Distributed energy storage aggregation regulation and control method and system based on improved clustering and group optimization

Abstract

The invention relates to a distributed energy storage aggregation regulation and control method and system based on improved clustering and group optimization, and the method comprises the steps: carrying out the clustering of distributed energy storage based on distributed energy storage regulation and control potential and parameter difference, and setting a distributed energy storage aggregator for each distributed energy storage group based on a clustering result; for a distributed energy storage economical efficiency regulation and control scene, a distributed energy storage aggregator obtains profit through electricity price arbitrage and participation in power grid auxiliary service, and regulates and controls distributed energy storage charging and discharging behaviors by considering energy storage operation cost; for a distributed energy storage auxiliary power grid peak clipping and valley filling scene, a distributed energy storage aggregator is distributed for transferring a peak load of a power system, the adjustable potential of distributed energy storage is tested, and the state of charge of distributed energy storage in the aggregator is distributed. Compared with the prior art, the load fluctuation of the power system can be effectively reduced, and meanwhile, the charge and discharge potential of distributed energy storage is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of distributed energy storage operation management, and particularly relates to a distributed energy storage aggregation regulation method based on improved clustering and group optimization. BACKGROUND

[0002] Distributed energy storage can convert electrical energy into a more stable form and store it in a device, and release it when needed. This feature enables distributed energy storage to simultaneously assume the roles of "source" and "load" in the power grid, effectively alleviating the pressure on the power grid caused by the access of new energy generation and rapid growth of load. By reasonably controlling distributed energy storage, the peak load can be reduced and the supply-demand contradiction can be alleviated at a lower cost while ensuring its controllable potential. Meanwhile, distributed energy storage can also store cheaper off-peak power and release it during peak hours to achieve economic benefits.

[0003] Distributed energy storage can effectively alleviate the pressure on the power grid caused by the access of new energy generation and rapid growth of load. Unlike traditional centralized operation of energy storage power stations, distributed energy storage has the advantages of small footprint and low investment cost, and can be flexibly installed at various locations on the user side. However, the charging and discharging power of a single distributed energy storage is small and the capacity is limited, which is difficult to attract the interest of power system management alone. In contrast, aggregated distributed energy storage is flexible and has great controllable potential, making it feasible to participate in the power market under reasonable regulation strategies.

[0004] Therefore, there is an urgent need for an aggregation management strategy for distributed energy storage to regulate the power market. SUMMARY

[0005] The purpose of the present application is to provide a distributed energy storage aggregation regulation method based on improved clustering and group optimization for the aggregation management of distributed energy storage.

[0006] The purpose of the present application can be achieved by the following technical solutions: As a first aspect of the present application, a distributed energy storage aggregation regulation method based on improved clustering and group optimization is provided, comprising the following steps: Clustering distributed energy storage based on regulation potential and parameter difference, and setting a distributed energy storage aggregator for each distributed energy storage based on the clustering results; For a distributed energy storage economic regulation scenario, the distributed energy storage aggregator participates in electricity price arbitrage and power grid auxiliary service to maximize the total economic benefit of the energy storage aggregator and considers the energy storage operation cost to regulate the charging and discharging behavior of each distributed energy storage; For the distributed energy storage auxiliary power grid peak shaving and valley filling scene, the distributed energy storage aggregator is used to smooth the power grid load curve, while maintaining the adjustable potential of the distributed energy storage itself, and the state of charge of the distributed energy storage in the energy storage aggregator is distributed.

[0007] As a preferred technical solution, the distributed energy storage is clustered, and the specific steps are as follows: Randomly initialize the positions and velocities of a group of particles, each particle representing a grouping method, and the particles are represented by a vector composed of all cluster centroids; Calculate the sum of errors between the data in the sample set and all cluster centroids as the fitness of the particles; Update the velocities and positions of all particles based on the inertia weight and acceleration factor to generate a new group of particles; if the velocity and / or position of the new particle exceeds the velocity and / or position boundary, the velocity and / or position of the particle is set to the boundary value of the corresponding velocity and / or position, and the particle with the minimum fitness in the current generation is stored; Repeat the iteration of updating the particle population until the optimal value of the historical fitness of the entire population does not change within a specified number of iterations, or the number of runs has reached the maximum value; Calculate the distance from all distributed energy storages to each cluster centroid, and classify the distributed energy storages into the group closest to the centroid; Calculate the average value of the regulation potential parameter vector of each energy storage in the cluster group, and replace the average value of the regulation potential parameter vector as the new centroid of the cluster group; Repeat the distributed energy storage division and centroid replacement process until the position change of the centroid in the two running processes is less than a preset value, and obtain the final clustering grouping result of the distributed energy storage.

[0008] As a preferred technical solution, the regulation potential parameter vector includes: Intrinsic physical parameters including rated capacity, rated power and charge-discharge efficiency; and initial state and constraint parameters including initial state of charge and upper and lower limits of state of charge operation.

[0009] As a preferred technical solution, the economic optimization target in the economic regulation scenario is to maximize the total economic benefit of the energy storage aggregator, and the charge and discharge power of each distributed energy storage at each time is used as a decision variable. By controlling the charge and discharge power, charging at low price and discharging at high price are realized, and the objective function is specifically represented as: In the formula, represents the time-of-use electricity price at time t, The total number of distributed energy storage participating in operation is represented, the first term of formula (10) is energy vending revenue, the second term is energy purchase cost, both of which constitute the operation revenue of distributed energy storage, the third term is the revenue obtained by energy storage participating in auxiliary service, and the fourth term is the degradation cost generated during the operation of energy storage.

[0010] As a preferred technical solution, the auxiliary optimization target for the distributed energy storage assisting the peak load regulation of the power grid includes two layers. The first layer takes the total output of the distributed energy storage of the entire aggregator group at each time as the decision variable, and minimizes the variance of the load curve as the optimization objective function: In the formula, is the time The load demand of the power grid, is the total output of the distributed energy storage at time The second layer optimization model distributes the state of charge of each distributed energy storage based on the total output of the distributed energy storage output by the first layer, and maximizes the sum of the smaller value of the maximum charging power and the maximum discharging power of each distributed energy storage as the optimization objective function: In the formula, represents the maximum charging power of the t th distributed energy storage at the current state of charge within the subsequent operation period; i represents the maximum discharging power of the th distributed energy storage at the current state of charge within the subsequent operation period. t i

[0011] As a preferred technical solution, the constraint conditions of the auxiliary optimization include: energy storage capacity constraint, energy storage power constraint and power grid power constraint, which constrain the charge amount, power and charging and discharging power of the distributed energy storage within the set upper and lower limits.

[0012] As a preferred technical solution, the two-layer optimization problem of the auxiliary optimization target adopts a decomposition-coordination strategy for solution: The optimization objective function of the first layer is solved by an optimization algorithm or an intelligent algorithm to obtain the optimal total power instruction curve; The optimization objective function solving and SOC distribution process of the second layer are as follows: For each time, the energy storage aggregator receives the total output instruction of the distributed energy storage from the output of the first layer optimization; ​​​The energy storage aggregator calculates the maximum power for continuous charging and the maximum power for continuous discharging of each energy storage unit under its jurisdiction based on the current state of charge of the energy storage; When the total output instruction of the distributed energy storage is less than zero, the energy storage with a larger maximum power for continuous charging and a smaller maximum power for continuous discharging is preferentially allocated for charging; When the total output instruction of the distributed energy storage is greater than zero, the energy storage with a larger maximum power for continuous discharging and a smaller maximum power for continuous charging is preferentially allocated for discharging; Through rolling optimization over multiple time steps, the state of charge of each distributed energy storage is optimally allocated.

[0013] As a second aspect of the present application, a distributed energy storage aggregation control system based on improved clustering and swarm optimization is provided, the system comprising: A distributed energy storage aggregation module: clustering the distributed energy storages based on the distributed energy storage control potential and parameter difference, and setting a distributed energy storage aggregator for each distributed energy storage group based on the clustering results; An economic control module, the distributed energy storage aggregator makes profits through electricity price arbitrage and participation in grid auxiliary services, and controls the charging and discharging behavior of the distributed energy storage by considering the operation cost of the energy storage; A grid peak load shifting scenario auxiliary control module, which allocates the distributed energy storage aggregators for shifting the power system peak load, and allocates the state of charge of the distributed energy storages in the aggregator based on the controllable potential of the distributed energy storages.

[0014] As a preferred technical solution, the step of clustering the distributed energy storages by the distributed energy storage aggregation module is as follows: Randomly initialize the positions and velocities of a group of particles, each particle representing a grouping method, and the particles are represented by vectors composed of all cluster centroids; Calculate the sum of errors between the data in the sample set and all cluster centroids as the fitness of the particles; Update the velocities and positions of all particles based on the inertia weight and acceleration factor to generate a new group of particles; if the velocity and / or position of a new particle exceeds the velocity and / or position boundary, the velocity and / or position of the particle is set to the boundary value of the corresponding velocity and / or position, and the particle with the minimum fitness in the current generation is stored; Repeat the iteration of updating the particle population until the optimal value of the historical fitness of the entire population does not change within a specified number of iterations, or the number of runs has reached the maximum value; Calculate the distances of all distributed energy storages to the cluster centroids, and classify the distributed energy storages into the group with the nearest centroid; The average value of the regulation potential parameter vector of each energy storage in each clustering group is calculated, and the regulation potential parameter average value vector is replaced as the new centroid of the clustering group; the regulation potential parameter vector includes intrinsic physical parameters including rated capacity, rated power and charging and discharging efficiency, and initial state and constraint parameters including initial state of charge and upper and lower limits of state of charge operation The distributed energy storage division and centroid replacement process is repeatedly performed until the position change of the centroid in two running processes is less than a preset value, and the final clustering grouping result of the distributed energy storage is obtained.

[0015] As a preferred technical solution, the economic optimization target in the economic regulation scenario is to maximize the total economic benefit of the energy storage aggregator, and the charging and discharging power of each distributed energy storage at each time is taken as a decision variable, and the charging at low price and discharging at high price are realized by controlling the charging and discharging power, and the objective function is specifically represented as: In the formula, represents the time-of-use electricity price at time t, represents the total number of distributed energy storages participating in operation, the first term of formula (10) is the energy selling benefit, the second term is the energy purchase cost, and the two together constitute the operation benefit of the distributed energy storage, and the third term is the benefit obtained by the energy storage participating in auxiliary service, and the fourth term is the degradation cost generated by the energy storage during operation The auxiliary type optimization target of the auxiliary regulation module of the power grid peak load shifting scenario includes two layers: The first layer takes the total output of the distributed energy storages of the entire aggregator group at each time as a decision variable, and minimizes the variance of the load curve as the optimization objective function: In the formula, is the time of the power grid, is the total output of the distributed energy storage at time t, The second layer optimization model is based on the total output of the distributed energy storages output by the first layer, and allocates the state of charge of each distributed energy storage, and takes the sum of the smaller value of the maximum charging power and the maximum discharging power of each distributed energy storage as the optimization objective function: In the formula, represents the maximum charging power of the first distributed energy storage at time t t under the current state of charge, i which can be continuously charged in the subsequent operation period; represents the maximum discharging power of the first distributed energy storage at time t t under the current state of charge, i which can be continuously discharged in the subsequent operation period.​

[0016] Compared with the prior art, the present application has the following beneficial effects: The present application can serve the power market in an easy-to-implement and easy-to-control mode under the fine and personalized control strategy through the aggregation management of distributed energy storage. This research has important theoretical and practical significance and helps to fully exploit the potential value of energy storage at the user side, improve its economy and accelerate its application. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 Flow chart of the distributed energy storage aggregation control method based on improved clustering and population optimization of the present application.

[0018] Figure 2 Flow chart of the clustering method based on the parameter difference of distributed energy storage of the present application.

[0019] Figure 3 Typical daily load demand in the specific embodiment of the present application.

[0020] Figure 4 Distributed energy storage economic benefits of different clustering groups in the specific embodiment of the present application.

[0021] Figure 5 Comparison of daily net benefits of a single energy storage using the optimal control strategy in the specific embodiment of the present application.

[0022] Figure 6 Life loss generated by the operation of distributed energy storage in one day in the specific embodiment of the present application.

[0023] Figure 7 Adjustable potential of distributed energy storage at t=6 in the specific embodiment of the present application.

[0024] Figure 8 Adjustable potential of distributed energy storage at t=11 in the specific embodiment of the present application.

[0025] Figure 9 Operation results of energy storage participating in load peak shaving of the power system in the specific embodiment of the present application.

[0026] Figure 10 Adjustable potential of distributed energy storage at t=6 in the peak shaving and valley filling scenario in the specific embodiment of the present application.

[0027] Figure 11 Adjustable potential of distributed energy storage at t=11 in the peak shaving and valley filling scenario in the specific embodiment of the present application. DETAILED DESCRIPTION

[0028] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0029] Example 1 This invention proposes a distributed energy storage aggregation and control method based on improved clustering and population optimization. For example... Figure 1 As shown, the specific steps are as follows: S1. Cluster distributed energy storage based on its regulation potential and parameter differences, and set a distributed energy storage aggregator for each distributed energy storage group based on the clustering results; S2. For the scenario of economic regulation of distributed energy storage, distributed energy storage aggregators profit from electricity price arbitrage and participation in grid ancillary services, and regulate the charging and discharging behavior of distributed energy storage by taking into account the operating costs of energy storage. S3. For distributed energy storage to assist the power grid in peak shaving and valley filling scenarios, distributed energy storage aggregators are allocated to transfer peak loads of the power system, and the controllable potential of distributed energy storage is considered to allocate the state of charge of distributed energy storage within the aggregator.

[0030] Specifically, in step 1, clustering is performed on distributed energy storage. The accuracy of traditional K-means clustering analysis on large sample sets largely depends on the accuracy of the sample center initialization method applicable to sample sets distributed across multiple machines. This invention further employs the Particle Swarm Optimization (PSO) algorithm to address this issue.

[0031] The Particle Swarm Optimization (PSO) algorithm treats each potential solution as a particle in a multidimensional space, and the process is as follows: d Particles exist in 1-dimensional space i Its location is The speed of movement is Each particle is updated iteratively using the following formula to seek the optimal solution: (1) (2) In the formula: w Inertial weight; As an acceleration factor; A random number between (0, 1); For the first i Individual particles The first time in its own historical optimal solution d Dimensional data; for The th time in the history of the best solution for all particles at time 1 d Dimensional data.

[0032] How to select a suitable number of categories in the clustering process has always been the focus of such research. For a large number of small and medium-sized energy storage clustering practical problems, too few categories will deviate from the expected regulation target, and too many categories will increase the difficulty of the aggregator's regulation, and only a moderate number of categories can better meet the regulation requirements. Therefore, in the actual regulation process, the load agent can determine the maximum K value range based on its regulation capacity, and further determine the optimal number of categories through the exhaustive method.

[0033] The present application proposes a clustering method based on the distributed energy storage regulation potential and parameter difference. The parameter difference on which the clustering is based refers to those physical parameters and state parameters that directly affect the operation performance and regulation capacity of the distributed energy storage and are relatively stable or initial state.

[0034] These parameters specifically include the following two categories: Inherent physical parameters (static difference): Rated capacity: the maximum electrical energy that the energy storage device can store (unit: kWh or MWh). It determines the continuous charging and discharging time of a single energy storage.

[0035] Rated power: the maximum charging and discharging power of the energy storage device (unit: kW or MW). It determines the instantaneous power regulation capacity of a single energy storage.

[0036] Charging and discharging efficiency: the proportion of loss in the energy conversion process.

[0037] Initial state and constraint parameters (quasi-static difference): Initial state of charge SOC: the power level at the beginning of the optimization period.

[0038] Upper and lower limits of state of charge SOC: the allowed SOC range (e.g. 0.1-0.9 as mentioned in the document) considering the life.

[0039] Based on the above inherent physical parameters and initial state and constraint parameters, a regulation potential parameter vector is constructed for each energy storage unit.

[0040] The clustering process of the present application does not directly use a complex calculation result named "regulation potential" as input, but indirectly but essentially considers the regulation potential through a "proxy" method. The essence of the regulation potential is the ability of an energy storage to safely and effectively absorb or release electrical energy in the future period of time. It mainly depends on two dynamic factors: 1. Power potential: the maximum allowed charging and discharging power at the current time (subject to rated power and SOC constraints).

[0041] 2. Energy potential: accumulative chargeable or dischargeable capacity before the end of the operation cycle at the current SOC.

[0042] In the clustering process, the consideration of regulation potential is reflected in the following two aspects: a) "Potential portrait" through static and quasi-static parameters Rated capacity and rated power directly define the theoretical maximum potential ceiling of energy storage. A 1 MWh capacity energy storage and a 10 MWh capacity energy storage are inherently different in their potential energy regulation capacity and must be divided into different groups.

[0043] Initial SOC and SOC constraints directly determine the initial available potential of energy storage. An energy storage with an initial SOC of 0.9 (almost full, large discharge potential but small charge potential) and an energy storage with an initial SOC of 0.2 (almost empty, large charge potential but small discharge potential) are completely opposite in their roles at the start of the regulation cycle. Dividing them into the same group will make it difficult for the aggregator to develop a unified and efficient regulation strategy.

[0044] Therefore, the "parameter difference" on which clustering is based actually constructs a "regulation potential portrait" for each energy storage unit. Clustering energy storage with similar "potential portraits" ensures: Interchangeability within the group: any energy storage within the group responds similarly when responding to the same power command, and the constraints it faces (such as whether it will soon reach the SOC limit) are similar.

[0045] Uniformity of regulation strategy: the aggregator can develop a unified charging and discharging strategy for the entire group without the need for complex and potentially conflicting individual calculations for each member of the group. This greatly simplifies the subsequent optimization problem.

[0046] b) The clustering result directly determines the "cluster regulation potential" after aggregation Unclustered (random grouping): large differences in potential portraits among group members. When receiving a charging command, energy storage with high SOC within the group may soon reach the upper limit and exit, causing the actual regulatable potential of the entire group to quickly decay.

[0047] Clustered (improved grouping): similar potential portraits among group members. When receiving a charging command, all members of the group can respond at a similar pace, without individual members prematurely "falling behind", thus maximizing the use of the group's overall regulatable potential.

[0048] As shown in Figure 2 , the clustering process based on distributed energy storage regulation potential and parameter difference is as follows: Step 1.1: Randomly initialize a set of particles, each of which is a solution to the actual problem. In the sample set, each particle is represented in vector form, which is constructed as follows: (3) where, is the centroid of the th cluster, K is the total number of clusters. Thus, each particle is equivalent to a grouping manner.

[0049] Step 1.2: Evaluate the fitness of the initial particles based on the following criteria: (4) where, is the th data in the sample set, is the total number of clustered data. By finding the minimum value of the fitness, the dispersion of the clusters can be minimized.

[0050] Step 1.3: If the number of runs has reached the maximum value, jump to Step 1.6, otherwise continue to Step 1.4.

[0051] Step 1.4: Store the best particle of the current generation. Update all particles according to equations (2), (3), and then generate a new set of particles. If the position of a new particle exceeds the boundary[ X min , X max ], the position of this particle will be set to X min or X max ; if the velocity of a new particle exceeds the boundary[ V min ,V max ], the velocity of this particle will be set to V min or V max .

[0052] Step 1.5: If the current entire population's historical optimal value has not changed in the specified iterations, then go to Step 1.6; otherwise, go to Step 1.3.

[0053] Step 1.6: Starting from the distributed energy storage numbered i =1, calculate its distance to the K th cluster centroid, and classify it into the cluster group to which the nearest centroid belongs.

[0054] Step 1.7: Calculate the average value of all components in the regulation potential parameter vector within each cluster grouping, and take this new regulation potential parameter average value vector as the new centroid of the cluster grouping. By clustering based on these parameters, energy storage units with similar capacity, similar power capability, and similar initial state can be grouped into the same group, forming aggregated units with highly consistent characteristics within the group.

[0055] Step 1.8: Repeat steps 1.6 to 1.7 until the position of the centroid changes by less than a preset value within two running processes.

[0056] One part of the user-side energy storage is the electricity revenue, which means that when the electricity price is low, the energy storage charges from the grid, and during the high electricity price period, it can replace the user's consumption by releasing stored electricity, thereby saving the user's electricity cost, or it can inject energy into the grid to obtain revenue, but such operation will cause the service life of the energy storage to decrease, so the revenue obtained must be higher than the cost brought by the operation of the energy storage. This part of the revenue can be represented as: (5) Another part of the user-side energy storage is the compensation obtained by the energy storage through serving the grid to help the grid reduce peak load, which can be calculated by the following formula: (6) In the formula, represents the revenue obtained by serving the grid, represents the revenue per unit capacity, represents the peak shaving electricity. The total revenue of the energy storage can be represented as the sum of these two parts of revenue: (7) The distributed energy storage is generally composed of energy storage devices and energy conversion devices, so the cost of the energy storage can be represented as: (8) In the formula, represents the cost of the energy storage device, represents the cost of the energy conversion device.

[0057] Due to the charging and discharging of the energy storage, a small part of its capacity is lost, which is commonly referred to as battery degradation. The aggregator bears all the costs related to the energy storage, so it must consider the degradation cost during the operation of the energy storage. In a market economy environment, distributed energy storage can be uniformly evaluated by the daily average cost per unit energy and per unit power: (9) Research on the regulation strategy of distributed energy storage under the energy storage aggregator: The setting of the optimization target is based on the analysis of the operation characteristics of the distributed energy storage, and the controllable potential of the distributed energy storage is excavated while meeting the regulation target. The former is realized by reasonably regulating the charging and discharging behavior of the distributed energy storage, and is the optimization focus of the scene; the latter is realized by reasonably distributing the energy of the internal energy storage of the aggregator.

[0058] Table 1 Target optimization scheme under different scenarios (1) Economic optimization target The economic benefits of distributed energy storage mainly consist of two parts: price arbitrage and participation in power grid auxiliary service. The former earns a price difference by absorbing power from the grid at low prices and releasing it at high prices; the latter obtains compensation by helping the power system reduce peak load and absorb new energy generation. Since the energy storage will inevitably produce wear and tear during operation, the running cost of the energy storage needs to be calculated when considering its economy. Therefore, based on the economy modeling, the target function can be set as: (10) In the formula, represents the time-of-use electricity price at time t, represents the total number of distributed energy storages participating in operation. The first term of formula (10) is the energy trading revenue, the second term is the energy purchase cost, and the third term is the revenue obtained by the energy storage participating in auxiliary services. The fourth term is the degradation cost of the energy storage during operation.

[0059] Auxiliary optimization target With the development of the power grid, research on reducing peak electricity demand, relieving line congestion, and improving system operation economy has gradually attracted attention. Using energy storage devices to transfer high peak power is an effective way for the power system to deal with peak load. Variance, as a commonly used indicator, can effectively reflect the deviation of random variables from the center of random variables. Therefore, the total output of the aggregator at 24 time points is used as the decision variable, represents discharging, represents charging; the variance of the minimum load curve is used as the first layer optimization target function: (11) In the formula, is the time load demand of the power grid, is the total output of the distributed energy storage at time t.

[0060] The total output instruction of the distributed energy storage obtained by the first layer is ​Decomposed into each specific distributed energy storage i Above, that is, determining the output command for each energy storage unit. , making To avoid adverse effects such as worsening of peak-valley trends due to the consistency of distributed energy storage systems during electricity price arbitrage, and for stability considerations, distributed energy storage should be able to maximize its controllability potential during operation. Based on the analysis of the controllability potential of distributed energy storage, the second-level optimization model uses the output commands of each energy storage system as decision variables, and the optimization objective function is expressed as follows: (12) in, Indicates the time. t No. i The maximum power that a distributed energy storage system can continuously charge during subsequent operating cycles under its current state of charge; Indicates at time t No. i The maximum power that a distributed energy storage device can sustainably discharge over subsequent operating cycles under its current state of charge. These two parameters are essentially dynamic, sustainable power capabilities geared towards future dispatch, rather than simple instantaneous maximum power. Their values ​​are jointly determined by the energy storage device's current state of charge (SOC), rated power, capacity, and operating constraints, reflecting the potential for continuous participation in regulation over a future period.

[0061] Energy storage capacity constraints: (13) Energy storage power constraints: (14) Power grid constraints: (15) (16) This two-level optimization problem is a typical sequential decision process, and this invention uses a decomposition-coordination strategy to solve it.

[0062] First level: System-level optimization (smoothing load curve) solution process: This optimization layer, viewed from the perspective of a "virtual central controller," determines the total power that the entire energy storage cluster should inject into (or absorb from) the grid at any given moment in order to smooth the load curve. How much is it?

[0063] This is a quadratic programming problem or nonlinear programming problem with constraints (equations 13-16) that can be solved using conventional optimization algorithms (such as interior point method, quadratic programming solver) or intelligent algorithms (such as genetic algorithm).

[0064] Output result: get an optimal total power instruction curve This curve is the ideal charge-discharge plan that makes the grid load most smooth.

[0065] Second layer: agent-level optimization (allocate SOC to maintain regulation potential), solve the SOC allocation process as follows: Receive instructions: for each time t , the energy storage aggregator receives the total output instruction of distributed energy storage from the output of the first layer optimization .

[0066] Evaluate potential: the energy storage aggregator calculates the current controllable potential of each energy storage unit i under its jurisdiction, that is, the remaining charging space and the remaining discharging space . These two values are directly determined by the current state of charge of the energy storage : Execute optimization allocation: The goal of this layer is to maximize the sum of the smaller of all energy storage "charging potential" and "discharging potential". This min(charge potential, discharge potential) is called "symmetric regulation capacity". Maximizing it means making the charge-discharge capacity of all energy storages as balanced and strong as possible.

[0067] For energy storage with low SOC: Charging potential is large: because the battery is empty, there is a large space to absorb electric energy (i.e. larger).

[0068] Discharge potential is small: because the battery stores little energy, the amount of dischargeable electric energy is limited (i.e. smaller).

[0069] Therefore, its min(charge potential , discharge potential) value is determined by the smaller discharge potential.

[0070] For energy storage with high SOC: Charging potential is small: because the battery is almost full, the remaining charging space is limited, i.e. smaller.

[0071] Discharge potential is large: because the battery stores sufficient energy, a large amount of electric energy can be released, i.e. Larger.

[0072] Therefore, its min (charge potential, discharge potential) value is determined by the smaller charge potential.

[0073] The specific operation for allocation to distributed energy storage SOC is as follows: Under the condition that... Under the premise that optimization algorithms tend to: When charging is needed In this scenario, priority is given to charging energy storage devices with lower State of Charge (SOC) (high charging potential but low discharging potential). Charging increases their SOC level, directly enhancing their currently weak discharging potential. Although the charging potential decreases slightly, the previously low discharging potential is significantly improved, ultimately reducing the minimum charging potential. , The value of the "shortcoming" (discharge potential) has increased.

[0074] When discharge is required When: Prioritize the discharge of energy storage devices with higher SOC (high discharge potential but low charging potential).

[0075] Discharging reduces its SOC level, thereby directly enhancing its currently weak charging potential. Although the discharge potential decreases slightly, the originally small charging potential is significantly improved, ultimately increasing the "weakest link" value of min(charging potential, discharge potential).

[0076] Achieving SOC allocation: Through the above strategy and rolling optimization over multiple time steps, the SOC of different energy storage systems will automatically evolve towards differentiation rather than convergence. Some energy storage systems will be deliberately maintained at a higher SOC to meet discharge demands, while others will be maintained at a lower SOC to meet charging demands. This achieves the allocation of the state of charge (SOC) of distributed energy storage within the aggregator, thereby meeting grid demands while maximizing the preservation of the entire cluster's subsequent regulation capabilities.

[0077] Example 2 As a specific implementation example of this application, this embodiment selects 500 distributed energy storage devices for cluster analysis. The energy storage capacities are 150 devices with a capacity of 1 MW·h, 150 devices with a capacity of 1.5 MW·h, and 200 devices with a capacity of 2 MW·h. All have a rated power of 100 kW and an efficiency of 95%. The unit power price of the energy conversion device shown is 3224 yuan / (kW), and the unit capacity price of the energy storage device is 1085 × 10⁻⁶ yuan / kW. 3 (Yuan / (MW·h)), the load demand for a typical day was simulated using the proposed model, and the results are shown in [reference needed]. Figure 3 For details on the electricity sales prices on the power grid, please refer to Table 2.

[0078] Table 2 Time-of-use electricity price table in February The compensation fee for each response to 1 kilowatt load is not more than 30 yuan, and the limit standard is 30 yuan for each response 8 times, 3.75 yuan for each kilowatt, and 2 times in the peak period of a day. In order to protect the service life of the energy storage, the state of charge is generally 0.1-0.9. The capacity attenuation of the energy storage is calculated according to the exponential model, and the capacity is calculated as scrap when it is reduced to 60%. The total cycle number is 2511 times, and the model parameters are shown in Table 3.

[0079] Table 3 Exponential model Economic regulation and control scene of distributed energy storage It is assumed that the energy storage aggregator needs to regulate and control 500 energy storages in its control range, and the initial state of charge of the distributed energy storage is randomly selected between 0.1-0.9 according to the normal distribution. The method proposed in the application is used to cluster and regulate the 500 energy storages, and the economic benefit results of the distributed energy storage with different cluster group numbers are shown in Figure 4 .

[0080] It can be seen that the proposed method can effectively reduce the total cost of the energy storage operation and improve the economic benefit of the energy storage. At the same time, since the parameters of each energy storage in the aggregation group are similar, each group has a similar operable space, which is conducive to the decision-making of the aggregator. With the increase of the cluster group number, the total cost of the distributed energy storage operation is continuously reduced, and the economic benefit is improved. However, when the group number reaches 10, the net benefit of the energy storage will not be significantly improved by increasing the group number, and the operation complexity will be greatly improved. Therefore, the energy storage is divided into 10 groups for regulation and control in this embodiment.

[0081] By Figure 5 comparing the daily net benefit of each energy storage with the adoption of the optimal regulation and control strategy, it can be seen that, compared with the distributed energy storage without the adoption of the regulation and control strategy, the net benefit of individual energy storage with the adoption of the regulation and control strategy decreases, the fluctuation of the benefit between the energy storages increases, but the total benefit of the energy storage is greatly improved. This is because through the control of the optimal regulation and control strategy, the energy storage with sufficient charging potential will act more frequently, and these energy storages inevitably bear the main profit.

[0082] The life loss generated by the operation of each distributed energy storage in a day is observed, and the results are shown in Figure 6 . It can be seen that the life loss of the energy storage without considering the operation strategy of the optimal regulation and control is relatively large, which is not conducive to the long-term operation of the distributed energy storage. In contrast, the life loss of the energy storage considering the operation strategy of the optimal regulation and control is relatively small, which shows that the regulation and control strategy proposed in the application helps to compress the cost of the distributed energy storage and improve the economic benefit.

[0083] A comparison of the controllable potential of energy storage before and after aggregation reveals that: unaggregated energy storage operates rigidly with limited controllable potential, while aggregated energy storage operates flexibly with greater controllable potential. This is because some energy storage systems have insufficient available capacity and temporarily lose their charging (discharging) ability before being recharged, leading to a decrease in controllable potential. In severe cases, this can cause aggregators to be unable to meet the demands of the power system. Figure 7 As shown, at 6 AM that day, the rechargeable potential of aggregated distributed energy storage was significantly greater than that of unaggregated distributed energy storage. However, since most distributed energy storage still retained a considerable amount of rechargeable capacity, the difference was not substantial. Figure 8 As shown, unlike at 6 o'clock, at 11 o'clock that day, because each energy storage unit was almost fully charged, a large portion of the randomly operating energy storage units lost their charging capacity, and their controllability potential decreased significantly. However, the aggregated energy storage units, because they prioritized charging the group with larger remaining capacity, preserved their controllability potential and still had some charging capacity. This shows that the control strategy is more effective in more severe situations.

[0084] In summary, the energy storage aggregation model proposed in this invention can effectively optimize and improve the economic benefits of the aggregated energy storage device, while also greatly enhancing the charging and discharging potential of distributed energy storage. Furthermore, the more severe the conditions, the better the effect of this control strategy.

[0085] Distributed energy storage assists in peak shaving and valley filling scenarios When distributed energy storage responds to peak avoidance scheduling, K The results of =10 group clustering of distributed energy storage participating in power system load peak reduction are shown in Figure 9 As can be seen, distributed energy storage can effectively reduce peak load demand and decrease load fluctuations.

[0086] like Figure 10 and Figure 11 As shown, similar to the economic regulation model, by rationally allocating the state of charge of distributed energy storage, the controllable potential of aggregated distributed energy storage is significantly improved compared to the unaggregated state. Since the peak-shaving regulation model does not consider economics and does not require full charging during the low electricity price period (1-10 pm), distributed energy storage can still absorb a considerable amount of power even during the most challenging charging times of the day. Therefore, the regulation model's improvement on the controllable potential of distributed energy storage does not reach the level of the economic model. However... t =11 o'clock, the improvement of the controllability potential of distributed energy storage by the controllability model is still greater than t =6 hours is large.

[0087] It can be seen that, after changing the application scenario, the control model still improves the controllability potential of distributed energy storage better under more severe conditions.

[0088] In summary, the energy storage aggregation model proposed in the present application can effectively reduce the load fluctuation of the power system, and the charging and discharging potential of the distributed energy storage can also be significantly improved.

[0089] Embodiment 3 As another embodiment of the present application, the embodiment also provides a system for performing the distributed energy storage aggregation regulation method based on improved clustering and population optimization as described in Embodiment 1 above, which specifically comprises: a distributed energy storage aggregation module: clustering the distributed energy storages based on the distributed energy storage regulation potential and parameter difference, and setting a distributed energy storage aggregator for each distributed energy storage group based on the clustering result; an economic regulation module, the distributed energy storage aggregator makes profits through electricity price arbitrage and participation in power grid auxiliary services, and regulates the charging and discharging behavior of the distributed energy storages by considering the energy storage operation cost; a power grid peak load shifting scenario auxiliary regulation module, which allocates the distributed energy storage aggregators for shifting the peak load of the power system, and allocates the state of charge of the distributed energy storages in the aggregator according to the regulatable potential of the distributed energy storages.

[0090] If the above functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the part of the prior art that essentially contributes to the present application or part of the technical solutions can be embodied in the form of a software product stored in a storage medium, including a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0091] The preferred embodiments of the present application are described in detail above. It should be understood that those skilled in the art can make many modifications and changes without creative labor based on the concept of the present application. Therefore, any technical solution obtained by logical analysis, reasoning or limited experiment based on the prior art according to the concept of the present application shall be within the protection scope determined by the claims.

Claims

1. A distributed energy storage aggregation and control method based on improved clustering and population optimization, characterized in that the steps are as follows: include: Distributed energy storage is clustered based on its regulation potential and parameter differences, and a distributed energy storage aggregator is set for each distributed energy storage group based on the clustering results. In the scenario of economic regulation of distributed energy storage, distributed energy storage aggregators participate in electricity price arbitrage and participate in grid ancillary services to profit. The goal is to maximize the total economic benefits of energy storage aggregators and take into account the energy storage operating costs, thereby regulating the charging and discharging behavior of each distributed energy storage unit. In the scenario of distributed energy storage assisting the power grid in peak shaving and valley filling, distributed energy storage aggregators are used to smooth the power grid load curve while maintaining the controllability potential of distributed energy storage itself, and to allocate the state of charge of distributed energy storage within the energy storage aggregator.

2. The distributed energy storage aggregation and control method based on improved clustering and population optimization according to claim 1, characterized in that, The specific steps for clustering distributed energy storage are as follows: A set of particles is randomly initialized with their positions and velocities. Each particle represents a grouping method, and the particles are represented by a vector composed of all the centroids of the classification. The sum of the errors of the data in the sample set and all classification centroids is calculated as the fitness of the particle. The velocity and position of all particles are updated based on inertia weight and acceleration factor to generate a new set of particles; if the velocity and / or position of the new particle exceeds the velocity and / or position boundary, the velocity and / or position of the particle is set to the boundary value of the corresponding velocity and / or position, and the particle with the smallest fitness in the current generation is stored. Repeatedly update the particle population until the current historical fitness optimum of the entire population has not changed within the specified number of iterations, or the number of runs has reached the maximum value; Calculate the distance of all distributed energy storage to each category centroid, and classify the distributed energy storage into the group to which the nearest centroid belongs; Calculate the average value of the regulation potential parameter vector for each energy storage unit within each cluster group, and replace the average value vector of the regulation potential parameter vector as the new centroid of the cluster group; Repeat the distributed energy storage partitioning and centroid replacement process until the change in the centroid position between two runs is less than a preset value, and obtain the final clustering grouping result of the distributed energy storage.

3. The distributed energy storage aggregation and control method based on improved clustering and population optimization according to claim 2, characterized in that, The regulation potential parameter vector includes: Inherent physical parameters, including rated capacity, rated power and charge / discharge efficiency; and initial state and constraint parameters, including initial state of charge and upper and lower limits of state of charge operation.

4. The distributed energy storage aggregation and control method based on improved clustering and population optimization according to claim 1, characterized in that, The economic optimization objective under the aforementioned economic regulation scenario is to maximize the total economic benefits for energy storage aggregators. The charging and discharging power of each distributed energy storage unit at each moment is used as a decision variable. By controlling the charging and discharging power, charging is achieved during low electricity prices and discharging is achieved during high electricity prices. The objective function is specifically expressed as follows: In the formula, Time-of-use electricity price indicating the time of day. The first term of formula (10) represents the revenue from energy sales, the second term represents the cost of energy purchase, and the two together constitute the revenue from the operation of distributed energy storage. The third term represents the revenue obtained by energy storage from participating in ancillary services, and the fourth term represents the degradation cost generated during the operation of energy storage.

5. The distributed energy storage aggregation and control method based on improved clustering and population optimization according to claim 1, characterized in that, The auxiliary optimization objective for distributed energy storage in the scenario of grid peak shaving and valley filling includes two layers: The first layer uses the total output of the distributed energy storage of the entire aggregator group at each moment as the decision variable, and minimizes the variance of the load curve as the optimization objective function: In the formula, It is time The load demand of the power grid For a moment Total output of distributed energy storage; The second-layer optimization model, based on the total output of the first-layer distributed energy storage, allocates the state of charge of each distributed energy storage unit, with the objective function being to maximize the sum of the smaller of the maximum charging power and the maximum discharging power of each distributed energy storage unit. in, Indicates at time t No. i The maximum power that a distributed energy storage system can continuously charge during subsequent operating cycles under its current state of charge; Indicates at time t No. i The maximum power that a distributed energy storage system can continuously discharge during subsequent operating cycles under its current state of charge.

6. The distributed energy storage aggregation and control method based on improved clustering and population optimization according to claim 5, characterized in that, The constraints of the auxiliary optimization include: energy storage capacity constraints, energy storage power constraints, and grid power constraints, which constrain the charge, power, and charging and discharging power of distributed energy storage to be within the set upper and lower limits.

7. The distributed energy storage aggregation and control method based on improved clustering and population optimization according to claim 5, characterized in that, The bi-level optimization problem with the auxiliary optimization objective is solved using a decomposition-coordination strategy: The objective function of the first layer is solved using optimization or intelligent algorithms to obtain the optimal total power command curve. The process of solving the objective function and allocating SOC for the second layer is as follows: At any given moment, the energy storage aggregator receives the total output command of the distributed energy storage from the first-level optimization output; Energy storage aggregators calculate the maximum continuous charging power and the maximum continuous discharging power of each energy storage unit under their jurisdiction based on the current state of charge of the energy storage. When the total output command of distributed energy storage is less than zero, priority is given to allocating energy storage with a larger maximum power for continuous charging and a smaller maximum power for continuous discharging for charging. When the total output command of distributed energy storage is greater than zero, priority is given to allocating energy storage with a larger maximum power for continuous discharge and a smaller maximum power for continuous charging for discharge. Through rolling optimization over multiple time steps, the state of charge of each distributed energy storage system is optimized and allocated.

8. A distributed energy storage aggregation and control system based on improved clustering and population optimization, characterized in that, The system includes: Distributed energy storage aggregation module: Based on the control potential and parameter differences of distributed energy storage, distributed energy storage is clustered, and a distributed energy storage aggregator is set for each distributed energy storage group based on the clustering results; The economic regulation module allows distributed energy storage aggregators to profit through electricity price arbitrage and participation in grid ancillary services, while also considering the operating costs of energy storage and regulating the charging and discharging behavior of distributed energy storage. The auxiliary control module for power grid peak shaving and valley filling scenarios allocates distributed energy storage aggregators to transfer peak loads in the power system, and considers the controllable potential of distributed energy storage to allocate the state of charge of distributed energy storage within the aggregators.

9. A distributed energy storage aggregation and control system based on improved clustering and population optimization according to claim 8, characterized in that, The distributed energy storage aggregation module performs the following steps for clustering distributed energy storage: A set of particles is randomly initialized with their positions and velocities. Each particle represents a grouping method, and the particles are represented by a vector composed of all the centroids of the classification. The sum of the errors of the data in the sample set and all classification centroids is calculated as the fitness of the particle. The velocity and position of all particles are updated based on inertia weight and acceleration factor to generate a new set of particles; if the velocity and / or position of the new particle exceeds the velocity and / or position boundary, the velocity and / or position of the particle is set to the boundary value of the corresponding velocity and / or position, and the particle with the smallest fitness in the current generation is stored. Repeatedly update the particle population until the current historical fitness optimum of the entire population has not changed within the specified number of iterations, or the number of runs has reached the maximum value; Calculate the distance of all distributed energy storage to each category centroid, and classify the distributed energy storage into the group to which the nearest centroid belongs; Calculate the average value of the regulation potential parameter vector for each energy storage unit within each cluster group, and replace the average value vector of the regulation potential parameter vector as the new centroid of the cluster group; The regulation potential parameter vector includes: inherent physical parameters, including rated capacity, rated power and charge / discharge efficiency; And initial state and constraint parameters, including initial state of charge and upper and lower limits of operation under state of charge. Repeat the distributed energy storage partitioning and centroid replacement process until the change in the centroid position between two runs is less than a preset value, and obtain the final clustering grouping result of the distributed energy storage.

10. A distributed energy storage aggregation and control system based on improved clustering and population optimization according to claim 8, characterized in that, The economic optimization objective under the aforementioned economic regulation scenario is to maximize the total economic benefits for energy storage aggregators. The charging and discharging power of each distributed energy storage unit at each moment is used as a decision variable. By controlling the charging and discharging power, charging is achieved during low electricity prices and discharging is achieved during high electricity prices. The objective function is specifically expressed as follows: In the formula, Time-of-use electricity price indicating the time of day. The first term of formula (10) represents the energy sales revenue, the second term represents the energy purchase cost, and the two together constitute the operating revenue of distributed energy storage. The third term represents the revenue obtained by energy storage participating in ancillary services, and the fourth term represents the degradation cost generated during the operation of energy storage. The auxiliary optimization objective of the auxiliary control module for the grid peak shaving and valley filling scenario includes two layers: The first layer uses the total output of the distributed energy storage of the entire aggregator group at each moment as the decision variable, and minimizes the variance of the load curve as the optimization objective function: In the formula, It is time The load demand of the power grid For a moment Total output of distributed energy storage; The second-layer optimization model, based on the total output of the first-layer distributed energy storage, allocates the state of charge of each distributed energy storage unit, with the objective function being to maximize the sum of the smaller of the maximum charging power and the maximum discharging power of each distributed energy storage unit. in, Indicates at time t No. i The maximum power that a distributed energy storage system can continuously charge during subsequent operating cycles under its current state of charge; Indicates at time t No. i The maximum power that a distributed energy storage system can continuously discharge during subsequent operating cycles under its current state of charge.

Images