Energy storage units Maximum permissible charging and discharging power (kW).

[0041] II. Economic Model

[0042] ① Time-of-use pricing response cost: The cost of energy storage charging and discharging is deeply coupled with the electricity price period. The electricity purchase cost function is: In the formula, For time period The cost of electricity purchase (in yuan); For time period The time-of-use electricity price (RMB / kWh).

[0043] ② Equipment wear and tear costs: Frequent switching of charging and discharging power leads to equipment wear and tear. The cost model is as follows: In the formula, Cost of equipment wear and tear (in yuan) caused by charging and discharging energy storage. This is the loss factor, which is related to the type of energy storage and historical operating data (yuan / kW).

[0044] Specifically, the multi-objective optimization problem is defined as follows:

[0045] I. Optimization Objectives

[0046] ① Maximizing grid regulation capacity: Suppressing load fluctuations through energy storage charging and discharging levels; the objective function is: In the formula, For time period Net load (kW); Average daily load (kW); For power grid regulation capacity; This represents the total number of time periods.

[0047] ② Optimize operational economy: Minimize the total lifecycle cost of energy storage. In the formula, The total cost of energy storage over its entire lifecycle (in yuan).

[0048] II. Constraints

[0049] ① Power grid security constraints: This ensures that the optimized load does not exceed the distribution network's carrying capacity limit.

[0050] ② Constraints on renewable energy consumption: Energy storage charging power must be compatible with renewable energy output. In the formula, For time period The new energy source can absorb the power.

[0051] As a feasible implementation method, the proposed solution framework based on swarm intelligence algorithms, addressing the high dimensionality and nonlinearity of multi-objective optimization problems, must simultaneously meet three requirements: Pareto optimality to simultaneously optimize multiple conflicting objectives and generate a non-dominated solution set; constraint compatibility to efficiently handle mixed-integer nonlinear constraints; and computational efficiency to meet the real-time requirements of large-scale distributed energy storage scenarios. Furthermore, it adheres to three principles: encoding energy storage charging and discharging power decision variables as particle positions to achieve solution space mapping; balancing global exploration and local exploitation through dynamic weight adjustment to avoid premature convergence; and employing an adaptive penalty function to construct a constraint relaxation mechanism to improve the feasibility of solutions under complex constraints.

[0052] Step 2: Based on the theoretical model of generalized distributed energy storage participating in distribution network operation, and in conjunction with a multi-objective optimization framework and a multi-objective particle swarm optimization algorithm, construct an aggregated control model for generalized distributed energy storage participating in distribution network operation.

[0053] In this embodiment, the objective function and constraints of the aggregation control model for generalized distributed energy storage participating in distribution network operation are as follows:

[0054] I. Objective Function

[0055] Objective function 1: Minimize the peak-to-valley difference: In the formula, This represents the maximum (peak) net load for the entire day. This represents the minimum (valley) value of the net load for the entire day. ,in, For time period Net load after internal optimization (kW); For time period The original load (kW).

[0056] Objective function 2: Minimize the total operating cost of energy storage: In the formula, The electricity price is fixed (RMB / kWh). A collection of energy storage units; This represents the total operating cost of the energy storage system.

[0057] II. Constraints

[0058] ① Peak-to-valley difference definition constraints:

[0059] ② New energy consumption constraints: Energy storage charging power shall not exceed the available surplus power of new energy. In the formula, For time period The surplus power of new energy sources can be used.

[0060] ③ Dynamic constraints of energy storage SOC: .

[0061] ④ Dynamic SOC lower limit model: In the formula, For energy storage units The inherent safety lower limit; For the region Backup power supply duration requirements for (industrial / residential / commercial areas); For the region During the period The load power; For energy storage units Rated capacity.

[0062] Energy storage SOC safety constraints: .

[0063] ⑤ Charge and discharge power limits: .

[0064] ⑥ Peak-shaving service demand constraints: Define peak periods Hegu period ,satisfy: In the formula, The minimum total discharge amount (kWh) required for peak shaving; The minimum total charging amount (kWh) required for peak shaving.

[0065] ⑦ Power purchase and sale restrictions: .

[0066] III. Optimized Output

[0067] After the optimization was completed, the scheduling layer output the following parameters as a reference for lower-level control:

[0068] ① Hourly charge / discharge power plan: In the formula, For energy storage units exist The charging power given by the time-schedule layer, in kW; Energy storage unit exist The discharge power given by the time-of-flight scheduling layer, in kW; This is a collection of times from the previous day.

[0069] ② Target state of charge at the top of the hour: In the formula, For energy storage units exist The target value of the remaining power percentage required by the time-scheduling layer.

[0070] The above parameters constitute the control baseline that each energy storage unit should follow at each time period of the next day. The lower aggregation layer and resource layer perform progressively refined adjustments under the constraints of this baseline.

[0071] Specifically, this invention divides the capability mapping of three types of heterogeneous energy storage units—5G base station energy storage, electric vehicle energy storage, and conventional independent energy storage—into three steps:

[0072] ① Heterogeneous parameter standardization: Standardize the differentiated physical parameters (rated capacity) of various energy storage systems. Power limits and ,efficiency and Dynamic SOC lower limit SOC limit A unified representation is provided as a standardized set of energy storage model variables, ensuring that all energy storage units are incorporated into a unified mathematical framework.

[0073] ② Capacity Boundary Aggregation: The aggregation layer transforms the standardized capacity parameters of each energy storage unit into an equivalent dispatchable capacity through superposition calculation.

[0074] Upward adjustable capability of aggregates: Downward adjustable capability of the polymer: In the formula, For polymers in The adjustable capacity at any given time (total amount of electricity that can continue to discharge), in kWh; For polymers in Downward adjustable capacity at any given time (total amount of electricity that can continue to be discharged), in kWh.

[0075] ③ Instruction Differentiation Decomposition: The aggregation layer decomposes global power instructions. The commands are broken down into individual instructions based on the adjustable capacity of each energy storage unit: In the formula, For energy storage units exist The current state of charge at any given moment; To be decomposed to energy storage units Individual power command, in kW; This refers to the global power command received by the aggregation layer, in kW.

[0076] This means that a proportional allocation method based on adjustable capacity is adopted, in which each unit is allocated instructions according to the proportion of its remaining adjustable capacity to the total adjustable capacity of the aggregate, thereby realizing the reverse mapping from unified scheduling instructions to differentiated execution of heterogeneous devices.

[0077] As a feasible implementation method, the multi-objective particle swarm optimization (MOPSO) algorithm for the aggregation and control model of generalized distributed energy storage participating in distribution network operation is designed as follows:

[0078] ① Inertia weight adjustment: In the formula, Inertial weights; These are the initial and final inertia weights (default 0.9 and 0.4). This represents the current iteration number.

[0079] ② Learning factor adjustment: In the formula, These are individual learning factors and social learning factors, respectively.

[0080] ③ Pareto front maintenance: External archive stores non-dominated solutions, and diversity is maintained by sorting by crowding distance.

[0081] ④ Algorithm Flow

[0082] Particle encoding: The objective function includes optimization of the sum of peak-to-valley differences at each time period, and adjusts the fitness function calculation logic.

[0083] Constraint handling: The penalty function method is adopted, and the penalty coefficient is tuned through pre-run (e.g., the penalty term is 100 times the sum of the constraint violations): with the goal of achieving a population feasibility rate of 60%-80% in the mid-cycle, the minimum penalty coefficient that meets this requirement is selected; the constraints violations are summed after dimensional normalization, and multiplied by this coefficient and included in the fitness function.

[0084] Iteration termination condition: maximum number of iterations 200 or Pareto solution set convergence.

[0085] Specifically, the parameters of the aggregation control model for generalized distributed energy storage participating in distribution network operation are as follows:

[0086] 1. Basic parameters of distribution network and energy storage

[0087] ① 5G base station energy storage: capacity Charge and discharge power limits: , , ; In the formula, For the region The backup power time (in hours) for (industrial areas, residential areas, and commercial areas) is calculated based on the average annual power supply time and the annual failure rate. For the region During the period The load power (kW); For energy storage units Rated capacity (kWh).

[0088] This article uses the benchmark of 5G base stations guaranteeing 3 hours of backup power in residential areas to derive the energy storage backup time for other load areas: In the formula, For base stations Minimum energy storage standby time; For base stations Annual failure rate of the land use type; For base stations Average annual power outage time for the land use type; and These are the average number of power outages and the average duration of power outages per year in residential areas, respectively. This is the baseline energy storage backup time.

[0089] ② Electric Vehicles (EVs): Charging Time Allocation: Private car charging behavior shows a significant time preference, with 30% of charging occurring during the day (8:00-18:00) and 70% during the night (18:00-8:00).

[0090] Access time modeling: Daytime charging probability follows a uniform distribution, concentrated at charging stations in workplaces; nighttime charging probability follows a normal distribution. (20:00,2 2 ), simulating user charging behavior after returning home; mathematical model analysis: mean ( The highest probability of charging is at 8 PM; standard deviation ( ): The degree of temporal dispersion of charging behavior; approximately 68% of charging events are concentrated between 18:00 and 22:00 (mean). ), 95% are concentrated between 16:00 and 24:00 (average) ).

[0091] Parameter settings: Capacity The initial SOC follows a normal distribution N(0.4, 0.1). 2 ); Expected off-grid power consumption Charge / discharge power limits: , .

[0092] ③ Conventional energy storage: capacity , , Charge / discharge power limits: , .

[0093] 2. Economic parameters: Time-of-use electricity price : Energy storage maintenance cost coefficient ;

[0094] 3. Algorithm parameters: Particle swarm size: Maximum number of iterations: Inertia weight range: , Learning factor adjustment: , .

[0095] As another feasible implementation, the aggregation layer performs intraday corrections on a rolling basis with a 15-minute cycle, and the correction model is as follows:

[0096] 1. Deviation detection: at each correction time. (in (This is a set of intraday correction times), the aggregation layer collects the real-time SOC of each energy storage unit and calculates the deviation from the target SOC of the day: In the formula, For energy storage units exist SOC deviation value at time; For energy storage units exist The actual remaining battery level at any given time; For energy storage units exist The target remaining power level that the time-scheduling layer is required to achieve; This is a daily time index, ranging from 0 to 95, representing 96 15-minute intervals in a day.

[0097] 2. Modify trigger conditions: If Correct the objective function: In the formula, The objective function value is corrected intraday; Indicates from the current moment By the end of the day Every moment in between; This is the current 15-minute point in time where the aggregation layer is making corrections; This is the last moment of the day, specifically the 95th 15-minute mark (23:45). For energy storage units exist The remaining battery power at any given time; For energy storage units exist The target remaining battery level at any given time; It is the collection of all energy storage units; The SOC deviation tolerance threshold is set within a reasonable range based on factors such as prediction accuracy, energy storage type, and communication conditions.

[0098] 3. Correct power limiting constraints: ,in, Take ±10% to ±20% of the planned power output for the day to ensure that intraday corrections do not deviate from the overall direction of the day's plan, in kW; The original charging power was determined recently, in kW; The newly calculated charging power after the polymer layer correction, in kW.

[0099] Specifically, the resource layer receives and executes power commands issued by the aggregation layer at a resolution of seconds: In the formula, For energy storage units exist The actual charging power delivered at any given time is the physical quantity actually output by the device, measured in kW. For energy storage units exist The actual discharge power delivered at any given moment is the physical quantity actually output by the device, measured in kW. The charging command issued after correction by the aggregation layer, in kW; This refers to the discharge command issued after the polymer layer has been corrected, in kW.

[0100] Meanwhile, the resource layer local controller continuously monitors the real-time SOC of each energy storage unit. When this occurs, autonomous protection is triggered: This means forcibly stopping the discharge; at the same time, reporting the violation status indicator. The energy is then transferred to the aggregation layer, where it reallocates the output of other online energy storage units.

[0101] When SOC recovers to: ( After taking 2%, a "recovery ready" signal is sent to the aggregation layer, and the data is reinstated into the scheduling sequence.

[0102] Specifically, the three levels are coupled on a time scale through the following constraints:

[0103] 1. Day-to-Day Coupling (Coarse-to-Fine Mapping): The State of Charge (SOC) at any point during the day must be within the allowable range of the day-to-day target value. In the formula, For energy storage units On the hour The target remaining battery power at any given time is a fixed value set by the scheduling layer; Tolerance for SOC deviation; For energy storage units 15 minutes The remaining battery power, a variable in the intraday adjustment plan; Index for 15-minute time intervals (0,1,2,...,95); That is, every 4 intraday time periods (15 minutes × 4 = 1 hour) correspond to the target for the hour before the day.

[0104] 2. Intraday → Real-time Coupling (Command Refinement): Real-time power is based on intraday corrected power. In the formula, For energy storage units In real time The actual charging power is the value that the device is actually going to perform, measured in kW; For energy storage units The intraday corrected charging power at the corresponding 15-minute interval is based on the instruction benchmark given by the aggregation layer, in kW. This is the proportional feedback coefficient. ; This refers to the actual SOC measured in real time.

[0105] 3. Real-time → Day-ahead Coupling (Feedback Closed Loop): After the end of each day, the tracking error is statistically analyzed and fed back to the next day. In the formula, This represents the average difference between the actual SOC of all devices and the target SOC at all times today. This represents the total number of time points within the day, specifically 96 15-minute time points. This represents the actual remaining battery power. The target remaining battery level.

[0106] The following day, the scheduling layer adaptively adjusts the planning margin based on this error, forming a daily iterative optimization closed loop. In the formula, This is the learning rate coefficient; Energy storage is part of tomorrow's daily plan. exist The target SOC at any given moment; Energy storage is part of today's plan. exist The target SOC at any given moment; This is the SOC that was actually implemented today.

[0107] Step 3: For typical distribution network scenarios, design a multi-timescale market model case study that includes medium- and long-term peak-shaving capacity contract constraints, day-ahead multi-objective optimization, and intraday rolling correction. Combine the control model with actual load data and renewable energy output curves to conduct simulation analysis and verify the effectiveness of the strategy.

[0108] In this embodiment, a multi-stakeholder integrated energy management simulation case is designed for a typical urban load area to verify the effectiveness of the multi-regional collaborative optimization strategy. Based on the already constructed generalized distributed energy storage aggregation and control model, simulation analysis is conducted using actual load data and new energy output curves. To verify the effectiveness of the multi-regional collaborative optimization strategy, a multi-stakeholder integrated energy management simulation case is designed for a typical urban load area: the system includes three typical load units: industrial area, commercial area, and residential area. To simulate the spatiotemporal uncertainty of electric vehicle charging behavior, the Monte Carlo method is used to generate 1001 sets of charging demand samples, with the charging power following a constant power mode of 6.6kW, and the grid connection time deviation meeting the requirements. The system follows a normal distribution over hours; there are a total of 50 5G base stations, with 20 in industrial and residential areas and 10 in commercial areas. The emergency backup time for residential areas is set to 3 hours; the conventional independent energy storage system consists of 3 load units, 3 power generation units, and energy storage units operating in conjunction, with an upper limit of 0.95 and a lower limit of 0.1 SOC.

[0109] Case Study 1: Using the aforementioned model and method, simulation analysis was conducted in a typical power distribution network scenario to verify the effectiveness of the generalized distributed energy storage aggregation and control strategy. The specific scenario is a typical daily load curve in a certain region during summer, which includes three forms of distributed energy storage: 5G base stations, electric vehicles, and conventional independent energy storage. The simulation time range is 24 hours, with a time interval of 15 minutes. This time scale is chosen to capture the dynamic changes in load and energy storage while ensuring computational efficiency.

[0110] Simulation results analysis: The optimized load curve achieves load smoothing and peak-valley balance. By adjusting the three energy storage charging and discharging strategies, the load utilization rate is increased during off-peak hours and the grid pressure is reduced during peak hours, thus achieving the goal of optimizing the grid load distribution.

[0111] In industrial areas: During peak load periods, energy storage is prioritized for discharge to smooth load fluctuations; during off-peak periods, energy storage is charged to absorb excess energy. This strategy significantly reduces the peak-to-valley difference in industrial areas and improves grid stability.

[0112] Commercial areas: Due to relatively stable loads, energy storage regulation is less frequent. Primarily, charging occurs during off-peak hours and discharging during peak hours to maximize profits.

[0113] Residential areas: The charge-discharge curves are relatively smooth, mainly responding to changes in electricity prices. User-side energy storage discharges during peak hours to reduce electricity costs.

[0114] The SOC curves in all regions fluctuated within a safe range. Industrial and commercial areas showed relatively small SOC changes, while residential areas experienced greater SOC fluctuations due to electricity price influences. Energy storage dispatch achieved dynamic optimization while ensuring safety. A comparison of SOC optimization across regions revealed that industrial areas experienced greater load fluctuations, and optimized dispatch improved SOC stability. Residential areas exhibited strong load regularity, resulting in efficient operation of the energy storage system. Commercial areas showed relatively stable loads, leading to more balanced energy storage optimization. Overall, the SOC variation characteristics across different regions reflect differences in load characteristics, and optimization strategies need to be adjusted according to regional characteristics to improve overall electricity efficiency.

[0115] Dynamic stability analysis of the state of charge (SOC) of energy storage systems in different scenarios: SOC mean curve of electric vehicle charging stations: Charging stations with large SOC fluctuations may need to optimize charging and discharging scheduling strategies to divert vehicles during high-load periods; while charging stations with smaller SOC changes can further improve the utilization rate of charging piles and improve operating efficiency.

[0116] 5G base station SOC mean curve: The energy storage system of the 5G base station can operate stably, and its charging and discharging behavior matches the power demand of the base station well, indicating that the battery energy management strategy is relatively effective. At the same time, the SOC curve is close to the middle range, indicating that the energy storage system has achieved a good dynamic balance between charging and discharging, which can meet the power demand of the base station while avoiding excessive battery discharge or energy waste after full charge. The stability and efficiency of the 5G base station energy storage system provide a reliable guarantee for the continuous operation of the base station, reflecting the rationality of the energy storage management strategy.

[0117] Conventional independent energy storage SOC curve: The curve reflects the dynamic charging and discharging characteristics of the energy storage system under independent operating conditions; the conventional independent energy storage system can effectively regulate charging and discharging under dynamic load conditions to maintain stable system operation; the fluctuation characteristics reflect the flexibility of the energy storage system in responding to the load, and the fact that the SOC curve does not exceed the upper and lower limits indicates that its management strategy is reasonable and its operation is reliable.

[0118] Economic efficiency assessment of multi-objective coordinated scheduling: Under the influence of time-of-use pricing: Off-peak charging amount: 601.97 kWh, charging cost: 324.38 yuan; Peak discharge amount: 243.39 kWh, discharge revenue: 243.39 yuan; Peak discharge amount: 208.65 kWh, peak discharge revenue: 152.72 yuan.

[0119] Sensitivity analysis of weight parameters in multi-objective optimization: See the table below. It is the penalty coefficient for photovoltaic curtailment; It is the penalty coefficient for fluctuations in purchased electricity; It is the peak-valley difference penalty weight.

[0120] The parameter settings and results are as follows:

[0121] Table 1 Parameter Settings 1

[0122]

[0123] Table 2 Comparison of economic efficiency and peak-valley difference before and after optimization 1

[0124]

[0125] Table 3 Parameter Settings 2

[0126]

[0127] Table 4 Comparison of economic efficiency and peak-valley difference before and after optimization 2

[0128]

[0129] Table 5 Parameter Settings 3

[0130]

[0131] Table 6 Comparison of economic efficiency and peak-valley difference before and after optimization 3

[0132]

[0133] By adjusting the peak-valley difference penalty term coefficient, the aggregation control strategy of generalized distributed energy storage can be effectively optimized, significantly reducing the peak-valley difference in the distribution network and improving the stability of distribution network operation. The simulation results show that after adopting the generalized distributed energy storage aggregation control strategy: the peak-valley difference is significantly reduced, the load curve is smoother, the energy storage economy is improved, and the total operating cost is reduced; the SOC of each region is stable, meeting the requirements for safe operation. Through simulation analysis, it can be seen that the generalized distributed energy storage aggregation control strategy has a significant optimization effect in the actual distribution network; energy storage not only effectively smooths the load curve and reduces the peak-valley difference, but also achieves economic improvement under the coordination of multiple objectives; the SOC of the three energy storage forms remains within the safe range, proving the reliability of the control strategy.

[0134] Example 2

[0135] like Figure 1-15 As shown, a generalized distributed energy storage aggregation control method for participating in distribution network operation is proposed, such as... Figure 1 As shown, it includes the following steps:

[0136] Step one involves establishing a three-tiered control architecture—resource layer, aggregation layer, and scheduling layer—through hierarchical control architecture, dynamic behavior modeling, multi-objective optimization problem definition, and algorithm framework design. The scheduling layer is responsible for setting medium- to long-term capacity baselines and day-ahead optimization; the aggregation layer is responsible for intraday rolling adjustments; and the resource layer is responsible for local security protection and autonomy. Its core layers and functions are as follows: Figure 2 As shown;

[0137] In this embodiment, a dynamic model and an economic model of the energy storage infrastructure are established; a multi-objective optimization problem is defined, and optimization objectives and constraints are proposed; furthermore, considering the high dimensionality and nonlinearity of the multi-objective optimization problem, a solution framework based on swarm intelligence algorithms is proposed, such as... Figure 3 As shown.

[0138] Step 2: Based on the theoretical model of generalized distributed energy storage participating in distribution network operation, an aggregated control model for generalized distributed energy storage participating in distribution network operation is constructed through a multi-objective optimization framework and algorithm design to achieve synergistic optimization of load smoothing, economic improvement and safe operation of energy storage.

[0139] In this embodiment, firstly, the objective function and constraints of the aggregation control model for generalized distributed energy storage participating in distribution network operation are established; then, the multi-objective particle swarm optimization (MOPSO) algorithm for the aggregation control model for generalized distributed energy storage participating in distribution network operation is designed, including inertia weight adjustment, learning factor adjustment, Pareto front maintenance, and algorithm flow; finally, the parameters of the aggregation control model for generalized distributed energy storage participating in distribution network operation are set, including basic parameters of distribution network and energy storage, economic parameters, and algorithm parameters; wherein the basic parameters of distribution network and energy storage include: 5G base station energy storage (reference data as follows) Figure 4 , Figure 5 As shown), electric vehicles (EVs), and conventional energy storage.

[0140] Step 3: To verify the effectiveness of the multi-regional collaborative optimization strategy, a comprehensive energy management case study involving multiple stakeholders was designed for a typical urban load area.

[0141] Case Study 2: This example, based on day-ahead optimization, additionally sets up medium- and long-term peak-shaving capacity contracts (e.g., minimum discharge targets for each aggregate during peak hours) and intraday disturbance correction scenarios (a sudden 30% drop in photovoltaic output). The dispatch layer first generates day-ahead dispatch boundaries based on medium- and long-term contracts, and then uses MOPSO for 24-hour optimization. The aggregate layer corrects the actual SOC deviation within the intraday rolling window. Compared with the day-ahead optimization scheme alone, this strategy improves medium- and long-term contract satisfaction, peak-valley difference improvement, and day-ahead plan tracking accuracy, verifying the effectiveness of the tiered progressive control strategy in the market model.

[0142] 1. Simulation environment and parameter settings: Energy output power data such as Figure 6 As shown, this invention uses a typical daily photovoltaic power output curve; the time-of-use electricity price is as follows: Figure 7 As shown, this electricity price is selected from the power purchase price disclosure table of a provincial power company of the State Grid Corporation of China; distributed energy storage includes 5G base station energy storage (100kWh), electric vehicle energy storage (50kWh), and conventional independent energy storage (200kWh).

[0143] 2. Regulation strategies: Figure 8 The convergence curve of Particle Swarm Optimization (PSO) is shown. The regulation strategy is optimized based on the MOPSO (Multi-Objective Particle Swarm Optimization) algorithm. The objective functions include minimizing the peak-to-valley difference and minimizing the total operating cost of energy storage. The algorithm balances the peak-to-valley difference and economic objectives by dynamically adjusting the particle positions, so that the energy storage scheduling strategy takes into account both load smoothing and cost minimization.

[0144] The simulation results are analyzed as follows:

[0145] 1. Comparison of before and after optimization: such as Figure 9 The comparison between the original load (blue line) and the optimized load (red line) shown in the data and images demonstrates that overall load smoothing and peak-valley balance have been achieved. By adjusting the three energy storage charging and discharging strategies, the goal of optimizing the power grid load distribution can be achieved.

[0146] 2. Optimization of charging and discharging power in different regions: such as Figure 10 As shown, this strategy significantly reduces the peak-valley difference in industrial areas, improving grid stability; in commercial areas, it enables charging during off-peak hours and discharging during peak hours, maximizing profits; and in residential areas, it allows energy storage on the user side to be discharged during peak hours, reducing electricity costs.

[0147] 3. Changes in SOC in different regions: such as Figure 11 As shown, the SOC curves for the three regions all fluctuate within a safe range. The SOC changes are relatively small in industrial and commercial areas, while the SOC in residential areas fluctuates significantly due to the influence of electricity prices. This indicates that energy storage dispatch achieves dynamic optimization while ensuring safety.

[0148] 4. Comparison of SOC optimization in different regions: Figure 12 The table shows the optimized SOC (State of Charge) curves for industrial, residential, and commercial areas, reflecting the dynamic changes in SOC of energy storage systems in each area. Industrial areas experience significant load fluctuations, and optimized scheduling improves SOC stability. Residential areas exhibit strong load regularity, resulting in efficient operation of energy storage systems. Commercial areas have relatively stable loads, leading to more balanced energy storage optimization. Overall, the SOC variation characteristics of different areas reflect differences in load characteristics, and optimization strategies need to be adjusted according to regional characteristics to improve overall electricity efficiency.

[0149] The dynamic stability analysis of the state of charge (SOC) of the energy storage system is as follows:

[0150] 1. Average SOC curve of electric vehicle charging stations: Figure 13 The curves showing the change of the mean SOC of electric vehicles at four charging stations over time are presented. By comparing the SOC change characteristics of different charging stations, it can be concluded that charging stations with large SOC fluctuations may need to optimize their charging and discharging scheduling strategies to divert vehicles during high-load periods; while charging stations with smaller SOC changes can further improve the utilization rate of charging piles and increase operating efficiency.

[0151] 2. Base station SOC mean curve: Figure 14 The curve showing the change of the average SOC of a 5G base station over time reflects the state changes of the energy storage system during base station operation. The average SOC curve (blue solid line) remains stable throughout the time range and is always between the upper and lower limits of SOC (black dashed line), without overcharging or over-discharging. This reflects the stability and efficiency of the 5G base station energy storage system, provides a reliable guarantee for the continuous operation of the base station, and demonstrates the rationality of the energy storage management strategy.

[0152] 3. Conventional standalone energy storage SOC curve: Figure 15 The curves showing the State of Charge (SOC) of a conventional independent energy storage system over time are presented, reflecting the dynamic charging and discharging characteristics of the system under independent operating conditions. The SOC curves exhibit significant overall fluctuations, demonstrating periodic charging and discharging processes, reflecting the system's flexibility in responding to loads and indicating a reasonable management strategy and reliable operation. Furthermore, the SOC remains consistently within the upper and lower limits, without overcharging or over-discharging. This demonstrates that the conventional independent energy storage system can effectively regulate charging and discharging under dynamic load conditions, maintaining stable system operation.

[0153] This invention presents a method for the aggregated regulation of generalized distributed energy storage participating in distribution network operation. In its application, this invention first establishes a theoretical framework for the collaborative optimization of generalized distributed energy storage systems in distribution networks by constructing a hierarchical regulation architecture, dynamic behavior modeling, defining multi-objective optimization problems, and designing an algorithm framework. Second, it constructs an aggregated regulation model for generalized distributed energy storage participating in distribution network operation using a multi-objective optimization framework combined with a multi-objective particle swarm optimization algorithm. Finally, it conducts simulation studies by designing various flexible resource management examples involving multiple stakeholders in typical urban load areas under typical distribution network scenarios. The example analysis is conducted under market modes including medium- and long-term peak-shaving contracts, day-ahead MOPSO optimization, and intraday disturbance correction. The results show that the proposed hierarchical progressive regulation strategy can effectively reduce the load peak-to-valley difference, improve the renewable energy absorption rate, ensure the safety of energy storage SOC, meet the requirements of multi-timescale market operation, and improve the economy and stability of the distribution network. This invention has the advantages of being based on a three-layer regulation architecture, constructing an aggregated regulation model, achieving collaborative regulation, and improving the stability and economy of the distribution network.

Claims

1. A generalized distributed energy storage aggregation and control method for participating in distribution network operation, using a three-layer architecture of "resource layer-aggregation layer-scheduling layer" as the physical carrier, setting hierarchical collaborative rules at three time scales: medium-to-long-term, day-ahead, and intraday, and employing a multi-objective particle swarm optimization algorithm as the core solution algorithm for day-ahead optimization of the scheduling layer, forming a strategy system that is architecture-supported, algorithm-driven, and time-series progressive, characterized by: The method includes the following steps: Step 1: Construct a three-layer collaborative hierarchical control architecture of "resource layer - aggregation layer - scheduling layer", complete the dynamic behavior modeling of generalized distributed energy storage, define the multi-objective optimization problem and build the algorithm framework, clarify the functional division of each layer in the three time scales of medium- and long-term, day-ahead and intraday, and establish the basic theoretical model for generalized distributed energy storage to participate in the operation of the distribution network. Step 2: Based on the aforementioned fundamental theoretical model, and combining the multi-objective optimization framework with the multi-objective particle swarm optimization algorithm, construct a generalized decentralized energy storage aggregation control optimization model for participating in distribution network operation; Step 3: Select a typical urban load area under a typical distribution network scenario, design a simulation case involving multiple entities and flexible resources, and conduct simulation verification and effect analysis of the control strategy.

2. The aggregation control method for generalized distributed energy storage participating in distribution network operation as described in claim 1, characterized in that: The resource layer in step 1 consists of distributed energy storage units and local loads, and is responsible for meeting local operational safety constraints at the execution layer. The distributed energy storage units include 5G base station backup batteries, industrial and commercial park energy storage systems, and electric vehicle energy storage. The aggregation layer is implemented by regional agents, which are responsible for rolling corrections of the deviation between the actual state of charge of the energy storage unit and the day-ahead plan every 15-30 minutes; the regional agents include base station smart terminals and park energy management systems. The scheduling layer consists of the distribution network scheduling center and the market trading platform. It is responsible for generating the charging and discharging capacity baseline of each aggregate based on the peak shaving contract in the medium and long term, and generating the intraday charging and discharging plan based on the source and load forecast data in the day-ahead period.

3. The aggregation control method for generalized distributed energy storage participating in distribution network operation as described in claim 2, characterized in that: The aggregation layer performs intraday SOC deviation correction on a rolling basis with a 15-minute cycle. The correction process is as follows: Deviation detection: at each correction time The aggregation layer collects the real-time state of charge of each energy storage unit and calculates its deviation from the day-ahead target value: In the formula, For energy storage units exist SOC deviation value at time; For energy storage units exist The actual remaining battery level at any given time; For energy storage units exist The target remaining power level that the time-scheduling layer is required to achieve; Intraday time index; Modify trigger conditions: if satisfied At that time, intraday correction and optimization are initiated, with the objective function being: In the formula, The objective function value is corrected intraday; Indicates from the current moment By the end of the day Every moment in between; The current moment; The last moment of the day; For energy storage units exist The remaining battery power at any given time; For energy storage units exist The target remaining battery level at any given time; It is the collection of all energy storage units; This is the SOC deviation tolerance threshold; Corrected power limiting constraint: The corrected charging power must meet the amplitude limit, expressed as: In the formula, This is the allowable adjustment range for the planned power output up to the day; This refers to the original charging power that was recently optimized and determined. The charging power after the polymer layer is corrected.

4. The aggregation control method for generalized distributed energy storage participating in distribution network operation as described in claim 2, characterized in that: The resource layer receives and executes power commands issued by the aggregation layer with a time resolution of seconds. The command execution relationship is as follows: In the formula, For energy storage units exist The actual charging power being executed at all times; For energy storage units exist The actual discharge power at any given moment; The charging command is issued after the aggregation layer has been corrected. The discharge command is issued after the polymer layer has been corrected; The resource layer local controller monitors the real-time state of charge of each energy storage unit, and when the conditions are met... At that time, the autonomous protection mechanism is triggered, forcibly stopping the discharge, that is, ordering Simultaneously, report the violation status indicator to the aggregation layer. The aggregation layer redistributes the output of the remaining online energy storage units; when the state of charge of the energy storage unit returns to: Then, a "recovery ready" signal is sent to the aggregation layer to rejoin the scheduling sequence; among which, This is the limit of the dynamic charging state; To restore margin.

5. The aggregation control method for generalized distributed energy storage participating in distribution network operation as described in claim 2, characterized in that: The dynamic behavior modeling in step 1 encompasses both dynamic characteristic modeling and economic modeling of the energy storage infrastructure, as detailed below: The dynamic characteristic modeling of the energy storage infrastructure includes the state-of-charge evolution equation, characterizing the energy storage unit. The variation pattern of SOC over time: In the formula, For energy storage units During the period The state of charge; These refer to the energy storage charging and discharging efficiencies, respectively. For time period Energy storage unit The charging and discharging power; The time step is specified; the state of charge evolution equation is equipped with SOC safety boundary constraints and charge / discharge power limits. The economic model includes time-of-use pricing response costs and equipment depreciation costs; whereby the time-of-use pricing response cost is the coupling cost of energy storage charging and discharging with the time-of-use pricing period, and the electricity purchase cost function is: In the formula, For time period The cost of purchasing electricity; For time period Time-of-use electricity pricing; Equipment depreciation cost refers to the cost of equipment aging caused by frequent switching of charging and discharging power. The cost model is as follows: In the formula, The equipment wear and tear costs incurred due to charging and discharging energy storage devices; This is the loss coefficient.

6. The aggregation control method for generalized distributed energy storage participating in distribution network operation as described in claim 5, characterized in that: The multi-objective optimization problem in step 1 consists of two parts: the optimization objective and the constraints. The optimization objectives include maximizing grid regulation capacity and optimizing operational economy; among which, maximizing grid regulation capacity focuses on suppressing load fluctuations through energy storage charging and discharging levels, and the objective function is: In the formula, For time period Net load; This represents the average daily load. For power grid regulation capacity; This represents the total number of time periods; Operational economic optimization aims to minimize the total lifecycle cost of energy storage. The objective function is: In the formula, The total cost of energy storage over its entire lifecycle; The constraints include grid security constraints and new energy consumption constraints.

7. The aggregation control method for generalized distributed energy storage participating in distribution network operation as described in claim 1, characterized in that: The objective function and constraints of the aggregation control model for generalized distributed energy storage participating in distribution network operation in step 2 are as follows: Objective function 1: Minimize the peak-to-valley difference in net load of the distribution network, expressed as: In the formula, This represents the maximum net load for the entire day. This is the minimum net load for the entire day; For time period Net load after internal optimization; Objective function 2: Minimize the total operating cost of the energy storage system, expressed as: In the formula, Fixed electricity price; A collection of energy storage units; The total operating cost of the energy storage system; The constraints include peak-valley difference definition constraints, new energy consumption constraints, energy storage SOC dynamic constraints, dynamic SOC lower limit constraints, charging and discharging power limit constraints, peak shaving service demand constraints, and power purchase and sale constraints.

8. The aggregation control method for generalized distributed energy storage participating in distribution network operation as described in claim 7, characterized in that: The day-ahead optimization output of the multi-objective optimization framework in step 2 serves as a reference benchmark for lower-level regulation, specifically including: Hourly charge / discharge power plan: In the formula, For energy storage units exist The charging power given by the time-scheduling layer; Energy storage unit exist The discharge power given by the time-of-flight scheduling layer; For the collection of times before the date; Target state of charge at the top of the hour: In the formula, For energy storage units exist The target value of the remaining power percentage required by the time-scheduling layer. The above output parameters constitute the control baseline for the operation of each energy storage unit the next day. The lower aggregation layer and resource layer perform progressively refined adjustments within the constraints of this baseline.

9. The aggregation control method for generalized distributed energy storage participating in distribution network operation as described in claim 7, characterized in that: The implementation rules of the multi-objective particle swarm algorithm used in step 2 are as follows: Inertia weight adjustment: A linearly decreasing inertia weight strategy is adopted, expressed as follows: In the formula, Inertial weight; These are the initial and final inertia weights, respectively. This represents the current iteration number; Learning factor adjustment: An asynchronous learning factor adjustment strategy is adopted, expressed as follows: In the formula, These are individual learning factors and social learning factors, respectively. Pareto front maintenance: Set up an external archive to store non-dominated solutions during the iteration process, and maintain the diversity of the solution set by crowding distance sorting; Particle encoding rule: A particle encoding method is designed for the dual-objective optimization problem of peak-valley difference and operating cost, and the calculation logic of fitness function is adjusted accordingly; Constraint handling mechanism: The penalty function method is used to handle the constraints. The penalty coefficient is tuned through pre-run: the goal is to achieve a population feasibility rate of 60%-80% in the middle of the iteration, and the minimum penalty coefficient that meets this condition is selected. The sum of each constraint violation after dimensional normalization is multiplied by the penalty coefficient and then included in the fitness function. Iteration termination conditions: The maximum number of iterations of 200 is reached, or the Pareto solution set converges.

10. The aggregation control method for generalized distributed energy storage participating in distribution network operation as described in claim 1, characterized in that: The typical distribution network simulation scenario in step 3 is set under a multi-timescale market model that includes medium- and long-term peak-shaving contracts, day-ahead multi-objective particle swarm optimization, and intraday disturbance correction. The simulation analysis is carried out by combining the constructed control model with actual load data and new energy output curves.

Images