Vehicle-network integration multi-mode intelligent control method and system for fast frequency response and medium

Abstract

This invention relates to the field of power grid interaction technology, and particularly to a multi-mode intelligent control method, system, and medium for vehicle-grid integration oriented towards rapid frequency response. The method includes: collecting state information of an electric vehicle (EV) cluster and establishing a hierarchical collaborative control relationship between the power grid dispatching terminal, the aggregation control terminal, and the charging terminal; constructing a heterogeneous EV aggregation model based on the state information and determining the real-time adjustable power range of the EV cluster; establishing a multi-state switching model based on the heterogeneous EV aggregation model; matching frequency adjustment commands and determining the corresponding state switching modes based on the multi-state switching model and the real-time adjustable power range; performing rolling prediction optimization based on the state switching modes; executing the state switching commands and updating the heterogeneous EV aggregation model and the multi-state switching model based on the execution results. This invention effectively solves the problem in existing technologies where clustered EVs cannot meet the rapid frequency response requirements of the power grid.

Description

Technical Field

[0001] This invention relates to the field of power grid interaction technology, and in particular to a multi-mode intelligent control method, system and medium for vehicle-grid integration with fast frequency response. Background Technology

[0002] With the continuous expansion of new energy power generation grid connection, the frequency stability of the power system faces greater uncertainty and volatility. Rapid and flexible frequency regulation resource allocation has become a crucial technical requirement for ensuring the safe and stable operation of the power grid. Electric vehicles, as mobile load resources with energy storage attributes and bidirectional power regulation capabilities, can participate in grid frequency regulation through vehicle-to-grid interaction. This means that, under the premise of meeting user vehicle needs and battery operating constraints, they can dynamically adjust charging power, discharging power, or operating status according to grid dispatch instructions, thereby providing frequency support services to the grid. In existing technologies, aggregating and controlling a large number of electric vehicles typically integrates dispersed, random electric vehicles with limited individual power into a flexible resource that can be uniformly dispatched. This is combined with charging and discharging control, state switching control, or predictive optimization control to enable electric vehicle clusters to participate in frequency regulation scenarios such as automatic power generation control.

[0003] However, existing technologies still have significant shortcomings in practical applications: On the one hand, due to the significant heterogeneity of electric vehicles in terms of battery capacity, state of charge, access behavior, and user habits, existing aggregation modeling methods often struggle to simultaneously achieve modeling accuracy and solution efficiency, resulting in insufficient representation of the cluster's true adjustment capabilities. On the other hand, existing state switching and control methods do not fully explore the adjustability of electric vehicle clusters, making it difficult to achieve efficient response and accurate tracking of frequency regulation commands in scenarios with rapid frequency fluctuations. Consequently, the overall electric vehicle cluster cannot fully meet the application requirements of rapid frequency response in the power grid.

[0004] The information disclosed in this background section is intended only to enhance the understanding of the general background of this disclosure and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0005] This invention provides a vehicle-to-grid (V2G) integrated multi-mode intelligent control method, system, and medium for fast frequency response, which can effectively solve the problems in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A vehicle-to-grid (V2G) integrated multi-mode intelligent control method for fast frequency response, the method comprising: Collect status information of electric vehicle clusters and establish a hierarchical collaborative control relationship between the power grid dispatching terminal, the aggregation control terminal and the charging terminal in order to receive frequency adjustment commands; Based on the state information, a heterogeneous electric vehicle aggregation model is constructed, and the real-time adjustable power range of the electric vehicle cluster is determined. A multi-state switching model is established based on the heterogeneous electric vehicle aggregation model to describe the state switching process and aggregation power response of the electric vehicle cluster. Based on the multi-state switching model and the real-time adjustable power range, the frequency adjustment command is matched and the corresponding state switching mode is determined. Rolling prediction optimization is performed based on the state switching mode to generate the state switching instructions for the electric vehicle cluster. The state switching command is executed, and the heterogeneous electric vehicle aggregation model and the multi-state switching model are updated based on the execution result to achieve a rapid response to the frequency adjustment command.

[0007] Furthermore, a heterogeneous electric vehicle aggregation model is constructed, including: The electric vehicle cluster is divided into multiple large intervals according to the upper and lower limits of the state of charge, and the state space dimension of the heterogeneous electric vehicle aggregation model is determined by the number of the large intervals. Each large interval is divided into multiple smaller intervals, and the modeling accuracy of the heterogeneous electric vehicle aggregation model is determined by the number of the smaller intervals. A double-layer nested discrete structure is constructed based on the large interval and the small interval to decouple the state space dimension and modeling accuracy of the heterogeneous electric vehicle aggregation model.

[0008] Furthermore, constructing a heterogeneous electric vehicle aggregation model also includes: The state transition process of the electric vehicle cluster during charging is described based on Markov chain theory. Based on the probability distribution of battery capacity of the electric vehicle cluster, the state transition probability between each of the smaller intervals is determined, and the expected one-step transition probability between each of the larger intervals is obtained. Based on the expected transition probability in the first step, construct the relationship between the change in the number of electric vehicles and the aggregated power output relationship within each of the large intervals; An idle process aggregation model and a discharge process aggregation model are established according to the modeling method corresponding to the charging process, and the charging process aggregation model, the idle process aggregation model and the discharge process aggregation model are uniformly transformed into a linear state-space expression.

[0009] Furthermore, a multi-state switching model is established, including: Define the charging state, discharging state, and idle state of the electric vehicle cluster; A bidirectional switching path is established between the charging state and the idle state, between the charging state and the discharging state, and between the idle state and the discharging state, so that the electric vehicle cluster can directly or indirectly switch between the charging state, the discharging state, and the idle state; A state vector is constructed based on the number of vehicles corresponding to the charging state, the discharging state, and the idle state, and a control vector is constructed based on the number of vehicles switching between each state. A multi-state switching model is established based on the principle of vehicle quantity conservation to describe the state changes and aggregate power output of the electric vehicle cluster.

[0010] Further, matching the frequency adjustment command and determining the corresponding state switching mode includes: Based on the multi-state switching model, the real-time adjustable power range is divided into power that can be reduced and power that can be increased. The power reduction capability is divided into three segments, corresponding to charging to idle, charging to idle and idle to discharging combination, and charging to discharging and idle to discharging combination, respectively. The power increase is divided into three segments, corresponding to discharge to idle, discharge to idle and idle to charge combination, and discharge to charge and idle to charge combination, respectively. The power boundaries for each segment are determined according to the rule that idle state is given priority for transition and a single electric vehicle is only allowed one state switch within the same control cycle.

[0011] Furthermore, the rolling prediction optimization based on the state switching mode includes: Collect current vehicle status data and select the corresponding control mode according to the power segment to which the frequency adjustment command belongs; Based on the multi-state switching model, a prediction model corresponding to the control mode is constructed; Rolling prediction optimization is performed using the tracking error of the frequency adjustment command, the number of state switching times, and the proportion of idle states as optimization targets. The state switching instruction sequence of the electric vehicle cluster is generated under the conditions of satisfying vehicle quantity constraints, power boundary constraints, and state switching constraints. Execute the current control instruction in the state switching instruction sequence, and update the prediction model for the next control cycle based on the execution feedback.

[0012] Furthermore, determining the expected transition probability in one step includes: The critical battery capacity that enables an electric vehicle to complete the interval transfer within a discrete time interval is determined based on the maximum charging power, the discrete time interval, and the interval boundary. The state transition probability between the intervals is determined based on the critical battery capacity and the battery capacity probability distribution. The average state transition probabilities corresponding to each of the smaller intervals within the same larger interval are taken to obtain the expected one-step transition probability between the larger intervals.

[0013] Furthermore, establishing a multi-state switching model also includes: The state transition matrix of the multi-state switching model is constructed from the state transition matrix of the charging process, the state transition matrix of the idle process, and the state transition matrix of the discharging process. A control matrix is ​​constructed based on the switching relationships between various operating states to describe the impact of state switching on the number of vehicles; An output matrix is ​​constructed from the aggregated power mapping relationship corresponding to each operating state to output the total aggregated power of the electric vehicle cluster.

[0014] A vehicle-to-grid (V2G) integrated multi-mode intelligent control system for rapid and high-frequency response, the system comprising: The collaborative acquisition module collects the status information of the electric vehicle cluster and establishes a hierarchical collaborative control relationship between the power grid dispatching terminal, the aggregation control terminal, and the charging terminal in order to receive frequency adjustment commands. The aggregation modeling module constructs a heterogeneous electric vehicle aggregation model based on state information and determines the real-time adjustable power range of the electric vehicle cluster. The state switching module establishes a multi-state switching model based on the heterogeneous electric vehicle aggregation model to describe the state switching process and aggregation power response of the electric vehicle cluster. The command matching module, based on a multi-state switching model and a real-time adjustable power range, matches frequency adjustment commands and determines the corresponding state switching mode. The prediction optimization module performs rolling prediction optimization based on the state switching mode to generate state switching instructions for the electric vehicle cluster. The execution update module executes state switching instructions and updates the heterogeneous electric vehicle aggregation model and multi-state switching model based on the execution results to achieve a rapid response to frequency adjustment instructions.

[0015] A computer-readable storage medium stores a computer program, the computer program including program instructions, which, when executed by a processor, can implement the aforementioned vehicle-to-grid fusion multi-mode intelligent control method for fast frequency response.

[0016] The technical solution of this invention can achieve the following technical effects: By constructing a hierarchical collaborative control architecture, and combining heterogeneous electric vehicle aggregation modeling, multi-state switching adjustment, adjustable power matching, and rolling predictive optimization control, the efficient and precise regulation of electric vehicle clusters to the grid's fast frequency response commands is achieved, effectively solving the problem that existing technologies make it difficult for clustered electric vehicles to meet the grid's fast frequency response requirements.

[0017] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart illustrating a multi-mode intelligent control method for vehicle-to-grid convergence oriented towards rapid frequency response; Figure 2 A flowchart illustrating the process of constructing a heterogeneous electric vehicle aggregation model; Figure 3 A flowchart illustrating the process of establishing a multi-state switching model. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0021] 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 to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0022] Example 1; like Figure 1 As shown, this application provides a vehicle-to-grid (V2G) integrated multi-mode intelligent control method for fast frequency response, the method including: S10: Collects status information of electric vehicle clusters and establishes hierarchical collaborative control relationships between the power grid dispatching terminal, aggregation control terminal and charging terminal to receive frequency adjustment commands; S20: Construct a heterogeneous electric vehicle aggregation model based on state information and determine the real-time adjustable power range of the electric vehicle cluster; S30: Establish a multi-state switching model based on the heterogeneous electric vehicle aggregation model to describe the state switching process and aggregation power response of the electric vehicle cluster; S40: Based on the multi-state switching model and real-time adjustable power range, the frequency adjustment command is matched and the corresponding state switching mode is determined. S50: Perform rolling prediction optimization based on state switching mode to generate state switching instructions for electric vehicle clusters; S60: Executes state switching instructions and updates the heterogeneous electric vehicle aggregation model and multi-state switching model based on the execution results to achieve a fast response to frequency adjustment instructions.

[0023] Specifically, in this embodiment, the vehicle-grid converged multi-mode intelligent control method for rapid frequency response is applied to the frequency regulation scenario of new energy power systems, and is completed collaboratively by the grid dispatching terminal, the aggregation control terminal, and the charging terminal. First, the status information of the electric vehicle cluster connected to the system is collected. The status information includes at least battery capacity, state of charge, and operating status. Based on this, a hierarchical collaborative control relationship is established between the grid dispatching terminal, the aggregation control terminal, and the charging terminal. The grid dispatching terminal is used to issue frequency regulation commands, the aggregation control terminal is used to receive vehicle status data uploaded by the charging terminal and execute aggregation modeling, state switching, power segmentation, and frequency regulation control algorithms, and the charging terminal is used to collect the status of connected vehicles and execute state switching commands, thereby forming a collaborative control system in which dispatching, control, and execution cooperate with each other. After obtaining the state information, a heterogeneous electric vehicle aggregation model is constructed based on the state information to characterize the state distribution evolution of the electric vehicle cluster during charging, discharging, and idle operation, and to determine the real-time adjustable power range of the electric vehicle cluster. Specifically, the heterogeneous electric vehicle aggregation model is established based on Markov chain theory and uses a double-layer nested discretization method of SOC to process the state of charge of electric vehicles. The first layer of discretization divides the SOC into multiple large intervals to determine the state space dimension of the model. The second layer of discretization further subdivides the large intervals into multiple small intervals to improve the modeling accuracy, thereby achieving independent design of model accuracy and state space dimension. Furthermore, the transition probability of the small intervals is determined based on the probability distribution of battery capacity, and the one-step expected transition probability between the large intervals is obtained from the small interval transition probability. Then, based on the one-step expected transition probability, the relationship between the number of vehicles and the aggregated power output relationship in each large interval are constructed, and aggregation models are established for charging, discharging, and idle states respectively. Finally, they are uniformly transformed into a linear state space expression to obtain the aggregation state and real-time adjustable power range of the electric vehicle cluster. After establishing the heterogeneous electric vehicle aggregation model, a multi-state switching model is established based on the heterogeneous electric vehicle aggregation model to describe the state switching process and aggregated power response of the electric vehicle cluster. Specifically, the electric vehicle cluster is defined to have three operating states: charging state, discharging state, and idle state. Bidirectional switching paths are established around the three operating states between charging and idle, charging and discharging, and idle and discharging, so that electric vehicles can switch directly between the three states or switch via transition states. On this basis, a state vector is constructed based on the number of vehicles in each state, and a control vector is constructed based on the number of vehicles switching between different states. Under the condition of satisfying the principle of vehicle number conservation, the state space equation of the multi-state switching model is established to output the total aggregated power response of the electric vehicle cluster during the state switching process. Furthermore, based on the multi-state switching model and the real-time adjustable power range, the frequency adjustment command is matched and the corresponding state switching mode is determined. In specific implementation, the real-time adjustable power range is divided into two categories: power that can be reduced and power that can be increased. The power that can be reduced is used for power adjustment when the grid frequency is too high, and the power that can be increased is used for power adjustment when the grid frequency is too low. At the same time, the power that can be reduced and the power that can be increased are processed in segments, with each segment corresponding to a different state switching combination. The switching priority principle of prioritizing the transition from idle state to direct switching between charging and discharging is followed. Moreover, the same electric vehicle is only allowed to perform one state switching within one control cycle to reduce battery damage and avoid mutual cancellation of charging and discharging power. Subsequently, the corresponding state switching mode is determined according to the power range to which the frequency adjustment command belongs. After determining the state switching mode, rolling prediction optimization is performed based on the state switching mode to generate state switching commands for the electric vehicle cluster. Specifically, the rolling prediction optimization adopts a multi-mode frequency modulation control strategy based on model predictive control. It uses a multi-state switching model as the prediction basis, establishes corresponding simplified prediction models for different power segments, and constructs a closed-loop control framework that includes "collecting state data, building prediction models, determining the power segment to which the frequency adjustment command belongs, selecting the corresponding control mode, rolling optimization to solve for the optimal control sequence, executing the current control command, and updating the vehicle state". The rolling optimization objective comprehensively considers the frequency adjustment command tracking error, the number of state switching times, and the proportion of idle states, while simultaneously setting constraints on the number of vehicles, power boundary constraints, and state switching constraints to ensure that the generated state switching commands are feasible and effective. Finally, the state switching command is executed, and the heterogeneous electric vehicle aggregation model and multi-state switching model are updated based on the execution results to achieve a rapid response to frequency adjustment commands. In specific implementation, the aggregation control terminal sends the current control command in the optimal state switching command sequence obtained by rolling prediction optimization to the charging terminal. The charging terminal controls the connected vehicles to switch between charging, discharging, and idle states according to the received state switching command. After execution, the vehicle state data is re-collected and fed back to the aggregation control terminal, which updates the heterogeneous electric vehicle aggregation model and multi-state switching model and enters the next control cycle to continue executing frequency adjustment command matching, mode selection, and rolling prediction optimization, thereby achieving a continuous, rapid, and accurate response to frequency adjustment commands. The technical solution of this invention constructs a hierarchical collaborative control architecture, which combines heterogeneous electric vehicle aggregation modeling, multi-state switching adjustment, adjustable power matching, and rolling predictive optimization control to achieve efficient and precise regulation of electric vehicle clusters to the grid's fast frequency response commands. This effectively solves the problem that existing technologies make it difficult for clustered electric vehicles to meet the grid's fast frequency response requirements.

[0024] Furthermore, such as Figure 2 As shown, a heterogeneous electric vehicle aggregation model is constructed, including: The electric vehicle cluster is divided into multiple large intervals according to the upper and lower limits of the state of charge, and the state space dimension of the heterogeneous electric vehicle aggregation model is determined by the number of large intervals. Each large interval is divided into multiple smaller intervals, and the modeling accuracy of the heterogeneous electric vehicle aggregation model is determined by the number of smaller intervals. A double-layer nested discrete structure is constructed based on large intervals and small intervals to decouple the state space dimension and modeling accuracy of the heterogeneous electric vehicle aggregation model; Furthermore, constructing a heterogeneous electric vehicle aggregation model also includes: The state transition process of an electric vehicle cluster during charging is described based on Markov chain theory. Based on the probability distribution of battery capacity of electric vehicle clusters, the state transition probability between each small interval is determined, and the one-step expected transition probability between the major intervals is obtained. Based on the one-step expected transition probability, the relationship between the change in the number of electric vehicles and the aggregate power output relationship in each major interval are constructed. Establish an aggregated model for the idle process and an aggregated model for the discharge process according to the modeling method corresponding to the charging process, and unify the aggregated models for the charging process, the idle process, and the discharge process into a linear state-space expression; Furthermore, determining the expected transition probability in one step includes: The critical battery capacity that enables an electric vehicle to complete the interval transfer within a discrete time interval is determined based on the maximum charging power, discrete time interval, and interval boundary. Determine the state transition probability between intervals based on the critical battery capacity and the battery capacity probability distribution; The average state transition probabilities of each subinterval within the same large interval are taken to obtain the expected one-step transition probability between large intervals.

[0025] As a preferred embodiment of the above, a cluster electric vehicle charging and discharging load aggregation modeling method considering battery capacity differences is established: Individual electric vehicles have low power and uncertain user behavior, making direct participation in frequency regulation difficult. Therefore, aggregation modeling is needed to transform a large number of heterogeneous electric vehicles into a controllable aggregate. This invention, based on Markov chain theory, designs a double-layer nested discretization method for SOC (System-on-Chips) to construct a clustered electric vehicle charging and discharging load aggregation model that considers differences in battery capacity, achieving independent design of model accuracy and dimensionality. 2.1 Markov property analysis of the charging process: The charging process of an electric vehicle exhibits a dynamic change in State of Charge (SOC) from low to high. Its discrete-time SOC recursive relationship satisfies the Markov property, meaning that the SOC at the next moment depends only on the SOC at the current moment and is independent of historical states. The SOC recursive formula is as follows: ; In the formula, for The state of charge of the battery at any given time; for The state of charge of the battery at any given time; for The charging power at any given moment; For charging efficiency; For discrete time intervals; This refers to the actual battery capacity, expressed in kWh. Based on the Markov property, the electric vehicle charging process can be described as a state transition process, that is, the transition of the electric vehicle in different SOC intervals is only related to the current interval, which provides a theoretical basis for aggregation modeling. 2.2 SOC double-layer nested discretization processing: Traditional single-layer discretization methods for SOC suffer from a "precision versus dimensionality contradiction": if the number of discrete intervals is small, the modeling accuracy is insufficient; if the number of discrete intervals is large, it will lead to an explosion in the dimensionality of the state space, increasing computational complexity. This invention designs a double-layer nested discretization method, which divides the SOC intervals into two layers: a large interval and a small interval, to achieve dimensionality control and accuracy improvement respectively. 2.2.1 Discretization Design Principles: First-level discretization (large-range discretization): Determine the charging range based on the upper and lower limits of the battery's State of Charge (SOC), and discretize the SOC into... There are several large state intervals (referred to as large intervals), and the number of large intervals is... Determine the dimension of the state space of the aggregation model; Second-level discretization (interval discretization): Subdivide each large interval into smaller intervals. There are several small state intervals (or simply small intervals), and the number of small intervals is... Determines the modeling accuracy of the aggregation model; 2.2.2 Definition of Discretization Parameters: The core parameters of the two-level discretization are defined as shown in Table 1: Table 1. Definition of SOC double-layer nested discretization parameters 2.2.3 Discretization Boundary Calculation: The lower boundary of a large interval is divided into equal intervals, and the calculation formula is as follows: ; In the formula, For the first The lower boundary of each large interval; This is the lower limit of SOC; This represents the upper limit of SOC. For large interval numbering; This represents the total number of items in the large interval. The lower boundaries of the smaller intervals within each larger interval are also divided at equal intervals, calculated using the following formula: ; In the formula, For the first Within each large interval The lower boundary between individual communities; For the first The lower boundary of each large interval; Numbering the intervals; The total number of sub-intervals within each large interval; 2.3 Calculation of the expected transition probability in one step of the charging process: The transition probability is the core of Markov chain aggregation modeling, used to describe the probability of an electric vehicle transitioning from one SOC interval to another; this invention derives the expected transition probability of one step in the charging process based on the probability distribution of battery capacity. 2.3.1 Basic Assumptions: To simplify calculations without affecting modeling accuracy, the following assumptions are proposed: All electric vehicles participating in the aggregation will use maximum charging power. Charge; Ignoring the differences in charging and discharging efficiency among different electric vehicles, a uniform charging efficiency is used. It is a fixed value; Battery capacity of electric vehicles Follows a known probability density distribution (The probability density distribution function of electric vehicle battery capacity), its range is as follows: , The minimum battery capacity (in kWh) of the electric vehicles participating in the aggregation. The maximum battery capacity refers to the maximum battery capacity of the electric vehicles participating in the aggregation (unit: kWh). 2.3.2 Definition of Critical Capacity Define critical capacity Its physical meaning is: when the charging power, charging efficiency, and time interval are determined, the electric vehicle can achieve charging within a discrete time interval. The maximum permissible battery capacity when transitioning from the current SOC range to the next SOC range; the calculation formula is: ; In the formula, This is the critical capacity; This is the maximum charging power; For the first The lower boundary of each large interval; For the first Within each large interval The lower boundary between individual communities; 2.3.3 Derivation of Interval Transition Probabilities: Electric vehicles from the first Within each large interval The interval is transferred to the first... The probability of a large interval It depends on the battery capacity. With critical capacity The relationship, specifically expressed as: ; In the formula: The zero-probability critical SOC is the minimum lower boundary value of the subinterval when the transition probability is 0. The critical SOC with a probability of 1 is the minimum lower boundary of the subinterval when the transition probability is 1. and The calculation formula is: ; ; In the formula, Minimum battery capacity; Maximum battery capacity; 2.3.4 Calculation of expected transition probability in one step: Since each large interval is divided into equally spaced smaller intervals, the expected one-step transition probability between large intervals can be obtained by averaging the transition probabilities of the smaller intervals. The calculation formula is as follows: ; In the formula, For the first The major intervals to the first The expected transition probability of a single step across a large interval; for The corresponding sub-interval number; for The corresponding sub-interval number; The total number of sub-intervals within each large interval.

[0026] Furthermore, such as Figure 3 As shown, a multi-state switching model is established, including: Define the charging state, discharging state, and idle state of the electric vehicle cluster; Establish bidirectional switching paths between charging and idle states, between charging and discharging states, and between idle and discharging states, so that electric vehicle clusters can directly or indirectly switch between charging, discharging, and idle states. A state vector is constructed based on the number of vehicles corresponding to charging, discharging, and idle states, and a control vector is constructed based on the number of vehicles switching between states. A multi-state switching model is established based on the principle of vehicle quantity conservation to describe the state changes and aggregate power output of electric vehicle clusters. Furthermore, establishing a multi-state switching model also includes: The state transition matrix of the multi-state switching model is constructed from the state transition matrix of the charging process, the state transition matrix of the idle process, and the state transition matrix of the discharging process. A control matrix is ​​constructed based on the switching relationships between various operating states to describe the impact of state switching on the number of vehicles; An output matrix is ​​constructed from the aggregated power mapping relationship corresponding to each operating state to output the total aggregated power of the electric vehicle cluster.

[0027] As a preferred embodiment of the above embodiments, in one embodiment: 2.4 Construction of the charging load aggregation model: Based on the one-step expected transfer probability, a dynamic transfer model of charging load is constructed to describe the change in the number of electric vehicles in different SOC ranges. 2.4.1 Load Dynamic Transfer Rules In a discrete time interval Inside, in the first There are only two possible transfer scenarios for electric vehicles within each large area: Maintain the original state: the transition probability is ; Transfer to the first Each large interval has a transition probability of [value]. ; In particular, in the first Once the electric vehicle has finished charging in each major area, it will exit the charging state. In addition, load variations caused by external factors (such as temporary access to or disconnection of electric vehicles from charging) need to be considered. ; 2.4.2 Equation for load quantity change: No. Within each large range Number of electric vehicles at any given time The calculation formula is: ; In the formula, for Time of the first The number of electric vehicles charging within each large region; for Time of the first The number of electric vehicles charging within each large region; for Time of the first The number of electric vehicles charging within each large region; For the first The major intervals to the first The expected transition probability of a single step across a large interval; for The first time caused by external factors Load variation within a large interval; For the first large interval, since there is no preceding interval, its load quantity change equation is: ; 2.4.3 Aggregate power output equation: The number of electric vehicles in each SOC range is converted into aggregated charging power, and the output equation is: ; In the formula, for Aggregated charging power of clustered electric vehicles at any given time; The maximum charging power for a single electric vehicle; 2.4.4 Linear state-space representation: To facilitate subsequent control strategy design, the charging load aggregation model is transformed into a linear state-space equation: ; ; In the formula, for 3D charging state vector; for A state transition matrix, the elements of which are determined by the expected transition probabilities of one step; for External perturbation vector; for 1-dimensional unit row vector; 2.5 Aggregation Model of Idle and Discharge Processes: The operating states of electric vehicles include charging, discharging, and idle. In the idle state, there is no charging or discharging power. The discharging process is reversible with the charging process. The method of constructing the aggregation model is the same as that of the charging process, only the parameter signs and physical meanings need to be adjusted. 2.5.1 Idle Process Aggregation Model: During idle periods, the SOC of the electric vehicle remains unchanged, and its state-space equation is: ; ; In the formula, for An idle state vector; for 3D identity matrix; for External perturbation vector; The polymerization power during the idle process is always 0; 2.5.2 Aggregation Model of Discharge Process: During discharge, the State of Charge (SOC) changes from high to low, and its recursive formula is as follows: ; In the formula, This refers to the discharge power. For discrete time intervals; For discharge efficiency; Battery capacity; The methods for constructing the one-step expected transfer probability, load quantity change equation, and state-space equation during the discharge process are consistent with those during the charging process; only the charging power needs to be considered. Replace with discharge power Charging efficiency Replace with discharge efficiency The state transition direction is reversed, and its aggregate power output equation is: ; In the formula, This represents the aggregated power during the discharge process; the negative sign indicates that the power flows to the power grid. This represents the maximum discharge power of a single electric vehicle. for Time of the first The number of electric vehicles discharging within each large interval; 3. Construction of the Multi-State Switching Model (MSM): Traditional state-switching models require electric vehicles to transition through an idle state when switching between charging and discharging states, which limits the adjustable power range. This invention constructs a Multi-State Switching Model (MSM), which allows direct switching between charging, discharging, and idle states, thus broadening the real-time adjustable power range of clustered electric vehicles. 3.1 State switching mechanism design: Three operating states of an electric vehicle are defined: charging state (C), discharging state (D), and idle state (I). Six state transition paths are designed to cover all direct and indirect transitions between states. The transition paths are as follows: Table 2 Correspondence between State Transition Paths and Vehicle Quantity Changes State transitions must satisfy the principle of vehicle quantity conservation, that is, state transitions only change the operating state of electric vehicles and do not change the total number of electric vehicles participating in aggregation. 3.2 State-space representation of the multi-state switching model: An aggregated model integrating charging, idle, and discharging states is constructed to build a state-space equation for a multi-state switching model, describing the changes in the number of electric vehicles and aggregated power output under the three states. 3.2.1 Definition of State Vector: The state vector of an MSM is defined as a combination of three state vectors: ; In the formula, for 3D state vector (N is the number of SOC large intervals, 3 corresponds to three state switching paths); for 3D charging state vector; for An idle state vector; for 3D discharge state vector; 3.2.2 Definition of Control Vector: Define the control vector of MSM as the number of vehicles undergoing state transitions: ; In the formula, for 3D control vector; for Vector of the number of vehicles switching between charging and idle periods; for Dimensional vector of the number of vehicles switching between charging and discharging; for Vector of the number of vehicles in the idle-discharge switching state; 3.2.3 Construction of state-space equations: The state-space equation of the MSM is: ; ; In the formula, for The state transition matrix is ​​composed of the state transition matrices for the charging, idle, and discharging processes, and its expression is: , for Zero-dimensional matrix; for A 3D control matrix is ​​used to describe the impact of state transitions on the number of vehicles, and its expression is: , for 3D identity matrix; for An external disturbance vector, consisting of external disturbance vectors for charging, idling, and discharging. It is assembled vertically. for The output matrix is ​​expressed as follows: , Let N be a unit row vector, and With consistent definition, all elements are 1; for The total aggregate power of the electric vehicles in the time cluster is equal to the algebraic sum of the charging power and the discharging power, where the charging power is positive and the discharging power is negative.

[0028] Furthermore, matching the frequency adjustment command and determining the corresponding state switching mode includes: Based on the multi-state switching model, the real-time adjustable power range is divided into power that can be reduced and power that can be increased. The power reduction capability is divided into three segments, corresponding to charging to idle, charging to idle and idle to discharging combination, and charging to discharging and idle to discharging combination, respectively. The power increase can be divided into three segments, corresponding to discharge to idle, discharge to idle and idle to charge combination, and discharge to charge and idle to charge combination respectively. The power boundaries for each segment are determined according to the rule that the idle state takes priority in transition and a single electric vehicle is only allowed to switch states once within the same control cycle.

[0029] As a preferred embodiment of the above, in order to reduce the number of direct switching between charging and discharging states and avoid mutual cancellation of charging and discharging power, the present invention designs an improved real-time adjustable power segmentation method based on a multi-state switching model. The real-time adjustable power of the cluster electric vehicles is divided into two categories: power reduction and power increase. Each category is divided into three segments. The idle state is used for transition first, and direct switching between charging and discharging is only enabled when necessary. 4.1 Power Segmentation Design Principles: Priority principle: The state transition priority is "charging ↔ idle" and "idle ↔ discharging" > "charging ↔ discharging". The idle state is used first for transition to reduce battery damage. Power range principle: Each power segment corresponds to a specific combination of state switching, and the upper limit of the power of the segment is determined by the number of vehicles in the corresponding state and the maximum charging and discharging power. Constraint principle: Within a control cycle (referring to the time interval between the secondary control center of the cluster electric vehicle executing one frequency modulation control command), the time interval is equal to the discrete time interval defined above. (Consistent, positive real number) Only one state switch is allowed for the same electric vehicle to avoid frequent switching affecting battery life; 4.2 Segmented design with scalable power: Reduceable power refers to the power absorbed from the grid or the power injected into the grid by the cluster of electric vehicles through state switching, in order to mitigate situations where the grid frequency is too high. Reduceable power is divided into three segments, and the state switching combinations and power ranges for each segment are shown in Table 3. Table 3 Correspondence between Power Reduction Segments and State Switching In the formula, for The number of electric vehicles that are constantly charging; for The number of electric vehicles that are always idle; This is the maximum charging power; This represents the maximum discharge power. 4.3 Power segmentation design can be added: Increased power refers to the power absorbed from the grid or reduced injected into the grid by the cluster of electric vehicles through state switching, used to mitigate situations where the grid frequency is too low. Increased power is also divided into three segments, and the state switching combinations and power ranges for each segment are shown in Table 4. Table 4 can be added as a power segment and state switching correspondence table. In the formula, for The number of electric vehicles in a constant state of discharge. The power limit of the first power segment can be increased, which is the maximum power increase value for that segment; 4.4 Real-time adjustable power calculation method: The real-time adjustable power limit of the cluster of electric vehicles is determined by the number of electric vehicles in each current state, and the calculation formula is as follows: ; ; In the formula, For maximum power reduction; To maximize the power that can be increased.

[0030] Furthermore, the optimization of rolling prediction based on state switching patterns includes: Collect current vehicle status data and select the corresponding control mode according to the power segment to which the frequency adjustment command belongs; Constructing a prediction model corresponding to the control mode based on a multi-state switching model; Rolling prediction optimization is performed using the tracking error of frequency adjustment commands, the number of state transitions, and the proportion of idle states as optimization targets. Generate a sequence of state switching instructions for an electric vehicle cluster, satisfying constraints on vehicle quantity, power boundary, and state switching. Execute the current control instruction in the state switching instruction sequence, and update the prediction model for the next control cycle based on the execution feedback.

[0031] As a preferred embodiment of the above embodiments, in a preferred embodiment, the present invention proposes a multi-mode frequency modulation control strategy based on model predictive control, using a real-time adjustable power segmentation method. Based on the power demand of the AGC command, the power segment to which it belongs is determined, the corresponding control mode is selected, and the optimal state switching command is generated through rolling optimization, thereby achieving precise tracking of the AGC command. 5.1 MPC Control Framework Design: The core of model predictive control is a closed-loop cycle of "prediction-optimization-correction", and its control framework is as follows: Predictive Model: Based on a multi-state switching model, predict the aggregated power output of electric vehicles in the cluster over a period of time in the future; Rolling optimization: With the goal of tracking AGC instructions, an optimization objective function is constructed, and the optimal state switching instruction is solved; Feedback correction: The first element of the optimal control sequence is sent out for execution, the vehicle state after execution is collected, the prediction model is updated, and the next control cycle begins; 5.2 Prediction Model Construction: For different power segments, corresponding simplified prediction models are constructed to reduce computational complexity. Taking the first segment with reducible power as an example... Taking (e.g.,) as an example, its prediction model only considers the state transition from charging to idle, and the state-space equation is: ; ; In the formula, To predict the step size, ; For the prediction time domain (the number of control steps the model predicts for the future, which is a positive integer); Based on Predicting based on time data Time-state vector; Based on Predicting based on time data Vector of vehicle quantity for constant charging and idle switching; For a simplified control matrix, the expression is: ; Based on Predicting based on time data time External perturbation vector; The prediction model construction method for other power segments is similar, only requiring adjustment of the control matrix to match the corresponding state switching combinations; 5.3 Design of the rolling optimization objective function: The goal of rolling optimization is to generate optimal state transition commands, ensuring that the aggregated power output of the cluster of electric vehicles tracks AGC commands as closely as possible, while minimizing the number of state transitions and avoiding power cancellation during charging and discharging. The objective function is: ; In the formula: The tracking error weighting coefficient is used to balance the aggregation power and the tracking accuracy of AGC commands; Based on Predicting based on time data Power of AGC commands at any time; This is the state transition weighting coefficient, used to reduce the number of state transitions in electric vehicles; For the control time domain (the length of the optimal control sequence obtained by rolling optimization, which is a positive integer and satisfies m≤p); This is the idle state weighting coefficient, used to increase the number of electric vehicles in the idle state and avoid the cancellation of charging and discharging power; for 1-dimensional unit row vector; Based on Predicting based on time data Idle state vector at any given time; 5.4 Setting Constraints: To ensure the feasibility of control commands, the following constraints are set: Vehicle quantity constraint: The number of vehicles transitioning between states must not exceed the number of vehicles in the current state, i.e. ; Power range constraint: The aggregated power output must not exceed the real-time adjustable power range, i.e. ; State transition count constraint: Only one state transition is allowed for the same electric vehicle within one control cycle, i.e. The element can take the value 0 or 1 (1 indicates toggling, 0 indicates no toggling); 5.5 Execution logic of multi-mode control strategy: Data Acquisition: The secondary control center of the cluster electric vehicle collects vehicle status data uploaded by the charging piles, including the number of vehicles in each status, SOC distribution, battery capacity, etc. Real-time adjustable power calculation: Based on the current vehicle status, calculate three segments of power that can be reduced and power that can be increased; AGC command matching: Receive AGC commands issued by the power grid dispatch layer, determine the power segment to which the command belongs, and define the power segment as follows: see 4.2 (power that can be reduced) and 4.3 (power that can be increased). Control mode selection: Select the corresponding control mode according to the power segment and load the corresponding prediction model; Rolling optimization solution: Construct the optimization objective function and solve for the optimal state switching instruction sequence (a 3N-dimensional control vector sequence of length m in the control time domain; the definition of the control vector is given in 3.2.2). Command issuance and execution: The first element of the optimal command sequence is issued to the charging pile to control the vehicle state switching. The state switching command corresponds to the 6 state switching paths in Table 2 and is issued from the cluster electric vehicle secondary control center to the charging pile terminal. State update and cycle: Collect the vehicle state after execution, update the prediction model, and enter the next control cycle.

[0032] Example 2; Based on the same inventive concept as the vehicle-to-grid fusion multi-mode intelligent control method for fast frequency response in the foregoing embodiments, the present invention also provides a vehicle-to-grid fusion multi-mode intelligent control system for fast frequency response, the system comprising: The collaborative acquisition module collects the status information of the electric vehicle cluster and establishes a hierarchical collaborative control relationship between the power grid dispatching terminal, the aggregation control terminal, and the charging terminal in order to receive frequency adjustment commands. The aggregation modeling module constructs a heterogeneous electric vehicle aggregation model based on state information and determines the real-time adjustable power range of the electric vehicle cluster. The state switching module establishes a multi-state switching model based on the heterogeneous electric vehicle aggregation model to describe the state switching process and aggregation power response of the electric vehicle cluster. The command matching module, based on a multi-state switching model and a real-time adjustable power range, matches frequency adjustment commands and determines the corresponding state switching mode. The prediction optimization module performs rolling prediction optimization based on the state switching mode to generate state switching instructions for the electric vehicle cluster. The execution update module executes state switching instructions and updates the heterogeneous electric vehicle aggregation model and multi-state switching model based on the execution results to achieve a rapid response to frequency adjustment instructions.

[0033] The system described above in this invention can effectively realize a vehicle-to-grid integrated multi-mode intelligent control method for fast frequency response, and the technical effects it can achieve are as described in the above embodiments, and will not be repeated here.

[0034] Example 3; Based on the same inventive concept as the vehicle-network converged multi-mode intelligent control method for fast frequency response in the foregoing embodiments, the present invention also provides a computer-readable storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor, can realize the vehicle-network converged multi-mode intelligent control method for fast frequency response.

[0035] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A multi-mode intelligent control method for vehicle-to-grid convergence oriented towards fast frequency response, characterized in that, The method includes: Collect status information of electric vehicle clusters and establish a hierarchical collaborative control relationship between the power grid dispatching terminal, the aggregation control terminal and the charging terminal in order to receive frequency adjustment commands; Based on the state information, a heterogeneous electric vehicle aggregation model is constructed, and the real-time adjustable power range of the electric vehicle cluster is determined. A multi-state switching model is established based on the heterogeneous electric vehicle aggregation model to describe the state switching process and aggregation power response of the electric vehicle cluster. Based on the multi-state switching model and the real-time adjustable power range, the frequency adjustment command is matched and the corresponding state switching mode is determined. Rolling prediction optimization is performed based on the state switching mode to generate the state switching instructions for the electric vehicle cluster. The state switching command is executed, and the heterogeneous electric vehicle aggregation model and the multi-state switching model are updated based on the execution result to achieve a rapid response to the frequency adjustment command.

2. The vehicle-to-grid integrated multi-mode intelligent control method for fast frequency response as described in claim 1, characterized in that, Constructing a heterogeneous electric vehicle aggregation model, including: The electric vehicle cluster is divided into multiple large intervals according to the upper and lower limits of the state of charge, and the state space dimension of the heterogeneous electric vehicle aggregation model is determined by the number of the large intervals. Each large interval is divided into multiple smaller intervals, and the modeling accuracy of the heterogeneous electric vehicle aggregation model is determined by the number of the smaller intervals. A double-layer nested discrete structure is constructed based on the large interval and the small interval to decouple the state space dimension and modeling accuracy of the heterogeneous electric vehicle aggregation model.

3. The vehicle-to-grid integrated multi-mode intelligent control method for fast frequency response as described in claim 2, characterized in that, Constructing a heterogeneous electric vehicle aggregation model also includes: The state transition process of the electric vehicle cluster during charging is described based on Markov chain theory. Based on the probability distribution of battery capacity of the electric vehicle cluster, the state transition probability between each of the smaller intervals is determined, and the expected one-step transition probability between each of the larger intervals is obtained. Based on the expected transition probability in the first step, construct the relationship between the change in the number of electric vehicles and the aggregated power output relationship within each of the large intervals; An idle process aggregation model and a discharge process aggregation model are established according to the modeling method corresponding to the charging process, and the charging process aggregation model, the idle process aggregation model and the discharge process aggregation model are uniformly transformed into a linear state-space expression.

4. The vehicle-to-grid integrated multi-mode intelligent control method for fast frequency response as described in claim 1, characterized in that, Establish a multi-state switching model, including: Define the charging state, discharging state, and idle state of the electric vehicle cluster; A bidirectional switching path is established between the charging state and the idle state, between the charging state and the discharging state, and between the idle state and the discharging state, so that the electric vehicle cluster can directly or indirectly switch between the charging state, the discharging state, and the idle state; A state vector is constructed based on the number of vehicles corresponding to the charging state, the discharging state, and the idle state, and a control vector is constructed based on the number of vehicles switching between each state. A multi-state switching model is established based on the principle of vehicle quantity conservation to describe the state changes and aggregate power output of the electric vehicle cluster.

5. The vehicle-to-grid integrated multi-mode intelligent control method for fast frequency response as described in claim 1, characterized in that, Matching the frequency adjustment command and determining the corresponding state switching mode includes: Based on the multi-state switching model, the real-time adjustable power range is divided into power that can be reduced and power that can be increased. The power reduction capability is divided into three segments, corresponding to charging to idle, charging to idle and idle to discharging combination, and charging to discharging and idle to discharging combination, respectively. The power increase is divided into three segments, corresponding to discharge to idle, discharge to idle and idle to charge combination, and discharge to charge and idle to charge combination, respectively. The power boundaries for each segment are determined according to the rule that idle state is given priority for transition and a single electric vehicle is only allowed one state switch within the same control cycle.

6. The vehicle-to-grid integrated multi-mode intelligent control method for fast frequency response as described in claim 1, characterized in that, Optimizing rolling prediction based on the state switching mode includes: Collect current vehicle status data and select the corresponding control mode according to the power segment to which the frequency adjustment command belongs; Based on the multi-state switching model, a prediction model corresponding to the control mode is constructed; Rolling prediction optimization is performed using the tracking error of the frequency adjustment command, the number of state switching times, and the proportion of idle states as optimization targets. The state switching instruction sequence of the electric vehicle cluster is generated under the conditions of satisfying vehicle quantity constraints, power boundary constraints, and state switching constraints. Execute the current control instruction in the state switching instruction sequence, and update the prediction model for the next control cycle based on the execution feedback.

7. The vehicle-to-grid integrated multi-mode intelligent control method for fast frequency response as described in claim 3, characterized in that, Determining the expected transition probability in one step includes: The critical battery capacity that enables an electric vehicle to complete the interval transfer within a discrete time interval is determined based on the maximum charging power, the discrete time interval, and the interval boundary. The state transition probability between the intervals is determined based on the critical battery capacity and the battery capacity probability distribution. The average state transition probabilities corresponding to each of the smaller intervals within the same larger interval are taken to obtain the expected one-step transition probability between the larger intervals.

8. The vehicle-to-grid integrated multi-mode intelligent control method for fast frequency response as described in claim 4, characterized in that, Establishing a multi-state switching model also includes: The state transition matrix of the multi-state switching model is constructed from the state transition matrix of the charging process, the state transition matrix of the idle process, and the state transition matrix of the discharging process. A control matrix is ​​constructed based on the switching relationships between various operating states to describe the impact of state switching on the number of vehicles; An output matrix is ​​constructed from the aggregated power mapping relationship corresponding to each operating state to output the total aggregated power of the electric vehicle cluster.

9. A vehicle-to-grid integrated multi-mode intelligent control system for high-frequency response, characterized in that: The system includes: The collaborative acquisition module collects the status information of the electric vehicle cluster and establishes a hierarchical collaborative control relationship between the power grid dispatching terminal, the aggregation control terminal, and the charging terminal in order to receive frequency adjustment commands. The aggregation modeling module constructs a heterogeneous electric vehicle aggregation model based on state information and determines the real-time adjustable power range of the electric vehicle cluster. The state switching module establishes a multi-state switching model based on the heterogeneous electric vehicle aggregation model to describe the state switching process and aggregation power response of the electric vehicle cluster. The command matching module, based on a multi-state switching model and a real-time adjustable power range, matches frequency adjustment commands and determines the corresponding state switching mode. The prediction optimization module performs rolling prediction optimization based on the state switching mode to generate state switching instructions for the electric vehicle cluster. The execution update module executes state switching instructions and updates the heterogeneous electric vehicle aggregation model and multi-state switching model based on the execution results to achieve a rapid response to frequency adjustment instructions.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which includes program instructions that, when executed by a processor, can implement the vehicle-to-grid fusion multi-mode intelligent control method for fast frequency response as described in any one of claims 1-8.

Images