Power distribution method and system for grid-forming energy storage system based on energy storage potential function

Abstract

The application discloses a network-constructing energy storage system power distribution method and system based on energy storage potential function, belongs to the technical field of network-constructing energy storage system control, and comprises the following steps: establishing a network-constructing energy storage system; establishing a VSG frequency dynamic model to execute primary control; executing secondary control on the network-constructing energy storage system to complete power distribution; obtaining a state of charge model of distributed energy; establishing an energy storage potential function of the network-constructing energy storage system; obtaining a dynamic updating law of corresponding distributed energy according to the energy storage potential function to adjust corresponding control input signals, realizing convergence of the network-constructing energy storage system to a minimum value of the energy storage potential function; when a cooperative control error is zero, deriving a capacity balance potential function and substituting into the state of charge model to obtain a proportion of actual output active power of each distributed energy equal to a proportion of corresponding damping coefficients, thereby completing power distribution. The application can guarantee stability of secondary frequency modulation and zero steady-state tracking error, improve load fluctuation resistance and external disturbance capacity, and has low communication cost.

Description

Technical Field

[0001] This invention belongs to the field of control technology for grid-type energy storage systems, specifically relating to a power allocation method and system for grid-type energy storage systems based on energy storage potential functions. Background Technology

[0002] With the continuous increase in the proportion of new energy power generation, the inertia level of the power system has decreased significantly, and power generation exhibits randomness, volatility, and uncontrollability. Traditional operating modes relying on synchronous generators face severe challenges. New power systems such as grid-based energy storage systems are gradually emerging, while simultaneously placing higher demands on the dynamic characteristics of power electronic interface devices. Virtual synchronous machine (VSG) technology is often used as one of the control strategies for grid-based energy storage. By introducing the motion equations of a synchronous generator into the control loop and simulating its rotational inertia and damping characteristics with the help of energy storage devices, the inverter can exhibit a dynamic response similar to that of a traditional synchronous machine under grid disturbances.

[0003] In grid-based energy storage systems, VSG control can enhance the inertial support and damping characteristics of the power supply, but it still inevitably leads to some deviation in system frequency and voltage. Therefore, it is necessary to introduce a secondary control loop into the system to achieve frequency and voltage recovery and stabilization. By controlling the active power, the corresponding voltage is also controlled. Grid-based energy storage systems independently supply power to their own loads, optimizing power quality and alleviating the power supply pressure on the power system. However, due to the highly nonlinear, complex dynamic, and MIMO characteristics of grid-based energy storage systems, traditional secondary frequency regulation control methods face many limitations: linear secondary frequency regulation relies on simplified models and is difficult to adapt to nonlinear coupling characteristics; nonlinear secondary frequency regulation is complex to design, difficult to tune parameters, and its performance depends on an accurate dynamic model, making it susceptible to model mismatch; although neural network-based control methods have nonlinear modeling capabilities, their black-box characteristics make it difficult to theoretically guarantee stability, and under actual operating conditions with communication constraints and limited computing resources, it is difficult to coordinate global stability and real-time control performance.

[0004] Existing secondary control methods for grid-based energy storage systems primarily involve constructing a two-layer secondary control system to achieve two-layer communication between distributed energy sources within the system and between distributed energy sources acting as proxies across the cluster. This approach not only involves large communication volumes and high communication costs, but also focuses on the cluster as a whole, uploading only the aggregated information and neglecting the real-time status, local constraints, and individual differences of individual distributed energy sources. Therefore, it easily overlooks the situation of individual distributed energy sources within the cluster. While traditional droop control can achieve power allocation, its accuracy decreases under scenarios of parameter mismatch or sudden load changes. Furthermore, existing neural network methods are mostly black-box models, lacking structured constraints and struggling to coordinate global stability with real-time control performance. Therefore, a coordinated control scheme for grid-based energy storage systems that combines frequency recovery and precise power allocation capabilities is urgently needed. Summary of the Invention

[0005] The purpose of this invention is to address the above-mentioned problems by proposing a power allocation method and system for grid-type energy storage systems based on energy storage potential functions. This method can ensure that the actual output active power of distributed energy sources is allocated proportionally, guarantee the stability of the secondary frequency regulation and zero steady-state tracking error of the grid-type energy storage system, improve the ability to resist load fluctuations and external disturbances, and reduce communication costs.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] The power allocation method for grid-type energy storage systems based on energy storage potential functions proposed in this invention includes the following steps:

[0008] S1. Establish a grid-based energy storage system, which includes... Cluster, of which, the first Each cluster contains A distributed energy source, , It is a positive integer;

[0009] S2. Establish a VSG frequency dynamic model to perform primary control on the grid-type energy storage system;

[0010] S3. Perform secondary control on the grid-type energy storage system to complete power distribution, as follows:

[0011] S31. Obtain the state of charge model of distributed energy resources;

[0012] S32. Establish the energy storage potential function of the grid-type energy storage system. The energy storage potential function is the sum of the virtual inertial potential function, the capacity equilibrium potential function, and the learnable nonlinear potential function. The input features of the energy storage potential function include the frequency deviation of each distributed energy source, the collaborative control error, and the energy storage capacity. The learnable nonlinear potential function is obtained by using a feedforward neural network to extract features from the input features of the energy storage potential function.

[0013] S33. Obtain the dynamic update law of the corresponding distributed energy source based on the energy storage potential function to adjust the control input signal of the corresponding distributed energy source, so as to realize the convergence of the grid-type energy storage system to the minimum value of the energy storage potential function.

[0014] S34. When the collaborative control error of the grid-type energy storage system is zero, the capacity equilibrium potential function is differentiated and then input into the state of charge model. This yields the ratio of the actual output active power of each distributed energy source to the ratio of the corresponding damping coefficient when the grid-type energy storage system is in steady-state operation, thus completing the power allocation of the grid-type energy storage system.

[0015] Preferably, the VSG frequency dynamic model is established as follows:

[0016] S21. Obtain the frequency model of each distributed energy source;

[0017] S22. Based on the frequency deviation of the distributed energy source, perform a consistent projection transformation on the corresponding frequency model to obtain the VSG frequency dynamic model of the corresponding distributed energy source, as shown in the following formula:

[0018]

[0019] In the formula, Let be the derivative of the phase angle of the distributed energy source with respect to time. It is the identity matrix. For dimension A column vector of all 1s For dimension A row vector consisting entirely of 1s. This refers to the total number of distributed energy sources in a grid-type energy storage system. The frequency deviation vector is the derivative of the frequency deviation vector with respect to time. , For the first In the cluster, the first Frequency deviation of distributed energy sources For column vectors, the equivalent moment of inertia vector is... , For the first In the cluster, the first The equivalent moment of inertia and damping coefficient vector of a distributed energy source. , For the first In the cluster, the first The damping coefficient of a distributed energy source controls the input signal vector. , For the first In the cluster, the first The control input signal for a distributed energy source, and the load power vector. , For the first In the cluster, the first The load power of a distributed energy source The rated active power of distributed energy resources. The rated angular velocity of distributed energy sources. This is the association matrix between nodes and lines in the time-varying communication topology diagram of a grid-type energy storage system. For the phase angle of distributed energy, Transpose, transformation matrix , No. diagonal matrices , For the first The number of lines in the cluster, the first diagonal matrices The Middle Line 1 diagonal items of the column This represents the power coupling coefficient at the corresponding location. It is a diagonal matrix. .

[0020] Preferably, the frequency model of each distributed energy source is obtained, then the first... In the cluster, the first The frequency model for distributed energy resources is shown in the following formula:

[0021]

[0022] In the formula, For the first In the cluster, the first The equivalent rotational inertia of a distributed energy source, For the first In the cluster, the first Frequency of distributed energy Seeking information about time The partial derivative, The rated active power of distributed energy resources. For the first In the cluster, the first The actual active power output of a distributed energy source The rated angular velocity of distributed energy sources. For the first In the cluster, the first The damping coefficient of a distributed energy source. For the first In the cluster, the first The frequency of a distributed energy source The rated frequency for distributed energy resources, ;

[0023] The frequency deviation of distributed energy resources is calculated using the following formula:

[0024]

[0025] In the formula, For the first In the cluster, the first Frequency deviation of distributed energy sources The rated frequency for distributed energy sources.

[0026] Preferably, the state-of-charge model of the distributed energy source is obtained, then the first... In the cluster, the first The state-of-charge model for a distributed energy source is shown in the following formula:

[0027]

[0028] In the formula, For the first In the cluster, the first The time derivative of the energy storage capacity of a distributed energy source. For the first In the cluster, the first The actual active power output of a distributed energy source For the first In the cluster, the first The rated energy storage capacity of a distributed energy source. ;

[0029] The collaborative control error of distributed energy resources satisfies the first... In the cluster, the first Coordinated control error of distributed energy sources The calculation is as follows:

[0030]

[0031] In the formula, For the first In the cluster, the first A control input signal for a distributed energy source For the first In the cluster, the first The control input signal for a distributed energy source, N k,i For the first In the cluster, the first A set of neighboring distributed energy sources of a distributed energy source.

[0032] Preferably, the energy storage potential function The formula is as follows:

[0033]

[0034] Among them, the virtual inertial potential function The formula is as follows:

[0035]

[0036] In the formula, For the first In the cluster, the first The equivalent rotational inertia of a distributed energy source, For the first In the cluster, the first The frequency of a distributed energy source The rated frequency for distributed energy resources;

[0037] Capacity equilibrium potential function The formula is as follows:

[0038]

[0039]

[0040] In the formula, For the first In the cluster, the first The weighting coefficients of distributed energy resources, For the first In the cluster, the first The energy storage capacity of a distributed energy source, This represents the average energy storage capacity of all distributed energy sources in a grid-type energy storage system. The total number of distributed energy sources in a grid-type energy storage system;

[0041] Learnable nonlinear potential function The formula is as follows:

[0042]

[0043] In the formula, It is a feedforward neural network. This is the input characteristic of the energy storage potential function.

[0044] Preferably, the dynamic update law of the corresponding distributed energy source is obtained based on the energy storage potential function, then the first... In the cluster, the first The dynamic update law for distributed energy resources is given by the following formula:

[0045]

[0046] In the formula, For the first In the cluster, the first A control input signal for a distributed energy source For the first In the cluster, the first The gain coefficient of a distributed energy source. For the energy storage potential function of a grid-type energy storage system Seeking information about The partial derivatives, Input characteristics of the energy storage potential function The first in One variable, , .

[0047] Preferably, the power allocation method for a grid-type energy storage system based on the energy storage potential function further includes obtaining the communication weights among distributed energy sources to construct the positive definite gain matrix of the grid-type energy storage system, then:

[0048] The communication weight between distributed energy sources is calculated using the following formula:

[0049]

[0050]

[0051] In the formula, Laplace matrix representing time-varying communication topology The Line 1 The communication weight of column elements, and , , This refers to the total number of distributed energy sources in a grid-type energy storage system. The Laplace matrix of the time-varying communication topology The Line 1 Communication weights of column elements The Laplace matrix of the time-varying communication topology The first line Frequency deviation of each node The Laplace matrix of the time-varying communication topology The first in the list Frequency deviation of each node The Laplace matrix of the time-varying communication topology The first line Energy storage capacity of each node, The Laplace matrix of the time-varying communication topology The first line Energy storage capacity of each node, The weighting coefficient for frequency deviation, Weighting coefficients for energy storage capacity deviation; time-varying communication topology diagram. , For a set of nodes, Let be a set of edges, where nodes are distributed energy sources and edges are lines;

[0052] The gain coefficient of each distributed energy source is obtained to form the positive definite gain matrix of the grid-type energy storage system. The gain coefficient of each distributed energy source is the sum of all associated communication weights of the corresponding distributed energy source.

[0053] Preferably, each cluster has a proxy distributed energy source, and each distributed energy source obtains the control input signal of the proxy distributed energy source within the corresponding cluster or the adjacent distributed energy source through one-way communication.

[0054] Preferably, the agent distributed energy source of each cluster is the root node of the time-varying communication topology of the grid-type energy storage system. When the grid-type energy storage system experiences external disturbances or load changes, the agent distributed energy source first updates the dynamic update law, and then transmits the control input signal adjusted by the dynamic update law to other distributed energy sources in its own cluster through the time-varying communication topology.

[0055] A power allocation system for a grid-type energy storage system based on an energy storage potential function includes a memory and a processor. The memory stores a computer program, which, when executed by the processor, implements the power allocation method for the grid-type energy storage system based on the energy storage potential function described above.

[0056] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0057] This method constructs a grid-type energy storage system and uses a VSG frequency dynamic model to perform primary control on the grid-type energy storage system. Based on frequency deviation, cooperative control error, and energy storage capacity, it performs dual-objective optimization of frequency recovery and energy storage capacity balancing through the energy storage potential function. This achieves frequency recovery and power distribution of the grid-type energy storage system, optimizes transient response performance, and improves resistance to load fluctuations and external disturbances. Compared with existing technologies, this method offers high stability and enhances adaptability to nonlinear disturbances by introducing a learnable nonlinear potential function, ensuring local asymptotic stability of the grid-connected energy storage system. Compared with traditional PI control and linear neural network control, it significantly improves transient response speed, steady-state error suppression, and disturbance resistance. It can flexibly adapt to centralized, distributed, and decentralized communication topologies, meeting the communication constraints in the actual operation of grid-connected energy storage systems, demonstrating strong communication adaptability. Unidirectional communication helps reduce communication volume and costs, and control can be performed with less communication requirement information, offering privacy protection advantages and wide applicability. It reduces dependence on training data, achieving higher training efficiency than existing neural networks and lowering training costs. This application is particularly suitable for frequency stability control and actual output active power distribution under grid-connected operation mode. Attached Figure Description

[0058] Figure 1 This is a flowchart of the power allocation method for a grid-type energy storage system based on the energy storage potential function according to the present invention;

[0059] Figure 2 This is a topology diagram of the grid-type energy storage system of the present invention;

[0060] Figure 3 This is a schematic diagram showing the frequency and actual output active power of the grid-type energy storage system of the present invention in a centralized configuration.

[0061] Figure 4This is a schematic diagram showing the frequency and actual output active power of the grid-type energy storage system of the present invention under a distributed configuration.

[0062] Figure 5 This is a schematic diagram showing the frequency and actual output active power of the distributed grid-type energy storage system of the present invention.

[0063] Figure 6 This is a control action response output diagram of the grid-type energy storage system of the present invention under centralized conditions;

[0064] Figure 7 This is a control action response output diagram for the distributed grid-type energy storage system of the present invention.

[0065] Figure 8 This is a diagram showing the control action response output of the distributed grid-type energy storage system of the present invention. Detailed Implementation

[0066] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0067] It should be noted that when a component is referred to as being "connected" to another component, it can be directly connected to the other component or there may be an intervening component. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application.

[0068] Example 1:

[0069] like Figures 1-8 As shown, the power allocation method for a grid-type energy storage system based on the energy storage potential function includes the following steps:

[0070] S1. Establish a grid-based energy storage system, which includes... Cluster, of which, the first Each cluster contains A distributed energy source, , It is a positive integer.

[0071] S2. Establish a VSG frequency dynamic model to perform primary control on the grid-type energy storage system.

[0072] In one embodiment, the VSG frequency dynamic model is established as follows:

[0073] S21. Obtain the frequency model of each distributed energy source;

[0074] S22. Based on the frequency deviation of the distributed energy source, perform a consistent projection transformation on the corresponding frequency model to obtain the VSG frequency dynamic model of the corresponding distributed energy source, as shown in the following formula:

[0075]

[0076] In the formula, Let be the derivative of the phase angle of the distributed energy source with respect to time. It is the identity matrix. For dimension A column vector of all 1s For dimension A row vector consisting entirely of 1s. This refers to the total number of distributed energy sources in a grid-type energy storage system. The frequency deviation vector is the derivative of the frequency deviation vector with respect to time. , For the first In the cluster, the first Frequency deviation of distributed energy sources For column vectors, the equivalent moment of inertia vector is... , For the first In the cluster, the first The equivalent moment of inertia and damping coefficient vector of a distributed energy source. , For the first In the cluster, the first The damping coefficient of a distributed energy source controls the input signal vector. , For the first In the cluster, the first The control input signal for a distributed energy source, and the load power vector. , For the first In the cluster, the first The load power of a distributed energy source The rated active power of distributed energy resources. The rated angular velocity of distributed energy sources. This is the association matrix between nodes and lines in the time-varying communication topology diagram of a grid-type energy storage system. For the phase angle of distributed energy, Transpose, transformation matrix , No. diagonal matrices , For the first The number of lines in the cluster, the first diagonal matrices The Middle Line 1 diagonal items of the column This represents the power coupling coefficient at the corresponding location. It is a diagonal matrix. .

[0077] In one embodiment, the frequency model of each distributed energy source is obtained, then the... In the cluster, the first The frequency model for distributed energy resources is shown in the following formula:

[0078]

[0079] In the formula, For the first In the cluster, the first The equivalent rotational inertia of a distributed energy source, For the first In the cluster, the first Frequency of distributed energy Seeking information about time The partial derivative, The rated active power of distributed energy resources. For the first In the cluster, the first The actual active power output of a distributed energy source The rated angular velocity of distributed energy sources. For the first In the cluster, the first The damping coefficient of a distributed energy source. For the first In the cluster, the first The frequency of a distributed energy source The rated frequency for distributed energy resources, ;

[0080] The frequency deviation of distributed energy resources is calculated using the following formula:

[0081]

[0082] In the formula, For the first In the cluster, the first Frequency deviation of distributed energy sources The rated frequency for distributed energy sources.

[0083] S3. Perform secondary control on the grid-type energy storage system to complete power distribution, as follows:

[0084] S31. Obtain the state of charge model of distributed energy resources;

[0085] S32. Establish the energy storage potential function of the grid-type energy storage system. The energy storage potential function is the sum of the virtual inertial potential function, the capacity equilibrium potential function, and the learnable nonlinear potential function. The input features of the energy storage potential function include the frequency deviation of each distributed energy source, the collaborative control error, and the energy storage capacity. The learnable nonlinear potential function is obtained by using a feedforward neural network to extract features from the input features of the energy storage potential function.

[0086] S33. Obtain the dynamic update law of the corresponding distributed energy source based on the energy storage potential function to adjust the control input signal of the corresponding distributed energy source, so as to realize the convergence of the grid-type energy storage system to the minimum value of the energy storage potential function.

[0087] S34. When the collaborative control error of the grid-type energy storage system is zero, the capacity equilibrium potential function is differentiated and then input into the state of charge model. This yields the ratio of the actual output active power of each distributed energy source to the ratio of the corresponding damping coefficient when the grid-type energy storage system is in steady-state operation, thus completing the power allocation of the grid-type energy storage system.

[0088] In one embodiment, the state-of-charge model of the distributed energy source is obtained, then the first... In the cluster, the first The state-of-charge model for a distributed energy source is shown in the following formula:

[0089]

[0090] In the formula, For the first In the cluster, the first The time derivative of the energy storage capacity of a distributed energy source. For the first In the cluster, the first The actual active power output of a distributed energy source For the first In the cluster, the first The rated energy storage capacity of a distributed energy source. ;

[0091] The collaborative control error of distributed energy resources satisfies the first... In the cluster, the first Coordinated control error of distributed energy sources The calculation is as follows:

[0092]

[0093] In the formula, For the first In the cluster, the first A control input signal for a distributed energy source For the first In the cluster, the first The control input signal for a distributed energy source, N k,i For the first In the cluster, the first A distributed energy source is a set of neighboring distributed energy sources. A neighboring distributed energy source refers to a set of distributed energy sources that have a direct communication or electrical connection with the corresponding distributed generation (DG).

[0094] In one embodiment, the energy storage potential function The formula is as follows:

[0095]

[0096] Among them, the virtual inertial potential function The formula is as follows:

[0097]

[0098] In the formula, For the first In the cluster, the first The equivalent rotational inertia of a distributed energy source, For the first In the cluster, the first The frequency of a distributed energy source The rated frequency for distributed energy resources;

[0099] Capacity equilibrium potential function The formula is as follows:

[0100]

[0101]

[0102] In the formula, For the first In the cluster, the first The weighting coefficients of distributed energy resources, For the first In the cluster, the first The energy storage capacity of a distributed energy source, This represents the average energy storage capacity of all distributed energy sources in a grid-type energy storage system. The total number of distributed energy sources in a grid-type energy storage system;

[0103] Learnable nonlinear potential function The formula is as follows:

[0104]

[0105] In the formula, is a feedforward neural network, is the input feature of the energy storage potential function.

[0106] In one embodiment, according to the energy storage potential function, the dynamic update law of the corresponding distributed energy is obtained. Then, for the th distributed energy in the

[0107]

[0108] In the formula, is the th control input signal of the th distributed energy in the th cluster, is the partial derivative of the energy storage potential function of the grid-connected energy storage system with respect to is the th variable in the input feature of the energy storage potential function, .

[0109] Among them, when , it represents the frequency deviation of the distributed energy; when , it represents the cooperative control error of the distributed energy;

[0110] In one embodiment, the grid-connected energy storage system power distribution method based on the energy storage potential function further includes obtaining the communication weights between distributed energies to construct a positive definite gain matrix of the grid-connected energy storage system. Then:

[0111] The communication weights between distributed energies are as follows:

[0112]

[0113]

[0114] In the formula, represents the th row and The communication weight of column elements, and , , This refers to the total number of distributed energy sources in a grid-type energy storage system. The Laplace matrix of the time-varying communication topology The Line 1 Communication weights of column elements The Laplace matrix of the time-varying communication topology The first line Frequency deviation of each node The Laplace matrix of the time-varying communication topology The first in the list Frequency deviation of each node The Laplace matrix of the time-varying communication topology The first line Energy storage capacity of each node, The Laplace matrix of the time-varying communication topology The first line Energy storage capacity of each node, The weighting coefficient for frequency deviation, Weighting coefficients for energy storage capacity deviation; time-varying communication topology diagram. , For a set of nodes, Let be a set of edges, where nodes are distributed energy sources and edges are lines;

[0115] The gain coefficient of each distributed energy source is obtained to form the positive definite gain matrix of the grid-type energy storage system. The gain coefficient of each distributed energy source is the sum of all associated communication weights of the corresponding distributed energy source.

[0116] The invention incorporates a Laplace matrix into the time-varying communication topology of the grid-type energy storage system. The feasibility of the method can be verified through a distributed consensus mechanism, a technique well-known to those skilled in the art and not elaborated upon here. Communication weights are time-varying parameters that influence the values ​​of the control input parameters (positive definite gain matrix). The diagonal elements of the positive definite gain matrix represent the sum of all associated communication weights for the corresponding distributed energy source. This determines the consensus convergence speed and information coupling strength of the grid-type energy storage system. Its value is adaptively adjusted according to the state differences of the distributed energy sources to enhance the coordinated control performance of the grid-type energy storage system.

[0117] Among them, by optimizing the control input signal vector This aims to improve the transient response of grid-based energy storage systems after disturbances while ensuring steady-state performance. Steady-state performance requires the grid-based energy storage system to stabilize at the desired value, while transient performance requires the system to converge to steady state with minimal control overhead. This involves restoring the frequency while ensuring proportional power distribution.

[0118] Specifically, differentiating the formula for the capacity equilibrium potential function, we get:

[0119]

[0120] In the formula, for The derivative, for The derivative, for The derivative;

[0121] Substituting into the state-of-charge model, we can see that when At that time, power will be allocated according to the proportion of energy storage capacity.

[0122] Under the condition of rated frequency synchronization, that is, when the cooperative control error is zero (the grid-type energy storage system converges to the minimum value of the energy storage potential function), such as Zero:

[0123]

[0124]

[0125]

[0126] In the formula, For the first In the cluster, the first The rated power of each distributed energy source can be determined from the VSG frequency dynamic model, where the input power of each distributed energy source equals its output active power. Combined with the consistency constraint satisfied by the secondary control input, the output active power of each distributed energy source satisfies the following coordination distribution relationship:

[0127]

[0128] In the formula, For the first In the cluster, the first The actual active power output of a distributed energy source For the first In the cluster, the first The actual active power output of a distributed energy source For the first In the cluster, the first The damping coefficient of a distributed energy source. For the first In the cluster, the first The damping coefficient of a distributed energy source. The power allocation can be completed according to the distribution relationship of the actual output active power of each distributed energy source.

[0129] In one embodiment, each cluster has a proxy distributed energy source, and each distributed energy source obtains control input signals from the proxy distributed energy source within the corresponding cluster or from adjacent distributed energy sources through one-way communication.

[0130] Each cluster has a proxy distributed energy source. The consistency of the secondary control input of each distributed energy source in steady state is achieved by obtaining the control input signals of the proxy distributed energy source or adjacent distributed energy sources within the corresponding cluster through one-way communication.

[0131] In one embodiment, the agent distributed energy source of each cluster is the root node of the time-varying communication topology of the grid-type energy storage system. When the grid-type energy storage system experiences external disturbances or load changes, the agent distributed energy source first updates the dynamic update law, and then transmits the control input signal adjusted by the dynamic update law to other distributed energy sources in its own cluster through the time-varying communication topology.

[0132] In this embodiment, any one of the distributed energy sources in each cluster is a proxy distributed energy source, which can be adjusted according to actual needs. The proxy distributed energy source acts as the root node in the time-varying communication topology diagram. When the grid-type energy storage system experiences external disturbances or load changes, the proxy distributed energy source prioritizes updating its secondary control dynamic update law, and then transmits the updated control input signal to other distributed energy sources in its cluster level by level through the time-varying communication topology diagram described by the Laplace matrix.

[0133] To visually verify the effectiveness of the method proposed in this invention, specific experiments are described below.

[0134] This experiment utilizes the interactive web-based computing environment Jupyter Notebook to build a VSG frequency dynamic model (primary control) and a controller (secondary control), and uses the open-source Python package Tensorflow 2.0 to train the controller. During the controller training phase, the number of training epochs is set to 400, meaning the entire training process will involve 400 complete iterative optimization rounds. In each training round, random power disturbance sequences are generated in batches. These disturbances simulate potential load surges or distributed energy source failures that may occur in a real power grid (grid-based energy storage system). Specifically, three distributed energy sources are randomly selected, and random power step changes are applied at random time points. These disturbance inputs are loaded into the initial state and then dynamically simulated, thus constructing diverse training samples. The state (frequency, phase angle, etc.) of the grid-based energy storage system evolves over time, generating dynamic response data. The training process does not rely on a fixed dataset but generates training samples online, ensuring the controller can adapt to random failures and improving robustness and generalization ability. Finally, simulation experiments are conducted on the VSG frequency dynamic model to verify the effectiveness of the designed controller. The hardware environment used in the experiment was: a 13th Gen Intel(R) Core(TM) i7-13620H processor, paired with 16GB of memory, and an NVIDIA GeForce RTX 4060 Laptop GPU. To verify the effectiveness of the method proposed in this invention, the specific experimental verification process is as follows:

[0135] Considering that most grid-based energy storage systems in real-world applications do not have fully connected real-time communication capabilities, the following performance comparison is presented under three different communication conditions—centralized, distributed (partial communication, as shown in the following examples), and decentralized—to verify the controller performance under communication constraints.

[0136] The grid-type energy storage system includes 10 distributed energy sources (DG), namely The first cluster includes three distributed energy sources ( , , , that is , , ), correspond The same applies to the rest; the second cluster includes four distributed energy sources ( , , , , that is , , , The third cluster includes three distributed energy sources (); , , , that is , , The first distributed energy source in each cluster is a proxy distributed energy source. The topology diagram of a grid-type energy storage system built with 10 distributed energy sources is shown below. Figure 2 As shown, Figure 2 The numbers 1-39 represent the corresponding grid bus node numbers. Black vertical or horizontal lines represent lines used to connect different nodes for power transmission and distribution. Downward-pointing black arrows represent loads, i.e., the electrical equipment connected to the corresponding node. The experiment simulates external disturbances by changing the load on the grid-type energy storage system at the 3rd and 6th seconds (load disconnected at the 3rd second and inserted at the 6th second), thereby examining the controller's dynamic response capability and frequency stability under sudden disturbances. To verify the robustness of the proposed controller under parameter changes in the grid-type energy storage system, some parameter settings of the VSG frequency dynamic model are shown in Table 1.

[0137] Table 1

[0138]

[0139] Depend on Figure 3 , Figure 4 , Figure 5 It is evident that under centralized, distributed, and decentralized communication conditions, the communication capability gradually decreases, and the convergence speed of the controller slows down accordingly. However, the stability of the grid-type energy storage system can be reliably guaranteed under all communication conditions. For example, if load disconnection and load insertion occur at the 3rd and 6th seconds respectively, the frequency of the grid-type energy storage system eventually reaches the rated frequency after a brief adjustment period, and the actual output active power returns to the required damping coefficient ratio. Figure 6 , Figure 7 , Figure 8 It can be seen that when responding to load disturbances, the control actions (i.e., the compensation amount of the control output) of the grid-type energy storage system can quickly correct deviations and then converge rapidly, demonstrating the rapid response capability of the grid-type energy storage system. Figure 6 In the above code, the control action before the load is disconnected is 0.60, the control action after the load is disconnected is 0.15, and the control action after the load is inserted is 0.60. Figure 7 , Figure 8Similarly, the controller not only fully leverages the advantages of adaptive learning but also achieves an optimal balance between rapid dynamic response and stability, demonstrating excellent control performance. In contrast, existing neural networks exhibit significant oscillations and even unstable behavior during dynamic processes. This is mainly due to the interference between proportional and integral control paths caused by their densely connected structure, the severe gain fluctuations caused by multi-layer nonlinear transformations, and the output resonance induced by the lack of physical constraints in the grid-type energy storage system, making effective frequency regulation difficult. This further highlights the crucial role of secondary control in ensuring the stability and dynamic performance of grid-type energy storage systems.

[0140] The experimental results above demonstrate that the method of this invention has significant advantages over traditional control methods in terms of steady-state error and frequency control, and maintains stability under three different communication conditions, thus improving controller performance. It not only achieves frequency recovery for the entire grid-type energy storage system but also effectively improves the economic efficiency of the system. This method can rapidly achieve frequency recovery in grid-type energy storage systems, ensuring that the actual output active power of each distributed energy source is distributed according to the damping coefficient ratio, with low communication costs and wide applicability.

[0141] Example 2:

[0142] A power allocation system for a grid-type energy storage system based on an energy storage potential function includes a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, it implements the power allocation method for a grid-type energy storage system based on an energy storage potential function as described in Example 1.

[0143] The memory and processor are electrically connected directly or indirectly to enable data transmission or interaction. For example, these components can be electrically connected to each other via one or more communication buses or signal lines. The memory stores a computer program that can run on the processor. By running the computer program stored in the memory, the processor implements the power allocation method for a grid-type energy storage system based on the energy storage potential function in Embodiment 1 of the present invention.

[0144] The memory can be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), and Electrically Erasable Programmable Read-Only Memory (EEPROM). The memory stores computer programs, and the processor executes these programs after receiving execution instructions.

[0145] The processor may be an integrated circuit chip with data processing capabilities. The aforementioned processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc. It can implement or execute the methods, steps, and logic block diagrams disclosed in Embodiment 1 of this invention. The general-purpose processor can be a microprocessor or any conventional processor.

[0146] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0147] The embodiments described above are merely specific and detailed examples of the embodiments described in this application, and should not be construed as limiting the scope of the application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the appended claims.

Claims

1. A power allocation method for a grid-type energy storage system based on energy storage potential function, characterized in that: Includes the following steps: S1. Establish a grid-based energy storage system, wherein the grid-based energy storage system includes... Cluster, of which, the first Each cluster contains A distributed energy source, , It is a positive integer; S2. Establish a VSG frequency dynamic model to perform primary control on the grid-type energy storage system; S3. Perform secondary control on the grid-type energy storage system to complete power distribution, as follows: S31. Obtain the state of charge model of distributed energy resources; S32. Establish the energy storage potential function of the grid-type energy storage system. The energy storage potential function is the sum of the virtual inertial potential function, the capacity equilibrium potential function, and the learnable nonlinear potential function. The input features of the energy storage potential function include the frequency deviation, collaborative control error, and energy storage capacity of each distributed energy source. The learnable nonlinear potential function is obtained by using a feedforward neural network to extract features from the input features of the energy storage potential function. S33. Obtain the dynamic update law of the corresponding distributed energy source based on the energy storage potential function to adjust the control input signal of the corresponding distributed energy source, so as to realize the convergence of the grid-type energy storage system to the minimum value of the energy storage potential function. S34. When the collaborative control error of the grid-type energy storage system is zero, the capacity equilibrium potential function is differentiated and then input into the state of charge model. This yields the ratio of the actual output active power of each distributed energy source to the ratio of the corresponding damping coefficient when the grid-type energy storage system is in steady-state operation, thus completing the power allocation of the grid-type energy storage system.

2. The power allocation method for a grid-type energy storage system based on the energy storage potential function as described in claim 1, characterized in that: The VSG frequency dynamic model is established as follows: S21. Obtain the frequency model of each distributed energy source; S22. Based on the frequency deviation of the distributed energy source, perform a consistent projection transformation on the corresponding frequency model to obtain the VSG frequency dynamic model of the corresponding distributed energy source, as shown in the following formula: In the formula, Let be the derivative of the phase angle of the distributed energy source with respect to time. It is the identity matrix. For dimension A column vector of all 1s. For dimension A row vector consisting entirely of 1s. This refers to the total number of distributed energy sources in a grid-type energy storage system. The frequency deviation vector is the derivative of the frequency deviation vector with respect to time. , For the first In the cluster, the first Frequency deviation of distributed energy sources For column vectors, the equivalent moment of inertia vector is... , For the first In the cluster, the first The equivalent moment of inertia and damping coefficient vector of a distributed energy source. , For the first In the cluster, the first The damping coefficient of a distributed energy source controls the input signal vector. , For the first In the cluster, the first The control input signal for a distributed energy source, and the load power vector. , For the first In the cluster, the first The load power of a distributed energy source The rated active power of distributed energy resources. The rated angular velocity of distributed energy sources. This is the association matrix between nodes and lines in the time-varying communication topology diagram of a grid-type energy storage system. For the phase angle of distributed energy, Transpose, transformation matrix , No. diagonal matrices , For the first The number of lines in the cluster, the first diagonal matrices The Middle Line 1 diagonal items of the column This represents the power coupling coefficient at the corresponding location. It is a diagonal matrix. .

3. The power allocation method for a grid-type energy storage system based on the energy storage potential function as described in claim 2, characterized in that: The frequency model of each distributed energy source is obtained, then the first... In the cluster, the first The frequency model for distributed energy resources is shown in the following formula: In the formula, For the first In the cluster, the first The equivalent rotational inertia of a distributed energy source, For the first In the cluster, the first Frequency of distributed energy Seeking information about time The partial derivative, The rated active power of distributed energy resources. For the first In the cluster, the first The actual output active power of a distributed energy source The rated angular velocity of distributed energy sources. For the first In the cluster, the first The damping coefficient of a distributed energy source. For the first In the cluster, the first The frequency of a distributed energy source The rated frequency for distributed energy resources, ; The frequency deviation of the distributed energy source is given by the following formula: In the formula, For the first In the cluster, the first Frequency deviation of distributed energy sources The rated frequency for distributed energy sources.

4. The power allocation method for a grid-type energy storage system based on the energy storage potential function as described in claim 1, characterized in that: The state of charge model for obtaining distributed energy resources is then... In the cluster, the first The state-of-charge model for a distributed energy source is shown in the following formula: In the formula, For the first In the cluster, the first The time derivative of the energy storage capacity of a distributed energy source. For the first In the cluster, the first The actual output active power of a distributed energy source For the first In the cluster, the first The rated energy storage capacity of a distributed energy source. ; The collaborative control error of the distributed energy source satisfies the following condition: In the cluster, the first Coordinated control error of distributed energy sources The calculation is as follows: In the formula, For the first In the cluster, the first A control input signal for a distributed energy source For the first In the cluster, the first The control input signal for a distributed energy source, N k,i For the first In the cluster, the first A set of neighboring distributed energy sources of a distributed energy source.

5. The power allocation method for a grid-type energy storage system based on the energy storage potential function as described in claim 1, characterized in that: The energy storage potential function The formula is as follows: Among them, the virtual inertial potential function The formula is as follows: In the formula, For the first In the cluster, the first The equivalent rotational inertia of a distributed energy source, For the first In the cluster, the first The frequency of a distributed energy source The rated frequency for distributed energy resources; Capacity equilibrium potential function The formula is as follows: In the formula, For the first In the cluster, the first The weighting coefficients of distributed energy resources, For the first In the cluster, the first The energy storage capacity of a distributed energy source, This represents the average energy storage capacity of all distributed energy sources in a grid-type energy storage system. The total number of distributed energy sources in a grid-type energy storage system; Learnable nonlinear potential function The formula is as follows: In the formula, It is a feedforward neural network. This is the input characteristic of the energy storage potential function.

6. The power allocation method for a grid-type energy storage system based on the energy storage potential function as described in claim 1, characterized in that: The dynamic update law for obtaining the corresponding distributed energy source based on the energy storage potential function is then... In the cluster, the first The dynamic update law for distributed energy resources is given by the following formula: In the formula, For the first In the cluster, the first A control input signal for a distributed energy source For the first In the cluster, the first The gain coefficient of a distributed energy source. For the energy storage potential function of a grid-type energy storage system Seeking information about The partial derivatives, Input characteristics of the energy storage potential function The first in One variable, , .

7. The power allocation method for a grid-type energy storage system based on the energy storage potential function as described in claim 6, characterized in that: The power allocation method for a grid-type energy storage system based on the energy storage potential function further includes obtaining the communication weights among distributed energy sources to construct the positive definite gain matrix of the grid-type energy storage system, then: The communication weight between distributed energy sources is calculated using the following formula: In the formula, Laplace matrix representing time-varying communication topology The Line 1 The communication weight of column elements, and , , This refers to the total number of distributed energy sources in a grid-type energy storage system. The Laplace matrix of the time-varying communication topology The Line 1 Communication weights of column elements The Laplace matrix of the time-varying communication topology The first line Frequency deviation of each node The Laplace matrix of the time-varying communication topology The first in the list Frequency deviation of each node The Laplace matrix of the time-varying communication topology The first line Energy storage capacity of each node, The Laplace matrix of the time-varying communication topology The first line Energy storage capacity of each node, The weighting coefficient for frequency deviation, The time-varying communication topology diagram is a weighting coefficient for the energy storage capacity deviation. , For a set of nodes, Let be a set of edges, where nodes are distributed energy sources and edges are lines; The gain coefficient of each distributed energy source is obtained to form the positive definite gain matrix of the grid-type energy storage system. The gain coefficient of each distributed energy source is the sum of all associated communication weights of the corresponding distributed energy source.

8. The power allocation method for a grid-type energy storage system based on the energy storage potential function as described in claim 1, characterized in that: Each cluster has a proxy distributed energy source, and each distributed energy source obtains control input signals from the proxy distributed energy source within the corresponding cluster or from adjacent distributed energy sources through one-way communication.

9. The power allocation method for a grid-type energy storage system based on the energy storage potential function as described in claim 8, characterized in that: Each of the aforementioned clusters' agent distributed energy sources serves as the root node of the time-varying communication topology of the grid-type energy storage system. When the grid-type energy storage system experiences external disturbances or sudden load changes, the agent distributed energy source first updates the dynamic update law and then transmits the control input signal adjusted by the dynamic update law to other distributed energy sources in its respective cluster through the time-varying communication topology.

10. A power distribution system for a grid-type energy storage system based on an energy storage potential function, comprising a memory and a processor, characterized in that: The memory is used to store a computer program, which, when executed by a processor, implements the power allocation method for a grid-type energy storage system based on the energy storage potential function as described in any one of claims 1-9.

Images